Introduction

The role of corallivorous sea stars in coral reef ecology and their impact on coral health have become increasingly important and investigated, due to the devastating effects from outbreaks of the crown-of-thorns sea star (COTS) (Acanthaster spp.) (Sweatman & Syms, 2011; De’ath et al., 2012; Pratchett et al., 2017). Under these circumstances, other species of coral-eating sea stars have been neglected, likely due to their apparent inability to cause population outbreaks or to represent an immediate and macroscopic risk to reef integrity. Among these less-recognized corallivorous organisms are the cushion sea stars of the genus Culcita.

Culcita spp. (Oreasteridae) are distributed across coral reefs in the Indian and Pacific Oceans. Currently, only three species are described; C. coriacea (Müller & Troschel, 1842), C. schmideliana (Bruzelius, 1805) and C. novaeguineae (Müller & Troschel, 1842). In the Indo-Pacific area, the predominant species are C. schmideliana and C. novaeguineae, with a large abundance of the former in the western Indian Ocean (Yasuda, 2012), whereas C. coriacea is considered endemic to the Red Sea (Clark, 1976). Like Acanthaster spp., Culcita spp. are also coral predators, which feed on corals by everting their stomachs and removing only the coral tissue without affecting the carbonate skeleton. Ultimately, the tissue is digested in situ, and this predation leaves scars that appear as large white spots (Goreau et al., 1972; Glynn & Krupp, 1986). However, both sea stars include a variety of other organisms in their diet (Thomassin, 1976; De’ath & Moran, 1998). The two sea stars also share similar spawning periods and larval ecology in many geographic areas (Yamaguchi, 1977; Yasuda et al., 2010; Ohta et al., 2011).

Nevertheless, differences in the feeding behaviour of Acanthaster spp. and Culcita spp. have been reported, with cushion sea stars consuming comparatively less coral and at slower rates. For example, while a Culcita sea star can remove up to 1 m2 year−1 of Pocillopora damicornis (Linnaeus, 1758) or up to 0.9 m2 year−1 of mixed coral prey (Glynn & Krupp, 1986), individual COTS have been estimated to consume an average of 5–6 m2 year−1 of coral surface, reaching values of 12–13 m2 year−1 in some locations (Birkeland & Lucas, 1990).

The main difference between these two sea stars is that Acanthaster spp. have a higher fecundity and cause large population outbreaks with disastrous consequences for coral reefs (Birkeland & Lucas, 1990; Kayal et al., 2012; Scott et al., 2017), whereas similar bursts of cushion sea star populations have not yet been reported. Nevertheless, average densities of Culcita spp. are typically higher than the number of individuals observed in non-outbreak populations of Acanthaster spp. (Goreau et al., 1972; Glynn & Krupp, 1986). This suggests that the cushion sea star, due to its continual removal of living coral, may represent a persistent force in affecting coral cover abundance and composition (Pratchett et al., 2011) and, therefore, similar to COTS, may influence broader ecological processes and interactions in coral reefs.

Although our knowledge of Acantahster spp. is continuously increasing (Pratchett et al., 2017), little information about the ecology and corallivorous behaviour of Culcita spp. in natural environments is currently available. To date, these often limited and preliminary field observations indicate that dietary preferences are highly variable, with Culcita spp. feeding on different coral species depending on location (Goreau et al., 1972; Thomassin, 1976; Glynn & Krupp, 1986; Quinn & Kojis, 2003). However, these studies often report a general preference for small coral colonies and/or recruits, potentially as the size and shape of Culcita individuals limit their ability to climb and feed on larger colonies (Glynn & Krupp, 1986; Quinn & Kojis, 2003).

In this study, the dietary ecology and corallivorous behaviour of the cushion sea star were investigated for the first time in a geographic area where this organism is poorly investigated: the Republic of Maldives. In the Maldives, the presence of the genus Culcita, and in particular, the species C. schmideliana, has been reported (Clark & Rowe, 1971; Jangoux & Aziz, 1984; Moosleitner, 1997; James, 2004); however, the Maldivian morpho-type shares common morphological features with both C. schmideliana and C. novaeguineae and, without a precise taxonomic identification, should therefore be referred to with the generic name Culcita sp.

