Introduction

The crown-of-thorns starfish (CoTS), Acanthaster spp. is widely known for its predation on Indo-West Pacific corals, especially the reef-building scleractinian corals1,2. Typically, this starfish has negligible impact at very low densities (< 1 starfish per hectare), but its population can suddenly increase (> 1000 starfish per hectare), causing outbreaks around the coral reefs3,4. Given that an individual adult CoTS can consume up to 12-m2 of coral tissue per year, they can kill up to 90% of corals during outbreaks and cause destructive structural changes to the ecosystem5,6.

In recent years, based on the theoretical modelling, the frequency of CoTS outbreaks has been considered to be likely to increase from once every 50–80 years to every ~ 15 years7,8. Furthermore, with the increased resolution of these new sampling and monitoring methods, the recent detection of renewed CoTS outbreaks have occurred earlier than expected9. Uthicke et al. used eDNA monitoring technology to detect a new outbreak wave of corallivorous seastar (Acanthaster cf. solaris) on Lizard Island, Great Barrier Reef, less than 10 years after the last outbreak10. There is still considerable debate about the causes of CoTS outbreaks. For example, the aggregation behaviour of CoTS leads to population outbreaks11, while nutrient enrichment creates favourable conditions for CoTS growth7,12,13,14,15. In addition, disturbance and pollution of coral reefs have led to declines in natural predators16,17,18. Although primary outbreaks are typically independent events, subsequent secondary outbreaks are more likely to be caused by a substantial supply of larvae12,13. As the most productive invertebrate, an adult female CoTS can produce up to 65 million eggs during a spawning season19, and even higher reproductive potential has been estimated (for example, a 480-mm diameter female is expected to produce 106 million oocytes)20. Furthermore, during the life cycle of CoTS, mortality rate is highest during the planktonic stage21. This means that even a small increase in the number of planktonic larvae can lead to a significant recruitment of new individuals, thereby increasing the baseline for CoTS outbreaks21.

Currently, the debate on how larval supply increases has focused on two hypotheses: the nutrient enrichment hypothesis and the predator removal hypothesis. The nutrient enrichment hypothesis, also known as the terrestrial runoff hypothesis, posits that high rainfall events flush terrestrial nutrients into the ocean, leading to phytoplankton blooms and subsequently increasing food availability for CoTS larvae12,13. This hypothesis appears to be supported by certain laboratory studies, which indicate that food availability plays a critical role in determining larval survival rate22,23,24. However, these laboratory studies lack support from field studies. Furthermore, CoTS larvae have also been found to be resilient against low nutrient conditions and the juveniles are particularly resilient25,26,27. This makes predation pressure on CoTS even more important. The predator removal hypothesis was originally defined based on the observation that CoTS populations are typically regulated by high levels of post-settlement or adult predation28. Recent studies have broadened the scope of this hypothesis and increased its credibility. Research has shown that the incidence or severity of CoTS outbreaks increases in areas subject to fisheries exploitation18,29. Sweatman et al. compared the likelihood of CoTS outbreaks on reefs open to fishing versus those closed to fishing within the Great Barrier Reef and found that 75% of reefs open to fishing experienced outbreaks, whereas only 20% of reefs closed to fishing for at least five years were affected30. Kroon et al. found that CoTS densities on reefs within no-take marine reserves were 2.8-fold lower than on reefs open to fishing17. Furthermore, certain planktivorous fish are able to consume CoTS larvae even when alternative prey is available, and they show a preference for CoTS larvae over co-occurring starfish (Linckia laevigata) larvae31. The results suggest that reef fish harvesting may increase the likelihood of CoTS outbreaks by removing one of the key regulatory mechanisms that prevent extreme population fluctuations. This suggests that fish predation may play a critical role in regulating CoTS populations21.

