Bioerosion by euendoliths decreases in phosphate-enriched skeletons of living corals

Abstract. While the role of microboring organisms, or euendoliths, is relatively well known in dead coral skeletons, their function in live corals remains poorly understood. They are suggested to behave like ectosymbionts or parasites, impacting their host's health. However, the species composition of microboring communities, their abundance and dynamics in live corals under various environmental conditions have never been explored. Here, the effect of phosphate enrichment on boring microorganisms in live corals was tested for the first time. Stylophora pistillata nubbins were exposed to 3 different treatments (phosphate concentrations of 0, 0.5 and 2.5 μmol l−1) during 15 weeks. After 15 weeks of phosphate enrichment, petrographic thin sections were prepared for observation with light microscopy, and additional samples were examined with scanning electron microscopy (SEM). Euendoliths comprised mainly phototrophic Ostreobium sp. filaments. Rare filaments of heterotrophic fungi were also observed. Filaments were densely distributed in the central part of nubbins, and less abundant towards the apex. Unexpectedly, there was a visible reduction of filament abundance in the most recently calcified apical part of phosphate-enriched nubbins. The overall abundance of euendoliths significantly decreased, from 9.12 ± 1.09% of the skeletal surface area in unenriched corals, to 5.81 ± 0.77% and 5.27 ± 0.34% in 0.5 and 2.5 μmol l−1-phosphate enriched corals respectively. SEM observations confirmed this decrease. Recent studies have shown that phosphate enrichment increases coral skeletal growth and metabolic rates, while it decreases skeletal density and resilience to mechanical stress. We thus hypothesize that increased skeletal growth in the presence of phosphate enrichment occurred too fast for an effective expansion of euendolith growth. They could not keep up with coral growth, so they became diluted in the apex areas as nubbins grew with phosphate enrichment. Results from the present study suggest that coral skeletons of S. pistillata will not be further weakened by euendoliths under phosphate enrichment.

croscopy, and additional samples were examined with scanning electron microscopy (SEM). Euendoliths comprised mainly autotrophic Ostreobium sp. filaments. Rare filaments of heterotrophic fungi were also observed. Filaments were densely distributed in the central part of nubbins, and less abundant towards the apex. Unexpectedly, there was a visible reduction of filaments abundance in the most recently-calcified apical 15 part of phosphate-enriched nubbins. The overall abundance of euendoliths significantly decreased, from 9.12 ± 1.09 % of the skeletal surface area in unenriched corals, to 5.81 ± 0.77 % and 5.27 ± 0.34 % in 0.5 and 2.5 µmol l −1 -phosphate enriched corals respectively. SEM observations confirmed this decrease. Recent studies have shown that phosphate enrichment increases coral skeletal growth and metabolic rates, while 20 it decreases skeletal density and resilience to mechanical stress. We thus hypothesize that increased skeletal growth in the presence of phosphate enrichment occurred too fast for an effective euendolith colonization. They could not keep up with coral growth, so they became diluted in the apex areas as nubbins grew with phosphate enrichment. The possible advantages and downsides of the reduction of euendoliths associated 25 with phosphate eutrophication in live corals are discussed in this article. 2426

