Outplanting of branching Acropora enhances recolonization of a fish species and protects massive corals from predation

Damselfish of the genus Stegastes inhabit territories and cultivate algal gardens on branching corals of the genus Acropora, aggressively protecting their territories from other fish and preventing predation upon corals within the territory. This behaviour has important ecological impacts and could also be useful in reducing predation on outplanted corals during reef restoration efforts. However, the degree of protection from predators may depend on the ability of Stegastes spp. to recolonise outplanted or newly established coral colonies. Protection of bleaching-resilient massive corals within territories may be of particular importance due to the role of these corals in maintaining coral cover following bleaching events. This study examined whether the presence of Stegastes spp. reduces predation on the massive bleaching-resilient coral Porites lutea in the Mauritian lagoon, and whether Stegastes spp. readily colonise outplanted branching coral fragments and provide adjacent massive corals with indirect protection from predation. Predation levels on wild-occurring and outplanted P. lutea within and outside Stegastes spp. territories were measured. In addition, Acropora muricata branches were outplanted adjacent to wild P. lutea colonies outside Stegastes spp. territories, and recolonisation of these outplants by Stegastes spp. and the impacts of recolonisation on predation were monitored. Both wild and outplanted P. lutea colonies within Stegastes spp. territories sustained less predation damage compared to colonies outside territories. Stegastes spp. recolonized outplanted A. muricata colonies within six months of outplanting, and in doing so returned predation protection to adjacent P. lutea colonies. The ability of Stegastes spp. to colonise outplanted corals and provide indirect protection to adjacent massive bleaching-resilient corals may inform coral outplanting efforts in systems where Stegastes spp. are common. Encouraging Stegastes spp. recolonisation may help to reduce predation damage to corals within territories and potentially improve the success of rehabilitation efforts.


Introduction
Coral reef ecosystems are threatened by numerous anthropogenic impacts including ocean temperature increases, pollution, and overfishing. Various conservation and management approaches have been utilised to combat this, including the use of coral outplanting to restore degraded sites. In particular, the reintroduction of Topical Editor Andrew Hoey & I. C. Tiddy izzy.seiwhale@gmail.com branching corals such as Acropora species has been utilised to increase habitat complexity and fish biodiversity through restoration of structural refugia (Lirman et al. 2010;Xin et al. 2016;Nithyanandan et al. 2018). Work has previously focussed on the effectiveness of coral restoration techniques and interspecific success rates in order to improve restoration success (Cabaitan et al. 2015). However, limited work has been done to analyse the contribution of other reef species to success in restoration efforts. Among the reef organisms that may affect the success of restoration are damselfish of the genus Stegastes (Gochfeld 2010;Johnson et al. 2011). Stegastes spp. form territories of * 1-8 m 2 (Hata and Kato 2004;Osorio et al. 2006;Dromard et al. 2018;Blanchette et al. 2019) on and between Acropora colonies, upon which they cultivate gardens of filamentous algae, which form part of their omnivorous diet (Ceccarelli et al. 2011). Stegastes spp. territories and their associated algal gardens may cover up to 100% of the benthos on and around branching Acropora colonies (White and O'Donnell 2010;Johnson et al. 2011;Quadros et al. 2019), and they may significantly alter the algal species present in their territory, reducing growth of larger macroalgae by selectively removing indigestible species (Hata and Kato 2002Ceccarelli et al. 2011). Stegastes spp. are highly territorial and will aggressively defend their territories against conspecifics and other fish species to prevent grazing of algae within territories (Johnson et al. 2011). Although territory size and degree of defence may vary among Stegastes species (Hata and Kato 2004), territorial defence is often limited to Acropora colonies and their underlying benthos (Jan et al. 2003). The territorial defence provided by Stegastes spp. may also indirectly protect organisms including corals within territories and increase coral biodiversity through the exclusion of coral predators (Rotjan and Lewis 2008;Gochfeld 2010;White and O'Donnell 2010). Predation by corallivorous fishes even at low levels is a significant stressor on coral colonies, resulting in tissue loss and potentially reduced coral growth, reproductive output, and survivorship of predated colonies at various developmental stages (Rotjan and Lewis 2008;Gochfeld 2010;Bonaldo and Bellwood 2011;Rice et al. 2019). While the presence of Stegastes spp. could have significant benefits to corals through predation reduction, Stegastes spp. may also negatively impact corals by encouraging algal growth in territories (Schopmeyer and Lirman 2015;Seraphim et al. 2020). Algae may compete with and overgrow corals, thus reducing coral growth and survivorship (Schopmeyer and Lirman, 2015), particularly with newly established and juvenile colonies (Casey et al. 2015). The effects of Stegastes spp. presence on hard coral cover