Construction and analysis of a coral reef trophic network for Qilianyu Islands, Xisha Islands

1. Introduction

The complex habitat composed of reef-building corals in the coral reef ecosystem is the main reason for its extremely high level of biodiversity and primary productivity (Knowlton, 2001). The large number of shelters, breeding grounds and nursery grounds in the ecosystem of coral reefs provide suitable habitats for various marine organisms with different living habits (Botha et al., 2013). However, coral reefs are currently under threat and deteriorating worldwide due to numerous influences of anthropogenic activities and climate change (Hoegh-Guldberg et al., 2007), which has prompted multifaceted conservation and restoration efforts (Baums, 2008).

Coral reefs in the South China Sea (SCS) cover an area of approximately 3.72×10⁴ km², with tropical reefs therein accounting for nearly 5% of the total global area of coral reefs (Wang and Guan, 2020). The richness of corals in this region (571 species) is comparable to that in the “Coral Triangle” (Huang et al., 2015). Coral reef ecosystems in the SCS have abundant biological resources, with at least 3 600 fish species (Shao et al., 2008) (more than one-third of these are reef fish) having been reported from them. Coral reefs on the Xisha Islands of the central part of SCS are located more than 400 km from mainland China and are isolated from terrestrial nutrient influence via runoff through rivers and drains. As the main tropical fishing ground in China, there are more than 400 species of oceanic fish and coral reef fish in the coral reefs of the Xisha Islands, it is the important fishing ground for catching commercial fish such as tuna, mackerel, snapper, bonito, flying fish, shark and grouper. Moreover, as part of the Xisha Islands, Qilianyu Islands also as the site of the largest spawning ground for green turtle (Chelonia mydas) in China (Zhang et al., 2020b), more than 40 species of birds occur there, and the surrounding waters are rich in high-quality and valuable seafoods (Yang, 2017). Despite all of this, the numbers and density of coral reef fish in the waters surrounding the Qilianyu Islands have recently trended down because of anthropogenic disturbance and natural environment variation (Li et al., 2017), such as overfishing, destructive fishing, ocean acidification, ocean warming, typhoon damage, and outbreaks of crown-of-thorns starfish (Wu et al., 2011), these factors also chiefly responsible for serious declines in the SCS coral reef fishery resources (Zhang et al., 2020a). Continuing degradation of coral reef ecosystems has generated substantial interest in how management and restoration can support reef resilience (MacNeil et al., 2015). Many measures can be taken to reduce the threats affecting coral reef ecosystems. Confronting large-scale threats requires a major scaling-up of management and restoration efforts based on an improved understanding of the ecological processes that underlie reef resilience (Bellwood et al., 2004). Focusing in ecosystem assessment of coral reef, the lack of knowledge of the structure and function about coral reef ecosystems is one of the main problems for conserving the marine organisms living in coral reefs. Thus, in an effort to assess the health state and sustainable fishery of coral reef, modeling the compositions of coral reef ecosystem and their interaction trajectories should be necessary.

Models are increasingly used to improve our understanding of marine ecosystem functioning and address applied questions in the field of fishery management (Walters and Martell, 2004). Although, it is customary in ecosystem models to mix together what is inherent in physiological growth with counter-effects imposed by interactions with other species and self-limitation, ecosystem-based modeling methods which take into consideration characteristics of both the environment and organisms provide an effective tool to appraise the status of coral reef ecosystems. Model development involves computer software modeling (Bello-Pineda et al., 2006), dynamics simulation (Ippolito et al., 2016; Melbourne-Thomas et al., 2011), hyper-spectral remote sensing and radiation transmission modeling (Petit et al., 2017), mathematics (Jokiel, 2016; Li et al., 2014; Spillman and Alves, 2009), and assessment and monitoring of the various components of the coral reef ecosystem (González-Rivero et al., 2014; Heenan et al., 2017). Such models may improve understanding of the present-day status and dynamics of an ecosystem, can improve scientific theories, and they can guide measures to protect and restore coral reef ecosystems. As there are no standardized criteria for designing food web models in terms of grouping strategy and pedigree, it is frequently difficult to compare different food webs (Pérez-Ruzafa et al., 2020). The reconstruction of trophic pathways plays a fundamental role in understanding the structure and function of ecosystems (Pimm et al., 1991). Ecopath with Ecosim (EwE) provides an overview of an ecosystem’s trophic state, and can represent aquatic food webs (Christensen and Walters, 2004). Ecopath model was first used to simulate a coral reef ecosystem by Polovina (1984), and has since been developed and integrated into the EwE software package; it has now been used extensively in various aquatic ecosystems (Christensen and Pauly, 1992; Heymans et al., 2016). The operating principle of the Ecopath component in EwE is based on thermodynamics. The ecosystem is simplified by constructing a food web, with various ecosystem characteristics quantified through modeling. This can both reflect the characteristics and nutritional relationships of a specific ecosystem in a certain period as the static state of the food web mass balance, and it also serve as a starting point for more dynamic ecosystem models (Odum, 1969; Christensen and Pauly, 1992). Exploring the trajectory of energy flow could facilitate our understanding of ecosystem respondence and resilience to external disturbance. The SCS provides an excellent natural laboratory to reveal the ecology responses and variation tendency of coral reef ecosystem under changeable environmental conditions over a large spatial scale.

Most research on coral reef habitat in the SCS has focused on benthic algae (Chen et al., 2019; Wu et al., 2021), hermatypic coral communities (Li et al., 2018), coral-associated organism (Li et al., 2020; Zhang et al., 2020a), symbiotic microorganism of coral (Chen et al., 2021; Qin et al., 2021), and sediment layers (Yang, 2019). In the previous studies, the main aims were to assess how environmental factors (e.g., seawater environmental parameters, coral reef fish diversity and zooxanthellae density) across a large spatial scale can correlate or control the geographical distribution of coral reef. Because ecosystem-level management of coral reef fish has received limited attention, knowledge of the structure and ecological significance of coral reefs in the SCS is fragmentary. Therefore, given that ecosystem-based modeling methods provide effective tools for assessing the status of coral reef ecosystems, our research aims became to: (1) generate a preliminary trophic model for the Qilianyu Islands coral reef (QICR) to identify the main ecosystem trophic interaction pathways and trophic functioning, and (2) assess the status of this ecosystem and its maturity.

