Coral Reefs

, Volume 29, Issue 3, pp 593–605 | Cite as

Three lines of evidence to link outbreaks of the crown-of-thorns seastar Acanthaster planci to the release of larval food limitation

Report

Abstract

Population outbreaks of the coral-eating crown-of-thorns seastar, Acanthaster planci, continue to kill more coral on Indo-Pacific coral reefs than other disturbances, but the causes of these outbreaks have not been resolved. In this study, we combine (1) results from laboratory experiments where larvae were reared on natural phytoplankton, (2) large-scale and long-term field data of river floods, chlorophyll concentrations and A. planci outbreaks on the Great Barrier Reef (GBR), and (3) results from A. planci—coral population model simulations that investigated the relationship between the frequency of outbreaks and larval food availability. The experiments show that the odds of A. planci larvae completing development increases ~8-fold with every doubling of chlorophyll concentrations up to 3 μg l−1. Field data and the population model show that river floods and regional differences in phytoplankton availability are strongly related to spatial and temporal patterns in A. planci outbreaks on the GBR. The model also shows that, given plausible historic increases in river nutrient loads over the last 200 years, the frequency of A. planci outbreaks on the GBR has likely increased from one in 50–80 years to one every 15 years, and that current coral cover of reefs in the central GBR may be 30–40% of its potential value. This study adds new and strong empirical support to the hypothesis that primary A. planci outbreaks are predominantly controlled by phytoplankton availability.

Keywords

Crown-of-thorns starfish Seastar Trophic limitation Great Barrier Reef Acanthaster planci Eutrophication Phytoplankton Chlorophyll 

Introduction

On most Indo-Pacific coral reefs, including the Great Barrier Reef (GBR), coral cover has been declining at rates of 0.2–1.5% per year since the 1960s (Bruno and Selig 2007). To date, predation of coral by the crown-of-thorns seastar (Acanthaster planci) accounts for a large proportion of the observed decline in coral cover on the GBR. Between 1985 and 1997, population outbreak of A. planci were observed on ~32% of monitored reefs on the GBR, with their coral cover averaging 9% 1 year after the outbreak, compared with a mean of 28% coral cover on reefs that had not experienced an outbreak in the same period (Lourey et al. 2000). These figures suggest a GBR-wide reduction in coral cover of 0.5% year−1 due to A. planci alone in this 12-year period. Population outbreaks of A. planci, i.e., the sudden emergence of a large population after a period of relative rarity (Moran 1986), were first recorded throughout the Indo-Pacific in the 1960s. The abrupt population increase by orders of magnitude from a small parent population is called a ‘primary outbreak’ (Birkeland and Lucas 1990), and questions concerning the cause(s) of such primary outbreaks, and whether or not human activities have changed their frequency, have to date remained unresolved. In contrast, secondary outbreaks are simply the consequence of the large numbers of gametes produced upstream by a primary outbreak or another secondary outbreak population. Such secondary outbreaks have been reconstructed using hydrodynamic models (Moran 1986; Dight et al. 1990).

There are many hypotheses that relate to the control of A. planci populations (reviewed in Birkeland and Lucas 1990; Brodie et al. 2005). Echinoderms that release large numbers of planktotrophic larvae such as A. planci have a propensity to population fluctuations (Uthicke et al. 2009), and some primary A. planci outbreaks have been recorded even on coral reefs that remained relatively unaltered by human activities (Birkeland 1982; Birkeland and Lucas 1990). However, the observed widespread decline in coral cover since the 1960s suggest that the recently observed frequency of primary outbreaks of once in ~15 years is unsustainable (Seymour and Bradbury 1999). Even before mass bleaching started to inflict additional mortality, most reefs were estimated to take 10–25 years for full recovery of coral cover, with 25% of reefs showing no signs of recovery from A. planci coral mortality (Lourey et al. 2000). Two hypotheses that specifically address the apparent increase in the frequency of primary outbreaks have been widely debated. They are: (1) the ‘terrestrial runoff hypothesis’ aka ‘larval starvation hypothesis’ that argues that nutrient-limited survival of the pelagic planktotrophic larvae of A. planci controls population outbreaks (Birkeland 1982; Lucas 1982; Brodie et al. 2005), and (2) the ‘predator removal hypothesis’, which postulates that more juveniles survive to maturity due to the removal of fish predators through human exploitation (reviewed in Birkeland and Lucas 1990). Both hypotheses are based more on circumstantial than empirical support.

