Fish Composition and Diversity of Four Coral Reefs in the South China Sea Based on Hand-Line Catch

Abstract

To improve the overall understanding of the fish diversity and spatial patterns of major coral reefs in the South China Sea, fish assemblage composition, dominant species, biodiversity indices, and multivariate analysis of community structure were reported for four major coral reefs based on hand-line survey data in May and September 2018. A total of five orders, 21 families, 45 genera and 121 species of fish were recorded with Perciformes (78.5%) being the most diverse. The highest number (5) of dominant species was found near Chenhang Island while the lowest (2) number was detected at Zhubi Reef. The highest abundance index (7.21) occurred at Zhubi Reef, while the Shannon–Wiener diversity (4.80), Pielou’s evenness (0.81), and Simpson’s dominance (0.95) indexes were all highest at Qiliangyu Island. Based on cluster analysis and non-metric multi-dimensional scaling (NMDS), fish communities varied more spatially than seasonally. Our results led us to hypothesize that the habitat complexity and level of anthropogenic disturbance were the main factors affecting the composition of reef-dwelling fish on each coral reef. Topography was likely responsible for most variation in the spatial pattern of fish diversity.

1. Introduction

Coral reefs are highly biodiverse and productive environments, and among the most obvious of the ecosystems to be affected by global climate change and anthropogenic disturbance. The conservation of coral reef ecosystems and reef fish diversity is a topical and internationally important issue in the field of marine environmental science [1], with the decreased biodiversity and functional degradation of coral reef habitats attracting considerable attention [2,3,4].

Coral reefs are the most important and distinctive ecosystems in the South China Sea (SCS), and are of great significance in the maintenance of biodiversity in this region and as fishing grounds for coral reef fisheries [5,6]. Various surveys of reef fish in this region have been performed [7,8,9,10], and extensive information regarding coral reef fish biodiversity is widely available [11,12,13,14]. New records of species and the feeding and biology of coral reef fish in the SCS have improved knowledge of species diversity, structure and function [10,12,15,16,17,18,19].

Climate change and ocean acidification can affect coral reef fish [20,21,22,23]. For example, coral bleaching can lead to diversity loss and changes in the structure of fish assemblages [24,25,26], and loss of coral reef fish biodiversity can affect ecosystem functioning and services [27]. Therefore, research on coral reef fish diversity improves our understanding of reef fish succession and their vulnerability to environmental changes and anthropogenic disturbance [28,29]. In particular, after repeat coral bleaching and long-term anthropogenic disturbance, the composition and diversity of coral reef fish in the SCS needs to be determined in order to conserve remaining coral reefs and their fish resources. Most previous studies on coral reef fish in the South China Sea have focused on the archipelagic scale, with few studies reporting the characteristics of fish diversity on the individual reef level and the differences among reefs [6,9,11,12]. To improve the overall understanding of coral reef fish diversity in the South China Sea, in this study we analyzed the composition, dominant species, biodiversity indices, and multivariate analysis of fish assemblages from four major coral reefs of the Nansha Islands and Xisha Islands in the SCS (Qilianyu and Chenhang Islands and Zhubi and Meiji Reefs), and investigated their spatial distribution and the differences among the coral reefs. In the four coral reefs of our study, there were two previous surveys on the coral reef fish of Meiji Reef, but the investigation efforts and sample sizes were very small, and the data were very poor and there was no diversity information available. For the other three study sites, there were no survey data available on fish biodiversity. Therefore, our research is very meaningful for understanding the diversity of fish in these coral reefs.

2. Materials and Methods

2.1. Study Site

Reef fish assemblages were surveyed at four coral reefs in the SCS (Qilianyu and Chenhang Islands and Zhubi and Meiji Reefs; Figure 1) in the boreal spring (May) and autumn (September) of 2018. The northeast monsoon prevails from October to March of the following year, and the southwest monsoon prevails from May to September. In Chenhang Island, Zhubi Reef and Meiji Reef, the water ecosystems are being measured, and commercial fishing is strictly prohibited except for scientific fishing, but in Qilianyu Island commercial coral reef fishing is allowed by gillnet, hand-line and diving fishing. The four study sites are not currently included in the MPA.