2. Materials and methods

2.1 Study area

Qilianyu Islands, comprising Zhaohu Island, West Shoal, North Island, Center Island, South Island, North Shoal, South Shoal, West-New Shoal and East-New Shoal, are located in the northeastern part of the Xuande Islands (part of the Xisha Islands) in the central SCS (16°55′–17°00′N, 112°12′–112°21′E) (Fig. 1), belong to tropical coral reef regions with a high reef coral diversity and numerous coral species. The islands area is about 1.32 km², the area of surrounding reef flat about 25 km², surrounding water depth is mostly shallower than 20 m, and mean annual surface sea temperature 26.8℃. Because of the tropical monsoon climate, these islands are affected by 8–10 typhoons annually, with precipitation concentrated in summer and autumn (May–November); the tide is irregularly diurnal (Yang, 2017). Currents around the Qilianyu Islands are primarily wind-driven, with their direction changing with that of monsoon winds; currents between islands are influenced by tides (Yang, 2019).

Figure  1.  The study area (a) and the locations of the sampling stations (b). Black triangle markers are the sampling stations; red dotted lines represent the rang of table reefs.

DownLoad: Full-Size Img PowerPoint

2.2 The Ecopath model

We construct an Ecopath model to represent an average annual situation (2018–2019) for the QICR ecosystem with EwE 6.5 software (Christensen et al., 2005). In the Ecopath model, functional groups with similar trophic level (TL), life history and niche characteristics must be comprehensively considered (Pauly et al., 2000), and all functional groups must include basic processes of energy and material flow in the ecosystem (Christensen and Pauly, 1992). Ecopath model comprises two main equations—the first, Eq. (1), defines the balance between the input and output of each functional group to ensure that the ecotrophic efficiency (EE) does not exceed 1; Eq. (2) defines the thermodynamics of the functional groups. The expressions of Eq. (1) and Eq. (2) are

where B_i is the biomass of group i; (P/B)_i is the production of group i per unit biomass, equal to the total mortality rate of group i; EE_i is the ecotrophic efficiency of group i, defined as the fraction of production that is consumed within the system or removed by fishers; Y_i is the total fishery catch rate of group i; (Q/B)_j is the consumption rate of group j per unit biomass; DC_ij is the fraction of prey i in the average diet of predator j; BA_i is the biomass accumulation rate of group i; E_i is the net migration rate (immigration and emigration); Q_i is the consumption of group i; R_i is the respiration of group i; and U_i is the unassimilated food of group i (Christensen et al., 2008).

Recognition of Ecopath functional groups requires consideration of shifts in the habitat of life history stages and/or diet. Thus, functional groups are represented as single biomass pools or multi-stanza functional groups (Walters et al., 2008). In this study, components in the Ecopath model for the QICR ecosystem can be divided into 21 functional groups (Table 1), 7 groups are fish groups; 7 groups are invertebrate groups, and the rest 6 groups are, turtles, zooplankton, phytoplankton, macroalgae, turf, and detritus, which cover the energy flow process of each ecosystem trophic level. Except for the results of field investigation, these functional groups are also selected based on characteristics of ecology, biosystematics, diets of species, and species distributions in the QICR area (Li et al., 2020). The species of fish are assigned to functional groups according to similar ecological characteristics (e.g., diet, predators, body sizes, and metabolic requirements), and the group names of carnivorous fish functional groups are assigned based on size (small, medium and large).

Table  1.  Qilianyu Islands coral reefs Ecopath model functional groups

Group name	Composition of dominant species
Chondrichthyes	ray, skate, shark
Large carnivorous fish	Aprion virescens, Pristipomoides filamentosus, Aphareus rutilans, large grouper, etc.
Medium carnivorous fish	Labridae, Lethrinidae, Priacanthidae, Mullidae, etc.
Small carnivorous fish	Holocentridae, Apogonidae, Cephalopholis, Epinephelus merra, etc.
Omnivorous fish	Pomacentridae, Balistidae, etc.
Coral-eating fish	Scaridae, Chaetodontidae, etc.
Herbivorous fish	Pomacanthidae, Acanthuridae, Siganus, etc.
Turtles	Chelonia mydas, Eretmochelys imbricata, Dermochelys coriacea, etc.
Crown-of-thorns starfish	Acanthaster planci
Giant triton	Charonia tritonis
Other echinoderms	urchin, cucumber, brittle star, starsish
Other mollusca	bivalve, snail, etc.
Crustaceans	crab & shrimp
Coral	Pocillopora damicornis, Pocillopora verrucosa, Acropora humilis, Porites lutea, etc.
Zooplankton	copepoda, planula, juvenile fish, etc.
Small benthic invertebrates	polychaeta, etc.
Macroalgae	coralline algae
Turf	turf
Phytoplankton	Bacillariopyta, Pyrrophyta, Chrysophyta, Cyanophyta, etc.
Detritus	particulate organic carbon & dissolved organic carbon

 | Show Table

DownLoad: CSV

2.3 Key Ecopath model input parameters

2.3.1   Basic input

Biomass data for QICR coral reef fish is based on a 2019 shallow-water fishery-resource survey by the R/V Nanfengin. Fish species richness is based on collections of species made from ground cages, hand fishing and other operations. Biomass of each coral reef fish functional group is estimated from acoustic signals. Acoustic detection used a Simrad EK60 echosounder with 38 kHz and 120 kHz split-beam transducers (Zhang et al., 2016); data were collected from the surface to the seabed. Besides, some model parameters used in this study would also apply to other coral reef Ecopath models with similar ecosystem characteristics. Chondrichthyes biomass was obtained from Huang et al. (2009). Phytoplankton biomass (g/m²) was calculated following Brown et al. (1991) and Shang et al. (2018), based on the SCS chlorophyll a (Chl a) concentration (mg/m³) (Ke et al., 2018). Benthic producer (turf and macroalgae) biomass was estimated from empirical data and Wabnitz et al. (2010). Detrital biomass was calculated following Pauly et al. (1993), as a function of primary production and euphotic depth (Tang et al., 2007).

A lack of biomass data for corals, Acanthaster planci and Charonia tritonis, necessitated using output parameters of similar models; to balance our model EE values for each were set at 0.70, 0.30 and 0.95, respectively (Wabnitz et al., 2010; Cáceres et al., 2016; Ruiz et al., 2016; Du et al., 2020). These values were input into the model to calculate the biomass of missing functional groups. The biomass of other functional groups (other invertebrates) referenced models for adjacent seas or similar ecosystems (Du et al., 2015, 2020; Xie et al., 2019).

Methods used to estimate P/B include assessment of population resources, and estimation using natural mortality (M) and fishing mortality. According to bioenergy theory, the ratio of production (P)/consumption (Q), P/Q, is assumed to range of 0.1–0.3. Under assumptions of mass-balance trophic models, the P/B of “fishes” equals the instantaneous rate of total mortality (Z) (Allen, 1971). The Q/B for each fish functional group was estimated following Palomares and Pauly (1998); P/B and Q/B of some functional groups were estimated based on parameters in Fishbase (https://www.fishbase.org) and other tropical coral reef models from similar latitudes (Chen et al., 2015; Liu et al., 2009) or with similar ecosystem characteristics (Bozec et al., 2004; Cáceres et al., 2016; Gaither et al., 2011; Ruiz et al., 2016).