The terrestrial runoff hypothesis has strong correlative evidence (reviewed in Brodie et al. 2005). More outbreaks occur on reefs near high Pacific islands or continental coasts from which terrestrial runoff occurs, compared to low atoll islands without terrestrial runoff, and most outbreaks follow large or drought-breaking floods that carry high nutrient and sediment loads (Birkeland 1982). The apparent increase in A. planci outbreak frequencies is attributed to increased nutrient levels resulting from the terrestrial runoff of fertilizers, sewage and eroding soils, as recorded throughout the Indo-Pacific in modern times (Brodie et al. 2005). The planktotrophic larvae of A. planci feed on nano- and microphytoplankton (>3 μm cells) (Okaji et al. 1997) that multiply at high nutrient levels. Earlier feeding experiments with cultured microalgae showed that larval development was optimal at 2–6.5 μg l−1 chlorophyll, while few larvae completed their development at <0.6 μg l−1 chlorophyll (Lucas 1982). However, the types of cultured microalgae also determined the developmental success of A. planci larvae (Lucas 1982), limiting inferences from these experiments about larval development in natural phytoplankton communities.

The predator removal hypothesis states that seastar populations are largely controlled by predation, and that increased human exploitation of fish predators has resulted in increased numbers of seastars surviving to maturity (McCallum 1987). Dulvy et al. (2004) have suggested a relationship between A. planci outbreaks and human population density (but not fish predator densities) on 7 of 13 investigated reefs in Fiji, and Sweatman (2008) has shown a relationship between reef protection status and A. planci outbreaks on pairs of reefs open and closed to fishing in the GBR. However, the more commonly fished large predatory fish do not usually prey upon A. planci (Sweatman 1995) and to date, no fish predator has been identified that can effectively regulate A. planci populations, although predation on juveniles undoubtedly occurs (Keesing et al. 1996). It is argued (but difficult to show empirically) that the removal of large predators may (1) suppress prey switching behavior, i.e., the fewer large predatory fish stop eating the less preferred A. planci, and/or (2) have caused some complex trophic cascades, eventually resulting in fewer bottom-dwelling invertebrates that eat juvenile seastar (Keesing et al. 1996; Sweatman 2008).

In this study, we re-examine the contribution of terrestrial runoff to limiting the frequency of primary outbreaks, by combining new empirical evidence from A. planci larval feeding experiments, large-scale and long-term river flood and chlorophyll monitoring data for the GBR, and model-based simulations of the population dynamics of A. planci and its main drivers. Extensive long-term and large-scale coral reef, water quality and river monitoring data are now available for the GBR (Brodie et al. 2007; Sweatman et al. 2008), thereby providing a new opportunity to resolve the A. planci issue that affects the health of many Indo-Pacific coral reefs.

Methods

Laboratory experiments

Eight separate laboratory experiments were used to quantify rates of development and survival of larvae reared at different concentrations of natural phytoplankton as detailed in Okaji (1996). Seawater was collected from the ocean and filtered through a 25-μm mesh to remove large zooplankton and detrital matter. This coarsely filtered seawater was used to create fresh batches daily of the following treatments:
  1. (a)

    0.45-FSW: filtration through a 0.45-μm GF/B filter;

     
  2. (b)

    2-FSW: filtration through a 2-μm polycarbonate membrane filter;

     
  3. (c)

    25-FSW: no further treatment of the coarsely filtered seawater;

     
  4. (d)
    NES: nutrient-enriched seawater: 2 ml of Guillard’s f/2 nutrient solution (Guillard 1975) were added to 20 l of freshly collected and coarsely filtered seawater. This nutrient-enriched seawater was incubated in a 500-l water bath outdoors without shading for 2 days prior to use to develop the phytoplankton communities. Nutrient enrichment increased the concentration of eukaryotic phytoplankton cells and the concentration of chlorophyll on average 20-fold in NES compared with 25-FSW, while the concentration of cyanobacteria cells increased to a lesser extent (2- to 3-fold, Table 1).
    Table 1

    Mean percentage of surviving A. planci larvae that completed their development (late brachiolarian stage or metamorphosed to juveniles) at age 22 days, in treatments with contrasting concentrations of natural phytoplankton