Zhubi (10°54′ N, 114°03′ E) and Meiji (9°55′ N, 115°32′ E) Reefs are both part of the Nansha Islands. Zhubi Reef is a closed and approximately pear-shaped atoll, with a central lagoon approximately 9.5 km² in area, mostly 20 m deep, and 24 m at its deepest [30]. Meiji Reef is a semi-closed, oval-shaped atoll with a lagoon of maximum depth 30 m, surrounded by a ring of reef flat of approximately 30.62 km² [31]. The straight-line distance between Zhubi and Meiji reefs is nearly 194 km, but both are located in tropical regions with no obvious seasonal fluctuation in weather conditions.

Qilianyu (16°55–17°00′ N, 112°12–21′ E) and Chenhang (16°27′ N, 111°43′ E) Islands are both part of the Xisha Islands. Qilianyu Island is composed of seven very small islets and sandbanks, including Zhaoshu, North, Middle, and South Islets, and North, Middle and South Sandbanks. Qilianyu Island is an open coral reef of maximum depth 40 m, with a lagoon of approximately 2 km² area. The center of Chenhang Island lies lower in latitude; it is surrounded by sandbanks, has a curved triangular shape and a land area of 0.43 km², and most of its coral reef is 5–30 m deep (maximum 45 m). The straight-line distance between Qilianyu Island and Chenhang Island is approximately 81 km. Both islands experience high temperatures throughout the year, without obvious seasonal distinction.

Water temperature and salinity were determined at each reef using an YSI ProPlus meter (YSI Inc., Yellow Springs, OH, USA). In spring and autumn, the mean temperatures of Zhubi Reef in the upper 5 m of the water column were 28.69 ± 0.09 °C and 29.08 ± 0.12 °C, respectively, and for Meiji Reef, 27.95 ± 0.10 °C, and 28.50 ± 0.05 °C, respectively. Mean temperatures in the upper 5 m at Qilianyu Island in spring and autumn were 28.88 ± 0.11 °C and 29.28 ± 0.08 °C, respectively, and at Chenhang Island, 28.82 ± 0.08 °C and 29.10 ± 0.07 °C, respectively.

2.2. Fish Surveys

The R/V NanFeng (66.66 m length, 12.40 m width, 1537 t GT), equipped with a motorboat, was used to perform surveys. Fish were collected by hand-line from an outboard gasoline-powered motorboat (7.85 m length, 1.50 m width).

Hand-line gear included small (hook size: 26.0 mm total length, 10.0 mm gape size, 1.0 mm thickness, 0.330 mm diameter nylon wire) and big (hook size: 35.0 mm total length, 15.0 mm gape size, 1.21 mm thickness, 0.405 mm diameter nylon wire) hooks with barbs. In both May and September, eight stations (5 m, 2 stations; 10 m, 4 stations, 20 m, 2 stations) were surveyed in the lagoon of each reef, covering the area of each reef. At each station, the same five fishers used 10 hooks (5 small, 5 big) and sampled synchronously for 3 h (07:00–08:00, 13:00–14:00, 19:00–20:00) using fresh shrimp as bait.

Each fish specimen caught was identified to the lowest taxonomic category using morphological characteristics, then frozen (−20 °C) for further shore-based analysis. The wet body mass was measured using electronic scales to the nearest 0.01 g (Sunny Hengping Instrument, Shanghai, China). The total number of catches collected was 2282 individuals.

2.3. Diversity Indices

The data were standardized according to the weight and number of the specimens per unit hook by time, and kg/(100 hooks·h) and individuals/(100 hooks·h) were used to represent the standardized catch per unit effort (CPUE) based on the weight and number of specimens. An index of relative importance [32], Margalef’s richness index (D) [33], Shannon–Wiener diversity index (H′) [34,35], Simpson’s diversity index (C) [36], Pielou’s evenness index (J′) [37], and Jaccard similarity coefficient (J_s) [38] were calculated to evaluate the structure of coral reef fish assemblages. Among them, in order not to be influenced by inter- and intra-species individual differences in fish species, and to calculate the results more closely relative to the inter-species energy distribution, a method proposed by Wilhm [34] was used to calculate fish species diversity by replacing the number of individuals with biomass.