Currently, an increasing number of fish species have been identified as predators of CoTS larvae by field observation or DNA detection in the GBR, Australia, and Japan16,32. The number of fish species reported to potentially prey on CoTS larvae has reached 84 species33. As a critical component of the food chain, fish primarily regulate prey populations through their predation behaviour and characteristics, thereby preventing excessive population growth that could lead to resource competition and ecological imbalance34. A few studies have explored the predation efficiency of certain fish species on CoTS larvae. For example, the planktivorous damselfish (Dascyllus reticulatus) has been shown to be an efficient predator of CoTS larvae, with predation efficiency increasing with larval density until saturation is reached. Furthermore, the presence of other morphologically similar prey does not affect their predation efficiency on CoTS larvae31,35,36. However, to better assess the regulatory role of fish predators on CoTS populations during their early developmental stages, more data are needed on the predation efficiency and characteristics of these fish species on CoTS larvae. This knowledge also has important implications for mitigating the impact of CoTS outbreaks on coral reef ecosystems through the establishment of marine protected areas, the conservation of fish biodiversity, and the formulation of sound fisheries management policies in the future.

This study focuses on fish species inhabiting reefs in the South China Sea affected by CoTS outbreaks, selecting five fish species: Chaetodon auripes (also known as Oriental butterflyfish), Arothron hispidus, Ostorhinchus taeniophorus, Amphiprion ocellaris, and Lates calcarifer as case studies to investigate their predation efficiency and characteristics. The first four species are common inhabitants of coral reef ecosystems in the South China Sea, some of which have been reported as potential predators of CoTS37,38. Although Lates calcarifer is not a typical reef fish, it is occasionally observed in waters adjacent to coral reefs, particularly when food is abundant39. To ensure comprehensive coverage and to account for this possibility, Lates calcarifer was included as a representative species in this study. By analysing the predation efficiency of these five fish species, the study identifies the most efficient predator for further investigation of its predation characteristics. In addition, the study examines whether the presence of an exogenous food source, such as Artemia (brine shrimp), influences the predation efficiency of the fish. Artemia is distinguished by its high protein, lipid, and essential amino acid content, and is widely recognised as one of the most commonly used live feeds in aquaculture40. It has also been reported as a preferred feed for the culture of several butterflyfish species41 and has been used in research on coral reef fish preying on CoTS larvae42. Although the likelihood of butterflyfish preying on artemia in the natural marine environment is extremely low, given its nutritional value and accessibility, artemia can be used as a substitute for the natural prey of butterflyfish in the experimental study. This study provide a theoretical basis for understanding the role of fish predators in regulating CoTS populations, and informed future strategies for mitigating CoTS outbreaks through marine conservation and fisheries management.

Materials and methods

Statement of enthics

The use of animals in this study was approved by the Animal Research and Ethics Committees of the South China Sea Institute of Oceanology, Chinese Academy of Sciences. All experiments were conducted following the guidelines of the committees (SCSIO-2024-010). The study was carried out in compliance with the Animal Research: Reporting of the Vivo Experiments (ARRIVE) guidelines.

Preparation of CoTS larvae

Adult CoTS captured on the reefs near Tanmen (Hainan, China) during the spawning season in May 2022 were subjected to artificial breeding. Spawning was induced by injecting 1 mL of 10−4 M 1-methyladenine into the junction of the arm and central disc of four female and five male CoTS, respectively. After 10 to 25 min, sperm and eggs were collected and fertilized by mixing the gametes in four ponds (1 m × 1 m × 1 m) filled with filtered seawater (FSW). The density of larvae in each cement pool was approximately 1 larva per 2 ml FSW, and the temperature of the culture water was 29 ± 0.5℃. The development of CoTS larvae was observed under a microscope (Table 1). When the CoTS larvae developed a complete digestive tract (about 15–48 h), they were fed with Chaetoceros muelleri at a concentration of 5000 cells/L once a day after the daily water change. The larvae developed to the bipinnaria stage in about 2–6 days and to the brachiolaria stage in 7–26 days. CoTS larvae at different stages (cleavage stage, gastrula, bipinnaria, brachiolaria) were identified using a microscope (Fig. 1) and collected through a 250 µm mesh.