are therefore complex (Emslie et al. 2012) but may be key to understanding the role of this widespread fish genus in coral conservation. The preference of Stegastes spp. for Acropora means that loss of live Acropora and their associated branching structure due to bleaching or storm events may significantly reduce the habitat available to Stegastes and other fish species, though some Stegastes spp. have been observed to colonise other benthic habitats in times of stress (White and O'Donnell 2010), and may maintain their ecosystem functions such as predation protection in such habitats. The loss of corals due to environmental stressors may also impact corallivore populations through loss of their preferred food sources. This may result in increased grazing pressure on remaining corals, particularly by obligate corallivorous species that are unable to feed on other items, though such suboptimal diets may lead to reduced fitness for specialist corallivores adapted to feed on bleaching-susceptible corals (Pratchett et al. 2004;Rice et al. 2019). The loss of Acropora, and thus Stegastes spp., in coral reef habitats may therefore result in negative consequences for bleaching-resilient corals, though these may benefit from reduced algal overgrowth (Schopmeyer and Lirman 2015). Currently, little work has been done to examine the effects of Stegastes spp. loss on predation on bleaching-resilient massive corals.
Restoration techniques such as coral outplanting have been employed in various degraded reef systems to restore damaged coral and fish assemblages, with varying success (Cabaitan et al. 2015;Nava and Figueroa-Camacho 2017). The overall effectiveness of outplanting is not yet clear, partly due to the small area over which studies have been conducted compared to the very large areas of reef degradation (Montoya-Maya et al. 2016). However, outplanting of branching corals such as Acropora spp. may increase structural complexity within degraded reef areas, potentially attracting fish assemblages including Stegastes spp. that may have left the area due to lack of structural refugia. If Stegastes spp. form colonies on outplanted Acropora spp. corals, this may restore predation protection to outplanted and adjacent corals within territories, potentially facilitating growth and survival of said corals (Suefuji and van Woesik 2001). Stegastes spp. may, however, aggressively bite both wild and outplanted corals within newly established territories to encourage algal growth and remove coral species that may compete with algal cultivation (Chasqui-Velasco et al. 2007;Seraphim et al. 2020). However, algal overgrowth may be reduced by herbivorous fish species attracted to outplanted corals (Frias-Torres et al. 2015), large shoals of which may overwhelm the defensive functions of Stegastes spp. (Johnson et al. 2011). While Stegastes spp. recolonisation may have mixed effects upon coral colonies, these species quickly recolonise outplanted corals (Schopmeyer and Lirman 2015;Seraphim et al. 2020) and it is thus likely that any restoration effort may be significantly affected by Stegastes spp. in areas where these fish are prevalent. It is therefore key to coral restoration and conservation in many reef habitats that the effects of Stegastes spp. presence on restoration projects are investigated, specifically with regard to their effects on bleaching-resilient massive corals in restoration areas, which remain understudied (Seraphim et al. 2020).
While the impacts of Stegastes spp. presence upon predation on the coral genera Pocillopora, Acropora, and Montipora have been previously studied (Gochfeld 2010;White and O'Donnell 2010), no such evidence yet exists for indirect protection of massive bleaching-resilient Porites spp. corals. Bleaching-resilient species such as massive Porites spp. often survive bleaching events that can kill their more susceptible coral counterparts (McClanahan et al. 2007;Pratchett et al. 2013) and are thus left behind in degraded areas, often with few or no Stegastes spp. (White and O'Donnell 2010), leading to a potential increase in predation upon these corals. In addition, while there is evidence that Stegastes spp. will recolonise and return predation protection to outplanted Acropora species (Schopmeyer & Lirman 2015), the ability of Stegastes spp. to return indirect predation protection to massive bleaching-resilient corals adjacent to outplanting areas is not understood.
We investigated these issues by surveying predation occurring on corals in the lagoon of Mauritius, an area known to contain abundant Stegastes spp. and subject to anthropogenic and climate-related stressors, causing loss of coral cover and biodiversity McClanahan and Muthiga 2020). Specifically, we quantified the extent of predation received by wild-occurring and outplanted Porites lutea in the absence of Stegastes spp. or in the presence of either of the two most common Stegastes spp. in this region-Stegastes punctatus and Stegastes nigricans. We also quantified predation on wild-occurring P. lutea adjacent to outplanted Acropora muricata, some of which was recolonised by S. nigricans. The study investigated the following predictions: (1) the presence of Stegastes spp. will reduce predation upon massive P. lutea within Stegastes spp. territories; and (2) Stegastes spp. will recolonise degraded areas following A. muricata outplanting into these areas, and recolonisation will reduce predation upon adjacent P. lutea within territories.