The index of relative importance (IRI) was calculated as:

$$IRI=(N+W)\times F$$

(1)

where N_i is a percentage, the number of the ith species divided by the total number of all individuals of all species; W_i is a percentage, the weight of ith species divided by the total weight of all specimens of all species; and F_i is a percentage, the number of stations with the ith species divided by the total number of surveyed stations. Species with IRI ≥ 500, 100 ≤ IRI < 500, 10 ≤ IRI < 100, and IRI < 10 are classified as dominant, common, general, and rare, respectively [39].

D, H′, C, J′, and J_s are calculated as follows [34,35,36,37,38]:

$$D=(S-1)/{log}_{2}N$$

(2)

$$H\prime =-{\displaystyle {\sum }_{i=1}^S(w_i}/W){log}_2(w_i/W)$$

(3)

$$C=1-{\displaystyle {\sum }_{i=1}^S(\frac{w_i}{W}}{)}^2$$

(4)

$$J\prime =H^{\prime }/H^{\prime }{}_{max}$$

(5)

$$J_s=\frac{c}{a+b-c}$$

(6)

where S is the total number of species; N is the total number of individuals in the sample; W_i is the wet weight of the ith species; W is total wet weight of the species divided by the total wet weight of all specimens of all species; H′_max is equal to log₂ S; a is the number of species on one reef (nominally “reef A”); b is the number of species on another reef (nominally “reef B”); and c is the number of species common to reefs A and B. When 0 ≤ J_s < 0.25, 0.25 ≤ J_s < 0.50, 0.50 ≤ J_s < 0.75, and 0.75 ≤ J_s < 1.00, the composition of species on reefs A and B are considered extremely dissimilar, moderately dissimilar, moderately similar, and very similar, respectively [40].

2.4. Multivariable Statistical Analysis of Community Structure

The data were processed and analyzed using ArcGIS 10.3 and R (version 3.6.3). The dominance and diversity indices of fish assemblages were calculated depending on regions and seasons to compare spatio-temporal differences. One-way analysis of similarities (ANOSIM) was used to test the differences in species composition among assemblage structures in different regions, and the significance among the communities was described by cluster analysis. ANOSIM was followed with a SIMPER routine to identify which species is/are the most responsible for the observed spatial or temporal differences. The Bray–Curtis similarity coefficients were calculated using the fourth root-transformed data to construct a similarity matrix among the islands in different seasons. Group average clustering and non-metric multidimensional scaling (NMDS) were used to analyze the heterogeneity of fish composition between seasons in the four regions [41]. Because of the complementary nature of these two methods, they could be used together to validate each other’s analytical results and to elucidate community patterns more effectively [42]. The above multivariate analysis was conducted in PRIMER 5.0 software. The strength of the NMDS analysis was measured by the stress coefficient (stress), where coefficients < 0.2, 0.1 and 0.05 indicated acceptable representation, good ranking and good representation of ranking, respectively [43].

3. Results

3.1. Species Composition

On the four coral reefs, 121 fish species were caught and identified, attributed to five orders, 21 families, and 45 genera (Appendix A). The most species occurred at Qiliangyu Island, where 60 species belonging to four orders and 14 families were recorded; Meiji Reef included 53 species belonging to 14 families and four orders, and Zhubi Reef encompassed 49 species from 15 families and five orders; 38 species from 21 families and five orders were registered near Chenhang Island. There were seasonal differences in fish composition between regions. In Meiji Reef, the number of species, genera and families was lower in May than in September. In Zhubi Reef, the number in May was equal to that in September. In Qilianyu and Chenhang Islands, the numbers in May were both higher than in September (

Table 1. Number of fish species in Meiji Reef, Zhubi Reef, Qilianyu Island and Chenhang Island in the South China Sea in May and September.

Island/Reef	Season	Order	Family	Genus	Species
Meiji Reef	May	2	10	12	26
	September	4	13	19	40
	All	4	14	22	53
Zhubi Reef	May	4	12	18	29
	September	3	12	19	29
	All	5	15	23	49
Qilianyu Island	May	4	14	26	37
	September	3	10	16	34
	All	4	14	28	60
Chenhang Island	May	4	12	20	35
	September	4	9	12	16
	All	5	14	22	38
Total	—	5	21	45	121

).