2.3.2   Catch and diet matrix

The catch of QICR coral reef fish and some invertebrates was estimated from fishery data. Because a diet matrix is important for Ecopath simulation, the diets of different functional groups (Table 2) were reconstructed from stomach content analysis, published literature, and Fishbase (https://www.fishbase.org). A “diet import” approach was needed for some functional groups given the Qilianyu Islands represents an open ecosystem (Christensen et al., 2008). Additionally, biological differences rendered unassimilated consumption (Un. Q) of herbivorous fish, benthic invertebrates, and zooplankton generally higher than that for other animals (Bozec et al., 2004); the Un. Q of crustaceans, zooplankton, small benthic invertebrates, and herbivorous fish was set at 0.25, 0.30, 0.25 and 0.40, respectively.

Table  2.  Estimates of Qilianyu Islands coral reefs ecosystem food intake for trophic groups

No.	Prey (predator)	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16
1	chondrichthyes	0.01	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0
2	large carnivorous fish	0.05	0.008	0	0	0	0	0	0	0	0	0	0	0	0	0	0
3	medium carnivorous fish	0.14	0.1	0.009	0	0	0	0	0	0	0	0	0	0	0	0	0
4	small carnivorous fish	0.12	0.12	0.13	0.03	0.019	0	0	0	0	0	0	0	0	0	0	0
5	omnivorous fish	0.12	0.05	0.06	0.02	0.01	0	0	0	0	0	0	0	0.005	0	0	0
6	coral-eating fish	0.1	0.05	0.04	0.01	0.06	0	0	0	0	0	0	0	0	0	0	0
7	herbivorous fish	0.1	0.19	0.16	0.03	0.05	0	0	0	0	0	0	0	0.005	0	0	0
8	turtles	0.001	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0
9	crown-of-thorns starfish	0	0.001	0.001	0	0.001	0	0	0	0	0.15	0	0	0.001	0	0	0
10	giant triton	1×10−5	0.000 5	0	0	0	0	0	0	0	0	0	0	0	0	0	0
11	other echinoderms	0	0.06	0.06	0.05	0.05	0	0	0	0	0.4	0	0	0.005	0	0	0
12	other mollusca	0.05	0.1	0.12	0.1	0.1	0	0	0.1	0	0.3	0	0.01	0.05	0	0	0
13	crustaceans	0.12	0.12	0.14	0.15	0.12	0	0	0.05	0	0	0	0	0.012	0	0	0
14	coral	0	0	0	0	0.05	0.4	0.1	0	0.8	0	0.05	0.06	0	0	0	0
15	zooplankton	0	0	0	0.28	0.1	0	0.05	0.1	0	0.15	0.15	0.25	0.2	0.1	0.01	0.1
16	small benthic invertebrates	0	0.05	0.18	0.2	0.1	0	0	0	0	0	0.05	0	0.133	0	0	0
17	macroalgae	0	0	0	0	0.13	0.3	0.3	0.3	0.07	0	0.2	0.2	0.1	0	0.05	0
18	turf	0	0	0	0	0.11	0.15	0.3	0.4	0.03	0	0.15	0.16	0.135	0	0.09	0
19	phytoplankton	0	0	0	0.03	0	0	0.1	0	0.1	0	0.12	0.22	0.154	0.3	0.6	0.15
20	detritus	0	0	0	0.1	0.1	0.15	0.15	0	0	0	0.28	0.1	0.2	0.4	0.25	0.75
Import		0.189	0.150 05	0.1	0	0	0	0	0.05	0	0	0	0	0	0.2	0	0
Sum		1.000 01	0.999 55	1	1	1	1	1	1	1	1	1	1	1	1	1	1

 | Show Table

DownLoad: CSV

2.4 Model diagnostics, balance, and verification

A pre-balance (PREBAL) diagnostic was used to check model consistency before balancing or running (Link, 2010). Checked indicators included biomass distribution, biomass ratios, vital rate distribution, vital rate ratios, and total production and removals.

According to thermodynamics, the Ecopath model after a PREBAL diagnostic is adjusted to reach a state of equilibrium. Output parameters requiring adjustment include: EE (<1.0), gross good conversion efficiency (P/Q, 0.1–0.3), net efficiency (NE, NE−P/Q>0), respiration/assimilation biomass (R/AS, <1.0), respiration/biomass (R/B, fish: 1–10; high conversion efficiency population: 50–100), and production/respiration (P/R, <1.0 or >1.0) (Link, 2010; Heymans et al., 2016).

A sensitivity analysis was performed to test how small variations in input data affect model output parameters (Steel et al., 2009). The sensitivity of output parameters to input data depends on degrees of association between functional groups. It is proposed that all basic input parameters change by steps of 10% within ±50%. Sensitivity was calculated following Christensen et al. (2005) as

2.5 Mixed trophic impact (MTI) analysis and keystone analysis

The MTI of the Ecopath model quantifies the direct and indirect trophic effects between functional groups, assesses the positive or negative impact of changes in biomass of one functional group on all other components in the coral reef ecosystem, and analyzes interactions between functional groups and fishing activity (Ulanowicz and Puccia, 1990); it is calculated as follows:

where MTI_ji is the interaction between the impacting functional group j and the impacted functional group i, DC_ji is the fraction of prey j in the diet of predator i, and FC_ij is a host composition term giving the proportion of the predation on i because of j as a predator.

Keystone analyses are based on MTI to identify functional groups in the food web with high overall effects and relatively low biomass. This index is calculated following Libralato et al. (2006),

where $\varepsilon_{i} $ is the total impact of functional group i on the ecosystem by estimating through the MTI, KS_i is the keystone index of functional group i, and p_i is the contribution of functional group i to the total food web biomass.

2.6 Model characteristics

According to the balanced results of the Ecopath model, the overall characteristic parameters of the ecosystem, calculated by network analysis to reflect the size, stability, and maturity of the QICR ecosystem, are used. Total system throughput (TST), which comprises the sums of all consumption, all exports, all respiratory flows, and all flows into detritus, can be used to quantify the ecological scale and “metabolic level” of an ecosystem (Finn, 1976; Ortiz et al., 2015). System descriptive indices such as ratios of total primary production/total respiration (TPP/TR) and total primary production/total biomass (TPP/TB), Finn’s cycling index (FCI), and Finn's mean path length (MPL), characterize ecosystem maturity (Odum, 1969; Finn, 1976; Christensen et al., 2008). The closer TPP/TR is to 1, the more stable is an ecosystem. Cycling of matter is also a critical process in natural ecosystem functioning because it can facilitate homeostatic control of the magnitude of flows (Odum, 1969); FCI and MPL increase with ecosystem maturity. Food web complexity and omnivory are often used to assess ecosystem stability (Landi et al., 2018); the connectance index (CI) is a ratio of the number of actual links to number of possible links for a given food web, and the system omnivory index (SOI) is the average omnivory index of all consumers weighted by the logarithm of each consumer’s food intake (Christensen et al., 2008). Both CI and SOI indicate the complexity of internal ecosystem connections; when their values approach 1, the relationship within the food web formed by functional groups is more likely to be complex, and the ecosystem more stable. Ascendency (A) integrates both size and organization (the number and diversity of interactions between ecosystem components) (Christensen et al., 2008); the relative overhead (O/C) is used as an index of ecosystem resilience (Heymans, 2003).