The Maldivian coral reef is the seventh largest reef system on earth, with 8920 km2 of total surface area (Spalding et al., 2001). Recently, its health status has been affected by multiple stressors, such as coral bleaching events (Tkachenko, 2015; Ibrahim et al., 2017; Perry & Morgan, 2017), disease outbreaks (Montano et al., 2015, 2016; Seveso et al., 2015, 2017), land reclamation and anthropogenic pressure (Brown et al., 2017; Fallati et al., 2017), algal overgrowth (Montano et al. 2012a, b; Seveso et al., 2012) and COTS outbreaks (Saponari et al., 2015, 2018). Hence, understanding the ecology of coral-consuming animals is of critical importance, in order to identify their role in mediating coral community structure and better predict their ecological impact under changing conditions. To this end, we examined a population of cushion sea stars within a Maldivian coral reef system to determine (a) whether the distribution and size of individuals are related to depth, habitat composition, structural complexity and reef zonation, and (b) what the diet of these sea stars consists of and if they exhibit preferences for specific coral genera, growth forms, or colony sizes.

Materials and methods

Study area and sampling design

Underwater surveys were performed on SCUBA diving during a number of field campaigns between February 2015 and April 2016 in the southern region of the Faafu Atoll, Republic of Maldives (Fig. 1). Our logistic base was the Marine Research and High Education Center (MaRHE) on Magoodhoo Island (3°4′49.08″N, 72°57′57.19″E, WGS84), an inhabited island that measures 900 × 450 m2 and is located on the south-eastern region of the atoll rim, approximately 140 km south of the capital Malè. Magoodhoo reef is approximately 3 km long and 1.55 km wide (Montano et al., 2012b).

Fig. 1
figure 1

Map of the study area. Seventeen sites (indicated by arrows) were selected both on the inner and outer side of the Southern part of the Faafu Atoll (a), in the Republic of Maldives (b, c). Scale bar: ~ 10 km. Modified from Montano et al. (2017b)

Full size image

Around the island, 17 different sites showing heterogeneous characteristics in terms of reef morphology and exposure were randomly selected among those accessible (Fig. 1). Some sites exhibited typical low-energy reef features with ample growth of coral and gentle slopes, whereas others were characterized by shallow patchy lagoon reefs or steep reef walls. Moreover, some sites were located inside the atoll, either as lagoon-patch reefs or as lagoon-facing sides of the atoll rim (inner reefs), whereas others were positioned on the external, ocean-facing side of the atoll rim (outer reefs), which are often subject to more intense hydrodynamic conditions. The presence of Acanthaster spp., whose signs of predation can be confused with those of Culcita spp., was not observed on any of the selected sites.

Determination of coral and substrate coverage in the study area

To determine the composition and structure of the benthic community in the study area as well as the diversity and the abundance of the scleractinian corals, a photo-quadrat method was used. For each sampling site, a minimum of 8 PVC quadrats of 1 × 1 m2 were randomly placed at different depths and spaced 10–20 m apart. Photographs were taken using a Canon G11 camera in an underwater housing (Canon WP-DC34), and the photographs were analysed using Coral Point Count with Excel extension software 4.0 (CPCe 4.0, Kohler & Gill, 2006). Substrate coverage was estimated through the identification of organisms below 100 randomly distributed points per quadrat. Data relative to each site were collected and analysed together as study areas, using the following benthic categories: corals (scleractinians), rock, sand, coral rubble, dead corals, coralline algae, algae, soft corals, sponges and other sessile invertebrates (tunicates, zoantharians and unknown). In addition, the genus level was added within the benthic category “corals” (identification according to Veron, 2000).

Field observations of Culcita sp.

At each study site, the presence of Culcita sp. was recorded by applying the roving SCUBA diving technique, which consists of a 1-h dive sampling unit, where the diver moves progressively from the maximum depth (approximately 30 m) to shallower water, swimming freely throughout the dive locality (Hoeksema & Koh, 2009; Montano et al., 2017a). To avoid resampling the same individuals, the divers swam in the same direction without turning back.

Distribution and size

For every cushion sea star encountered, the depth at which it was observed (< 10 m, 10–20 m and > 20 m) and its location in terms of reef zones (reef flat, reef crest, reef slope) were recorded. To determine the coral cover of the area where each individual was found, the living coral cover percentage was evaluated visually within a 2-m radius calculated through a tape measure around each sea star. A coral covering score from 1 to 4 was assigned, where “1” indicated a coral coverage between 0 and 10%, “2” a coral coverage between 10 and 30%, “3” a coral coverage between 30 and 50% and “4” a coral coverage > 50%.