Perciformes dominated, with 95 species in 16 families accounting for 78.5% of all species, followed by Beryciformes (13 species), Tetraodontiformes (7 species), Aulopiformes (3 species), and Clupeiformes (2 species). Some 21 families were reported (Figure 2).

3.2. Dominance Degree

For all four coral reefs, the seasonal differences in the composition of dominant species of fish assemblages were significant (

Table 2. Dominant species at four coral reefs in the South China Sea in different seasons.

Month	Meiji Reef		Zhubi Reef		Qilianyu Island		Chenhang Island
	Specie	IRI	Specie	IRI	Specie	IRI	Specie	IRI
May	Epinephelus merra	2575	Parupeneus trifasciatus	2148	Gnathodentex aureolineatus	2897	Pentapodus caninus	8288
	Lethrinus atkinsoni	1039	Cheilinus fasciatus	1908	Lethrinus rubrioperculatus	1781	Cephalopholis spiloparaea	3227
	Pentapodus caninus	1029	Pristicon trimaculatus	1891	Lethrinus lentjan	1706	Gnathodentex aureolineatus	2045
	Rastrelliger kanagurta	788	Parapercis hexophtalma	1332	Lethrinus erythracanthus	931	Parupeneus trifasciatus	1697
			Sufflamen chrysopterum	1119	Aprion virescens	901
			Cephalopholis urodeta	628	Myripristis murdjan	742
			Parapercis tetracantha	519	Lethrinus obsoletus	656
					Sargocentron caudimaculatum	501
September	Epinephelus merra	2659	Parupeneus trifasciatus	1768	Lethrinus semicinctus	3087	Pentapodus caninus	5639
	Lutjanus kasmira	1396	Cheilinus fasciatus	884	Lethrinus rubrioperculatus	2319	Cephalopholis spiloparaea	4779
	Lutjanus gibbus	672	Monotaxis grandoculis	680	Gnathodentex aureolineatus	2139	Parupeneus trifasciatus	2761
	Pentapodus caninus	594	Rhinecanthus aculeatus	635	Gymnocranius griseus	1739	Gnathodentex aureolineatus	2739
			Cephalopholis urodeta	613	Lutjanus fulviflamma	1367	Epinephelus fasciatus	945
			Sufflamen chrysopterum	544	Epinephelus quoyanus	1081	Epinephelus quoyanus	749
			Scolopsis bimaculata	544	Gymnocranius grandoculis	748
					Lethrinus miniatus	581

). Reef fish with IRI ≥ 100 are presented in

Table 3. Dominant and common species at four coral reefs in the South China Sea regardless of the season.

Meiji Reef	IRI	Zhubi Reef	IRI	Qilianyu Island	IRI	Chenhang Island	IRI
Epinephelus merra	5190	Parupeneus trifasciatus	971	Gnathodentex aureolineatus	2452	Pentapodus caninus	7479
Pentapodus caninus	1539	Bolbometopon muricatum	509	Lethrinus rubrioperculatus	1565	Cephalopholis spiloparaea	3632
Lutjanus kasmira	1503	Sufflamen chrysopterum	496	Lethrinus semicinctus	847	Gnathodentex aureolineatus	2207
Lutjanus gibbus	801	Cephalopholis urodeta	394	Gymnocranius griseus	717	Parupeneus trifasciatus	1967
Lethrinus atkinsoni	425	Parapercis hexophtalma	372	Aprion virescens	431	Lethrinus rubrioperculatus	627
Parupeneus trifasciatus	382	Scolopsis bimaculata	197	Lethrinus lentjan	371	Epinephelus merra	339
Rastrelliger kanagurta	319	Monotaxis grandoculis	192	Lutjanus fulviflamma	369	Epinephelus fasciatus	280
Lethrinus obsoletus	315	Rhinecanthus aculeatus	184	Epinephelus quoyanus	289	Cheilinus fasciatus	199
Lethrinus rubrioperculatus	305	Cheilinus fasciatus	131	Lethrinus obsoletus	285	Epinephelus quoyanus	186
Scolopsis lineata	218			Gymnocranius grandoculis	216	Cephalopholis urodeta	121
Cheilinus fasciatus	177			Cephalopholis spiloparaea	207	Aprion virescens	115
Scolopsis affinis	156			Lethrinus erythracanthus	206	Variola albimarginata	115
Apogonichthyoides taeniatus	152			Parapercis hexophtalma	190
Cephalopholis spiloparaea	127			Myripristis murdjan	167
Lethrinus microdon	118			Lethrinus miniatus	163
				Lutjanus kasmira	156
				Cephalopholis urodeta	148
				Abalistes stellatus	134
				Myripristis vittata	125
				Dispinus ruber	117
				Priacanthus hamrur	110
				Sufflamen chrysopterum	106