3. Results

3.1 Quality of Ecopath model

According to PREVAL diagnostics, QICR biomass estimates span 5 orders of magnitude (biomass in an aquatic ecosystem typically covers 5–7 orders of magnitude (Link, 2010)). Additionally, because the log-scale biomass slope falls by −0.117 across all trophic levels, QICR ecosystem primary productivity and functional group biomass of low TLs are high (a general range is −0.05 to −0.1 (Link, 2010)). P/B, Q/B and R/B values for each functional group decrease with increased TL, and biomass and production values for each functional group of consumers also exceed those of primary producers. Total consumption removals for each consumer functional group are lower than production values, and all have total consumption removals below the sum of human removals (e.g., catch), except for turtles and carnivorous fish (medium and large).

Sensitivity analysis reveals relationships between EE and biomass for each functional group to be most sensitive, with a 50% increase in biomass of a functional group resulting in an average decrease of about 30.11% in its estimated EE value. A linear relationship is also apparent between input parameters (i.e., P/B, Q/B, EE) of each functional group. Compared with other sensitivity relationships, the EE of phytoplankton is most sensitive when its biomass decreases, while the EE of turf has the highest relative increase with an increase in its biomass.

3.2 Trophic structure and flow

QICR Ecopath model functional group input and output parameters are listed in Table 3. The TL for each functional group is consistent with the fundamental laws of ecology, ranging 1.00–3.80 (Fig. 2), with half concentrated in TLs II or III. Among consumers, chondrichthyes (3.80) and large carnivorous fish (3.57) occupy the top two food web positions. The mean TL for coral reef fish is 2.92, with the interval between functional groups 2.16–3.57. For invertebrates, the giant triton (3.34), crown of thorns starfish (2.90), and coral (2.13) have the highest TL values; except for primary producers and detritus, the TL of zooplankton is the lowest. The EE of functional groups ranges 0.187–0.981, with the highest value for omnivorous fish and the lowest for detritus. The range in GE for each functional group is 0.053–0.446, with “other mollusca” (0.446) and zooplankton (0.313) having the highest values.

Table  3.  Basic Ecopath model functional group input and estimated output parameters

No.	Group name	TL	B/(t·km−2·a−1)	P:B/(a−1)	Q:B/(a−1)	Un. Q	Catch	EE	P/Q	NE	OI
1	chondrichthyes	3.80	0.097	0.25	4.72	0.20	−	0.189	0.05	0.07	0.41
2	large carnivorous fish	3.57	0.430	0.79	6.64	0.20	0.200	0.723	0.12	0.15	0.30
3	medium carnivorous fish	3.42	3.781	1.00	8.50	0.20	2.500	0.830	0.12	0.15	0.16
4	small carnivorous fish	3.07	1.700	4.00	13.50	0.20	0.800	0.942	0.30	0.37	0.22
5	omnivorous fish	2.84	1.103	4.50	16.30	0.20	1.500	0.981	0.28	0.35	0.40
6	coral-eating fish	2.45	1.415	2.40	20.00	0.20	0.200	0.878	0.12	0.15	0.30
7	herbivorous fish	2.16	8.156	3.00	28.00	0.40	1.800	0.397	0.11	0.18	0.15
8	turtles	2.32	0.020	0.14	3.50	0.20	0.002	0.878	0.04	0.05	0.29
9	crown-of-thorns starfish	2.90	0.485	1.20	5.00	0.20	−	0.300	0.24	0.30	0.20
10	giant triton	3.34	0.002	1.22	4.08	0.20	0.001	0.950	0.30	0.37	0.07
11	other echinoderms	2.26	3.035	2.20	7.80	0.20	0.500	0.786	0.28	0.35	0.21
12	other mollusca	2.33	8.700	2.50	5.60	0.20	6.000	0.955	0.45	0.56	0.24
13	crustaceans	2.46	4.300	3.20	28.00	0.25	0.500	0.904	0.11	0.15	0.31
14	coral	2.13	19.574	3.00	10.00	0.20	−	0.700	0.30	0.38	0.13
15	zooplankton	2.01	3.510	76.00	242.50	0.30	−	0.416	0.31	0.45	0.01
16	small benthic invertebrates	2.10	3.911	12.00	60.00	0.25	−	0.629	0.20	0.27	0.09
17	macroalgae	1.00	22.000	18.00	−	−	−	0.375	−	−	−
18	turf	1.00	30.000	25.00	−	−	−	0.239	−	−	−
19	phytoplankton	1.00	8.817	231.00	−	−	−	0.324	−	−	−
20	detritus	1.00	315.000	−	−	−	−	0.187	−	−	0.24
Note: B: biomass; OI: omnivory index; TL: trophic level; P/B: production/biomass; Q/B: consumption/biomass; Un. Q: unassim. consumption; EE: ecotrophic efficiency; P/Q: production/consumption; NE: net efficiency. Values in bold are estimated by the present model. − reprsents no data.

 | Show Table

DownLoad: CSV

Figure  2.  Qilianyu Islands coral reefs ecosystem flow diagram. Circles are proportional to the functional group biomass in the system.

DownLoad: Full-Size Img PowerPoint

Network analysis confirmed compartmental throughput of the 20 functional groups in a Lindeman spine with 5 integrated TL (Fig. 3). The “all flow” of each TL decreases substantially with increased TL (Table 4). Flows within TLs I and II generate 77.24% and 19.93% of throughput, respectively, consistent with principles of an energy pyramid. As the major source of energy, the flow of consumption by TL I predators accounts for 87.40% of the sum for consumption by predators, and the flow to detritus accounts for 75.27% of the sum for flow to detritus, demonstrating fairly inefficient energy utilization of TL I; most energy is stored as detritus in low TLs, which cannot be transferred upward.

Figure  3.  Lindeman spine representing of trophic flows from Qilianyu Islands coral reefs Ecopath model. P: primary producers; D: detritus; II–V: trophic levels; TST: total system throughput; TE: transfer efficiencies; TL: trophic level.