Study Sites
Six sites were selected for study around Mauritius island; these were three sites under no legal protection, one fishing reserve site in which harvesting is prohibited, and two marine park sites in which harvesting is limited and closely regulated (Fig. 1).

Survey of fish and coral assemblage
Transects were surveyed at each site every two months from September 2018 to June 2019 to record coral, Stegastes, and corallivorous fish assemblages at each site. Corallivorous invertebrates were not included in the study, as little invertebrate predation was observed on P. lutea during pilot surveys, with the vast majority of existing scars on P. lutea colonies originating from fish predation. Four transects were assessed per site visit, with consistency between months ensured through the use of markers on the shore. Each transect was 30 m long and fish were recorded 5 m either side of the transect line, for a total area of 300 m 2 per transect (Bonaldo and Bellwood 2011;Friedlander et al. 2014). A measuring tape was used to measure the length of the transects, and human estimate for the width. Immediately after laying the tape, a free diver swam the length of the transect twice, once on each side of the transect line to ensure full coverage, filming ahead of them with an underwater camera. Footage was viewed in its entirety, with videos ranging from 1 to 4 min in length according to current and weather conditions. From the footage, the presence and numbers of corallivorous fish and Stegastes species and the benthic coral coverage at each site were determined. Corallivorous fish and Stegastes species observed off transects were also photographed to record all species present at each site. Corallivorous fish were defined as all fish species present in Mauritian waters known to predate corals (Supplementary Table 1) (Cole et al. 2008;FishBase 2020). To reduce the likelihood of individual corallivores being counted more than once, an underwater slate was used to make a note if it was suspected that the same individual had been captured more than once on the footage. The note was shown in the video recording to allow notes to be synced with recordings during later analysis. Coral cover included both live coral and recently deceased coral that retained its three-dimensional structure and therefore would still provide potential habitat for fish species. Coral cover was estimated to the nearest 5%, with 5% used as the minimum value where any individuals of a coral type were present.

Survey of predation on wild P. lutea
At each site, 11-15 P. lutea colonies within Stegastes spp. territories and 11-15 outside Stegastes spp. territories were surveyed every two months from September 2018 to June 2019 (Fig. 2). The surface area of surveyed colonies varied significantly (\ 3 to [ 9000 cm 2 ), due to low P. lutea abundance at some sites; colony surface area was thus included in statistical analysis. Colonies outside Stegastes spp. territories were defined as those occurring outside A. muricata aggregates and having no Stegastes spp. present within * 0.5 m of the colony (Jan et al. 2003). Aggregates were defined as continuous expanses of A. muricata, potentially consisting of multiple colonies whose branches overlapped. P. lutea colonies within territories were defined as those within A. muricata aggregates with Stegastes spp. within * 0.5 m of the colony (Jan et al. 2003). This was based on the observation that Stegastes spp. rarely ventured outside of A. muricata aggregates, with Stegastes observed to venture * 0.3 m outside aggregates while chasing fish during pilot surveys. This value was increased to 0.5 m based on previous observations of the distance that Stegastes spp. will often venture outside their territories while chasing fish (Jan et al. 2003). In cases where fewer than 11 P. lutea colonies of either category were present, all available P. lutea were surveyed, and results were pooled across months. The minimum value of 11 was determined using a power analysis carried out on the pilot data from September, with up to four additional corals surveyed to increase power. P. lutea corals were not tagged, due to their shape making it difficult to reliably tag them. Instead, surveying areas were designated at each site, and corals were haphazardly selected within these areas each month. All observed corals fitting the criteria mentioned were surveyed up to a maximum of 15 colonies to reduce selection bias. Whether a coral colony was within the territory of S. punctatus or S. nigricans was recorded from photographs and transects. For each coral colony, all sides were photographed to show all bites, and to be able to measure coral colony surface area. Photography and analysis were carried out by the same person each time to ensure consistency. The app ''SketchAndCalc (2019)'' was used to measure coral area, with a tape measure included in each photograph for scale. Surface areas acquired were approximate due to the uneven nature of the P. lutea surface.