Table 1 Development time for CoTS larvae.
Full size table
Fig. 1
figure 1

Life stages of Crown-of-Thorn Starfish larvae. (a) fertilized egg; (b–d) cleavage stages; (e) blastula; (f) gastrula; (g) early bipinnaria; (h) mid-bipinnaria; (i) late bipinnaria; (j) early brachiolaria; (k) mid-brachiolaria; (l) late brachiolaria.

Full size image

Experiment 1: evaluation of the predation efficiency of different reef fish species

Four available species of common reef fishes were purchased from Xiamen Jincheng Wave Technology Co., Ltd. (Xiamen, China), including Chaetodon auripes, Arothron hispidus, Ostorhinchus taeniophorus, and Amphiprion ocellaris, which or groups within them, have been reported to prey on CoTS. Lates calcarifer was supplied by the Tropical Aquatic Research and Development Center (Hainan, China) (size and weight see Table 2). Five replicates of each fish species were maintained together in a 12L aquarium containing an oxygen stone and 10 L of 0.5-µm filtered seawater (FSW), with daily water changes. All were fed with commercially available food pellets and were not used for experiments until they were fed normally. A total of five fish of each species were used in each experimental tank, and there were five experimental tanks in total and twenty-five fishes. All were fasted for 24 h prior to the predation experiment.

Table 2 Measurements of five different species fishes (n = 5; expressed as the mean ± standard deviation [SD]).
Full size table

To test the consumption rate (which was inferred based on loss of CoTS larvae in presence of these putative larval predators) of each species, 400 brachiolaria were added to the five 10 L FSW systems with five replicates of each different species, together with a blank FSW system without the presence of fish as a control group. After 1 h, the fish were removed and the number of CoTS larvae remaining was counted after being separated by a 250 µm mesh. After 2 days of temporary nursing, the experiment was repeated once. During the respite period, all fish were fed normally with the commercial food pellets. The consumption rate of each fish was obtained by calculating the average of the two replicates, and the consumption rate of the fish in each experiment was calculated as [(number of initial larvae – number of remaining larvae—background value)/time]/ number of fish replicates, with the unit (larvae h−1). The background value was calculated from the consumption of the control group as (initial larval number—remaining larval number). Fish species with the highest consumption rate of CoTS larvae were selected for the following study.

Experiment 2: effect of CoTS larvae life stages and densities on the reef fishes’ predation

To evaluate the effect of life stage and density of CoTS larvae on the consumption rate of predatory fish, this experiment considered four distinct larval stage (cleavage stage, gastrula, bipinnaria, and brachiolaria larvae) and four densities (25 larvae L−1, 50 larvae L−1, 100 larvae L−1, and 200 larvae L−1) of each. Since the four distinct larval stages were conducted separately (in accordance with the development time of the larvae), four controls were set up for each concentration at each stage. Larvae of each life stage at different densities were fed to C. auripes for 1 h in separate 1.2 L FSW systems with four replicates. One fish was placed in each replicate tank, and four replicates of the same system without fish were used as controls at each density. Thus there were 128 trials and 64 unique specimens of C. auripes. The consumption rate of the predatory fish at each density was obtained by calculating the average of the four replicates. The consumption rate of predator fish for each larval density at each stage was obtained by calculating as [(number of initial larvae—number of remaining larvae—average value of the control group)/time]/ number of fish replicates, with the unit (larvae h−1). The average value of the control group was calculated by adding together the consumption of the four controls and then dividing by the number of replicates. The consumption of each replicate was calculated as the difference between the initial and remaining larval numbers. When larvae die, they gradually degrade and whole individuals cannot be seen. Therefore, in the control group, complete larval morphology could not be observed under a microscope and was considered a loss.