    Experiment

    Treatment

    Chl. a (μg l−1 ± SD)

    Eukaryote density (103 ml−1 ± SD)

    Cyanobacteria density (103 ml−1 ± SD)

    Completion (% of survivors ± SD)

    1

    0.45-FSW

    0.07 ± 0.03

    (Not detected)

    25 ± 16

    0 ± 0

    1

    2-FSW

    0.17 ± 0.10

    0.004 ± 0.006

    55 ± 36

    0 ± 0

    1

    25-FSW

    0.40 ± 0.20

    0.214 ± 0.112

    65 ± 36

    0 ± 0

    2

    0.45-FSW

    0.08 ± 0.03

    (Not detected)

    31 ± 16

    0 ± 0

    2

    2-FSW

    0.25 ± 0.11

    0.004 ± 0.004

    75 ± 34

    0 ± 0

    2

    25-FSW

    0.52 ± 0.21

    0.234 ± 0.086

    83 ± 30

    0 ± 0

    3

    2-FSW

    0.08 ± 0.03

    0.163 ± 0.125

    6.4 ± 5.9

    0 ± 0

    3

    25-FSW

    0.29 ± 0.10

    0.437 ± 0.222

    7.1 ± 6.1

    18.7 ± 8.5

    4

    25-FSW

    0.28 ± 0.08

    0.385 ± 0.178

    6.7 ± 5.1

    0 ± 0

    5

    2-FSW

    0.19 ± 0.10

    0.004 ± 0.002

    56 ± 40

    0 ± 0

    5

    25-FSW

    0.28 ± 0.10

    0.207 ± 0.077

    62 ± 41

    88.3 ± 8.6

    5

    50% NES

    2.91 ± 1.35

    2.435 ± 0.564

    142 ± 90

    100 ± 0

    5

    100% NES

    5.25 ± 2.32

    4.441 ± 0.989

    202 ± 157

    100 ± 0

    6

    2-FSW

    0.19 ± 0.10

    0.004 ± 0.002

    56 ± 40

    0 ± 0

    6

    25-FSW

    0.28 ± 0.10

    0.207 ± 0.077

    62 ± 41

    97.2 ± 1.7

    6

    50% NES

    2.91 ± 1.35

    2.435 ± 0.564

    142 ± 90

    99 ± 0.7

    6

    NES

    5.25 ± 2.32

    4.441 ± 0.989

    202 ± 157

    100 ± 0

    7

    NES

    0.10

      

    0 ± 0

    7

    NES

    0.20

      

    0 ± 0

    7

    NES

    0.40

      

    0 ± 0

    7

    NES

    0.80

      

    32.2 ± 0.5

    7

    NES

    1.60

      

    50.2 ± 26.6

    8

    NES

    0.01

      

    0 ± 0

    8

    NES

    0.25

      

    0 ± 0

    8

    NES

    0.50

      

    6.8 ± 0.9

    8

    NES

    0.75

      

    38.6 ± 4.1

    8

    NES

    1.00

      

    61.6 ± 4.4

    Treatments: 0.45-FSW, 2-FSE, and 25 FSE seawater filtered using 0.45, 2 and 25-μm filters, respectively; NES nutrient-enriched seawater

     

The treatment levels of the 8 experiments are listed in Table 1. Each treatment was done in triplicates for Experiments 1–6, and in duplicates for Experiments 7 and 8. Experiments 1, 2, 5 and 6 were conducted at Lizard Island Research Station in the northern Great Barrier Reef (GBR), which is surrounded by clear offshore waters. Experiments 3 and 4 were conducted at the University of the Ryukyus (Okinawa, Japan), using water sampled from the front of Chatan Reef, Okinawa, from often clear coastal waters. Experiments 7 and 8 were conducted at the Australian Institute of Marine Science (central GBR), with the often turbid coastal seawater collected off Cape Bowling Green once a week and stored outdoors in aerated 500-l tanks. For Experiments 1–4, duplicate seawater samples were taken from each container every day, and for Experiments 5 and 6 every second day. The chlorophyll a concentration of these samples was determined using fluorometry, and the densities of eukaryotic algal and cyanobacteria cells were counted with an epifluorescence microscope (Table 1). In Experiments 7 and 8, chlorophyll a of the NES was determined fluorometrically before use, and NES was diluted with 0.45-FSW to obtain the final chlorophyll concentrations. The use of total chlorophyll a concentrations overestimates food availability for A. planci, because the contribution of nano- and microphytoplankton, the preferred food of A. planci larvae, is typically <50% of chlorophyll a in Indo-Pacific waters (Charpy and Blanchot 1999; Crosbie and Furnas 2001). The rest is picoplankton (<3 μm cells) that constitutes <10% of the diet of A. planci larvae (Okaji et al. 1997). However, the use of natural phytoplankton communities in our experiments, in which the relative contribution of nano- and micro-phytoplankton to chlorophyll reflects that found in the field, enabled us to relate the experimental results to the GBR long-term chlorophyll data.

