Nocturnal substrate association of four coral reef fish groups (parrotfishes, surgeonfishes, groupers and butterflyfishes) in relation to substrate architectural characteristics

Fish survey and study species

This study was conducted at Sekisei lagoon and Nagura Bay in the Yaeyama Islands, Okinawa, Japan (Fig. 1). Nocturnal underwater observations (1830 h–23:00 h) were conducted at 19 sites between November 2021 and March 2022. Using SCUBA and flashlights, the first diver swam in a zigzag pattern and searched for inactive individuals that were associated with substrates (Fig. 2), taking special care not to overlap with previous courses. The second diver followed the first diver with a data collection sheet. When the first diver found a focal fish, the second diver recorded the species, total length (TL) and substrate with which the focal fish individual was associated. In some cases, the whole body of the fish was not completely observed due to hiding behavior within the substrate. In this case, the focal fish individual was collected by spear and the TL was measured. Over 40 min observations were conducted at each site (ranging from 40 to 72 min, average μ = 52.3 ± 9.2 s.d. minutes). According to Nanami (2021), average distance of 1-minute swimming was 17.4 m. Thus, the estimated distance of each time survey was 17.4 m × survey minutes. Since the width of the time transect was 5 m, the estimated area was distance ×5 m².

During the observation period, 19, two, nine, and 12 parrotfish, surgeonfish, grouper and butterflyfish species were identified, respectively (Table 1). Among them, nine, two, two, and four species showed higher frequencies (total number of individuals was 10 or more) for parrotfishes, surgeonfishes, groupers and butterflyfishes, respectively. Thus, the data analysis was conducted in two steps. The first step was to clarify the species-level substrate associations by using above-mentioned 17 most frequent species (nine, two, two and four species for parrotfishes, surgeonfishes, groupers and butterflyfishes, respectively). The second step was to clarify the family-level substrate associations by using all species including both frequent and less-frequent species (19, two, nine, and 12 species for parrotfishes, surgeonfishes, groupers and butterflyfishes, respectively).

Data collection of substrate availability

Substrate availability at the 19 study sites was recorded during daytime. The locations of sites where nocturnal observations were conducted were recorded using a portable GPS receiver (GARMIN GPSMAP 64csx). Then, video recordings were used to record substrates on the seafloor during 20 min at each site. Static images were extracted at 10-second intervals by QuickTime Player software (version 7.6), yielding 121 static images for each site. For each image, the substrate at the center of the static image was recorded.

Substrate categorization and definition of substrate architectural characteristics

Substrates were categorized into 25 types and substrate architectural characteristics (physical structure) were categorized into seven types with some modification from several previous studies (Gardiner & Jones, 2005; Wilson et al., 2008; Nanami, 2020; Doll et al., 2021) as follows (Table 2, Fig. 3): (1) eave-like space, (2) large inter-branch space, (3) overhang provided by protrusion of fine branching structure, (4) overhang by coarse structure, (5) uneven structure without large space or overhang, (6) flat and (7) macroalgae.

Data analysis for substrate association

The analyses were conducted in two steps. The first step was to clarify the associations between fish species and the seven types of substrate architectural characteristics (physical structure). The second step was to clarify the associations between fish species and the 25 substrate types.

Fish associations were analyzed by using “resource selection ratio” (Manly et al., 2002). The approach follows previous studies that have applied this index to examine the quantitative degree of substrate association of coral reef fishes to specific substrate characteristics (e.g., Gardiner & Jones, 2005; Wilson et al., 2008; Doll et al., 2021; Nanami, 2023). This index also shows 95% confidence intervals, which can be used to test the statistical significance of the substrate association of fishes for each substrate type.