Experiment 3: effect of feeding durations on the reef fishes’ predation

Sixty CoTS bipinnaria larvae were fed to C. auripes in 1.2 L FSW (50 larvae L−1) for 10 min, 30 min, 60 min, and 120 min. Four replicate experimental and four replicate control trials were conducted simultaneously for each treatment period. The number of larvae consumed was counted from the larvae remaining in the 250 µm mesh after removal of each fish at the end of the feeding period.

Experiment 4: effect of food competition on the reef fishes’ predation

To evaluate fish predation on CoTS larvae with alternative food, healthy and active Artemia larvae purchased together with CoTS larvae were fed to the representative fish (C. auripes) at five different ratios in 1.2 L FSW as shown in Table 3. The number of larvae consumed was counted from the remaining larvae 1 h later and the consumption rate was calculated as described above.

Table 3 Ratio of CoTS and Artemia larvae used.
Full size table

Data analysis

The Kruskal–Wallis H test was used to analyze variance in data that did not conform to a normal distribution and homogeneity, such as in Experiments 1 and 4. Experiment 2 used two-factor ANOVA after logarithmic transformation of the data. Based on the results of normal distribution and homogeneity of variance tests, one-way ANOVA analysis was performed using SPSS software (version 22, IBM, USA) to calculate the p-value for significance (the difference is considered significant if p-value < 0.05), as in Experiment 3. Significance was labelled with lowercase letters in alphabetical order. Origin 8.1 software was used for plotting graph, and Adobe Illustrator 2020 for graph layout and beautification.

Results

Different stages of larvae were bred (Fig. 1), collected and counted for further experiments. Five reef fish species were tested for their ability to prey on CoTS larvae, including C. auripes, A. hispidus, O. taeniophorus, A. ocellaris, and L. calcarifer (Fig. 2). In particular, C. auripes consumed significantly more larvae than the other four fish species (ANOVA, Fig. 2, P < 0.05) with an average of 69.9 ± 3.0 larvae h−1 (n = 2). There was no significant difference among the consumption rates of the other four fish species (Fig. 2, P > 0.05). O. taeniophorus showed the lowest predation (11.9 ± 14.3 larvae h−1) on CoTS larvae, followed by L. calcarifer (19.9 ± 11.5 larvae h−1), A. hispidus (28.8 ± 14.4 larvae h−1) and A. ocellaris (32.0 ± 2.3 larvae h−1) in increasing order of consumption rate.

Fig. 2
figure 2

Consumption of five fishes on CoTS larvae. Error bars represent the standard deviation of the mean (replicates = two per treatment). Letters indicate significant differences (p < 0.05).

Full size image

Based on the optimal predation efficiency of C. auripes, the consumption rate of larvae at different densities and life stages was further determined. The results showed that C. auripes had no preference for a particular life stage of the CoTS larvae, as evidenced by the non-significant differences among the consumption rates of cleavage stage, gastrula, bipinnaria, and brachiolaria within the same prey densities (Fig. 3, P > 0.05). However, the predation rate increased dramatically with the density of brachiolaria larvae (Fig. 3, P < 0.05). Specifically, the predation rate increased from 26.50 ± 9.56 larvae h−1 to 97.25 ± 13.50 larvae h−1 when the density of CoTS larvae was increased to 100 larvae L−1 compared to 50 larvae L−1, and peaked at 164.00 ± 52.57 larvae h−1 when the larval density was 200 larvae L−1 (Fig. 3, P < 0.05).

Fig. 3
figure 3

Consumption of C. auripes on CoTS larvae with different life stages and different densities. Error bars represent the standard deviation of the mean (replicates = four per treatment). Letters indicate significant differences (p < 0.05).

Full size image

To characterize the effect of feeding duration on predation, we examined the number of CoTS larvae preyed upon by C. auripes for four durations (10 min, 30 min, 60 min, and 120 min). The results showed that predation for 120 min resulted in a significantly higher consumption (Fig. 4, P < 0.05), which was not different among the other three groups (Fig. 4, P > 0.05).