.

At Meiji Reef, four species were dominant, 11 were common, 19 were general and 19 were rare. Catches of dominant, common, general, and rare species accounted for 46%, 36%, 14%, and 4% of the total catch, respectively. There were four dominant species in both May and September, and two species among them were recurrent.

At Zhubi Reef, two species were dominant, seven were common, 23 were general and 18 were rare. Catches of dominant, common, general, and rare species accounted for 58%, 22%, 16%, and 4% of the total catch, respectively. There were seven dominant species in both May and September, and four species among them were recurrent.

At Qilianyu Island, four species were dominant, 18 were common, 34 were general and four were rare. Catch of the dominant, common, general, and rare species accounted for 35.1%, 48.5%, 16.1% and 0.3% of the total catch, respectively. There were eight dominant species in both May and September, and two species among them were recurrent.

At Chenhang Island, five species were dominant, seven were common, 12 were general and 14 were rare. Catches of dominant, common, general, and rare species accounted for 77%, 15%, 6% and 2% of the total catch, respectively. There were four dominant species in May, six dominant species in September, and four species among both were recurrent.

3.3. Diversity Indices

In terms of season (

Table 4. Richness, diversity and evenness indices of fish assemblages in Zhubi Reef, Meiji Reef, Qilianyu Island and Chenhang Island in May and September.

Index	Time	Zhubi Reef	Meiji Reef	Qilianyu Island	Chenhang Island	Mean
D	May	5.041	3.049	5.051	3.780	4.230
	September	4.667	4.553	4.297	1.955	3.868
H′	May	4.182	3.669	4.302	3.057	3.803
	September	4.160	4.148	4.066	2.913	3.822
J′	May	0.861	0.781	0.814	0.596	0.763
	September	0.856	0.774	0.799	0.728	0.789

), the mean D and H′ of the four coral reefs were higher in May than September for the four reefs, but the mean J′ of the four coral reefs was lower in May than September. In terms of spatial distribution (

Table 5. Diversity indices of fish in Meiji Reef, Zhubi Reef, Qilianyu Island and Chenhang Island.

Reef	D	H′	J′	C
Meiji Reef	5.459	4.372	0.763	0.919
Zhubi Reef	7.064	3.255	0.580	0.729
Qilianyu Island	6.840	4.801	0.813	0.946
Chenhang Island	3.915	3.158	0.602	0.795
Mean	5.820	3.897	0.690	0.847

), D values among reefs ranged 3.915–7.064, with that for Zhubi Reef being highest, and that for Chenhang Island lowest; Hʹ values ranged 3.158–4.801, with that for Qiliangyu Island being highest and that for Chenhang Island lowest; Jʹ values ranged 0.580–0.813, with that for Qiliangyu Island being highest and that for Zhubi Reef being lowest; and C values ranged 0.729–0.946, with that for Qiliangyu Island being highest and that for Zhubi Reef being lowest. J_s values ranged 0.172–0.273, with that between Chenhang and Qiliangyu islands being the greatest, and that between Qiliangyu Island and Zhubi Reef being the least (

Table 6. Jaccard similarity indices (J_s) of fish assemblages among Meiji Reef, Zhubi Reef, Qilianyu Island and Chenhang Island.