Batches of A. planci larvae were reared in the laboratory (Okaji 1996). Actively swimming and healthy early bipinnaria larvae with fully developed alimentary canal were collected 2 days after the in vitro fertilization of gametes, and 100 larvae (150 larvae in Experiments 7 and 8) were added to each treatment chamber (1 l volume for Experiments 1 and 2; 2 l for all others), which were gently aerated and kept in the temperature range 26.5–29°C. Larvae were sieved with a 60-μm mesh and transferred to a clean set of containers of freshly prepared seawater every day. Every second to forth day, larvae were individually examined under a dissecting microscope, and their developmental stages recorded following Lucas (1982). For Experiments 7 and 8, the body lengths of 10–20 randomly selected larvae per chamber were measured along their longest axes every fourth day with an ocular micrometer. When the first larvae reached brachiolaria stage, aeration was reduced and a few small chips of crustose coralline algae were introduced as settlement substratum. These chips were checked daily in order to count metamorphosing larvae and settled juveniles.

The diameters of juvenile seastar after completion of metamorphosis were measured with an ocular micrometer in Experiment 5. The rate of successful completion of development was defined as the proportion of surviving larvae that were either in late brachiolaria or juvenile stages after 22 days. These two stages were combined, since late brachiolaria larvae are competent to metamorphose to juveniles. Absolute survivorship was not analyzed because abnormal or regressed larvae can remain alive for extended periods of time without any prospect of further development (‘living ghosts’), and there were no rigorous criteria to distinguish these from healthy larvae. Clearly regressed or abnormal larvae were scored as bipinnaria.

For the statistical analyses, the results of all 8 experiments were combined (Table 1). The relationship between rate of successful completion of development after 22 days and chlorophyll concentration was analyzed using a generalized linear mixed model; a logistic regression model where the response was the proportion of successful developmental completion. Two severe outliers (88 and 97% survivorship in E5, 25-FSW and E6, 25-FSW; Table 1) were excluded from the final model. Runs in Experiment 2 that were terminated after 18 days due to a lack of development in all treatments were scored as ‘zero completion’. All analyses were done using the statistical software package R (R Development Core Team 2009).

GBR flood history and chlorophyll data

Data of the cumulative discharge volumes of the Burdekin River since 1922 and the five largest Wet Tropics rivers (Herbert, Tully, Johnstone, Russell and Barron Rivers, latitude 16.5°–18.5°S) were obtained from the Queensland Department of Environment and Resource Management, and the values for the Burdekin River plotted for each ‘water year’ (1st October–30th September). The Burdekin River is the largest river entering into the GBR lagoon, and the most important factor determining inter-annual variability in flood plumes, since the annual discharge from its dry subtropical catchment varies by two orders of magnitude between its wettest and driest years (long-term annual mean discharge: 8.5 km3 year−1; CV = 106% of annual mean). In contrast, discharges from the many annually flooding rivers in Wet Tropics catchments, which jointly supply ~40% of the total annual runoff to the GBR (Furnas 2003), vary less between years (CV = 34–37% of annual means for Tully, Johnstone and Russell River, and 75–76% for the Herbert and Barron Rivers). Continuous monitoring of the Burdekin River started in 1922, and of the Wet Tropics Rivers between 1967 and 1983.