Microsoft photo editor was used to count bites, which were identified visually based on literature descriptions of their appearance. Bites were grouped into four categories: parrotfish, butterflyfish, damselfish, and other. Parrotfish bites appear as paired grooves, which can be single or up to five parallel grooves and may have a space in the middle where the jaws have closed, hence paired. All paired grooves were classed as parrotfish bites, as well as unpaired grooves if C 2 9 as long as they were wide (Rotjan and Lewis 2008). Butterflyfish bites are not always visible, as some species remove only a single polyp. Where they are visible, they appear as small circular white marks with up (1) Flic en Flac; (2) Belle Mare; (3) Blue Bay 1; (4) Balaclava; (5) Blue Bay 2; (6) Blue Bay 3. Red triangles indicate sites under no legal protection, blue squares indicate marine parks, and green circles indicate fishing reserves. Sites were selected for accessibility and presence of wild Porites lutea corals within and outside Stegastes spp. territories (figure adapted from Enchanted Learning, 2018 andd-maps, 2020) to 16 polyps removed, and an area of B 1 cm 2 . Damselfish bites form small circular marks with an undamaged centre (Rotjan and Lewis 2008). ''Other'' bites were defined as areas of damage that resembled bite marks in depth and extent of damage but did not fit the shape profile. This included single bites that were \ 2 9 as long as wide, and areas that may have been the result of parrotfish biting the same area several times (focussed biting) but where no individual grooves could be detected, meaning it could not definitively be said that parrotfish were responsible (Rotjan and Lewis 2008). These marks could have been made by trigger or pufferfish, but no literature on the shape of these bites on massive Porites spp. could be found. Average area of the different bite types was calculated from averages of * 50 bite scars of each bite type on different P. lutea colonies, or from the maximum number of scars available if fewer than 50 were found throughout the survey. This was used to calculate percentage surface area damaged. This and bite density (bites cm -2 ) were used as measures of corallivory upon P. lutea. It is likely that there was a tendency to underestimate the surface area of P. lutea due to its uneven nature. This meant that occasionally surface area damaged by bites would be calculated as covering a greater area than that of the coral itself. In these cases, the surface area damaged was recorded as 100%. Also included were other damaged areas that did not appear to result  from fish corallivory, such as areas damaged by disease, borers, or other predators. These were included in the total area of the coral but marked out as additional damage not due to fish bites. Areas covered in sediment or algae were not included in the total area of the coral so as not to bias calculation of bitten areas, as these areas were not detectably bitten by corallivorous fishes.
Survey of predation on P. lutea outplanted within and outside Stegastes spp. territories In addition to the surveys on haphazardly selected wild colonies, ten P. lutea colonies per site at each of the unprotected sites were outplanted from degraded areas with no Stegastes spp. into Stegastes spp. territories in December 2018 to January 2019 (Fig. 2). Colonies growing in sediment in isolated areas were outplanted directly to A. muricata aggregates and placed among the branches.
Outplanted P. lutea were not fragmented but were picked up whole from the sediment to prevent damage to corals that could reduce the success of outplanting. Outplanted colonies were relatively small (surface area range 21-207 cm 2 ; average 80 cm 2 ) to allow colonies to be placed among A. muricata aggregates without damaging branching coral colonies. P. lutea colonies were not secured to minimise the amount of foreign material (e.g. plastic ties) introduced to the environment, but were placed between A. muricata branches to minimise displacement risk. Outplanting could not be carried out at protected sites due to licencing restraints. Outplanting was carried out into A. muricata aggregates found along transects, and A. muricata adjacent to outplants were tagged to identify the outplants in subsequent surveys. At sites two and three, corals were outplanted into S. nigricans territories, while at site one, corals were outplanted into S. punctatus territories. By way of a control, a further five to ten P. lutea from degraded areas were outplanted into degraded areas adjacent to other outplanted colonies (Fig. 2). These colonies were simply moved from degraded areas where they were found to those near to other outplanted colonies. Control colonies were also positioned along transect lines and tagged where allowed by their shape to allow re-identification. Corals were re-surveyed, and degree of predation measured in February-March, April, and June 2019. The photography and analysis methods used for outplanted P. lutea were the same as those described above for wild P. lutea. Data from wild and outplanted P. lutea were combined into a single dataset for analysis to increase statistical power and control for any effects of outplanting. Data from outplanted corals that died during the course of the study were collected up to the point of coral death. In addition, bites from a sub-sample of 28 corals randomly selected from the wild/outplant dataset were re-counted ''blind'', i.e.
with the counter not aware of the position of the coral within or outside territories; these were then compared to the original dataset to ensure that the counter was not biased.