The resource selection ratio was calculated as: ${\displaystyle w_i=o_i/{\pi }_i}$ where w_i is the resource selection probability function, o_i is the proportion of the ith substrate that was used by a focal fish species, and π_i is the proportion of the ith substrate that was available in the study area (Manly et al., 2002). For multiple comparisons, Bonferroni Z corrections were used in order to calculate the 95% confidence interval (CI) for each w_i. The formula used to calculate the 95% CI was: ${\displaystyle 95\text{\%}\mathrm{CI}=Z_{a/2k}\surd \left[o_i\left(1-o_i\right)/\left(U_{+}{{\pi }_i}_2\right)\right]}$ where Z_a/2k is the critical value of the standard normal distribution corresponding to an upper tail area of a/2k, a is 0.05, k is the number of substrate categories, and U₊ is the total number of individuals of the focal fish species. Substrates with w_i ±95% CI above and below 1 indicate a significantly positive and negative association, respectively. Substrates with w_i ±95% CI encompassing 1 had no significant positive or negative association.

In addition, the standardized selection ratio that indicates relative degree among substrates for habitat selection was calculated as follows: ${\displaystyle B_i=w_i/\sum w_i.}$ If a focal species shows Bi and Bj for i th and jth substrates, ith substrate is selected with Bi/Bj times the probability of jth substrate.

Both species level (17 species) and family level (four families) data analyses were performed.

Variations in substrate associations among different fish size classes

To investigate the variations in substrate associations among different fish size classes, fish individuals were divided into three size classes as follows: (1) TL ≤ 29 cm (smaller-sized); (2) TL = 30–39 cm (medium-sized) and (3) TL ≥ 40 cm (larger-sized). Then, their degree of substrate association was analyzed. Five species (Scarus schlegeli, Chaetodon trifascialis, C. lunulatus, C. ephippium and C. auriga) were excluded from the analysis, since total length of the all individuals were 29 cm or less for the five species.

Data preparation prior to analysis

All data for substrate associations by fish were obtained from the 19 study sites were pooled for the analysis. Although all data for substrate availability from the 19 sites were also pooled for the analysis, a modification was applied due to the difference in observation time among the 19 sites (see substrate availability raw data; Supplemental Information). Namely, substrate compositions at sites with longer fish observation durations should be included with greater proportions whereas substrate compositions at sites with shorter time observation durations should be included with lower proportions. The degree of the proportion was assigned by the observation duration at the site. Thus, the modification was as follows: ${\displaystyle \text{Overall proportion of}i\text{th substrate}=\sum _{j=1}^{19}{P}_{ij}{T}_j/\sum _{i=1}^k\sum _{j=1}^{19}{P}_{ij}{T}_j}$ where P_ij is the proportion of ith substrate at site j, T_j is the observation duration (minutes) at site j, and k is the number of substrate types (k = 7 for seven types of substrate architectural structure and k = 25 for twenty-five substrate types).

Overall trend in substrate association

To summarize species-specific differences in substrate association, a principal component analysis (PCA) and cluster analysis using the group average linkage method with the Bray–Curtis similarity index was applied based on the number of fishes by including data from the seventeen fish species. Analyses were performed using PRIMER (version 6) software package (Clarke & Warwick, 1994). For plotting the PCA score of each fish species, data about nocturnal substrate association were also shown by pie charts. Additional PCA was performed to clarify the variations in substrate associations among the above-mentioned three fish size classes.

Parrotfishes

Chlorurus microrhinos was primarily associated with large inter-branch space (staghorn Acropora) and overhang by coarse structure (rock) (Fig. 4A). Significant positive associations with large inter-branch space and overhang by coarse structure were found (Table 3, Table S1). However, no significant substrate associations were found for any types of 25 substrates (Table 4, Table S2). For size difference, smaller-sized and medium-sized individuals were primarily associated with large inter-branch space (staghorn Acropora), whereas larger-sized individuals were primarily associated with overhang by coarse structure (rock: Fig. S1).