Data of summer chlorophyll a concentrations were extracted from the GBR long-term Chlorophyll Monitoring Program Data Base, which has sampled chlorophyll concentrations along fixed transects across the continental shelf monthly since 1993 (Brodie et al. 2007). Most A. planci on the GBR spawn in December to January (Lucas 1973; Babcock and Mundy 1992). Under experimental conditions, the duration of the pelagic larval phase is 12–22 days for well-fed larvae (Birkeland and Lucas 1990), and >50 days for larvae reared in filtered seawater (Lucas 1982). Thus, chlorophyll records from November to March were used to quantify long-term average summer values for the far northern (FN, 12.0°–15.0°S) and central/northern regions (CN, 15.1°–19.2°S). The data from each region were further split into the inner <25 km of the shelf (containing inshore and midshelf reefs in CN and FN) and outer locations, as most small to medium-sized river flood plumes remain within the inner 25 km of the shelf and travel along the coast toward the north due to the prevailing hydrodynamic patterns and Coriolis forcing (King et al. 2001; Devlin and Brodie 2005). Differences in the probabilities of larvae completing their development were then calculated for FN and CN based on the estimated differences in larval survival rates for given chlorophyll levels using the response curve from the laboratory experiments and long-term average chlorophyll concentrations in the GBR.

The A. planci: coral simulation model

The model was used to simulate spatial–temporal distributions of A. planci on the GBR. By varying the drivers and parameters of the model, running various scenarios and conducting sensitivity analyses, we investigated the temporal dynamics and the spatial patterns of the outbreaks. The simulation model comprised four linked sub-models:
  1. (1)
    The A. planci model: This is an age-structured meta-population model with two juvenile and six adult stages, each of 1 year duration. Life history parameters included age-dependent size, rates of survival across age cohorts, age-dependent fertility and feeding rates on corals (Table 2; Moran 1986; Birkeland and Lucas 1990; Scandol 1993, 1999).
    Table 2

    Model parameters for the A. planci—coral simulation model

    Life stage

    Survival

    Fecundity

    Coral consumed

    J1

    0.02

    0

    0.01

    J2

    0.1

    0.1

    1.3

    A1

    0.25

    0.3

    3.7

    A2

    0.5

    0.5

    7.3

    A3

    0.6

    0.7

    11

    A4

    0.6

    0.7

    11

    A5

    0.6

    0.7

    11

    The values were based on Scandol (1993). The temporal dynamics and patterns of the model results were relatively insensitive to variation of these parameters. Abbreviations: J1, J2: juveniles aged 1 and 2 years, A1–A5: adults aged 3–7 years

     
  2. (2)

    The coral model: The parameters included the rates of growth of coral as the prey of juvenile and adult A. planci (Scandol 1993; Sweatman et al. 2001; Wolanski and De’ath 2005).

     
  3. (3)

    The chlorophyll model: The spatial–temporal variation in chlorophyll was based on data from the GBR Long-Term Chlorophyll Monitoring Program (Brodie et al. 2007). Compared to long-term change, temporal variation was large on seasonal and short time scales. Both observed chlorophyll data (i.e., field observations) as well as simulated data including gradients and spikes originating from floods were used in the model.

     
  4. (4)

    The connectivity model: The connectivity of a reef to other reefs determines its capacity to provide larvae to itself (i.e., to self-seed), to other reefs (i.e., be a source), and to receive larvae from other reefs (i.e., be a sink). Hydrodynamic models provided estimates of the self-seeding, source and sink levels for 321 reefs in the central and northern GBR (James et al. 2002).

     

A justifiable criticism of population modeling is that, given enough parameters, one can reproduce most observed data. We have safeguarded against this problem by (1) focusing on relative not absolute effects (determining the ratios and 90% confidence ranges of values for A. planci and corals for FN and CN), and (2) using sensitivity analyses. The robustness of the model assumptions were tested by adding various levels of stochastic noise to any of the A. planci and coral life history variables, and by varying many of the model parameters within reasonable ranges (e.g., variations in the rates of coral replenishment and consumption, and juvenile and adult survival and fecundity). The models and all analyses of outputs were programmed in the statistical software package R (R Development Core Team 2009).