Outplanting of A. muricata to degraded areas and survey of predation on adjacent P. lutea To gauge the ability of Stegastes spp. to recolonise degraded ecosystems following loss of A. muricata and any effect this may have on predation of adjacent P. lutea, forty A. muricata fragments (four per P. lutea colony), each consisting of a forked branch measuring approximately 30-40 cm in length, were outplanted to ten isolated P. lutea colonies at each of the unprotected sites in December 2018 (Fig. 2). A. muricata and P. lutea were fixed with cement and string to concrete blocks to reduce the chance of corals being swept away. P. lutea to which A. muricata branches were outplanted ranged in surface area from 4 to 313 cm 2 (mean 102 cm 2 ), as smaller corals were easier to affix to blocks. A. muricata branches were tagged and blocks positioned along transect lines to make it easier to return to outplants in subsequent months. Outplants were placed a minimum of 0.5 m from healthy A. muricata stands to reduce the likelihood that outplants would be within previously established Stegastes spp. territories (Jan et al. 2003). This was further ensured by observing the response of Stegastes spp. to outplants at the time of outplanting and moving outplants further away from territories if aggression (biting or chasing) was shown to the outplant or to the individual carrying out outplanting. Corals were re-surveyed in February-March, April, and June 2019. Predation was measured upon P. lutea adjacent to A. muricata corals, with wild P. lutea outside Stegastes spp. territories used as controls. The controls were not placed upon blocks to minimise the amount of foreign material introduced to the environment. The methods of P. lutea photography and analysis were the same as those described above. The presence or absence of Stegastes spp. inhabiting outplanted colonies was recorded at each re-surveying to assess whether recolonization occurred within the outplant survey period of six months, and any effect this may have upon predation of P. lutea. Data from each month of re-surveying were pooled for analysis due to limited sample size caused by death of P. lutea adjacent to outplanted A. muricata. Data from P. lutea that died during the study were collected up to the point of coral death. A sub-sample of 15 P. lutea adjacent to outplanted A. muricata was reanalysed ''blind'', i.e. with the counter not aware of the recolonisation status of the outplant; these were then compared to the original dataset to ensure that the counter was not biased.

Statistical analysis
Data were analysed using the data analysis software R (version 4.0.2). Linear mixed-effects models (LMEs) in the package lme4 were used to examine the data, and visual inspection of residuals from these models was used to verify normality and homogeneity of variance. Following inspection of model outputs, bite density and percentage surface area damaged were cube root transformed to increase normality and homogeneity of variance. LMEs were then constructed for survey and P. lutea outplant data, and for A. muricata outplant data. LMEs were constructed for each dataset using the response variables ''bite density'' and ''percentage surface area damaged'', respectively, while explanatory variables were altered according to the dataset. For the survey and P. lutea outplant data, the explanatory variables were ''Stegastes species'', ''outplant status'', and ''P. lutea surface area'', and the interaction term ''Stegastes species*outplant status''. Likelihood ratio tests were used to compare models without this interaction. The ''Stegastes species'' variable included ''no Stegastes present'' as a level, allowing predation level to be compared between corals within and outside territories. For A. muricata outplant data, the response variables were ''A. muricata outplanting'' (presence or absence of outplanted A. muricata adjacent to P. lutea), ''Stegastes spp. recolonisation'' (presence or absence of Stegastes spp.), and ''P. lutea surface area''. Month, site, and individual coral identity were included as random effects in all LMEs. Coefficient values from final LMEs were utilised to determine significance and direction of effects. The coefficient of determination, or r 2 , was calculated to examine the predictive capacity of models. The r 2 values included conditional and marginal r 2 values (r 2 c and r 2 m ), which described the proportion of variance explained by fixed factors, and by both fixed and random factors, respectively (Nakagawa and Schielzeth 2013). Graphs were plotted using the ggplot2 package in R (4.0.2) (R Core Team 2020). All data for this study are available at Mendeley Data, https://doi.org/10.17632/ 572svz6nvb.3.

Results
Fish and coral assemblage at study sites Mean hard coral cover ranged from * 41% (site one) to * 79% (site six). A. muricata was the most prevalent branching coral species at all sites, with the mean percentage of total benthic cover consisting of A. muricata ranging from * 23% (site five) and * 28% (site one) to * 58% (site six). Mean P. lutea cover ranged from \ 1% (site six) to 5% (site one). S. nigricans and S. punctatus were present at all sites except site five, where only S. punctatus was present. A total of 143 individuals of 20 corallivorous fish species were observed on transect surveys and in photographs from all six sites (Table 1).

