Intraspecific and habitat-mediated responses to chemical cues play key roles in structuring populations of marine species. We investigated the behaviour of herbivorous-stage juvenile crown-of-thorns sea stars (COTS: Acanthaster sp.) in flow-through choice chambers to determine if chemical cues from their habitat influence movement and their transition to becoming coral predators. Juveniles at the diet transition stage were exposed to cues from their nursery habitat (coral rubble-crustose coralline algae -CCA), live coral, and adult COTS to determine if waterborne cues influence movement. In response to CCA and coral as sole cues juveniles moved toward the cue source and when these cues were presented in combination, they exhibited a preference for coral. Juveniles moved away from adult COTS cues. Exposure to food cues (coral, CCA) in the presence of adult cues resulted in variable responses. Our results suggest a feedback mechanism whereby juvenile behaviour is mediated by adult chemical cues. Cues from the adult population may deter juveniles from the switch to corallivory. As outbreaks wane, juveniles released from competition may serve as a proximate source of outbreaks, supporting the juveniles-in-waiting hypothesis. The accumulation of juveniles within the reef infrastructure is an underappreciated potential source of COTS outbreaks that devastate coral reefs.

Specimens and rearing condition

As the taxonomy of the Pacific species of Acanthaster is uncertain (1), the species investigated here is referred to as Acanthaster sp. or COTS. Adults were collected near Cairns (northern GBR) by COTS control divers and shipped to the Sydney Institute of Marine Science where they were maintained in flow-through aquaria (filtered sea water, FSW 5 µm) at 26 °C, the temperature of their habitat (http://data.aims.gov.au/aimsrtds/yearlytrends.xhtml). For each of the three fertilisations, the gametes of two different males and two females were used, and the larvae were reared to the settled juvenile stage following routine methods (2). Larvae with a well-developed juvenile rudiment (14–16 days post-fertilization) were induced to settle using coralline algae (Amphiroa sp.). The juveniles were maintained in a constant temperature room (26 °C, 12:12 hr light cycle) in UV sterilised 1 µm FSW with ambient salinity of ~ 34 % (Merck salinity probe) and fed small fronds of Amphiroa sp. Juveniles, up to 3mm in diameter, were placed in 12 well dishes (5 ml wells). Food and water were renewed every 3 days. As the juveniles grew, (>3 mm diameter) they were transferred to 6 well (10 ml well) dishes and when they reached 10 mm diameter were transferred to large culture dishes (100-200 ml). For the large juveniles, culture dishes, food, and FSW were renewed twice a week.

Behaviour experiments

The behaviour experiments were conducted using a suite of cues including live coral, coral rubble encrusted with crustose coralline algae (CCA), and adult COTS. The CCA-covered rubble was collected from the shallow reefs (< 2m depth) at One Tree Island, GBR (23.5076° S, 152.0916° E) and maintained in a tropical aquarium at 26 °C. Corals (Acropora sp.) that are the preferred prey of COTS (3) were sourced from aquarium suppliers. The CCA and live corals were maintained in separate 20 L closed system aquaria at 26 °C. Adult COTS were maintained in a 160 L tropical aquarium system at 26 °C. Before the start of the coral and CCA rubble experiments these cues were maintained for 2h in the header tank (25 L) source water before turning on the flow. The amount of rubble and live coral used covered the base of the header tank as similarly as possible between trials to emulate a layer of habitat in nature. Adult COTS were placed in the header source for 30 min before the start of the experiment. During this time water circulation was maintained using a water pump.

The behavioural responses of the juveniles were tested in a constant temperature room (26 °C) using a two-channel choice flume as used in behavioural studies of fish larvae and sea urchin juveniles (4, 5). Steady gravity-driven flow (8.3 cm³/s per channel) in the flume was controlled by flow meters (Dwyer MMA series) and laminar flow was established at the inflow point by course filter pad lined with a mesh screen. To ensure that behaviour was not influenced by vision (6), the experiments were done under red light 625 λ (measured by an Ocean Optics, Spectrophotometer USB4000; Ocean View 2.0 Software) which exceeds the photoreception range of COTS eyes (7). Before use in trials, the juveniles were placed in clean containers of 1.0µm UV FSW with no food for 24 hrs. These are also the conditions they experienced when first placed in the choice flume before the turn-on of flow. All experiments were conducted during the day.

The water flow on each side of the flume was segregated by a divider which ensured parallel flow into the main chamber supporting a distinct distribution between the cue source flows and eliminated backflow. Each side of the flume was supplied with water from a 25L header tank, that contained a treatment cue. At the start of each experimental run, a single juvenile was placed in the centre of the main chamber where it would receive even flow from both sides of the flume. At the point of water outflow, a gate was installed to allow the juvenile to acclimate for one minute in static water. The gate was then removed, and the flow was turned on. Although the exact position of the juveniles when the gates were lifted, varied somewhat, we did not adjust to avoid a physical disturbance. None of the juveniles were in a location where they would only experience flow from one side of the flume.

Juveniles were allocated 20 min to respond as demonstrated by upstream movement towards the cue source or downstream movement. The treatment cues, live coral, CCA reef rubble, and COTS were tested individually against the no cue FSW only flow and against each other (CCA/live coral; CCA/COTS; COTS/live coral). The same juvenile was not used more than once per treatment. Treatment cues were rotated between the left and right sides of the flume to remove a side bias. Images taken at 20-second intervals after the acclimation period were captured with an Olympus Tough TG-6 (Olympus) camera and were then autogenerated into time-lapse videos. The camera was mounted above the choice flume with the entire arena and scale in view. Three juveniles climbed the chamber wall and “surfed’ at the air-water interface oral side up and so were not used for the experiment.

