Variable food alters responses of larval crown-of-thorns starfish to ocean warming but not acidification

Marine ecosystems are increasingly exposed to multiple stressors, notably ocean warming and acidification driven by anthropogenic CO2. Surface oceans have already warmed by ~0.75 °C and acidified by ~0.1 pH units since the mid-20th century, with further changes projected by 2100. In tropical oceans, warming and nutrient limitation are reducing phytoplankton abundance while also increasing its temporal variability due to changing climatic and upwelling dynamics. Phytoplankton are the primary food for most marine benthic invertebrate larvae, so food availability likely modulates larval capacity to cope with ocean change. Prior work shows high food can bolster larval echinoderms and molluscs under low pH or warming within thermal limits, whereas low food hampers resilience to acidification and combined stressors. Warming may transiently speed development but later exacerbate effects of low food. For crown-of-thorns starfish (CoTS, Acanthaster sp.), outbreaks have been linked to nutrient-driven phytoplankton blooms that enhance larval success. CoTS larvae typically develop rapidly under optimal food but can persist under low food. Previous studies found fertilization and early development are generally robust to warming/acidification, though temperatures above ~29 °C and pH decreases ≥0.2 can impair later stages; combined stressors can exacerbate early-stage effects but show mixed interactions at later stages. Only one prior study assessed food–warming interactions in CoTS larvae; none examined food effects on responses to acidification or combined stressors. This study tests how three food regimes (low, low-to-high switch, and high) alter larval CoTS responses to present-day (pH 8.0, 26 °C), warming (30 °C), and acidification (pH 7.6). We expected high food to increase resilience to warming and acidification relative to low food. We further hypothesized that under present-day conditions, a short low-food period would slow growth/development but not reduce survival and that switching to high food would allow recovery; under warmed and acidified conditions, we expected carry-over effects of early low food and increased energetic demands to limit recovery compared to constant low food.

Outbreaks of Acanthaster spp. have caused major coral declines in the Indo-Pacific and are hypothesized to stem partly from anthropogenic nutrient inputs that fuel phytoplankton blooms, enhancing larval survival. There is a positive relationship between food abundance and CoTS larval growth and development up to ~1 × 10^5 cells mL−1, above which performance declines. Under optimal food, larvae commence feeding at 2–3 days post-fertilisation (dpf) and can reach late brachiolaria by 13–17 dpf, yet larvae also show resilience to nutrient-poor conditions, surviving weeks without phytoplankton. Regarding ocean change, CoTS fertilisation and early development are generally unaffected by warming and acidification, but temperatures above ~29 °C negatively affect later stages. Acidification of ≥0.2 pH units can slow development and increase mortality with mixed effects on size. Simultaneous warming and acidification can exacerbate gastrulation impacts but have inconsistent interactions at later stages; parental exposure can also influence larval responses. Prior studies typically used food rations supporting normal development (0.1–2 × 10^4 cells mL−1). Only one study showed that with abundant food, CoTS larvae grew ~30% faster at 30 °C than at 28 °C. No prior work examined how varying food availability modifies larval responses to acidification or combined warming and acidification.