Chlorurus spilurus was primarily associated with eave-like space (corymbose Acropora and tabular Acropora), large inter-branch space (staghorn Acropora) and overhang by fine branching structure (non-acroporid branching coral) (Fig. 4B). Significant positive associations with eave-like space and large inter-branch space were found (Table 3, Table S1). For eave-like space, no significant substrate-specific associations were found (Table 4, Table S2). For large inter-branch space, significant positive association with staghorn Acropora was found (Table 4, Table S2). In contrast, a significant negative association with overhang by coarse structure (rock) was found (Tables 3 and 4, Tables S1, S2). By size, smaller- and medium-sized individuals showed relatively greater proportion of association with eave-like space (corymbose and tabular Acropora) and large inter-branch space (staghorn Acropora), respectively (Fig. S2).

Hipposcarus longiceps was primarily associated with large inter-branch space (staghorn Acropora) and overhang by coarse structure (rock) (Fig. 4C). Significant positive and negative associations with large inter-branch space (staghorn Acropora) and overhang by fine branching structure were found, respectively (Tables 3 and 4, Tables S1, S2). By size, smaller-, medium- and larger-sized individuals showed relatively greater proportion of association with large inter-branch space (staghorn Acropora), overhang by coarse structure (rock) and eave-like space (tabular and dead tabular Acropora), respectively (Fig. S3).

Scarus ghobban was primarily associated with eave-like space (corymbose Acropora) and overhang by coarse structure (massive coral and rock) (Fig. 5A). Although this species showed respectively significant positive and negative associations with overhang by coarse structure and overhang by fine branching structure (Table 3, Table S1), no significant substrate-specific associations were found (Table 4, Table S2). All three size classes showed relatively greater proportion of association with overhang by coarse structure (massive coral and rock: Fig. S4).

Five species (Scarus forsteni, S. niger, S. oviceps, S. rivulatus and S. schlegeli) were primarily associated with eave-like space (corymbose Acropora and tabular Acropora) (Figs. 5B, 5C, 6A–6C) and showed a significant positive association with the eave-like space (Table 3, Table S1). Three species (S. forsteni, S. oviceps and S. rivulatus) and one species (S. schlegeli) showed positive associations with tabular Acropora and corymbose Acropora, respectively (Table 4, Table S2). In contrast, S. niger did not show any substrate-specific associations (Table 4, Table S2). For size difference, two size classes (smaller- and larger-sized) individuals of S. forsteni showed greater proportion in association with eave-like space (tabular Acropora) while medium-sized fish were associated with overhang by coarse structure (rock) , respectively (Fig. S5). Smaller-sized individuals of S. niger showed greater proportion in association with eave-like space (mainly tabular Acropora) and medium-sized with large inter-branch space (staghorn Acropora), respectively (Fig. S6). In contrast, all size classes of the two species (S. oviceps and S. rivulatus) were primarily associated with eave-like space (mainly tabular Acropora: Figs. S7, S8).

Surgeonfishes

Naso unicornis was primarily associated with overhang by coarse structure (rock: Fig. 7A) and showed a significant positive association with the substrate (Tables 5 and 6, Tables S3, S4). A significant negative association with overhang by fine branching structure was also found (Table 5, Table S3). By size, smaller- and larger-sized individuals were primarily associated with eave-like space (dead tabular Acropora) and overhang by coarse structure (rock), respectively (Figs. S9A, S9C). Medium-sized individuals was associated with both eave-like space (tabular Acropora) and overhang by coarse structure (rock: Fig. S9B).

Naso lituratus was primarily associated with eave-like space (tabular Acropora) and overhang by coarse structure (rock: Fig. 7B). Significant positive association with eave-like space was found (Table 5, Table S3). However, no significant substrate associations were found for any types of 25 substrates (Table 6, Table S4). For size difference, smaller- and medium-sized individuals showed greater proportion in association with eave-like space (mainly tabular Acropora) and overhang by coarse structure (rock), respectively (Fig. S10).