Furthermore, the size of each Culcita sp. encountered was noted by measuring with a ruler the maximum diameter of the specimen to the nearest centimetre (cm).

Feeding activity and coral prey preferences

To define the behaviour of the organism, three main activities of the cushion sea star were considered: “moving”, when tube feet were visible and the motion of the sea stars was easily detectable; “preying”, when sea stars with everted stomachs were feeding on corals or other organic matter; and “stationary,” when specimens were not moving or preying.

Within the investigated area (2 m radius around the sea star), we also counted the number of corals consumed by the sea star. In particular, corals showing clear signs of external digestion and the absence of coral tissues were considered as preyed corals (Fig. 2). Importantly, the size of the investigated area (2 m radius) was selected considering the slow movement of the sea stars. This ensured that the coral colonies, showing fresh and bright white scars due to the total removal of living coral tissue as well as a visible thin layer of mucus, had been recently preyed upon by Culcita sp. (Fig. 2). The diameter of the targeted colonies was measured to the nearest centimetre (cm) considering their longest axis, and these corals were classified into three size-classes (< 5 cm, 5–10 cm and > 10 cm). Furthermore, the dimensions of scars on the coral surface were noted, and the coral was classified as partially or totally consumed. In addition, the genus of the targeted coral colonies was recorded as well as their growth morphology, according to the following categories: branching, encrusting, massive, foliaceous and mushroom.

Fig. 2
figure 2

Small adult colony of Pocillopora spp. consumed by Culcita sp. (a). Note the absence of living tissue. Cushion sea star (indicated by the arrow) close to recently consumed coral colonies (b). Scale bars: ~ 4 cm in (a) and ~ 10 cm in (b)

Full size image

Data analysis

Data are presented as the mean ± standard error (SE) unless otherwise stated. Data normality were verified using the Shapiro–Wilk test. Three separate one-way ANOVAs, followed by a Tukey’s HSD post hoc test, were used to evaluate significant differences in the abundance of sea stars at different depths, in different reef zones (reef flat, crest and slope) and in relation to different coral cover in the area around the sea star (2 m radius). Likewise, three separate one-way ANOVAs, followed by a Tukey’s HSD post hoc test, were used to evaluate significant differences in the sizes of sea stars using the same three previous factors.

Spearman’s rank correlation order test was used to examine whether the size of the scars produced by Culcita sp. on the preyed coral colonies was correlated with the size of the colonies. All statistical analyses were performed using SPSS version 24 (IBM, New York) and R version 3.2.3 (R Development Core Team, 2015).

Sea stars’ preferences for different groups of prey, here considered as coral genera, were calculated following Chesson (1983), as the standardized forage ratio (Si) and defined for a group i as:

$$S_{i} = \frac{{\left( {\frac{{r_{i} }}{{P_{i} }}} \right)}}{{\mathop \sum \nolimits_{1}^{n} \left( {\frac{{r_{i} }}{{P_{i} }}} \right)}}$$

where Pi is the prey relative abundance in the environment, ri is the relative abundance of the prey in the predators’ diet, and n is defined as the number of groups in the system. The values of the standardized forage ratio ranged between 0 and 1, with Si= 0 indicating avoidance and Si = 1 indicating exclusiveness for a group of prey. We calculated Pi as the mean percentage cover for a coral genus in all the sites obtained from photo-quadrats analysis and ri as the percentage of predatory actions on a coral genus in relation to all preyed colonies.

The Ivlev’s electivity index, Ei (Ivlev, 1961), was obtained by the following calculation:

$$E_{i} = {{\left( {r_{i} - \, P_{i} } \right)} \mathord{\left/ {\vphantom {{\left( {r_{i} - \, P_{i} } \right)} {\left( {r_{i} + \, P_{i} } \right)}}} \right. \kern-0pt} {\left( {r_{i} + \, P_{i} } \right)}} \,$$

where ri and Pi are defined as above. Values of Ei ranged between − 1 and 1, such that Ei = − 1 represents total avoidance, Ei= 0 means non-selective feeding ,and Ei = 1 represents exclusive feeding on a given prey item. As reported by many authors, this index is not independent of prey density and thus has a major shortcoming for comparative purposes versus the standardized forage ratio (Jacobs, 1974; Chesson, 1983). Both indices have been used in feeding studies on Asteroidea and hence provide an efficient tool for the comparative analyses of sea star feeding behaviour (Farias et al., 2012; Saponari et al., 2018).