Results

Laboratory experiments

The laboratory experiments, in which freshly hatched A. planci larvae were reared in seawater at 0.01–5.25 μg l−1 chlorophyll a, showed that the proportion of larvae completing their development increased rapidly with increasing natural phytoplankton concentration (Fig. 1). At 0.01–0.25 μg l−1 chlorophyll, few larvae developed from the bipinnaria to early brachiolaria stage, none developed beyond the early brachiolaria stage, and most regressed at days 10–14. The odds of completion of development increased by a factor 8.3 (95% CI = 4.7, 17.7) for each doubling of concentrations of chlorophyll (Fig. 2a; Table 1). At low to moderate chlorophyll concentrations (<0.5 μg l−1), this was equivalent to increasing the probability of completing development by a factor 7–8 for each doubling of chlorophyll. At higher concentrations, the rate of increase in the probability of completion slowed, and plateaued at >3 μg l−1 where completion was certain.
Fig. 1

Developmental success of A. planci larvae that were exposed to different phytoplankton concentrations in 8 experiments (E1–E8). Larval developmental stages: white bipinnaria, wide diagonal hatch early brachiolaria, narrow diagonal hatch mid brachiolaria, gray shade late brachiolaria, black metamorphosed juvenile seastar. The latter two stages together were scored as ‘completed development’. 0.45-FSW, 2-FSE and 25 FSE seawater filtered using 0.45, 2 and 25-μm filters; NES nutrient enriched seawater

Fig. 2

a Relationship between chlorophyll a concentration and the proportion of A. planci larvae completing their development. b Body length of A. planci larvae at 17–20 days of age. Each point represents the mean results of duplicate or triplicate deployments per treatment. Black lines are model fits, the thin black lines are 2 SE of the mean

Growth rates, developmental speed and final body sizes of the larvae and early seastar also depended on the availability of phytoplankton. Larvae reared at ≥0.8 μg l−1 chlorophyll reached the maximum observed size of 1.2–1.3 mm at 17–20 days of age, suggesting growth was not food limited (Fig. 2b). At <0.5 μg l−1 chlorophyll, larvae initially grew from 0.8 to 1.0 mm but growth arrested after days 12–15 (Fig. 3). The time for 50% of surviving larvae to complete development decreased from 41 to 14 days when chlorophyll increased from 0.5 to >2 μg l−1 (not shown). Similarly, larvae that developed at 0.28 μg l−1 chlorophyll metamorphosed into seastars with significantly smaller mean diameter (0.44 mm, SE = 0.07 mm) than larvae reared at 2.9 or 5.2 μg l−1 chlorophyll (0.66 mm, SE = 0.05 mm; 0.64 mm, SE = 0.09 mm).
Fig. 3

Patterns of growth of A. planci larvae at increasing concentrations of chlorophyll a (Experiments 7 and 8). Shrinkage observable around days 16–20 reflects contraction for metamorphosis

Combining the data on rates of developmental completion and growth suggests the following chlorophyll thresholds. At <0.25 μg l−1 (<220 eukaryotic cells ml−1; Table 1), a negligible proportion of larvae complete development, suggesting starvation. At 0.25–0.8 μg l−1 (220–670 eukaryotic cells ml−1), this proportion is moderate, but development is slow and body sizes of larvae and juveniles remain small, suggesting severe food limitation. Finally, at >2 μg l−1 (>1700 eukaryotic cells ml−1), larval developmental success is high, developmental speed is fast, and both larvae and juveniles grow to their maximum observed size, suggesting release from trophic limitation.

Temporal and spatial correlations between chlorophyll availability and A. planci primary outbreaks

The patterns in Burdekin River discharges showed strong temporal and spatial agreement with the timing and location of primary outbreaks of A. planci in the GBR. On the GBR, primary outbreaks were first observed at 16.75°S in 1962 and 1979, and between 14.7° and 16.1°S in 1993/94 (Moran et al. 1992; Miller 2002; Sweatman 2008). The three largest recorded floods of the Burdekin River yielded freshwater discharges of 28, 54 and 40 km3 in 1958, 1974 and 1991 (Fig. 4). In 1974, all Wet Tropics rivers except the Herbert also produced >90th percentile floods, and in 1991, the Herbert and Barron produced >90th percentile floods and the other three rivers were above median levels. Therefore, the 1979 and 1994 outbreaks occurred three to 5 years after the two wettest years on record. The 1962 outbreak is difficult to interpret since the 1958 flood occurred in February–March and hence may have been too late in the season to feed A. planci larvae, and because few data exist from the Wet Tropics rivers. A large flood also occurred very early in the 1950/51 wet season, but no A. planci data exist from that period (Fig. 4). The region north of Latitude 16.75°S is the only section of the whole GBR where the dense matrix of large mid- and outer-shelf reefs such as Green Island frequently encounters river plumes (Fig. 5; Devlin and Brodie 2005; Brodie et al. 2005). A satellite image from a moderate flood event on the central and northern GBR (Fig. 5) illustrates: (1) the Burdekin plume extending >200 km to the north where it merges with the plumes from the Herbert and many Wet Tropics rivers; (2) the plumes intersect mid- and outer-shelf reefs around latitude 16°–17°S due to offshore diversion by the Cape Grafton headland and a narrow continental shelf; and (3) the plume waters do not intersect with any large reefs elsewhere as the remaining reef tract is too far offshore and all inshore reefs are very small.
Fig. 4