Groupers

Plectropomus leopardus was primarily associated with eave-like space (corymbose and tabular Acropora) and overhang by coarse structure (rock: Fig. 8A). This species showed a significant positive association with eave-like space (Table 5, Table S3), although no significant substrate-specific associations were found (Table 6, Table S4). In contrast, a significant negative association with flat (coral rubble) was found (Tables 5 and 6, Tables S3, S4). By size, medium-sized individuals were primarily associated with eave-like space (mainly corymbose and tabular Acropora: Fig. S11B). However, no clear trends were found for smaller- and larger-sized individuals (Figs. S11A, S11C).

Epinephelus ongus was primarily associated with overhang by fine branching structure (non-acroporid branching coral) and overhang by coarse structure (rock: Fig. 8B). A significant positive association with overhang by coarse structure were found (Table 5, Table S3). However, for substrate-specific associations, significant positive and negative associations with non-acroporid branching coral and branching Acropora were respectively found (Table 6, Table S4). All size class individuals showed greater proportions in association with overhang by coarse structure (rock: Fig. S12). Some individuals were also associated with overhang by fine branching structure (branching coral) and this trend was observed for all size classes (Fig. S12).

Butterflyfishes

Chaetodon trifascialis was primarily associated with eave-like space (tabular Acropora) and large inter-branch space (staghorn Acropora: Fig. 9A) and showed significant positive associations with these substrates (Tables 5 and 6, Tables S3, S4). This species also showed a significant negative association with overhang by coarse structure (rock: Tables 5 and 6, Tables S3, S4).

Chaetodon lunulatus was primarily associated with large inter-branch space (staghorn Acropora: Fig. 9B) and showed a significant positive association with the substrate (Tables 5 and 6, Tables S3, S4). This species also showed a significant negative association with overhang by coarse structure (rock: Tables 5 and 6, Tables S3, S4).

Chaetodon ephippium was associated with large inter-branch space (staghorn Acropora) and overhang by coarse structure (dead massive coral and rock: Fig. 9C) and showed a significant positive association with large inter-branch space (Table 5, Table S3). However, no significant substrate associations were found for any types of 25 substrate types (Table 6, Table S4).

Chaetodon auriga was primarily associated with eave-like space (corymbose Acropora and dead tabular Acropora) and overhang by coarse structure (rock: Fig. 9D). However, no significant associations with any structural characteristics and substrate types were found (Tables 5 and 6, Tables S3, S4).

Family-level substrate associations

Parrotfishes were primarily associated with eave-like space (corymbose Acropora and tabular Acropora), and some individuals were also associated with large inter-branch space (staghorn Acropora), overhang by fine branching structure (non-acroporid branching coral) and overhang by coarse structure (rock: Fig. S13A). Parrotfishes showed significant positive associations with eave-like space (corymbose Acropora, tabular Acropora and dead tabular Acropora) and large inter-branch space (staghorn Acropora) were found, whereas showed a significant negative association with overhang by fine branching structure (bottlebrush Acropora: Tables S5–S8).

Surgeonfishes were primarily associated with overhang by coarse structure (rock), and some individuals were also associated with eave-like space (tabular Acropora) and overhang by fine branching structure (non-acroporid branching coral) (Fig. S13B). Surgeonfishes showed significant positive associations with eave-like space (tabular Acropora) and overhang by coarse structure (rock: Tables S5–S8). A significant negative association with overhang by fine branching structure was also found (Tables S5, S7).

Groupers were primarily associated with overhang by coarse structure (rock), and some individuals were associated with eave-like space (corymbose Acropora and tabular Acropora), large inter-branch space (staghorn Acropora) and overhang by fine branching structure (non-acroporid branching coral: Fig. S13C). For seven types of substrate architectural characteristics, groupers showed significant positive and negative associations with overhang by coarse structure and flat, respectively (Tables S5–S8). However, for 25 substrate types, a significant positive associations with non-acroporid branching corals was found (Tables S6, S8). In contrast, significant negative associations with branching Acropora, bottlebrush Acropora and coral rubble were found (Tables S6, S8).