Reef	Meiji Reef	Zhubi Reef	Qilianyu Island	Chenhang Island
Meiji Reef		0.272	0.228	0.264
Zhubi Reef			0.172	0.177
Qilianyu Island				0.273
Chenhang Island

).

3.4. Community Patterns

The results of the cluster analysis among the different seasons and coral reefs showed that the fish could be divided into three communities: Community I, Community II and Community III (Figure 3). Moreover, differences among seasons and coral reefs were also visualized through non-metric multi-dimensional scaling (NMDS) ordination (Figure 4). The overall stress coefficient was 0.14, and the stress coefficient of sorting result was less than 0.2, which indicated that the graphs had interpretative meaning. The ANOSIM test showed that the difference between communities was significant (R² = 0.729, p < 0.05), indicating that community division was feasible.

The species contribution was calculated by combining the ANOSIM and SIMPER programs, and species contributing more than 10% were identified as being most responsible for the observed spatial or temporal differences. Epinephelus merra (11.41%) was the main contributor at Meiji Reef; four species of Parupeneus trifasciatus (18.94%), Cheilinus fasciatus (18.19%), Parapercis hexophtalma (14.83%) and Cephalopholis urodeta (13.54%) were the main contributors at Zhubi Reef; two species of Lethrinus rubrioperculatus (11.70%) and Gnathodentex aureolineatus (11.55%) were the main contributors at Qilianyu Island; and four species of Pentapodus caninus (11.57%), Cephalopholis spiloparaea (10.67%), Gnathodentex aureolineatus (10.51%) and Parupeneus trifasciatuswere (10.39%) the main contributors at Chenhang Island.

4. Discussion

4.1. Species Composition

Species richness is the most direct and fundamental expression of the degree of species diversity [44]. We reported the total number of fish species on four coral reefs in the SCS to be much lower than that previously recorded for these areas [15,16,45,46], but higher than the number of species in the Weizhou Island Coral Reef Sea [47], possibly because of differences in the timing, duration, frequency, and season of sampling. Qilianyu Island had the most species, and Chenhang Island the least, with the number of species at Meiji and Zhubi reefs being similar.

Differences in the numbers of fish species at different coral reefs may be related to reef size, habitat complexity, reef status, and different levels of anthropogenic disturbance [48,49,50]. Sandin et al. [51] found classic positive relationships between reef fish abundance and habitat area, and others have shown that increased isolation from terrestrial disturbance leads to increased biomass and abundance [52,53,54,55,56]. Meiji Reef is a semi-enclosed lagoon, with water exchange within and outside the lagoon. Zhubi Reef is a closed atoll, which, together with the reef flat barrier, prevents seawater exchange between the lagoon and the sea. Qilianyu Island comprises a series of smaller open islands. Chenhang Island is an atoll, on the same reef as Guangjin Island, with a lagoon within the atoll, with the northern part of the atoll connected to the outer sea. Differences in habitat structure among coral reefs and human activities (including pollution and overfishing) strongly influence the abundance and distribution of fish [57,58,59,60,61]. Our results led us to hypothesize that the habitat complexity and level of anthropogenic disturbance were the main factors affecting the composition of reef-dwelling fish on each coral reef. At the same time, compared with Zhubi Reef and Meiji Reef, there was obvious seasonality in the fish compositions at Qiliangyu Island and Chenhang Island, which also belong to the Xisha Islands. Qiliangyu Island and Chenhang Island are located in the northwestern part of the South China Sea, and are influenced by southwesterly and northeasterly winds and have obvious seasonal climate changes, while the sea temperature changes are jointly influenced by the ENSO and East Asian monsoon (EAM), so the fish composition showed obvious seasonal differences [62,63].

Theory suggests that the community in an undisturbed habitat will often include morphologically distinct species belonging to different phyla, while a heavily disturbed habitat has communities often comprising a few closely related species [64]. The influence of animal bait (i.e., shrimp) may bias sampling toward carnivorous fish and exclude herbivorous or omnivorous species, which can also affect fish composition in biodiversity surveys [65]. The fish species of southeastern Brazil and the Brazilian coast are dominated by the order Perciformes [66,67]. Fish in our study were also dominated by Perciformes, with the proportion on each reef being relatively high. This may be an important feature of coral reef fish community composition in the SCS. However, to determine whether the high proportion of Perciformes on the reef is a natural feature or because the habitat or sampling deviation has been disturbed would need further exploration.