Experimental design: Full-factorial combinations of temperature (26, 30 °C), pH (8.0, 7.6), and food regime (low, switch, high) in a purpose-built flow-through seawater system. Food regimes using Proteomonas sulcata (cryptomonad) fed three times daily: low = 1 × 10^3 cells mL−1; switch = 1 × 10^3 cells mL−1 from 3–11 dpf then 5 × 10^4 cells mL−1 thereafter; high = 5 × 10^4 cells mL−1. System and water chemistry: pH manipulated via automated CO2 injection in 60-L reservoirs (pH 7.6) or unmanipulated control (pH 8.0), with aeration to maintain DO >98%. Water warmed in 20-L reservoirs (26 or 30 °C) and delivered to replicates at ~2.9 L h−1 (~60 turnovers/day). Continuous recirculation maintained stable pH and temperature. pH (total scale), temperature, and salinity were measured 4–6 days/week; total alkalinity measured by potentiometric titration; carbonate system parameters computed using CO2SYS with Mehrbach constants refit by Dickson and Millero. Rearing: Larvae reared in flow-through downwellers (1.16 L volume; 74-µm mesh floor). Inner PVC inserts exchanged ~every 10 days. Study species and spawning: Adult Acanthaster sp. collected from Great Barrier Reef near Cairns; maintained at 25–27 °C. Gametes from 4 females and 4 males. Eggs pooled and fertilized in experimental seawater at sperm:egg ~100:1, achieving >90% fertilization. Embryos stocked into downwellers at 7.8 embryos mL−1. At ~72 hpf, when feeding began, densities standardized to 1.46 ± 0.06 larvae mL−1, then randomly assigned to food treatments. Microalgae culture: P. sulcata grown in aerated 20-L carboys at 25–32 °C with F media; food was rapidly flushed from downwellers, but larvae consistently had algae present in gut 0–3 h post-feeding. Measurements: Survival tracked weekly from start of feeding to 56 dpf; survival expressed as % of initial density. At 11 and 18 dpf, ~20–50 larvae per replicate were relaxed (7% MgCl2) and preserved (10% formaldehyde-FSW). All larvae scored as normal or abnormal (irregular shape, small size, arrested development). Normal larvae (median 20 per replicate) were photographed and measured for length and width (ImageJ), and staged (gastrula to late brachiolaria). Competency to settle assessed every 1–2 days from 14 dpf in subsamples (~20–100 larvae per replicate): larvae deemed competent if late brachiolaria with a prominent rudiment and settlement behaviors. Endpoint: >50% of surviving larvae competent or all larvae dead. Replication: Downweller (n = 7 per treatment) used as replication unit for analyses unless low replication due to mortality/abnormalities (noted for pH 7.6 at 18 dpf). Statistics: Three-way PERMANOVAs/ANOVAs tested effects of temperature, pH, and food on length, width, L:W ratio (11 dpf), abnormal proportion (11, 18 dpf), and early brachiolaria proportion (11 dpf). Survival during weeks 1–3 analyzed by repeated measures ANOVA (temperature, pH, food fixed; week random). Late brachiolaria at 18 dpf analyzed by two-way ANOVA (temperature, food) for pH 8.0 only due to low replication at pH 7.6. For 18 dpf size metrics with missing cells, one-way ANOVA with combined factor and planned contrasts supported presentation as three-way ANOVA without 3-way interaction. Assumptions evaluated; some variables non-normal with limited effect of transformation; outliers retained as results unchanged. Competency analyzed by binary logistic regression with predictors temperature, pH, food, and their interactions; best-fitting model selected by model χ²; effects assessed by Wald χ² with 1000 bootstrap BCa CIs. Data on time to competency not formally analyzed.

- Food effects: Low food (1 × 10^3 cells mL−1) produced smaller, slower-developing larvae with more abnormalities than high food (5 × 10^4 cells mL−1). Switching from low to high food at 11 dpf improved development rate and reduced abnormalities to levels comparable to continuously high food but larvae remained smaller by ~16–17% at 18 dpf versus continuous high food. At 11 dpf, larvae in low/switch were ~23% shorter and ~18–20% narrower than high food; no difference between low and switch at that time. At 18 dpf, low and switch were ~35% and ~16–17% shorter and ~33% and ~17–19% narrower than high food, respectively; switch larvae were ~19–22% larger than low. - Temperature and pH main effects on size and development: At 11 dpf, under 26 °C, pH 7.6 larvae were 25% shorter and 21% narrower than pH 8.0; under 30 °C there was no pH effect on size. At pH 8.0, 30 °C larvae were ~24–25% smaller in length/width than 26 °C; at pH 7.6, 26 and 30 °C did not differ. At 18 dpf, independent main effects: 30 °C larvae were ~28–29% shorter/narrower than 26 °C; pH 7.6 larvae were ~29–32% shorter/narrower than pH 8.0. - Abnormalities: At 11 dpf, abnormal proportion depended on temperature × food: at low and switch food, abnormalities were ~3.2× and ~1.7× higher at 30 °C vs 26 °C, respectively; at high food, no temperature effect. At 30 °C, low food had ~1.9× more abnormal larvae than high; at 26 °C, no food effect. At 18 dpf, abnormal proportions (7.1–79.4%) showed a significant temperature × pH × food interaction with complex patterns. - Developmental stage composition: At 11 dpf, early brachiolaria proportion was higher at 26 °C vs 30 °C (~2.1×), at pH 8.0 vs 7.6 (~2.6×), and in high food vs low/switch (~3.8–4.9×). At 18 dpf (pH 8.0 only), late brachiolaria proportion showed temperature × food interaction: at 26 °C, high food had ~3× more late brachiolaria than low; switch did not differ from low or high; at 30 °C, no food effect. - Survival (weeks 1–3): Significant temperature × weeks and pH × weeks interactions; at 3 weeks survival was greater at 30 °C vs 26 °C and at pH 8.0 vs 7.6; food had little effect on survival during weeks 1–3. In warm (30 °C) conditions with low food, all larvae experienced arrested development and died by week 6; with high/variable food, some completed development by 40 dpf. - Competency to settle: Competent late brachiolaria observed from 16 dpf in some treatments. The ambient 26 °C, pH 8.0, high-food treatment had the most replicates with competent larvae (6/7 at 16 dpf; mean surviving larvae/replicate 1108 ± 83 SE; 54.9% ± 6.8 SE late brachiolaria at 16 dpf). Acidified (pH 7.6), warmed (30 °C), and reduced food (low/switch) treatments had fewer replicates with competent larvae (<4/7) and took up to 2.4× longer to reach competency (28–35 dpf; 9–1022 surviving larvae/replicate). In 5/12 treatments, all larvae died before competency. Logistic regression (Model χ²(5) = 27.27, p < 0.00005; Nagelkerke R² = 0.42; 84.5% correct): competent larvae were ~143× less likely at 30 °C vs 26 °C; ~250× less likely at pH 7.6 vs 8.0 (β CI crossed zero; interpret cautiously). Low vs high food reduced odds by ~100×; switch vs high showed no significant difference. Temperature × pH interaction was in the best model but not significant (p = 0.071). Overall, warming slowed growth/development and increased abnormalities, effects mitigated by high food; acidification consistently slowed growth/development and increased abnormalities irrespective of food.