Results

Coral and substrate coverage in the study area

Considering the community structure and composition of the investigated area (Fig. 3a), benthic community composition was dominated by corals (29.9 ± 3.6%), followed by rock (18.9 ± 2.1%) and sand (13.5 ± 1.6%). The coral genus Acropora was the most abundant (13.2 ± 6.4%), followed by Porites (6.7 ± 4%), Pocillopora (1.2 ± 1%), Favites (0.9 ± 0.7%), Montipora (0.8 ± 1.3%), Pavona (0.8 ± 0.5%) and other less represented genera (Fig. 3b).

Fig. 3
figure 3

Overall coverage of the benthic categories (a) and coral genera (b) in the whole study area. Coral genera showing a coverage % lower than 0.5 were included in the category “Other” (represented by 30 coral genera). Data are expressed as the mean ± SEM

Full size image

Distribution and size of Culcita sp.

During the study period, 163 specimens of Culcita sp. were found in the area examined.

The abundance of sea stars showed significant variations depending on the depth (one-way ANOVA, F(2,50) = 24.825, P = 0.000), with the majority of organisms (approximately 70%) found between 0 and 10 m. Below 10 m, the number of observed sea stars decreased progressively with increasing depth, even if no significant difference was detected between the depth ranges 10–20 m and > 20 m (Fig. 4a). No significant difference was observed in the number of sea stars in the different reef zones (one-way ANOVA, F(2,50) = 0.268, P = 0.766, Fig. 4b), whereas their abundance varied significantly depending on the coral cover percentage recorded in the 2 m radius area around the sea star (one-way ANOVA, F(2,66) = 6.219, P = 0.001). In particular, significant differences were detected between areas characterized by a coral coverage between 10 and 30%, where the highest number of Culcita sp. was found (~ 41.5%) and those with a coral coverage > 50%, where the lowest abundance of specimens was observed (~ 9.5%), (Fig. 4c).

Fig. 4
figure 4

Mean percentage of abundance of Culcita sp. in three depth ranges (a), in three reef zones (b), and in 2 m radius areas around sea stars, showing 4 different coral cover percentages (1: coral coverage between 0 and 10%, 2: 10–30%, 3: 30–50%, 4: > 50%) (c). Data from 17 sites were analysed and percentages expressed as the mean ± SEM (one-way ANOVA followed by Tukey’s HSD multiple pair-wise comparisons). Average size of Culcita sp. (cm) in three depth ranges (d), in the three reef zones (e) and in areas showing different coral cover percentages (f). Data are expressed as the mean ± SEM (one-way ANOVA followed by Tukey’s HSD multiple pair-wise comparisons). In each graph, letters denote Tukey’s significant difference among the different groups (P < 0.05); thus, the same letter indicates no significant difference (P ≥ 0.05)

Full size image

The average size of all sea stars sampled was 18.74 ± 0.15 cm, which did not show significant variation depending on the depth (one-way ANOVA, F(2,153) = 1.624, P = 0.107, Fig. 4d). However, the average size of the individuals showed significant variation depending on the location of the organisms in the different reef zones (one-way ANOVA, F(2,153) = 3.217, P = 0.0427, Fig. 4e) and the coral cover abundance in the 2-m radius area around each sea star (one-way ANOVA, F(3,153) = 5.933, P = 0.000, Fig. 4f). In particular, significant differences in average sizes were detected between the reef crest and the reef slope (Tukey HSD post hoc, P < 0.05; Fig. 4e) and between areas where the coral coverage was more than 50% compared to areas where this percentage was less than 50% (Tukey HSD post hoc, P < 0.05; Fig. 4f).

Feeding activity and coral prey preferences of Culcita sp.

Only 14 sea stars (~ 9%) out of a total of 163 were observed to perform feeding activities during the sampling, and all were found to be preying on scleractinian corals. Most sea stars (~ 68%) were found stationary on the reef, whereas approximately 23% of the specimens were moving.