Fig. 4
figure 4

Consumption of C. auripes on CoTS larvae with different feeding duration. Error bars represent the standard deviation of the mean (replicates = four per treatment). Letters indicate significant differences (p < 0.05).

Full size image

To determine whether the presence of other foods would affect the predatory efficiency of C. auripes towards CoTS larvae, Artemia larvae—which are commonly used in the commercial artificial rearing of aquaculture fish—were mixed with CoTS larvae at five ratios. The result showed that the consumption rate of CoTS larvae by C. auripes was 22.5 ± 3.7 larvae h−1 when no Artemia larvae were presented. When the same amount of Artemia larvae was added to the system, the consumption rate of CoTS increased to 38.3 ± 14.0 larvae h−1 without a significant change (Fig. 5, P > 0.05). When the number of Artemia larvae was further increased to twice the number of CoTS larvae (a ratio of 1:2), the feeding rate C. auripes’s on Artemia reached 115.5 ± 3.4 larvae h−1, compared to about 50 larva h−1 in the other trials (Fig. 5), but there was still no significant change in the consumption rates of CoTS larvae (Fig. 5, P > 0.05).

Fig. 5
figure 5

Consumption of C. auripes on CoTS larvae with the presence of Artemia. Error bars represent the standard deviation of the mean (replicates = four per treatment). Letters indicate significant differences (p < 0.05).

Full size image

Discussion

The outbreak of crown-of-thorns starfish is undermining the resilience of coral populations and has become one of the threats to the health of coral reef ecosystems in the Indo-Pacific, including the South China Sea43. Direct fish predation on CoTS larvae is considered to be one of the major limiting factors for CoTS population size17. An increasing number of fish species have been reported to prey on CoTS16,33. However, it is unclear to what extent fish play a role in mitigating CoTS outbreaks. In this study, we extended these findings by examining the predation potential of five coral reef fish species commonly found on outbreak reefs in the South China Sea on CoTS larvae, and further identified the predation characteristics of the most effective fish species. This work contributes to a deeper understanding of the intricate interactions between fish predators and CoTS larvae, and provides insights into potential strategies to enhance coral reef fish conservation and support efforts to mitigate CoTS outbreaks.

Our study shows that the oriental butterflyfish (C. auripes) significantly outperforms the other four coral reef fishes in predation on CoTS larvae, suggesting that butterflyfish predation on pelagic CoTS may be easier to observe in the field. Similarly, butterflyfish (Chaetodontidae) have been reported to be able to consume pelagic CoTS17, and Keesing and Halford found that the C. auripes was the only planktivorous fish predator of CoTS sperm observed in the wild38, probably because it shares a similar habitat with CoTS44. However, all of these studies focused on predation on a single life stage of CoTS larvae, neglecting the potential difference in consumption rate caused by different life stages. It has been reported that with increasing age or size of CoTS, predator deterrent chemicals such as saponins accumulate in the body, which may reduce the consumption rate of predators16,32,33,45. However, our study shows that the life stages of CoTS larvae have a negligible effect on the consumption rate of the butterflyfish tested. This suggests that the butterflyfish may be insensitive to the toxic chemicals of CoTS larvae. In addition, the butterflyfish has been reported to feed on injured and dead adult CoTS, which may further support its predation on CoTS16. The increasing consumption rate resulting from the increased prey density suggests that the butterflyfish could increase its predation to satiation. Even in the presence of other alternative prey (e.g. Artemia), the butterflyfish did not appear to switch prey. Furthermore, when the same quantity of Artemia larvae was added to the system, the consumption rate of CoTS increased to 38.3 ± 14.0 larvae h−1 without any significant change (Fig. 5, P > 0.05). It seems that an appropriate mixture of both types of larvae makes the fish a better appetizer. In addition, when the number of CoTS larvae was doubled relative to the number of Artemia larvae (a ratio of 2:1), the feeding speed of C. auripes for CoTS larvae did not differ significantly compared to a ratio of 1:1 (Fig. 5, P > 0.05). This suggests that the predation of butterflyfish on other species may have little or no effect on their predation of CoTS larvae, possibly due to their generalist predator nature, feeding on a variety of invertebrates46.