Juvenile size (diameter mm) was similar, but number of replicates varied somewhat among treatments (FSW x̄ = 10.32, SE = 0.40, n = 42; COTS x̄ = 10.24, SE = 0.42, n = 50; Coral x̄ = 9.67, SE = 0.21, n = 30; Coral/COTS x̄ = 9.67, SE = 0.21, n = 41; CCA x̄ = 9.40, SE = 0.19, n = 27; CCA/Coral x̄ = 12.97, SE = 0.25, n = 35; CCA/COTS x̄ = 12.77, SE = 0.36, n = 22). We conducted more runs for the adult COTS cue to facilitate the interpretation of juvenile responses because this treatment was the most variable.

Juvenile movement in the time-lapse files was analysed using Tracker (Version 6.0.1). Each file was calibrated in Tracker to display at a framerate of 1.5/s. Path length (m) was manually calculated from the starting point until the juvenile moved towards or away from the direction of flow. Tracking was used for single and two-cue treatments for the juveniles that remained in the field of view for the entire time and so could be completely followed (n: coral = 11, CCA = 14, COTS = 20, FSW = 8, CCA/Coral = 13, CCA/COTS = 12, Coral/COTS = 12). The paths taken by the juveniles were traced and illustrated as in a previous study of sea star behaviour (8).

Statistical analyses  

All analyses were performed using R (v1.2.1335) (9). Chi-squared goodness of fit test was used to analyse the percentage frequency of the juvenile choice responses with equal expected proportions. Choice response data were analysed with a 25%:25%:25%:25% probability of upstream movement to either treatment cue, downstream, or no movement. For the FSW only (control) the data were analysed with a 33%:33%:33% probability of upstream movement, downstream movement, or no movement. The choice data were illustrated as the percentage of individuals in each response (e.g., to cue, downstream, no movement).

Tracking data on speed (mm/min), distance moved (mm) and time to make a choice (min) were analysed. For the single cue experiments, only the significant responses (to or away from cue) as indicated by Chi-squared analysis were tracked. The speed data were analysed by one-way analysis of variance (ANOVA) with treatment as a fixed factor with four levels (attraction to coral, attraction to CCA, upstream movement in FSW only, avoidance to adult COTS) with the FSW as the control. Homogeneity of variance (HOV) and normality were confirmed by Levene’s and Shapiro-Wilk tests, respectively (significance α = p < 0.05). Tukey’s HSD post hoc-pairwise comparison was performed to identify treatments that differed. Data on distance moved and time to make a choice did not meet the assumption of HOV, and so these data were analysed using Kruskal-Wallis non-parametric ANOVA with treatment type as a fixed factor. Dunn Test (R Package FSA) with Bonferroni corrected p-values were used post hoc to identify significant pairwise treatment effects.

For the two-cue experiments, data on distance moved, time to make a choice and speed were analysed by one-way analysis of variance (ANOVA) with treatment as a fixed factor with three levels (CCA/live coral, CCA/COTS, live coral/COTS). Assumptions of HOV and normality were confirmed as above, except for the normality of the distance moved data. As ANOVA is robust to this (10) the analysis was undertaken and a Tukey HSD post hoc-pairwise comparison was performed.

References:

1. Uthicke S, Pratchett MS, Bronstein O, Alvarado JJ, Wörheide G. The crown-of-thorns seastar species complex: knowledge on the biology and ecology of five corallivorous Acanthaster species. Marine Biology. 2024;171(1):32. DOI: 10.1007/s00227-023-04355-5
2. Deaker DJ, Agüera A, Lin H-A, Lawson C, Budden C, Dworjanyn SA, et al. The hidden army: corallivorous crown-of-thorns seastars can spend years as herbivorous juveniles. Biology Letters. 2020;16(4):20190849. DOI: 10.1098/rsbl.2019.0849

3. Keesing JK, Thomson DP, Haywood MD, Babcock RC. Two time losers: selective feeding by crown-of-thorns starfish on corals most affected by successive coral-bleaching episodes on western Australian coral reefs. Marine Biology. 2019;166(6):72. DOI: 10.1007/s00227-019-3515-3

4. Gerlach G, Atema J, Kingsford MJ, Black KP, Miller-Sims V. Smelling home can prevent dispersal of reef fish larvae. Proceedings of the National Academy of Sciences. 2007;104(3):858-63. DOI: 10.1073/pnas.0606777104
5. Nishizaki MT, Ackerman JD. A secondary chemical cue facilitates juvenile‐adult postsettlement associations in red sea urchins. Limnology and Oceanography. 2005;50(1):354-62. DOI: 10.4319/lo.2005.50.1.0354

6. Petie R, Hall MR, Hyldahl M, Garm A. Visual orientation by the crown-of-thorns starfish (Acanthaster planci). Coral Reefs. 2016;35(4):1139-50. DOI: 10.1007/s00338-016-1478-0

7. Hall MR, Kocot KM, Baughman KW, Fernandez-Valverde SL, Gauthier ME, Hatleberg WL, et al. The crown-of-thorns starfish genome as a guide for biocontrol of this coral reef pest. Nature. 2017;544(7649):231-4. DOI: 10.1038/nature22033

8. Dale J. Coordination of chemosensory orientation in the starfish Asterias forbesi. Marine & Freshwater Behaviour and Physiology. 1999;32(1):57-71. DOI: 10.1080/10236249909379037

9. Team RC. RStudio. 2020. p. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria.

10. Underwood AJ. Experiments in ecology: their logical design and interpretation using analysis of variance: Cambridge University Press; 1997. DOI: 10.1017/CBO9780511806407