Introduction
Euendoliths are boring autotrophic and heterotrophic microorganisms that include cyanobacteria, chlorophytes, rhodophytes, and fungi (Tribollet, 2008a). They develop in a large variety of carbonate substrates, including crustose coralline algal thalli and coral skeletons (Tribollet and Payri, 2001;Tribollet and Golubic, 2005), in which they actively 5 penetrate through the active process of dissolution (Golubic et al., 1981;Garcia-Pichel et al., 2010). They colonize live and dead substrates, although colonization has been shown to be more intense in dead ones (Le Campion-Alsumard et al., 1995a;Tribollet and Payri, 2001). In live coral skeletons, euendoliths grow from the inside of the skeleton towards the surface, trying to keep up with coral growth (Le Campion-Alsumard et 10 al., 1995a). In dead skeletons on the contrary, they penetrate from the outside and bore towards the inside of the substratum (Tribollet, 2008b). In dead corals, euendoliths have been shown to be important primary producers, and major agents of reef bioerosion and sediment production (Schneider and Torunski, 1983;Chazottes et al., 1995;Perry, 2000;Tribollet et al., 2002Tribollet et al., , 2006Tribollet and 15 Golubic, 2005). Various environmental factors have been reported to affect rates of dissolution by euendoliths in dead substrates. Zubia et al. (2001), Chazottes et al. (2002), and Carreiro-Silva et al. (2005, 2009 reported enhanced rates of dissolution under eutrophicated conditions, while Tribollet and Golubic (2005) showed that terrigenous inputs can mitigate the effects of eutrophication by limiting settlement and penetration 20 of euendoliths. Moreover, it was shown recently that rates of bioerosion by euendolithic communities are positively affected by elevated pCO 2 (Tribollet et al., 2009).
In live corals, besides their role as skeleton bioeroders, euendoliths are known to have different activities. Boring heterotrophic fungi appear to inflict damages to their live hosts (Bentis et al., 2000;Alker et al., 2001;Domart-Coulon et al., 2004)  of euendoliths and the balance between damages and benefits in live corals remain however poorly known (Ferrer and Szmant, 1988;Tribollet, 2008a). In particular, the role of environmental factors on bioerosion of live coral skeletons has been seldom addressed. It was shown that elevated light leads to a photoacclimation of phototrophic euendoliths when increased progressively, and makes them more susceptible to ther-5 mal photoinihibition and photodamages when increased rapidly, while concomitant increases in light and temperature lead to a decrease of their photosynthetic efficiency (Fine and Loya, 2002;Fine et al., 2004Fine et al., , 2005. But the roles of other factors such as nutrient concentrations have never been formally examined. Since corals are becoming increasingly impacted by eutrophication due to continuous nutrient release from 10 sewage discharges, rainfall, rivers and ground waters (Tomascik and Sander, 1985;Bell and Tomascik, 1993;McCook, 1999), the impact of nutrients on boring euendolithic communities of live corals deserves more attention. Based on studies carried out on dead coral skeletons, we hypothesize that nutrient enrichment stimulates euendoliths in live corals and thus, rates of dissolution. 15 The aim of the present study was to test the impact of enrichment by a single nutrient, phosphate, under controlled conditions in aquaria, using the tropical coral S. pistillata. Phosphate was chosen because it has been reported to affect the skeletal composition and structure of corals (Godinot et al., 2011a;Dunn et al., 2012). Indeed, a recent study showed that phosphate decreased skeletal density in the coral Acropora muri-20 cata, which was suggested to possibly cause live corals to be more colonized by euendoliths (Dunn et al., 2012). S. pistillata was selected for the present study because in this species, phosphate was already shown to increase tissue and skeletal growth, phosphate incorporation into the mineral fraction of the skeleton, as well as zooxanthellae specific growth rate, photosynthetic efficiency and phosphorus content (Godinot et

Experimental design
The experimental setup used in this study has already been described in a previous paper (Godinot et al., 2011a). Briefly, live nubbins (initial size of 1.3 ± 0.4 cm long and 0.6 ± 0.3 cm in diameter) of S. pistillata were cultured in duplicated aquaria under three 5 continuous phosphate enrichments (0, 0.5 and 2.5 µmol l −1 ). The 0.5 µmol l −1 enrichment represented a phosphate concentration which has been reported on some eutrophicated reefs (Kinsey and Davies, 1979), whereas the 2.5 µmol l −1 enrichment was used to highlight the effect of phosphate on coral physiology. Corals were kept unfed to control for phosphorus enrichment, and light, temperature, salinity, algal development 10 and nutrient concentrations were controlled in each aquarium (Godinot et al., 2011a). Three nubbins per treatment (9 nubbins in total) were sampled for euendolith observations after 15 weeks of phosphate enrichment, and were immediately fixed in a 4 % solution of formaldehyde in buffered seawater. 15 Nubbins were cut transversally in two halves for observation of euendoliths with light microscopy and scanning electron microscopy (SEM) respectively. The first halves were used to prepare longitudinal petrographic thin sections for light microscopy observations. The samples were dehydrated in a series of ethanol and acetone baths, then embedded in araldite as described by Tribollet (2008b). Several 20 millimeter thin slabs of skeleton were cut using a diamond circular saw, and were then mounted on microscope slides, ground to the quality of petrographic thin sections, briefly etched with 5 % HCl, rinsed carefully, and stained with 5 % toluidine blue to reveal the euendolithic filaments (a total of 5 slides of good quality were finally obtained per phosphate treatment). Sections were observed with a Nikon Eclipse LV100 micro- The second halves of samples were used to prepare SEM sections. Samples were bleached with sodium hypochlorite before embedding, then cut longitudinally, shortly etched with 5 % HCl, rinsed and dried carefully, and then platinum-coated. Three samples per treatment were observed with a ZEISS Evo.LS.15 environmental SEM. 5