DownLoad: Full-Size Img PowerPoint

Table  4.  Distribution of Qilianyu Islands coral reefs ecosystem flow

Trophic level
	Consumption by predators	Export	Flow to detritus	Respiration	Throughput
V	0.070	0.125	0.406	1.196	1.798
IV	1.650	1.220	6.121	15.040	24.030
III	23.730	5.445	56.840	110.500	196.500
II	194.700	7.207	658.500	713.600	1 574.000
I	1 533.000	2 371.000	2 195.000	0.000	6 099.000
Sum	1 754.000	2 385.000	2 916.000	840.400	7 896.000

 | Show Table

DownLoad: CSV

The transfer efficiency of a TL is the ratio of output and ingested energy to its throughput. Transfer efficiency indicates the energy efficiency of a TL in an ecosystem. The geometric mean of the transfer efficiency (mTE) for the food web from TLs II–IV is 13.15% (above the Lindeman trophic transfer efficiency (10%)) (Lindeman, 1942), with 13.15% from primary producers and 13.14% from detritus (Table 5). The proportion of total flow originating from detritus (45%) and primary producers (55%) indicates that the QICR ecosystem is dominated by a grazing food chain.

Table  5.  Transfer efficiency of Qilianyu Islands coral reefs ecosystem trophic levels

Source	Trophic level
	II	III	IV	V
Producer	12.37	15.53	11.84	11.1
Detritus	13.64	13.73	12.13	10.43
All flows	12.83	14.84	11.94	10.87
Note: Proportion of total flow originating from detritus: 0.45; transfer efficiencies (mTE, calculated as geometric mean for TL II–IV); from primary producers: 13.15%; from detritus: 13.14%; total: 13.15%.

 | Show Table

DownLoad: CSV

3.3 Niche overlap, MTI, and keystone analysis

The niche overlap index mainly describes niche partitioning between functional groups; it is based on similarities in, for example, predation, competition, and environmental adaptation. Corals and zooplankton have the lowest degree of predatory overlap of all functional groups, but a high degree of prey overlap; herbivorous and small carnivorous fish have similar predators, but their prey is quite different; crustaceans and other echinoderms have high prey and predator overlaps, indicating their very similar niches, especially when competing as predators (Fig. 4).

Figure  4.  Qilianyu Islands coral reefs Ecopath model plot of functional group niche overlap. Point colors represent geometric mean of “prey overlap index” and “predator overlap index” (color scale to right); functional groups: 4, 7 small carnivorous and herbivorous fish; 14, 15 coral and zooplankton; 11, 13 other echinoderms and crustaceans.

DownLoad: Full-Size Img PowerPoint

MTI analysis (Fig. 5) reveals primary producers have a positive effect on most functional groups, benthic algae (turf and macroalgae) have a negative effect on coral, and coral has an obvious positive effect on coral-eating fish and crown of thorns starfish. The negative effects of fishing activity and the functional group “crustaceans” on the giant triton are the result of direct and indirect impacts, respectively; the giant triton also has a significant negative effect on the crown of thorns starfish, which it preys upon. Fishing activity has an obvious negative effect on medium and large carnivorous fish, while an increase in fishing intensity is associated with a positive impact on the biomass of omnivorous fishes, small carnivorous fishes, and crown of thorns starfish. This indirectly suggests that fishing activities in this area affect the coral reef ecosystem.

Figure  5.  Qilianyu Islands coral reefs model mixed trophic impact analysis. Positive (blue) and negative (red) values of mixed trophic impact index represent positive and negative effects, respectively. LCF, large carnivorous fish; MCF, medium carnivorous fish; SCF, small carnivorous fish; OF, omnivorous fish; CEF, coral-eating fish; HF, herbivorous fish; CTS, crown of thorns starfish; OE, other echinoderms; OM, other mollusca; SBI, small benthic invertebrates.

DownLoad: Full-Size Img PowerPoint

Based on the keystone index (KS) and relative total impact (RTI), it is determined that the key functional groups in the QICR ecosystem include corals (KS=−0.197, RTI=1.00), crustaceans (KS=−0.144, RTI=0.981), and herbivorous fish (KS=−0.159, RTI=0.981) (Fig. 6). The high KS and RTI of these functional groups imply that they play important roles in the QICR ecosystem.

Figure  6.  Keystone index for Qilianyu Islands coral reefs model functional groups. For each functional group, the keystone index (y-axis) is reported against their relative total impact on the trophic web (x-axis). Overall effects are relative to the maximum effect measured; the x-axis scale is between 0.0 and 1.0. The functional groups are ordered by decreasing keystone index; therefore, the key functional groups are those ranking among the first groups. Circles are proportional to the functional group biomass in the system.

DownLoad: Full-Size Img PowerPoint

3.4 Summary statistics

QICR ecosystem characteristics are presented in Table 6. TST was 7 952 t/(km²·a), of which 22.76% is due to consumption and 10.57% to respiration; 29.99% and 36.68% of this originates in backflows to exports and detritus, respectively. TPP/TR is 3.79, TPP/TB 26.30, FCI 3.64%, and MPL 2.47. The mean trophic level of catches (TLc) is relatively low (2.62). The total overhead for this ecosystem is 22 977 t/(km²·a), with a total ascendency (A) accounting for 27.71% of total capacity (C).

Table  6.  Comparison of Qilianyu Islands coral reefs ecosystem characteristics with other coral reefs