P. lutea survey and outplanting
The most common bite type on both wild and outplanted P. lutea was parrotfish (82.5% of bites on wild and 78.3% of bites on outplanted), followed by butterflyfish (wild 8.79%, outplant 15.8%), other (wild 8.73%, outplant 5.92%), and damselfish (wild 0.02%, outplant 0.04%). Throughout the study, a total of 28 outplanted P. lutea were lost or died, mainly due to becoming dislodged from Stegastes spp. territories or covered with sediment, respectively. The Stegastes species*outplant status interaction was not retained in either model. Bite density and percentage surface area damaged were lower within the territories of both S. nigricans and S. punctatus compared to outside territories, but there were no significant differences in either metric between the territories of S. nigricans and S. punctatus (Table 2, Fig. 3, Fig. 4). Neither bite density nor surface area damaged varied between outplanted and wild corals, but both were greater on smaller corals (Table 2).

A. muricata outplanting
Out of 30 total outplants at three sites, ten were recolonised by S. nigricans within six months, with four colonies at site two being recolonised within two months of outplanting. No recolonisation was observed by Stegastes spp. other than S. nigricans. Recolonisation only occurred at outplants placed within 1 m of established A. muricata colonies containing S. nigricans territories. Within six months, recolonisation had occurred at 89% of outplants (eight out of nine) placed within one metre of healthy A. muricata stands at site two and at 20% of outplants (two out of ten) placed within one metre of healthy A. muricata at site one. A total of 12 P. lutea adjacent to outplanted A. muricata died during the course of the study. Three died following recolonisation by S. nigricans; of these two were buried by sediment and one was overgrown with filamentous algae. Of those that died prior to or without recolonisation, four were overgrown with larger macroalgae, two were overgrown with filamentous algae, and three died of unknown causes. An additional three P. lutea were partially overgrown with filamentous algae during the study; one following recolonisation by S. nigricans.
The most common bite type found on P. lutea adjacent to A. muricata outplants and on control colonies was parrotfish (78.6% of bites on colonies adjacent to outplants and 82.3% of bites on controls), followed by butterflyfish (outplant 12.1%; controls 12.2%) and other (outplant 9.4%; controls 5.5%). No damselfish bites were observed on P. lutea adjacent to outplants, or on control corals.
Bite density and percentage coral surface area damaged were lower on corals recolonised by S. nigricans compared to corals where S. nigricans was absent (Table 3, Fig. 5). Neither bite density nor surface area damaged differed between P. lutea colonies with and without outplanted A. muricata, but both were higher on smaller colonies (Table 3).

Discussion
This study found that the presence of Stegastes spp. significantly reduces predation upon the massive bleachingresilient coral P. lutea, in terms of both bite density and percentage surface area damaged. This was observed for both wild and outplanted P. lutea within Stegastes spp. territories. In addition, S. nigricans was capable of recolonising outplanted A. muricata and returning predation protection to adjacent P. lutea. This study therefore builds upon the results of Gochfeld (2010) and White and (Schopmeyer and Lirman 2015), though our findings suggest that the ability of Stegastes spp. to colonise outplanted corals may vary among sites and Stegastes species. The novel finding that Stegastes spp. can restore predation protection to massive corals within their territories through recolonisation of adjacent outplanted branching A. muricata may have significant conservation implications with regard to bleaching-resilient massive corals, and the reef restoration potential of Stegastes spp.