Butterflyfishes were primarily associated with large inter-branch space (staghorn Acropora), and some individuals were also associated with eave-like space (corymbose Acropora and tabular Acropora), overhang by fine branching structure (branching Acropora) and overhang by coarse structure (rock: Fig. S13D). Butterflyfishes showed a significant positive association with large inter-branch space (staghorn Acropora), whereas a significant negative association with overhang by fine branching structure (bottlebrush Acropora) (Tables S5–S8). A significant negative association with massive coral was also found (Tables S6, S8).

Overall trend of substrate association including the seventeen fish species

For the seven types of substrate architectural characteristics, PCA revealed that three architectural characteristics (eave-like space, large inter-branch space and overhang by coarse structure) showed major contributions for nocturnal fish associations (Fig. 10A). Cluster analysis revealed the 17 species could be divided into six groups (Fig. 10B, Fig. S14A). Two species (Scarus ghobban and Naso unicornis: group B), one species (Chaetodon lunulatus: group D) and five species (Scarus forsteni, S. niger, S. oviceps, S. rivulatus and S. schlegeli: group F) showed greater proportions in association with overhang by coarse structure, large inter-branch space and eave-like space, respectively. Other fishes belonging to three groups (group A, C and E) did not show greater proportion in association with any particular architectural characteristics. For fish size difference, four species (Ch. microrhinos, H. longiceps, S. niger and N. unicornis) showed relatively clear variations in substrate associations among difference size classes (Fig. S15). For the two species (Ch. microrhinos and H. longiceps), the main associated substrates changed from large inter-branch space to overhang by coarse structure as fish size increased (Figs. S15B, S15D). In contrast, the other two species (S. niger and N. unicornis) showed that the main associated substrates changed from eave-like space to large inter-branch space (Fig. S15G) and from eave-like space to overhang by coarse structure as fish size increased (Fig. S15K), respectively.

For 25 substrate types, PCA revealed that three substrate types (tabular Acropora, staghorn Acropora and rock) showed major contributions for nocturnal fish associations (Fig. 10C). Cluster analysis revealed 17 species could be divided into eight groups (Fig. 10D, Fig. S14B). Naso unicornis (group A), Chaetodon lunulatus (group D), Scarus schlegeli (group F) and two species (Scarus oviceps and S. rivulatus: group H) showed greater proportions in association with rock, staghorn Acropora, corymbose Acropora and tabular Acropora, respectively. Other fishes belonging to four groups (group B, C, E, G) and one species (Chaetodon trifascialis: group D) did not show greater proportions in association with any particular substrate type. For fish size difference, two species (Ch. microrhinos and H. longiceps,) showed that the main associated substrates changed from staghorn Acropora to rock as fish size increased (Figs. S16B, S16D). Two species (S. niger and N. unicornis) showed that the main associated substrates changed from tabular Acropora to staghorn Acropora (Fig. S16G) and from dead tabular Acropora to rock as fish size increased (Fig. S16K: dead tabular Acropora was shown as “other substrates” in Fig. S16K. See also Fig. S9), respectively.

Parrotfishes

Most previous studies have conducted diurnal observations to clarify the spatial distribution in relation to topographic and substrate characteristics (Hoey & Bellwood, 2008; Hernández-Landa et al., 2014; Nanami, 2021) and foraging substrates (Nanami, 2016; Bonaldo & Rotjan, 2018). However, substrate associations for parrotfish species have not been sufficiently examined due to their highly diurnal activity (e.g., Welsh & Bellwood, 2012). Pickholtz et al. (2023) examined nocturnal substrate associations of seven parrotfish species in the Indian Ocean (Gulf of Aqaba), in which substrates were categorized into five types (branching coral, massive coral, soft coral, rock and artificial structure). In contrast, the present study conducted in the North Pacific (Okinawa) and categorized substrates into seven types in terms of architectural characteristics and 25 types in terms of more precise aspects (e.g., coral morphology, live coral or dead coral, and other non-coralline substrates).