Petrographic slides
Two semi-quantitative methods were selected to (i) determine if phosphate had an effect on the overall distribution of microboring filaments in the skeletons, and to (ii) quantify the abundance of filaments in each treatment. In this latter technique, only non-porous areas of the skeleton (i.e. microscopic fields fully covered by skeleton) 10 were selected for repeatability and accuracy of the abundance quantifications. This choice was made because of the highly porous structure of S. pistillata skeletons, which was of ca. 50 ± 8 % (estimated on pictures of the petrographic slides with the software ImageJ).
The first semi-quantitative method consisted in selecting one representative slide out 15 of the five per phosphate treatment to map in details the spatial distribution of filaments across the entire sections of skeleton, i.e. porous and non porous areas. For this new approach of euendoliths distribution, pictures of the entire thin sections selected were taken and assembled using the software NIS-Elements D (Nikon). These pictures were converted to binary black and white pictures with ImageJ. The outlines of the skeletons 20 were recovered with the software Adobe Illustrator, and colored distribution maps were drawn within those outlines. Maps were based on estimations of the abundance of euendolithic filaments, visually ranked from 1 to 5 by the same observer (respectively lowest and highest filament abundances encountered across all the samples). Abundances were estimated on 0.14 mm 2 fields, at ca. 500 µm intervals across the whole The second semi-quantitative method consisted in ranking the abundance of filaments on 30 randomly selected non-porous microscopic fields of 0.14 mm 2 per slide (5 slides, thus 150 measurements per phosphate treatment), in order to quantify and to compare the abundance of filaments among phosphate treatments. We thus observed a total surface area of 0.041 cm 2 per slide out of 0.85 cm 2 on average, with a porosity 5 of 50 ± 8 %. Thus, quantifications were performed on ca. 10 % of the total surface area of the samples. The same scale as described above (ranks from 1 to 5) was used.
To statistically compare the abundance of filaments between phosphate treatments, ranks were matched to percentages of surface area covered by euendoliths. These percentages were determined for each rank of abundance as a preliminary step, using 5 representative photographs per rank which were analyzed with the software ImageJ. The minimum and maximum values found for each rank gave the range of percentages of bioeroded surface area attributed to that rank (presented in Table 1). The 30 abundance observations performed per slide were thus used to calculate the range of surface area covered by euendoliths on each slide. Medians of these ranges were 15 compared among the three treatments using non-parametric Kruskal-Wallis tests, followed by U Mann-Whitney post-hoc paired tests, performed with the software StatView. Non-parametric tests were selected since the normality assumption was not respected.

SEM sections
SEM sections were observed to confirm (iii) the specific diversity of euendoliths ob-20 served on petrographic slides, and (iv) the semi-quantitative analyses performed. For that latter part, ten pictures were randomly taken per section (30 pictures per phosphate treatment) to quantitatively measure the surface area bioeroded by euendoliths using the software ImageJ (expressed in percent of the total surface area of the picture). The effect of phosphate enrichment was tested using a Kruskal-Wallis test followed by U

Results
Nubbins measured on average ca. 0.8 ± 0.3 cm in diameter at the end of the experiment, with a length of 3.3, 3.5, and 3.7 ± 1.0 cm long (respectively for the 0, 0.5 and 2.5 µmol l −1 treatments). Euendolithic communities observed in the skeletons of live S. pistillata were mainly 5 made of Ostreobium sp. filaments (Fig. 1), with possibly fungi filaments as well. No other species were observed. Those filaments were rather densely distributed in the middle part of the nubbins (yellow to red colors on Fig. 2), while they were less abundant at the apex of the corals (blue to yellow colors on Fig. 2). Differences were observed between the unenriched and the two phosphate-enriched corals (Fig. 2a): in the latter, filaments were even less abundant towards the most recently-calcified apical part of the nubbins (large blue and green areas on the right of Fig. 2b and c).