The distribution and abundance of CoTS larvae, as well as their effective settlement and recruitment, are fundamental to determining further patterns of adult abundance. Once the larvae have settled, the success rate of lethal predation will be greatly reduced. According to Wilmes’ model, only 3.5% of fertilized zygotes (up to a mortality rate of 15.4% d−1 within the 21-day pre-settlement phase) will survive to settlement21, but the morality rate is only 2.6% d−1 for 3 mm diameter CoTS juveniles, with this rate decreasing with increasing age47,48. Thus, the pelagic phases provide a valuable window for CoTS population control. This study showed that the consumption rate of CoTS larvae by oriental butterflyfish (C. auripes) peaked at 154.1 ± 57.28 larvae h−1 when the density of CoTS larvae was 200 larvae L−1. Coupled with field observations that CoTS larval densities during outbreaks ranged from 0.037 larvae L−1 to 0.053 larvae L−1 in the natural reef environment49,50, which is much lower than the experimental parameters. It is hypothesized that an increase in the number of surrounding butterflyfish to one per cubic metre is likely to result in a significant reduction in the number of CoTS larvae and a subsequent reduction in the outbreak base. Thus, predatory fishes, especially those with high predation efficiency such as the oriental butterflyfish, may play an important role in regulating CoTS outbreaks. This further emphasises the paramount importance of protecting these fish and their habitat, as healthy and protected reefs with more intact fish assemblages may be less susceptible to CoTS outbreaks51.

Two lines of future research should be pursued. Firstly, further excavation of fish species that can prey on CoTS, based on existing reports. Secondly, research into the predation characteristics of fishes reported to prey on CoTS should be intensified. This will help to identify the most effective predators of CoTS in the natural marine environment. It is recommended that these predators be included in local reef fish conservation management lists. In addition, related work, such as artificial stocking could be considered with the aim of protecting and rehabilitating their habitats and restoring their populations, but also need to think about the most likely factors that are regulating their population.

In recent years, the number of naturally occurring fish predators of CoTS in the South China Sea, including butterflyfish, has been gradually declining52. Therefore, it is of great importance to isolate the coral reef community by artificially releasing target fish species to supplement unnatural losses of fish to increase predation pressure on CoTS. This may not only improve the environment that favours the survival of planktivores, but may also allow the planktivorous fishes to play a more influential role in controlling CoTS populations and eventually maintain population stability at normal densities53. However, it is important to note that fish stocking should preferably be carried out during the spawning season of the CoTS population outbreak, and prey fish stocking should be conducted in accordance with best practice guidelines. This is because CoTS spawning in the wild is usually aggregated, producing a large number of gametes in a short period of time (10–30 min)54. Such intensive spawning could easily overwhelm local predators in a normal stock, leading to inefficient predation of the CoTS gametes and zygotes and further survival of the newborns. Release during the spawning season can mitigate the effects of the overcrowding and allow it to play a fuller and more efficient role in CoTS population control.

Conclusions

Overall, the results of our study indicate that butterflyfish have great potential to regulate the population of CoTS larvae, particularly at CoTS spawning sites where the fish would experience an intensive influx of gametes. Concurrently, our study reinforces the fish-predator hypothesis of CoTS outbreaks, indicating that coral reef fish may play a pivotal role in regulating CoTS populations. Further endeavours should be undertaken to safeguard and restore keystone coral reef fishes and their habitats, particularly the most effective predators of crown-of-thorns starfish.