Area	Qilianyu Islands	Hawaii Island	Cocos Island	Darwin &Wolf Island	Nanwan Bay	Uvea Atoll	Caribbean Sea
							Cayos Cochinos	Media Luna
Group number	20	26	31	32	18	25	22	21
Sum of all consumption/(t·km−2·a−1)	1 810	5 332	22 978	8 880	8 373	292	31 013	27 381
Sum of all exports/(t·km−2·a−1)	2 385	520	20	344	16 200	185	81 700	9 779
Sum of all respiratory flows/(t·km−2·a−1)	840	3 477	12 050	5 278	4 629	86	17 096	16 264
Sum of all flows into detritus/(t·km−2·a−1)	2 916	1 700	6 136	2 151	20 115	346	90	17 881
TST/(t·km−2·a−1)	7 952	11 030	41 184	16 652	49 317	909	220 232	71 305
Sum of all production/(t·km−2·a−1)	3 642	−	978	5 235	21 553	325	10 6510	31 684
TLc	2.62	2.59	3.55	3.02	2.40	3.50	3.64	2.97
GE	0.00 4	0.000 09	2.00×10−6	0.001	0.000 36	0.000 15	1.01×10−9	1.20×10−5
Net p.p./(t·km−2·a−1)	3 182.73	3 895.09	4 583.59	3 408.00	20 199.00	265.30	98 796.00	26 043.00
TPP/TR	3.79	1.12	0.38	0.65	4.40	3.10	5.78	1.60
NSP/(t·km−2·a−1)	2 342.30	417.78	−7 466.37	−1 898.00	15 570.00	179.40	81 700.00	9 779.00
TPP/TB	26.30	5.57	2.32	3.64	9.90	19.95	11.82	10.21
TB:TST/(a−1)	0.015	0.06	0.05	0.06	0.04	0.01	0.04	0.04
Total catch/(t·km−2·a−1)	14.003	0.350	0.010	3.841	7.300	0.039	1.00×10−4	0.311
CI	0.33	−	0.17	0.15	−	−	0.30	0.26
SOI	0.21	−	0.40	0.32	−	−	0.21	0.20
FCI/%	3.64	6.13	6.50	4.75	3.50	−	1.60	6.95
MPL	2.47	−	−	−	−	−	6.65	5.65
A:C/%	27.71	31.50	24.80	29.80	−	−	47.00	31.00
mTE/%	13.15	−	12.20	10.70	7.80	−	−	−
Note: GE: Gross efficiency (total catch/net p.p.); net p.p.: calculated total net primary production; NSP: net system production; TST: total system throughput; TLc: mean trophic level of the catch; TPP/TR: total primary production/total respiration; NSP: net system production; TPP/TB: total primary production/total biomass; TB/TST: total biomass/total throughput; CI: connectance index; SOI: system omnivory index; FCI: Finn's cycling index; MPL: Finn's mean path length; A/C: ascendency/capacity; mTE: mean transfer efficiencies. −represent no data.

 | Show Table

DownLoad: CSV

To evaluate the status of the QICR ecosystem, we refer to coral reef Ecopath models for the Atlantic and Pacific. The QICR ecosystem has a relatively reduced total biomass, but its gross efficiency and total catch are higher than reported for other coral reef ecosystems. The mean trophic level of QICR catch (2.62) is significantly lower than values for coral reef ecosystems in the Caribbean Sea (Cayos Cochinos Islands and Media Luna Archipelago) (Cáceres et al., 2016), Cocos Island (Gaither et al., 2011), Galapagos Islands (Darwin and Wolf Islands) (Ruiz et al., 2016), and Uvea Atoll (Bozec et al., 2004), but similar to Hawaii Island (2.59) (Wabnitz et al., 2010) and Nanwan Bay, Taiwan (2.40) (Liu et al., 2009). The proportion of TST components in the Ecopath model of QICR ecosystem differs in some regards from those in other Ecopath models. The sum of flows into detritus represents the highest proportion of TST (36.68%) in the QICR, similar to Ecopath models for coral reef ecosystems in the Nanwan Bay and Uvea Atoll. The QICR TPP/TR was also closest to a value for Uvea Atoll, however, TPP/TB and total biomass/total system throughput (TB/TST) are significantly higher or lower than values for most other coral reef ecosystem models. Comparison of model results indicates that the QICR CI and SOI are closest to those of the Caribbean Sea. The A/C of this study was in the range of 24.8%–31.0% reported for other coral reef ecosystem models, excepting the Cayos Cochinos.

4. Discussion

4.1 Abundance of detritus and herbivorous fish in QICR ecosystem

The QICR ecosystem was supported primarily by input of flows and materials from detritus and primary producers (phytoplankton, turf and macroalgae) (Table 4), with the four functional groups at TL I all having lower EE values (Table 3). Although only a small flow of detritus and primary producers enter to other TL, but a substantial proportion flow of them was exported from the ecosystem or was returned to the detritus pool. Coral reef ecosystems are among the most biodiverse ecosystems in the world and have great ecological importance and economic value (McCook et al., 2009). The oligotrophic nature and high biological productivity of seawater in coral reef habitats are notable features of coral reef ecosystems (Adey and Goertemiller, 1987), and their biological productivity is 50–100 times that of the surrounding tropical oceans. As the second largest producer in coral reef ecosystems, detritus is an essential feature of ecosystem development (Arias-González et al., 1997). Natural sources of marine detritus include animals, phytoplankton, bacteria and blue green algae, periphyton, submerged aquatic vegetation, intertidal macrophytes, river borne detritus, beach and shore material, terrestrial detritus and atmospheric deposition (Darnell, 1967). Accumulated detritus is buried in sediment and is further decomposed or mineralized by micro-benthic heterotrophs (Nixon et al., 1986; Chen and Qiu, 2010). Sediment is also widely considered to influence coral reef ecosystems (Bahr et al., 2020), significantly affecting turf productivity (Tebbett and Bellwood, 2020), and regulating the feeding of fish and spatial scales of ecosystem function delivery (Tebbett et al., 2020). In the other way, bacteria can either be attached to particulate detritus, or aggregated interstitially within the spaces of sediment and particulate matter (Mason and Varnell, 1996). Furthermore, like many coral reef fisheries, the catch in the reef fish of QICR is dominated by herbivorous fishes (Tebbett and Bellwood, 2020) (Table 3) , which provide important grazing functions for coral reef ecosystems and improve the ability of coral reefs to self-repair under climate change-related disturbances, they are generally considered to be one of the key contributors to maintaining coral reef ecosystem health (Bellwood et al., 2004; Bozec et al., 2016; Hughes et al., 2007). In addition, the higher biomass and catch of herbivorous fish in our model suggest that the QICR ecosystem may be due to coral reef degradation (such as coral–algal phase shift) and overfishing, which leading to a decrease in the biomass of traditional top predators in habitats, while the proportion of herbivorous fish biomass increased (Ainsworth and Mumby, 2015; Arias-González et al., 2004; Darling and D’Agata, 2017).

4.2 The QICR ecosystem dominated by epifauna

The niche overlap analysis provides evidence that the functional group of crustaceans and other echinoderms both have a high similarity for their prey and predator (Fig 4). They are the main components of benthic invertebrate communities in coral reef, both have the highest degree of similarity in the niche in the QICR ecosystem. Different with reef fish, benthic invertebrates have relatively single food sources due to the limitation of habitat topography and their own mobility, the benthic food web structure of coral reef formed by them is the main way for detritus to enter the energy flow of ecosystem. In general, niche overlap results in functional redundancy (Mouillot et al., 2013) and intensifies competition among functional groups with similar niches. However, in a coral reef ecosystem with high biodiversity and bottom-up control (Frank et al., 2007), we suggest that the combination of niche overlap between crustaceans and other echinoderms improves resistance and repairability of QICR ecosystem.