Scars typical of predation by Culcita sp. were found on 433 coral colonies belonging to 16 different coral genera (Table 1). Most of the preyed corals belonged to the genera Acropora (38.1%) and Pocillopora (~ 36.3%). A considerable number of colonies belonging to the genus Pavona were also found to be preyed upon by sea stars (~ 12.5%), whereas predation on other reef-building genera represented only a small percentage of the total predation (Table 1). Considering the growth morphology of the consumed colonies, Culcita sp. predation was highest on corals with a branching growth morphology, whereas the predation on massive forms was less intense, although it encompassed nine different coral genera (Table 1).

Table 1 Relative percentages of colonies belonging to each coral genus that have been eaten by Culcita sp. and their average size (± SE)
Full size table

Most of the sampled sea stars consumed small adult corals with a diameter between 5 and 10 cm (260 colonies out of 433, ~ 60%) followed by juvenile corals with a diameter < 5 cm (129 colonies out of 433, ~ 30%), whereas large adult colonies with a diameter > 10 cm (average size: 15.6 ± 0.84 cm; maximum size: 38 cm) appeared to be the least predated (44 colonies out of 433, ~ 10%).

Furthermore, considering all the preyed colonies, both totally and partially, a significant positive correlation (Spearman’s rho = 0.897, P < 0.001) was detected between the size of these colonies and the size of the scars created by the predatory activity of the sea star (Fig. 5). However, as the size of the predated corals increased, the number of partially consumed colonies also increased. In fact, partial predation by Culcita sp. occurred on only approximately 7% of all consumed juvenile colonies and on approximately 19.5% of small and 25% of the larger (> 10 cm) predated colonies (Fig. 6a). As shown in Fig. 6b, the coral genus Montipora was the most partially consumed (three colonies out of 7, ~ 42%) followed by Pavona (17 colonies out of 54, ~ 31%), Porites, Acropora, Galaxea and finally Pocillopora (only 12 colonies out of 157, ~ 8%).

Fig. 5
figure 5

Relationship between the size of the scars produced by Culcita sp. on the preyed coral colonies and the size of the eaten colonies

Full size image
Fig. 6
figure 6

Relative abundance (%) of coral colonies eaten totally (in black) or partially (in white) by Culcita sp. based on their size range (diameter < 5 cm, between 5 and 10 cm and > 10 cm) (a). Numbers above each bar indicate the total number of eaten colonies (both totally and partially) per size range. Relative abundance (%) of coral colonies eaten totally or partially by Culcita sp. based on their genus (b). The coral genera reported in the graph are the only ones for which partially consumed colonies were observed. Numbers above each bar indicate the total number of eaten colonies (both totally and partially) per coral genus

Full size image

The use of the standardized forage ratio (Si) and the Ivlev’s index (Ei) allowed us to compare the abundance of prey categories in the environment and in the predator’s diet. The values obtained for both of these indices are shown in Table 2. Remarkably, the standardized forage ratio produced values that indicate a preference for corals of the genus Pocillopora and Pavona, with values of 0.592 and 0.298, respectively. In contrast, the value generated for the genus Porites (0.005) was close to zero, indicating avoidance. The Ivlev’s index showed a similar pattern, with values corresponding to preferences for Pocillopora and Pavona (0.761 and 0.576, respectively) and a negative value, indicating strong avoidance for Porites (− 0.885), (Table 2).

Table 2 Ivlev’s index (Ei) and standardized forage ratio (Si) referring to the six-most abundant coral genera in the study area
Full size table

Discussion

The direct consumption of live coral, or corallivory, represents an important biotic stressor for reef-building corals that can accelerate the rate of coral reef decline (Knowlton et al., 1990; Rotjan et al., 2006; Lenihan et al., 2011). However, although a wide variety of species from numerous taxa consume living coral (Carpenter et al., 1997; Cole et al., 2008; Rotjan & Lewis, 2008), to date, most studies have focused on a few specific organisms (e.g., COTS, butterflyfishes and Drupella spp.), indicating that the actual impact of corallivores on reefs may be underestimated. This study increases the knowledge of the little-investigated cushion sea star Culcita sp. in the Maldives, showing how this sea star is widely distributed over the reef, although exhibiting a preference for a specific depth range, reef zone and coral coverage of the habitat, and has a predilection for specific colony sizes and genera.