High phytoplankton food availability substantially improved CoTS larvae performance and resilience under warming, allowing some larvae at 30 °C to complete development and achieve competency, whereas low food at 30 °C led to developmental arrest and mortality by week 6. This aligns with energy-limited stress tolerance theory: warming elevates metabolic demand, and sufficient food maintains positive aerobic scope; inadequate food under heat stress precipitates failure. Acidification (pH 7.6) consistently slowed growth and development and increased abnormalities, and these effects were not alleviated by increased food, suggesting different physiological constraints (e.g., impaired digestion or acid–base regulation) that are less sensitive to external energy supply. Variable food (initial low then high) enabled compensatory growth and recovery in development rate relative to constant low food, with larvae reaching late brachiolaria at frequencies comparable to continuously high food under ambient pH, though with persistent size deficits (16–17% smaller at 18 dpf), indicating carry-over effects of early nutritional limitation. The study highlights that larval outcomes depend not only on mean food availability but also on its temporal pattern, with implications for bloom-driven outbreak hypotheses: short phytoplankton pulses may partially rescue cohorts experiencing early scarcity, especially under warming. Methodological differences among studies (flow-through vs static systems, pulse vs constant feeding) may underlie variability in reported size responses, pointing to the need to relate laboratory rations to natural phytoplankton dynamics. Overall, food availability modulates sensitivity to thermal stress near upper limits but has limited capacity to buffer acidification impacts at the levels tested.

This work demonstrates that phytoplankton food availability critically shapes larval CoTS responses to ocean warming: high or initially low then increased food supports development and competency under elevated temperature, whereas sustained low food under warming leads to arrested development and mortality. In contrast, acidification to pH 7.6 slows growth and development and increases abnormalities regardless of food regime, indicating limited mitigation by nutrition at this pH. These findings refine outbreak hypotheses by emphasizing that both the abundance and temporal variability of phytoplankton can determine larval success in warming tropical seas. Future research should: (1) quantify how natural, episodic phytoplankton pulses (magnitude, duration, timing) translate to larval feeding under realistic flow; (2) test broader pH/temperature ranges and interactions with parental conditioning; (3) compare pulse versus continuous feeding regimes and multi-species algal diets; and (4) assess post-settlement consequences of early nutritional histories to link larval performance with juvenile survival.

- Some datasets were non-normal (e.g., abnormal proportions, survival weeks 1–3), and transformations had little effect; results should be interpreted with this in mind. - High mortality and elevated abnormality in pH 7.6 treatments at 18 dpf led to low replication (n < 3) and omission of size and developmental data for those cells; late brachiolaria at 18 dpf analyzed only for pH 8.0. - The logistic regression indicated a strong negative pH effect on competency odds, but the 95% BCa CI for the β crossed zero, warranting caution. - Only one microalgal species and specific ration levels were tested under laboratory flow-through conditions, which may not capture natural diet diversity, feeding dynamics, or environmental variability. - Parental environmental histories were not manipulated here, though they can influence larval responses in CoTS.