4.2. Dominant Species

There were obvious differences in the dominant species of each island reef in different seasons, especially in Qilianyu Island. Dominant species occupy an important position in the ecosystem, and any changes in them can affect community structure, status and energy flow, and stability [68,69]. We reported that the dominant species on Meiji and Zhubi reefs belonged to the families Serranidae, Lutjanidae, Pentapodidae, Lethrinidae, Mullidae, and Scaridae, which is generally consistent with the types of dominant taxa found in coral reef habitats around the Nansha Islands [70]. However, the number of dominant species on Meiji Reef was lower than reported by [71], possibly because of differences in methods, duration and survey effort; Li et al. [15] did not report representatives of either the Labridae (which play an important role in maintaining reef stability) or Pomacentridae (which indicate living coral cover) on Zhubi Reef. Being a closed atoll, Zhubi Reef is more susceptible to deterioration, decline of coral cover, and habitat destruction.

Dominant species at Qiliangyu and Chenhang islands belonged to the Pentapodidae, Lethrinidae, Mullidae, and Serranidae. Based on a gillnet survey, Sun et al. [72] reported more coral reef fish at the Xisha Islands in families such as the Chaetodontidae, Labridae, Scaridae, and Acanthuridae than we did. Compared with Sun et al. [72], we reported an obvious difference in dominant taxa, especially in the Chaetodontidae, probably because of differences in sampling methods. The smaller number of advantageous categories and higher IRI at Chenhang Island affects the survival of less competitive fish, and reduces the ability of the fish community to cope with external threats, to the detriment of community stability [73,74]. Recent studies have shown that coral reefs in the Xisha Islands have deteriorated, with reduced habitat and food resources for reef fish, and with reef fish densities generally declining because of anthropogenic disturbance [15,75]. The loss of coral cover was very serious in the South China Sea. According to studies, live coral cover declined from an average of approximately 65% to approximately 20% during 1998–2007 in the major offshore atolls of the Nansha Islands and Xisha Islands [76]. In the Xisha Islands, the coral cover in 2016 was approximately 5.44% [77]. The carrying capacity of the Xisha Islands ecosystem may have been reduced because of environmental change and anthropogenic disturbance, and fish assemblages may have responded to habitat change by altering community structure, especially of dominant species.

4.3. Coral Reef Fish Diversity

Species diversity is influenced by a number of factors, and is necessary for maintaining ecosystem stability [78,79,80,81]. Fish diversity varies among regions and years [29]. We reported significant differences in fish diversity among coral reefs in the SCS, which may be a function of spatial differences among reef habitats and their fish assemblages in this region. Values of H′ exceeding 3 usually occur only in healthy ecosystems with high biodiversity [82]. We reported H′ values ranging 3.16–4.80, which was indicative of extremely high coral reef biodiversity in the SCS.

The highest values of species richness, H′, J′ and C, occurred at Qilianyu Island, and D was second only to that at Zhubi Reef, on which basis we inferred that this environment was highly heterogeneous with many ecological niches, resulting in higher species diversity. Species richness and evenness in distribution influence biodiversity indices: the higher the evenness and richness, the higher the biodiversity [37,83]. Additionally, according to “island effect” theory, habitat area is extremely important for determining the diversity of coral reef fish [84]. As an open coral reef comprising a series of islets, Qilianyu Island has more habitat than the other enclosed or semi-enclosed atolls, and can accommodate more species. Fish assemblages at Zhubi Reef had the highest D, but H′, J′ and C were lower. As a closed atoll, Zhubi Reef is vulnerable to environmental change and anthropogenic disturbance [16], its ecosystem is fragile, and it warrants protection. Zhang [71] and Chen et al. [39] reported fish assemblage H′ values at Meiji Reef of 3.58 and 0.92, while we reported a higher value of 4.37. It has been suggested that a proportion of both non-seasonal and seasonal resident fish occurs at Meiji Reef, and H′ values are also associated with a large number of seasonal and incidental species [29,71]. Further investigation is necessary to determine whether the higher H′ values of fish assemblages at Meiji Reef were because of seasonal effects. Fish assemblages at Chenhang Island had low D, H′, J′ and C values, and the lowest species richness among reefs. A recent tendency for some Xisha Island coral reefs to eutrophicate because of anthropogenic disturbance may affect fish diversity [18,85,86]. Differences in H′ values across islands suggested that topographic structure might be one of the main causes of differences in spatial distribution.