Regarding the distribution patterns of the cushion sea star, our results indicate that in the study area Culcita sp. can be found at a wide depth range (from 0 to at least 30 m) and in all reef zones (flat, crest, slope), even though a preference for reef slopes characterized by a shallow depth, within 10 m, was evident. In previous studies, C. novaeguineae was observed on reef slopes to 20 m depth, on reef flats, and on patch reefs in protected areas (Goreau et al., 1972; Yamaguchi, 1975; Grosenbaugh, 1981; Glynn & Krupp, 1986). However, in line with our observations, C. novaeguineae always displayed a higher density on the reef slopes below 3 and 7 m (Goreau et al., 1972; Glynn & Krupp, 1986). In contrast, C. schmideliana mainly colonizes coral reef flats (Thomassin, 1976). In this study, the reef slope may have been preferred over the reef flats and crests due to lower water movement and light exposure, a higher presence of palatable coral prey, or an abundance of coralline algae that induce the recruitment of sea star larvae and provide juveniles with food and shelter, which has been proposed to explain similar patterns of distribution of COTS (Johnson et al., 1991; De’ath & Moran, 1998; Pratchett, 2005; Kayal et al., 2012). Variations in seawater temperature could also affect cushion sea star distribution. For example, C. novaeguineae was observed in areas characterized by a temperature of between 29°C and 31°C, with highest abundances recorded at approximately 29°C (Yokley, 2016), which corresponds to the average surface temperature recorded in our study area (Seveso et al., 2015, 2017). Our data also showed that most of the sea stars were located on substrates characterized by a low live coral coverage (10–30%), whereas fewer specimens were located in reef zones with a coral coverage > 50%. This observation suggests that Culcita sp. may prefer areas where in addition to corals, several types of prey are found, increasing the food resource diversity. In fact, Culcita spp. are not obligate corallivores because they feed on other organisms, such as the epilithic algal matrix growing on dead coral, sponges, soft corals, bryozoans, small fauna (nematodes, crustaceans) and algae (Thomassin, 1976; Glynn & Krupp, 1986).

Regarding the size of Culcita individuals in relation to the coverage of live coral, our results suggest as in areas characterized by elevated coral coverage, the average size of the Culcita sp. population appeared to be the lowest. Considering that corals represent the main framework builders of the Maldivian reef systems, areas characterized by a high coral abundance generally also show a high structural complexity. Small sea stars may prefer a complex three-dimensional habitat that allows them to reduce both the pressure of predation, through hiding and crypsis, and the effects of waves and currents that could easily transport them away from the corals into deeper parts of the reef. During the sampling activities conducted in daylight conditions, very few Culcita sp. specimens were actively feeding upon corals, in line with previous evidence suggesting that this animal is preferentially a nocturnal feeder (Glynn & Krupp, 1986), and the majority of individuals were observed hidden and stationary inside ravines, crevices or under coral colonies.

In addition to habitat preferences, our study suggests that Culcita sea stars on Maldivian reefs exhibit strong patterns of prey preference and feeding behaviour. Based on previous observations, the feeding preferences of Culcita spp. appear to vary depending on geographical area. C. novaeguineae in Guam fed mainly on small heads of Acropora, Pocillopora, various faviids, Porites lichen, and possibly Hydnopora and Montipora (Goreau et al., 1972). In Hawaii, the same sea star species mainly preyed on Pocillopora meandrina (Dana, 1846), followed by Montipora verrucosa (Lamarck, 1816) and Porites lobata (Dana, 1846), showing the same food preferences when offered a single coral prey species in captivity (Glynn & Krupp, 1986). In Madagascar, C. schmideliana has been observed to more frequently eat coral colonies of small sizes and large polyps, such as Galaxea fascicularis (Linnaeus, 1767) or Goniopora stokesi (Milne Edwards & Haime, 1851), and less commonly small Acropora (Thomassin, 1976). Our results showed that in the Maldives a wide range of coral genera are preyed upon by Culcita sp.; however, the majority of predation occurred on Acropora and Pocillopora corals, which are both characterized by a branching growth morphology. Interestingly, no predatory event on colonies of Acropora displaying a tabular morphology was observed. All other corals preyed on had either an encrusting (Pavona) or massive growth form; however, this accounted for only ~ 25% of total predation events. The preference for particular coral morphologies may be in part due to the surface complexity of the coral (De’ath & Moran, 1998). For example, Keesing (1990) found that branching forms of Acropora have the greatest complexity, thereby providing the sea star A. planci (Linnaeus, 1758) with greater surface area and hence more tissue per feeding. However, the availability in the investigated area of the different coral genera targeted as prey should be considered in order to detect real food preferences. While sea stars fed on a number of coral general, Si and Ei values suggest that sea stars have a preference for the genus Pocillopora followed by Pavona, which are both characterized by a low abundance in the area. Considering that Culcita sp. has a small stomach and low feeding rate (Glynn & Krupp, 1986; Birkeland, 1989), its preference for Pocillopora could be related to the high digestibility or nutritional quality of this coral genus. In fact, the tissues of Pocillopora are very thin and, because they are not intricately connected and deeply penetrating in the skeleton, they form a superficial layer that makes them easily available for the sea star (Glynn & Krupp, 1986; Loya et al., 2001). Moreover, this genus is characterized by a low level of protective mucus, absence of large nematocysts and ineffective protection of the juvenile colonies by the Trapezia spp. symbiotic crabs, which often fail to prevent the predatory action of C. novaeguineae (Glynn & Krupp, 1986; Rouzé et al., 2014).