Discussion
This study is, to the best of our knowledge, the first to report on the distribution and   , 1995a), in which the ubiquitous chlorophyte Ostreobium quekettii dominated assemblages, with occasional filaments of fungi and of the cyanobacterium Plectonema terebrans. In S. pistillata however, fila-5 ments of the cyanobacteria P. terebrans were not observed, but may have been overlooked or confounded with fungi hyphae. The low diversity of euendoliths in S. pistillata confirms that only a few species can penetrate into skeletons of live corals. Euendoliths in S. pistillata were however distributed differently than in Porites colonies. They were localized across the skeletons of S. pistillata (unenriched treatment) while they were 10 condensed in a green band beneath the surface of Porites colonies. The above variations in distribution probably result from differences in structure and porosity among coral species. This strongly suggests that all coral species are not colonized the same way by euendoliths, as is also the case for dead carbonate substrates (Perry, 1998;Tribollet, 2008a). 15 The decrease of euendolithic filament abundance, and thus bioerosion, reported here was somewhat unexpected, as it was in contradiction with the assumption of Dunn et al. (2012). These authors indeed suggested that bioerosion by euendoliths increases with phosphate eutrophication. If skeletal microdensity decreased in S. pistillata corals due to the continuous 15 weeks phosphate enrichment, as was the case 20 of A muricata corals enriched for 16 weeks with phosphate (Dunn et al., 2012), this decrease did not lead to a faster colonization of the skeleton by euendoliths. On the contrary, euendoliths abundance decreased in phosphate-enriched corals, especially in the apexes (Fig. 2). This result may be linked to the increase in skeletal growth rates observed with phosphate enrichment in S. pistillata (Godinot et al., 2011a). An 25 inverse relationship was found between the abundance of euendoliths (expressed as percentages of surface area bioeroded) reported in the present study and the skeletal growth rates reported by Godinot et al. (2011a) (Fig. 4). We hypothesize that increased skeletal growth in the presence of added phosphate was too fast for the euendoliths to actively follow coral growth, and that euendoliths became diluted as nubbins grew with phosphate enrichment. Contrary to the negative effect of phosphate enrichment on euendoliths growth observed in the present study in live corals, eutrophication has been reported to increase bioerosion by euendoliths in dead substrates (Zubia and Peyrot-Clausade, 2001;Cha-5 zottes et al., 2002;Carreiro-Silva et al., 2005, 2009. Chazottes et al. (2002) and Carreiro-Silva et al. (2005) highlighted the confounding roles of grazing and organic matter release in this positive response, which led to changes in euendolithic communities with eutrophication in dead substrates. They hypothesized that increased nutrients can initiate a feedback loop, where bioerosion by euendoliths and by grazers reinforce one another, leading to accelerated bioerosion of the reef framework. However, the latter confounding roles of grazing and organic matter release were absent in the present controlled study on live corals. Furthermore, the processes of bioerosion in dead substrates and live coral skeletons are likely to be very different (Le Campion-Alsumard et al., 1995a). Indeed, in dead substrates, euendoliths are in contact with the ambi-15 ent seawater, where they can possibly benefit from high nutrient concentrations. In live corals, polyps and their zooxanthellae form a protective barrier and actively take up the nutrients (D'Elia, 1977;D'Elia et al., 1983;Bythell, 1990;Godinot et al., 2009Godinot et al., , 2011a. Even though phosphate did reach the skeleton in the present study, as evidenced by the higher P:Ca ratio and phosphorus content of the mineral fraction of the skeleton 20 of phosphate-enriched nubbins (Godinot et al., 2011a), it is not granted that this phosphate was available to euendoliths. In fact, phosphate was incorporated as calcium phosphate in the crystal lattice of the skeleton, and was probably not accessible by euendoliths. Another source of phosphate for euendoliths might have been localized within the pores of the skeleton. Indeed, skeletal pore water has been reported to be 25 nutrient rich in some massive corals (Risk and Müller, 1983;Ferrer and Szmant, 1988), with phosphate concentrations elevated by 0.39 µmol l −1 above those encountered in ambient seawater. However, the very poorly connected pore structure of S. pistillata may have prevented this enrichment of skeletal water. Euendoliths in massive corals such as Porites colonies may however, respond completely differently as those corals have a slow growth rate (Pätzold, 1984) and a structure allowing a better circulation of seawater inside skeleton (Knackstedt et al., 2006). Results of the present study need to be confirmed by carrying further enrichment experiments (with different nutrients, combined nutrients, in various concentrations), with diverse coral species and over longer 5 periods of time.
By addressing issues of bioerosion by euendoliths in the context of nutrient enrichment in live corals, this study adds to the growing body of evidence on the impacts of phosphorus on live corals, and adds to the understanding of euendoliths dynamics in those live substrates. Results from the present study indicate that coral skeletons of 10 S. pistillata will not be further weakened by euendoliths under phosphate enrichment. A decrease of bioerosion rates in polluted areas could therefore be positive for living corals facing eutrophication, as it would represent one less stressor to cope with. Indeed, it was shown that when the corals Porites lobata, Pocillopora eydouxi, Acropora cytherea, Acropora humulis, and Montipora studeri are attacked by fungi filaments, 15 they actively resist fungal penetration by depositing conical structures of dense repair aragonite in growing calices (Le Campion-Alsumard et al., 1995b;Bentis et al., 2000). This process is energetically costly, but will likely be reduced if euendoliths become diluted in skeletons of fast growing corals under phosphate enrichment. On the other hand, such dilution of euendoliths in live coral skeletons may have a negative impact 20 on corals during bleaching events as euendoliths partially replace zooxanthellae, by providing food to their host, and thus a better resistance to thermal stress (Schlichter et al., 1995;Fine and Loya, 2002). The delicate balance between benefits and disadvantages provided by euendoliths to the various species of live corals therefore needs further investigations.   Skeletal growth rates are from Godinot et al. (2011a) and were measured over 8 weeks of phosphate enrichment. Abundance of euendoliths was estimated based on the percentages of surface area bioeroded in each phosphate treatment after 15 weeks of exposure to phosphate treatments. Data are presented as the means ± SE, with n = 5 samples per phosphate treatment for the abundance, and n = 10 for growth rates.