The recent concept of “ecological memory” maintains that coral adaptation may be difficult because of the cumulative thermal stresses of long-term warming [61]. Long-term seawater warming has not promoted thermal adaptation of corals in the SCS, and future coral growth will be less suited to warmer conditions [87]. The cumulative thermal stress of long-term warming and regional environmental stresses caused by anthropogenic disturbance (increased terrestrial sedimentation, deteriorating water quality, etc.) may weaken coral thermal adaptation, reduce resistance and resilience, exacerbate declines in growth, and contribute to an overall decline in coral growth in the SCS. This inevitably weakens the functions of coral reef ecosystems that provide food, habitat, and breeding grounds for fish, and reduces the diversity of coral reef fish. Concurrently, changes in evenness appear to drive differences in species richness on different reefs—that is, the effects of anthropogenic disturbance may be reversible over time but may also simultaneously accumulate and significantly alter diversity at various spatial scales [88].

4.4. Community Structure

Fish species composition showed some spatial and temporal heterogeneity, which is very closely related to complex physicochemical factors and seafloor geomorphology [89]. The community structure is directly related to the ecological function of the habitat [90]. Community I was in the Zhubi Reef, which is a typical closed atoll in the northern part of the Nansha Islands in the South China Sea, with no connection to the outer sea except limited water exchange between the lagoon and the outer ocean at high tide [91]. The special geographical, topographical and hydrological environment brought a unique community pattern. Community II was in the Meiji Reef and Chenhang Island. Although they were relatively far apart and differed in fish composition, there were many common dominant species. Although fish aggregation patterns can define the composition of aggregations in different areas, changes are gradual and there are no clear aggregation boundaries, with most fish occurring in two or more communities simultaneously [92]. Community III was at Qilianyu Island, which consists mainly of a series of smaller open reefs and is distinctly different from Zhubi Reef. Among the four coral reefs, the habitat of Qilianyu is relatively more favorable. Therefore, the species richness of coral reef fish in the Qilianyu was also relatively high. In the study, for the four coral reefs, the variation among fish communities by space was greater than that by season, based on cluster analysis. The four coral reefs are all located in tropical waters, and their natural environment, such as water temperature, monsoon, salinity, has relatively small seasonal differences, but the differences among coral reefs in the complexity, depth and area are relatively greater. Therefore, the influence of geomorphology on fish communities was greater than that of season.

5. Conclusions

We reported fish composition, dominant species, biodiversity indices, and assemblage structures at four coral reefs in the Nansha Islands (Meji Reef and Zhubi Reef) and Xisha Islands (Chenhang Island and Qilianyu Island) in the South China Sea based on hand-line survey data in May and September 2018. Of the four reefs, a total of five orders, 21 families, 45 genera and 121 species of fish were recorded with Perciformes (78.5%) being the most diverse. The highest number (5) of dominant species was found near Chenhang Island and the lowest (2) number was detected on Zhubi Reef. The highest abundance index (7.21) occurred at Zhubi Reef, while the Shannon–Wiener diversity (4.80), Pielou’s evenness (0.81), and Simpson’s dominance (0.95) indexes were all highest at Qiliangyu Island. Based on cluster analysis and NMDS, the spatial distribution of fish assemblages among the reefs in different seasons could be divided into three communities. The ANOSIM test showed that the differences in fish composition among different assemblages were significant (R² = 0.729, p < 0.05). The variation in assemblage structures by space was greater than that by season. Our results led us to hypothesize that the habitat complexity and level of anthropogenic disturbance were the main factors affecting the composition of reef-dwelling fish on each coral reef. Topography was likely responsible for most variation in the spatial pattern of fish diversity.