Three species (Chlorurus microrhinos, C. spilurus and Hipposcarus longiceps) showed significant positive associations with large inter-branch space (staghorn Acropora). Pickholtz et al. (2023) revealed nocturnal substrate associations for three closely related species in the Indian Ocean (C. gibbus, C. sordidus and H. harid) and showed some individuals of the three species were associated with branching corals. These results suggest that substrates that were positively associated with parrotfishes are similar among closely related species.

Scarus ghobban and Chlorurus microrhinos showed significant positive associations with overhang by coarse structure. Nanami & Nishihira (2004) showed smaller-sized fish species (pomacentrids and juveniles of labrids of less than 10 cm in length) were associated with the base of massive corals as shelter due to their overhang structure. In contrast, Kerry & Bellwood (2012) suggested that massive corals showed less contribution for concealment of larger-sized fishes (over 10 cm in length), although a possibility that large massive corals might provide canopy effects by overhang at the base of the colony. The results of this study support this suggestion. Namely, overhangs provided by coarse structure serve to some degree as sleeping sites for larger-sized parrotfish individuals (TLs were 24 cm and over).

The remaining five species (Scarus forsteni, S. niger, S. oviceps, S. rivulatus and S. schlegeli) and C. spilurus showed significant positive associations with eave-like space (primarily provided by corymbose Acropora and tabular Acropora). As Kerry & Bellwood (2012) suggested, it was revealed that tabular corals provide concealment for some parrotfish species as sleeping sites due to their canopy structure.

Surgeonfishes

Naso unicornis and N. lituratus showed significant positive associations with overhang by coarse structure mainly provided by rock and eave-like space being mainly provided by tabular Acropora, respectively. Some N. unicornis were also associated with eave-like space provided by tabular Acropora. These findings suggest that canopy structure (overhangs and tabular structure) should be conserved as sleeping sites for these species.

Naso unicornis and N. lituratus are main fishery targets in coral reefs (Bejarano et al., 2013; Taylor et al., 2014) and nighttime spear fishing is a common method to catch inactive individuals of these species (Taylor et al., 2014). Conservation of critical substrates as sleeping sites could serve as fishing locations that can be utilized by fishermen.

Groupers

Plectropomus leopardus is diurnally active and nocturnally inactive (Matley, Heupel & Simpfendorfer, 2015). Broad-scale diurnal survey (several and several-tens of kilometer scale) have shown that a greater coverage of branching Acropora was positively related with greater density of this species (Nanami, 2021). In contrast, this species showed a significant positive association with eave-like space mainly provided by corymbose and tabular Acropora as sleeping sites. These results suggest that substrate types that affect the spatial distribution of the species may be different between daytime and nighttime. Plectropomus leopardus is a carnivore and its main prey items are small-sized fishes (St John, 1999). Since such small-sized fishes were often associated with branching Acropora, this species might occur at sites with greater coverage of branching Acropora for foraging during daytime but utilize eave-like space as sleeping sites during nighttime. Thus, multiple substrate types are needed to satisfy the ecological requirements of this species during both daytime and nighttime.

Diurnal observations revealed that large-sized Epinephelus ongus individuals (over 18 cm TL) showed a significant positive association with large inter-branch space that was created by staghorn Acropora (Nanami et al., 2013). In contrast, nocturnal observations by this study showed positive associations with overhang by coarse structure. Nanami et al. (2018) suggested that this species is nocturnally active since a greater home range size was observed at nighttime than daytime. This species might be associated with overhang by coarse structure for ambush foraging at nighttime.

Butterflyfishes

Chaetodon trifascialis showed positive associations with eave-like space (tabular Acropora) and large inter-branch space (staghorn Acropora). This species is an obligate coral polyp feeder and mainly feeds on polyp of tabular Acropora and corymbose Acropora (Pratchett, 2005; Nanami, 2020). This suggests that coral species providing large inter-branch space are important architectural structure as sleeping sites for this species, which was not indicated by diurnal observations for the clarifying foraging behavior. In contrast, tabular Acropora was also utilized as sleeping sites, suggesting that tabular Acropora is essential as both foraging and sleeping sites for this species.