A keystone functional group is a species or species assemblage that has a controlling effect in a key position of food web. There is little doubt that corals represent a keystone group in the QICR ecosystem. Historically, coral cover in the Xisha Islands exceeded 50%. However, because of coral disease, ocean warming and acidification, overfishing, and crown of thorns starfish outbreaks, coral cover has dropped to about 15% (Li et al., 2019b). Present-day coral cover in QICR has recovered slowly since being substantially reduced in 2007 (Li et al., 2018). Combined with results of keystone analysis, corals continue to dominate the QICR ecosystem trophic structure, despite the reef habitat being seriously degraded. This indicates that corals have provided functional complementarity (Brandl et al., 2016) for the coral reef ecosystem. Due to the nature of Ecopath model and the limitation of dataset, microbiomes that symbiosis with corals are not set as functional groups in this study, such as endosymbiotic Symbiodiniaceae (formerly named zooxanthellae), bacteria, viruses, fungi, and archaea. In the construction of Ecopath model, therefore, we empirically integrated these microbiomes within the coral functional group, treating the two as a whole. As the critical parts of the coral reef ecosystem, the symbiotic microbiomes of coral can transfer their photosynthetically derived nutrients for daily metabolism of corals. But corals can also be heterotrophic, as they feed on a variety of sources, including suspended particulate matter (Mills et al., 2004), dissolved organic compounds (Godinot et al., 2011), phytoplankton (Ferrier-Pagès et al., 2011), and zooplankton (Houlbrèque et al., 2003). During the growth of corals, the heterotrophic feeding of corals provides adequate phosphorus, nitrogen and other micronutrients that are deficient in photosynthetic products (Xu et al., 2021). This is one of the main reasons why we consider corals as consumer rather than primary producer. Moreover, according to the output results of QICR model, the TL for functional group of coral (TL=2.13) may not be able to completely reflect the characteristics that corals possess the capability for mixotrophy (autotrophic and heterotrophic), but at least we can better understand the rough trophic status of corals between primary producers and primary consumers in ecosystem from the energy flow diagram. Different carbon and nitrogen sources have distinct stable isotopic compositions (Wu et al., 2021), stable isotope analysis has emerged as one of most powerful tools for tracing organic matter in food webs (Thomas Larsen) and indicating organism diet and trophic position for organisms. In the future study, using stable isotope analysis to reveal the food source of corals and the role of symbiotic microbiomes (Luo et al., 2019), and accurately reflect the trophic level of corals in the ecosystem, is one of the important tasks to improve the construction of coral reef ecosystem model.

4.3 Fishing pressure is the main reason that cause the degradation of QICR ecosystem

MTI analysis reveals coral has a significant effect on coral-eating fish and the coral-eating crown of thorns starfish (Li et al., 2019a) (Fig. 5), while the giant triton has a significant negative effect on the crown of thorns starfish through predation. Early reports of large-scale outbreaks of this starfish were made on the Great Barrier Reef, Australia (Chesher, 1969). Waters in and surrounding the QICR and throughout the SCS have been severely damaged by outbreaks of this starfish (Huang et al., 2012; Reimer et al., 2019). Because of this, many coral reefs may collapse through bioerosion and currents, which possibly influences their ecological function by reducing reef structural complexity (Fabricius et al., 2010). The giant triton, one of few natural predators for crown of thorns starfish, occurs mainly throughout Indo-Pacific tropical coral reef waters (Zhang et al., 2013). The market value of these tritons has resulted in the overexploitation of wild populations, potentially leading to local extinction (Russo et al., 1990). This has affected crown of thorn starfish populations (Zhang et al., 2013). Li et al. (2019b) reported large-scale outbreaks of this starfish on other coral reefs around the Qilianyu Islands from 2018, despite the existence of potential barriers to its dispersal such as strong currents and depths between islands (Muallil et al., 2020). Because the density of crown of thorns starfish is inversely proportional to the number of living corals (Wilmes et al., 2020), the Ecopath model described herein may underestimate crown of thorns starfish biomass and overestimate that of the giant triton. MTI analysis also demonstrated fishing activity to have variable negative effects on corals and the giant triton, but a positive effect on the crown of thorns starfish. Consequently, fisheries can both directly and indirectly affect the function and status of corals. Overfishing degrades biological resources and modifies coral reef food webs (e.g., the giant triton plays an important role in maintaining a complex balance in the coral reef ecosystem (Glynn and Enochs, 2011)). A reduction in coral biomass leads to algal overgrowth (Hodgson, 1999), which reduces reef structure complexity and potentially affects the feeding of coral reef fish (McCormick et al., 2017), and the coral–algal phase shift would reduce biodiversity and ecosystem maturity (Ainsworth and Mumby, 2015). Although, there is no doubt that the potential importance of individual herbivorous species in removing macroalgae from coral reefs (Mantyka and Bellwood, 2007), the process of macroalgal removal would be strongly influenced by coral reef conditions (Chong-Seng et al., 2014). Cheal et al. (2010) suggested that coral reefs could lose resilience even under relatively low fishing pressure on herbivorous fishes. Nevertheless, coral communities in the SCS have dramatically declined over the past several decades (Yu, 2012). The coral cover in the central SCS has declined from over 70% in the 1980s (Hughes et al., 2007) to 16% in 2015 (Chen et al., 2019). Likewise, investigation data show that the coral cover in the ecological monitoring area of the Xisha Islands decreased from about 70% to less than 15% from 2001 to 2019 (Li et al., 2019b), as the consequence of frequent anthropogenic activities (overfishing and destructive fishing) and recurrent natural events (ocean warming, typhoon damage, and outbreaks of crown-of-thorns starfish). Consequently, it is necessary to take some measures for identification and protection of ecosystem components that are critical for the prevention of coral–algal phase shift in QICR ecosystem.

4.4 Compared with other ecosystems, QICR ecosystem is being strongly influenced by the external impacts

To evaluate the status of the QICR ecosystem, results from it are compared with Ecopath models for coral reefs elsewhere throughout the Atlantic and Pacific (Table 6). Features of TST components for these models vary. The sum of flows into detritus for the QICR ecosystem represented the highest proportion of TST (36.68%), similar to models for coral reefs in the Nanwan Bay and Uvea Atoll. The estimated QICR TPP/TR (2.981) is closest to a value for Uvea Atoll, which indicates that increased organic matter caused the TPP to exceed the TR; the QICR TPP/TB ratio (26.1) is significantly higher than values for most other coral reef ecosystems, which also indicates that the QICR ecosystem is tending to develop towards resource accumulation. Both TPP/TR and TPP/TB values suggest that the QICR ecosystem is immature, or in a state of “poor stability”. Low CI and SOI values imply that redundancy in and the resistance of the food web are weak; both values are closest to those for the Caribbean Sea. FCI (4.80) and MPL (2.58) values indicate that the proportion of productivity devoted to material cycle is low. Compared with other models, the QICR has the highest gross efficiency, which indicates that the production efficiency of fishery resources is higher. Collectively, the QICR ecosystem maturity level is generally lower than for most other coral reef ecosystems, being most similar to the Nanwan Bay. The QICR is situated in an open water area, and of all other reefs for which Ecopath models are available, is geographically closest to the Nanwan Bay. Therefore, QICR and the Nanwan Bay coral reef ecosystems may be similar because of comparable anthropogenic disturbance (mainly intensive fishing), but potentially also because the coral reef ecosystems in and around the SCS might have relatively similar ecological characteristics.