The genus Pavona showed a high index of preference but also a high percentage of partially eaten colonies. The genus Acropora, despite the large number of predatory acts, did not appear to be preferentially chosen by Culcita sp., and therefore, its high rates of predation may be due to the large abundance of this genus in the study area. The genus Porites, although one of the most abundant in the area, was among the least eaten, and all these few colonies targeted by Culcita showed a massive growth form. Porites could be less palatable due to its thick tissue, which penetrates deeply into the skeleton and is characterized by a low protein content, as well as the numerous and large nematocysts that can cause arm rearing, tube feet retraction and stomach withdrawal in sea stars (Brauer et al., 1970; Moore & Huxley, 1976; Glynn & Krupp, 1986; Keesing, 1990; Lough & Barnes, 2000; Loya et al., 2001). Porites is also known to host organisms that may prevent feeding (DeVantier et al., 1986; DeVantier & Endean, 1988). In addition, Porites is generally considered a large growing genus, and it is possible that Culcita sp. would have difficulties climbing it (Glynn & Krupp, 1986). Not surprisingly, the average size of the few colonies of Porites eaten by Culcita sp. was relatively small and the largest ones were only partially consumed.

In this context, small colonies (< 10 cm in diameter) have been found to be preferred by Culcita sp., which is in line with previous studies, suggesting that the lack of prehensile arms prevents Culcita spp. from climbing onto large or high growing corals (Goreau et al., 1972; Thomassin, 1976; Glynn & Krupp, 1986). However, Culcita sp. also preyed upon larger coral colonies, which were only partially eaten. In general, we observed that as the size of the colonies targeted by the sea stars increased, both the size of the scars on the colonies as well as the amount of partially consumed colonies increased. This result suggests that Culcita sp. is seldom able to kill an entire large colony at one time, perhaps preferring to change coral once a certain amount of its tissue has been assimilated, rather than consuming an entire large colony during multiple feeding attacks. We hypothesized that Culcita sp. might be more easily deterred by the developed defensive mechanisms of large adult colonies, which could also be more effective in their defence than smaller colonies due to the higher number of host symbiotic organisms (Abele & Patton, 1976; Glynn, 1980). However, even the partial mortality of coral colonies can affect reef health because it reduces coral growth rate and reproduction, due to the expenditure of the energy stores used for tissue regeneration processes (Veghel & Bak, 1994; Henry & Hart, 2005). In addition, if tissue regeneration is incomplete, portions of the colony could become susceptible to colonization by spatial competitors such as algae, sponges, other colonial invertebrates or disease-causing agents (Bak & Van Es 1980; Nugues & Bak, 2009).

Conclusion

This study provides an overview of the feeding behaviour of a corallivorous organism, Culcita sp., that has received scarce attention, but whose impact on coral reef composition and health should not be underestimated. Considering its preferential predation on recruits and juvenile coral colonies and its predilection for some specific coral genera, Culcita sp. may affect reef recovery and resilience following other stress events and generate local shifts in coral community composition and structure. In this regard, further ecological investigations on larger scales and quantitative analyses on population dynamics should be considered.