Chaetodon lunulatus showed a significant positive association with large inter-branch space being provided by staghorn Acropora. In contrast, diurnal observations revealed that this species mainly feeds on polyps of encrusting, massive and non-acroporid corals, which do not provide large inter-branch space (Pratchett, 2005; Nargelkerken et al., 2009; Nanami, 2020). This indicates that C. lunulatus depends on staghorn Acropora as sleeping sites but it is not utilized as a foraging substrate, suggesting that various types of corals are essential for this species.

Chaetodon ephippium showed a significant positive association with large inter-branch space being provided by staghorn Acropora. In contrast, this species showed frequent bites on the surface of coral rubble, dead coral and rock (Nargelkerken et al., 2009; Nanami, 2020), probably due to catch invertebrates (Sano, Shimizu & Nose, 1984; Pratchett, 2005). This indicates that substrates utilization by C. ephippium was different between daytime and nighttime.

Chaetodon auriga did not show any significant associations with substrates. This species is facultative coral polyp feeder (Sano, Shimizu & Nose, 1984) and showed a greater number of bites on coral rubble and rocks (Nanami, 2020). Since this species was mainly associated with four types of substrate architectural characteristics (eave-like space, large inter-branch space, overhang by fine branching structure and overhang by coarse structure) but not associated with other three types of architectural characteristics (uneven surface, flat and macroalgae), this species utilized substrates with complex physical structure as sleeping sites. Since these four types of substrate architectural characteristics are provided by both live corals and rock, such substrates with greater complexity should be conserved as sleeping site for the species.

Overall, this study revealed large inter-branch spaces that created by staghorn Acropora was important physical structure as sleeping sites for the three species (C. trifascialis, C. lunulatus and C. ephippium) and substrates with complex physical structure were also important as sleeping site for C. auriga, which have not been revealed by diurnal observations in previous studies.

Variations in substrate association among different fish size classes

Four species showed clear variations in nocturnal substrate associations among different size classes. The two species (Ch. microrhinos and H. longiceps,) and one species (N. unicornis) showed that their main associated substrates changed from large inter-branch space (staghorn Acropora) to overhang by coarse structure (rock), and from eave-like space (dead tabular Acropora) to overhang by coarse structure (rock) as fish size increased, respectively. These results suggest that smaller- and larger-sized individuals were respectively associated with fine and coarse habitat structures, and various types of substrate architectural characteristics are needed for the various sizes of the three species as nocturnal sleeping sites. In contrast, S. niger showed that the main associated substrates changed from eave-like space (mainly tabular Acropora) to large inter-branch space (staghorn Acropora) as fish size increased, suggesting that various types of acroporid corals are needed for the various sizes of the species as nocturnal sleeping sites.

Implication about coral community degradation induced by climate change

Numerous studies have shown that coral species belonging to the genus Acropora is highly susceptible to coral bleaching by climate change (e.g., Marshall & Baird, 2000; Loya et al., 2001; McClanahan et al., 2004) and such degradation of the acroporid coral community causes significant declines of fish populations in coral reefs (Pratchett et al., 2008). All 17 species were nocturnally associated with acroporid coral, although the degree of association was species-specific. Especially, five species (Scarus oviceps, S. rivulatus, S. schlegeli, Chaetodon lunulatus and C. trifascialis) showed a greater proportion in association with acroporid corals. Some other species (Chlorurus microrhinos, C. spilurus, Hipposcarus longiceps, S. forsteni, S. niger, Naso lituratus, Plectropomus leopardus, Chaetodon ephippium) also showed positive associations with acroporid corals to some extent. In contrast, almost all fish species (except for one individual of P. leopardus) showed no associations with uneven structure without large space or overhang, flat and macroalgae, indicating fish avoidance of the three substrate architectural structural categories. These results suggest that the effects on coral degradation would negatively impact on the availability of sleeping sites for some fish species. This degradation would also cause a decline in fishing grounds for night spear fishing.