Furthermore, in the Ecopath model of coral reef ecosystems involved in the comparative analysis, A/C in other models ranges 24.8%–31.0% of output results, except for the Cayos Cochinos model; the A/C of the QICR Ecopath model is 27.71%, which indicates that this ecosystem has considerable development potential, and spaces for buffer interference. However, about the source of overhead, the internal flow and import accounted for 59.07% and 0.39% of total capacity, respectively. This indicates that the ecosystem has strong internal buffer spaces to effectively buffer it from external interference, but the buffer capacity for external importation is weak. The QICR ecosystem is likely vulnerable to external importation. Eutrophication is a main driver of algal distribution on coral reefs (Huang et al., 2020). Seawater around the Qilianyu Islands is affected by complex and strong hydrodynamics and convergences can easily form (Yang, 2019). Therefore, in the ecosystem of QICR, if a large amount of nutrients input exceeds the threshold, it may cause large fluctuations in the structure and function of the ecosystem. Although some researches highlight anthropogenic nutrient deposition as an important nutrient source for coral reef (Chen et al., 2019; Kim et al., 2014), there is increasing evidence that anthropogenic activities have bring additional material inputs to coral reefs (Zhang et al., 2020a). Anyway, those are undoubtedly direct evidence that climate change and anthropogenic activities are affecting the habitat of coral reef and should be a cause for concern.

Compared with other coral reef models, although the total biomass of the QICR ecosystem is relatively reduced, its total efficiency and total catch are higher than those reported by other models, indicating that the fishery resource exploitation efficiency in the study area is higher. In general, differences in ecosystem type influence ecological characteristics (Heymans et al., 2011), however, TLc can indicate major differences in fishery protection between ecosystems (Albouy et al., 2010). The TLc of the QICR model (2.62) is significantly lower than that for coral reefs in the Caribbean Sea (Cayos Cochinos Islands and Media Luna Archipelago), Galapagos Islands (Darwin and Wolf Islands), and Uvea Atoll, while those for Hawaii Island (2.59) and Nanwan Bay (2.40) are relatively close. According to the results of Nanwan Bay model, Liu et al. (2009) suggests an overfished status would lead to the mean trophic level of the catch, matter cycling, and trophic transfer efficiency are extremely reduced. In addition, given that coral reefs are often overfished (Mumby et al., 2006), climate change induced degradation will limit the ability of coral reefs to exert its ecosystem functions (Dove et al., 2020). The reasons that TLc of the QICR model is also lower than values from the northern SCS (2.93) (Chen and Qiu, 2010) are possibly that the combined effects of climate change (Alva-Basurto and Arias-González, 2014) and fishing activities (Pauly et al., 1998) have down the food web. On the one hand, climate change is a big one that one of the most important threats facing coral reefs on a global scale (Lough, 2011), ocean acidification and ocean warming are two climate-related impacts to coral reefs. If large amounts of anthropogenic CO₂ from the atmosphere are dissolved in seawater that would causes the lowering of the ocean’s pH (ocean acidification) (Hoegh-Guldberg et al., 2007). And ocean acidification reduces the calcium carbonate in the seawater, making it difficult for corals to form skeletons (Anthony et al., 2008). The results of Eyre et al. (2018) showed that under the action of ocean acidification, it is expected that by circa 2050 CE, reef sediments globally will transition from net precipitation to net dissolution. In addition, ocean warming water caused by global warming would prompt corals to release the zooxanthellae, and then make corals experience different degrees of bleaching (Gates, 1990), all of which would eventually lead to large-scale death of corals and the degradation of coral reef ecosystem (Cook et al., 1990). According to the Status of Coral Reefs of the World: 2020 report produced by the Global Coral Reef Monitoring Network, the total amount of corals in the world’s coral reefs gradually declined by 14% since 2009, which is more than all the coral currently living on Australia’s coral reefs. There are evidences that sea surface temperatures in the SCS increased in the last several decades since the middle of the twentieth century (Jiao et al., 2015). On the other hand, “Fishing down the food web” refers to the decreasing of the catch size due to the depletion of previous catch size (Pauly et al., 1998; Pauly and Palomares, 2005), for example, excessive artisanal fishing pressure may be the main hindrance to develop ecosystem quality in the shelf slope area (Karim et al., 2019). Except for overfishing, destructive fishing such as cyanide fishing and blast fishing are often considered one of the most destructive human activities on coral reef ecosystems (Fox and Caldwell, 2006; Mak et al., 2005). Of the more than 1 000 species of coral reef fish assessed by the International Union for Conservation of Nature, 8% of them are threatened with extinction (Hixon and Randall, 2019). Consequently, anthropogenic activities can indirectly and/or directly act to homogenize and simplify ecosystems, by the climate change and fishing activities that aforementioned discussions, artificially favoring stress-tolerant species and forcing the ecosystem into an earlier successional state (Williams et al., 2015).

5. Conclusions

EwE is used to construct an Ecopath model for the QICR ecosystem. This represents the first study to construct an Ecopath model for a coral reef ecosystem in the SCS and proffers the first comparative analysis of ecosystem structure and function in the QICR and other coral reefs. The QICR ecosystem has a high energy transfer efficiency (mTE=13.15%) compared with other models, but it is also degrading and has low ecological stability. Direct and indirect effects of climate change and anthropogenic disturbance (e.g., fishing) are the main reasons for QICR degradation. Policy should incorporate an ecosystem-based approach to fisheries management for coral reef fish resources and for protection and reparation of reefs to improve their resilience and resistance, to realize sustainable development of coral reef fishery resources. Although this study has conducted a comprehensive assessment and analysis of the ecological characteristics of the QICR ecosystem, there are still some deficiencies. As a static model, Ecopath model does not consider the impact on the model caused by changes in environmental factors, such as the temporal and spatial factors in seasonal changes, therefore, the results predicted by the model must be used with caution in combination with the field survey results. In addition, the ecological status simulated based on Ecopath model mainly emphasizes the theoretical boundary, while lack the consideration about the impact of the growth and development of organisms at various trophic levels on the ecosystem structure, so it is difficult to scientifically reflect the actual conditions of the ecosystem. The study here can provide a background for the follow-up research on the dynamic changes of the QICR ecosystem. Long-term monitoring research and precise quantitative techniques are needed in order to provide a more comprehensive scientific theoretical basis for the development and management strategies of coral reef fisheries.