Branching Acropora spp. are often targeted for outplanting projects, due to their ecological importance, bleaching susceptibility, and the relative ease of propagating and outplanting branching compared to massive corals (Lirman et al. 2010;Young et al. 2012;Forsman et al. 2015;Xin et al. 2016;Nithyanandan et al. 2018). This restoration of branching coral cover has also been shown by this and previous studies to attract territorial Stegastes spp. (Schopmeyer and Lirman 2015). Several previous studies of the effects of Stegastes spp. on coral restoration efforts have highlighted the negative impacts of these fish on the survival and growth of corals within their territories, due to algal overgrowth and biting of corals to establish territories and facilitate algal lawn propagation (Chasqui-Velasco et al. 2007;Ceccarelli et al. 2011;Casey et al. 2015;Schopmeyer and Lirman 2015;Seraphim et al. 2020). While these studies provide strong evidence for the negative impacts of Stegastes spp. on coral restoration projects, the present study adds to the body of previous lutea within a Stegastes nigricans territory at site one showing minimal bite damage, and territorial behaviour by S. nigricans towards the surveyor (approaching tape measure to bite/chase) work (Suefuji and van Woesik 2001;Gochfeld et al. 2010;White and O'Donnell 2010;Johnson 2011) that has identified ways in which Stegastes spp. may facilitate restoration, namely through predation and erosion reduction. Specifically, the present study highlights the effects of Stegastes spp. recolonisation on massive bleaching-resilient corals such as P. lutea. These represent a group that may prove vital to reef conservation efforts using bleaching-resilient strains or species (Bhagooli and Taleb-Hossenkhan 2012;Pratchett et al. 2013;Morikawa and Palumbi 2019;Louis et al. 2020). Stegastes spp. have been previously shown to aid survival of juvenile corals following bleaching events by reducing incidental erosion from herbivores (Suefuji and van Woesik 2001), and the current study provides evidence that the benefits of Stegastes spp. to corals that survive bleaching events may also extend to predation reduction. Predation has the potential to be a significant stressor to massive bleaching-resilient corals following bleaching events, due to the loss of many branching coral species that would normally be Table 3 Linear mixed-effects model results for factors contributing to bite density and percentage surface area damaged by bites on P. lutea corals with or without adjacent A. muricata coral outplants, and with or without recolonisation by S. nigricans. For the factor 'treatment', 'control' (no outplanted A. muricata) is the reference level; for the factor 'Stegastes', 'absent' is the reference level. In each model, sampling month, site, and individual coral were included as random effects  Fig. 5 a Bite density and b surface area damaged on Porites lutea adjacent to outplanted Acropora muricata two to six months after outplanting according to whether recolonisation by Stegastes spp. has occurred, and on controls preferentially fed upon (Pratchett et al. 2014;Lewis 2006, 2008). Therefore, encouraging Stegastes spp. to recolonise outplanted Acropora spp. following bleaching events may facilitate growth and survival of remaining corals, including both massive and bleaching-resilient branching strains within territories (Bhagooli and Taleb-Hossenkhan 2012;Louis et al. 2020). As bleaching events become more common, this may represent a way to utilise ecological processes to facilitate a shift to bleaching resilience among coral assemblages.
While S. planifrons (Schopmeyer and Lirman 2015), and now S. nigricans, have been shown to be capable of recolonising degraded coral habitats, the variation in recolonisation success among sites in the current study indicates that variation in local conditions may affect the ability of S. nigricans to recolonise outplanted corals. At site three, high levels of mortality of outplants and adjacent P. lutea may have reduced the number of potential habitats for colonisation. While outplant size has been previously shown to impact survival and recolonisation rates (Lirman et al. 2010;Schopmeyer and Lirman 2015), in this study, outplanted A. muricata fragments were all of similar size. Still, an effect of size cannot be ruled out and it would be beneficial for future studies to quantify this effect. Similarly, at site one high tourist activity resulting in mechanical damage to outplants may have reduced recolonisation success. Low existing branching coral cover at site one may also have reduced the density of Stegastes spp. present and thus the potential for recolonisation. The lack of recolonisation by S. punctatus was surprising but may have been due to congeneric competition resulting in exclusion of S. punctatus from established colonies close to transplant sites by S. nigricans.
In addition to inter-site variation, the lack of recolonisation of outplanted A. muricata placed more than one metre from established territories suggests that S. nigricans may be distance-limited in terms of their ability to recolonise. Alternatively, it is possible that it takes longer than six months for individuals to recolonise outplants placed further away. This may increase the difficulty associated with rehabilitating larger areas of degraded coral reefs, however could potentially be overcome through the direct reintroduction of Stegastes spp. individuals from established territories to outplanting sites. This may help juvenile outplanted corals to become established (Suefuji and van Woesik, 2001); however, introducing Stegastes spp. to new areas may also increase the incidence of coral biting by Stegastes spp. as individuals establish territories (Chasqui-Velasco et al. 2007). Given the impact of Stegastes spp. on reef ecosystems with regard to both algal farming and predation reduction potentially leading to increased diversity and survival of bleaching-resilient corals, such a project may be valuable to further examine the relative effects of Stegastes spp. on successful rehabilitation of degraded areas. An additional understudied factor is the influence of Stegastes spp. on invertebrate corallivores such as Drupella spp., which may also significantly affect coral health (Rotjan and Lewis 2008). This study did not analyse invertebrate corallivory due to low invertebrate predation upon P. lutea at the study sites, with only the sea snail species Coralliophila violacea and C. radula previously recorded as predating P. lutea in the Mauritian lagoon (Kaullysing et al. 2017. However, Stegastes spp. are known to dislodge herbivorous invertebrates from their territories (Hata and Kato 2004) and may do the same with invertebrate corallivores, representing another potential benefit of encouraging Stegastes spp. recolonisation as part of reef rehabilitation.
An important source of mortality to P. lutea colonies adjacent to outplanted A. muricata in this study was algal overgrowth, which is well documented for its negative effects on outplanting success and may be increased by Stegastes spp. presence (Dizon and Yap 2006;Young et al. 2012;Schopmeyer and Lirman 2015;van Woesik et al. 2018). In this study, although one P. lutea died and a further one was partially overgrown due to filamentous algae following recolonisation by S. nigricans, two P. lutea died and two were partially overgrown due to filamentous algae prior to or without recolonisation. In addition, several of those not recolonised were overgrown with larger macroalgae, which is more likely to occur where Stegastes spp. are absent (Hata and Kato 2002Ceccarelli et al. 2011). Despite the evidence of algal cultivation by Stegastes spp. negatively impacting corals within territories (Schopmeyer and Lirman 2015), the low rate of algal overgrowth on P. lutea within recolonised territories in this study indicates that the benefits of Stegastes spp. presence with regard to reduced predation may outweigh the negative impacts of algal overgrowth for bleachingresilient massive corals. For future studies, regular cleaning of outplanted corals and structures to which they are fixed could reduce the negative effects of algal overgrowth (Young et al. 2012). The loss of outplanted P. lutea colonies that occurred in the present study could also be reduced by fixing outplanted corals securely to outplanting sites, or through the use of larger coral colonies that may be less likely to be dislodged or buried. In addition, the size of outplanted colonies may also impact predation rates. While coral colony size is often unrelated to the amount of predation that they incur (Roff et al. 2011), for some coral species larger colonies receive greater amounts of predation per unit area (Burkepile 2012). The opposite effect was found in the current study, as smaller colonies received the most predation, with fewer bites being required to damage a large proportion of the total surface area of smaller corals. In addition, a smaller surface area means bites are more likely to overlap, resulting in a higher bite density but not necessarily contributing to a large increase in surface area damaged. In any case, the size of outplanted corals may be an important factor to consider in future studies in order to optimise coral survival as well as minimising the impacts on donor reefs.
The findings of this study indicate that Stegastes spp. play a significant role in predation protection of corals within their territories, including protection of bleachingresilient corals which represent a significant asset in maintaining coral reef structure in the face of climate change. While few previous studies on species recolonisation and population restoration have focussed on coral reef fish, this study reaffirms that Stegastes spp. have the potential to recolonise outplanted A. muricata aggregates, although their success rate may vary between sites, and shows for the first time that in doing so they may return predation protection to bleaching-resilient corals. While the overall impact of Stegastes spp. on corals within their territories requires further investigation, this study adds to the growing body of work showing the positive impacts of these fish on coral assemblages and provides the first evidence that they may contribute to the preservation of bleaching-resilient corals, with potentially significant implications for reef conservation.
Acknowledgements We would like to thank the University of Mauritius (Department of Biosciences and Ocean Studies) for equipment provision including boat hire for site access, and the Ministry of Blue Economy, Marine Resources, Fisheries, and Shipping (Mauritius) for the provision of licencing to allow coral surveys and outplanting. We are grateful to Ecosud NGO for equipment provision and assistance with fieldwork, and to Wilhelmiina Toivo and Kevin Schneider for their assistance with statistical analysis. RB is grateful to the University of Mauritius (R214 and RF002) and Higher Education Commission (T0721 and T0105) for research funds to study coral reefs. For funding to allow conference attendance, we are grateful to the University of Mauritius (Department of Biosciences and Ocean Studies) and the University of Glasgow (School of Life Sciences). SSK was supported by Natural Environment Research Council Advanced Fellowship (NE/J019100/1) and European Research Council starting grant (640004). We are grateful to the topic editor and anonymous reviewers for insightful comments that significantly improved the manuscript.

Declarations
Conflict of interest On behalf of all authors, the corresponding author states that there is no conflict of interest.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons. org/licenses/by/4.0/.