Cumulative discharge volumes of the Burdekin River into the GBR for each year since 1922. Red lines indicate the three large floods that preceded the three recorded primary outbreaks of A. planci in 1966, 1979 and 1994. The dark gray line shows an early large flood in 1951, but no data exist from that period. The blue lines show the large 2008 and 2009 Burdekin floods, potentially predicting the onset of a fourth primary outbreak

Fig. 5

Satellite image of the central GBR (Modis, 10th February 2007), also showing the locations of the mouths of the main rivers, and towns (filled square). All inshore reefs, and the mid- and outer-shelf reefs north of latitude 17°S (the presumed location of source reefs for primary A. planci outbreaks on the GBR, red box) are inundated by flood waters from the merged plumes of several rivers, while the remaining mid- and outer-shelf reefs are not intercepted by the flood plumes during this moderate flood event

Strong regional differences in the long-term average summer chlorophyll concentrations are also apparent on the GBR. Along the inner 25 km of the GBR, chlorophyll values were on average twice as high in the central/northern GBR (CN) compared to the far northern GBR (FN) (0.54 vs. 0.26 μg l−1; Table 3). Assuming our experimental results were indicative of food limitation in the field, this ~2-fold difference in chlorophyll concentrations between CN and FN would translate into an ~8-fold higher rate of successful larval development in the former. Additionally, levels of chlorophyll exceeding 0.5 μg l−1 occur for ~37% of summer values in the inner CN, compared to 5.7–6.4% in the remaining sectors.
Table 3

Summer chlorophyll a concentrations on the inner (<25 km off the coast) and outer section of the continental shelf in the far northern (FN, latitude 12°–15.0°S) and central/northern (CN, latitude 15.1°–19.2°S) regions of the Great Barrier Reef (Brodie et al. 2007)

Shelf

Region

N

Mean (μg l−1 ± SE)

Median (μg l−1)

<0.25 μg l−1 (%)

>0.5 μg l−1 (%)

>0.8 μg l−1 (%)

Inner 25 km

FN

104

0.26 ± 0.01

0.25

50.0

5.8

0.0

CN

619

0.54 ± 0.02

0.38

32.0

37.0

17.8

Offshore

FN

235

0.27 ± 0.01

0.25

49.4

6.4

0.004

CN

352

0.24 ± 0.01

0.19

63.9

5.7

2.8

N = number of samples. Shown are means, medians, and the percentage of water samples with chlorophyll concentrations below 0.25 μg l−1 and exceeding 0.5 and 0.8 μg l−1

A re-analysis of AIMS Long-Term Monitoring Program data (Sweatman et al. 2008) of A. planci outbreaks and coral cover shows that in the period 1985–2007, 12.9% ± 1.7% (SE) of reefs in CN were in a state of ‘active or incipient outbreak’ at anyone time, and coral cover averaged 16.5% ± 0.8%. In contrast, in FN only 5.5% ± 0.01% of reefs had A. planci outbreaks, no outbreak waves have been observed, and coral cover averaged 28.0% ± 1.0%.

A. planci: coral spatial–temporal simulation model

We used the A. planci—coral spatial–temporal simulation model to investigate and quantify the relationships between inshore chlorophyll, seastar populations and coral cover (Figs. 6, 7; Table 2). The two principal drivers of the A. planci populations were both food-resource related and comprised the following:
Fig. 6

Relationship of Acanthaster planci population dynamics and chlorophyll in the Great Barrier Reef (GBR) off the NE of Australia. a Map of the GBR. b Long-term average chlorophyll concentrations in the GBR in the far northern (FN, blue) and central/northern (CN, red) region, monitored near-monthly since 1992. Applying the results from the laboratory experiments c showed that the odds for survival of A. planci larvae was ~ 8-fold higher at chlorophyll levels found in CN compared with FN. Simulations of A. planci and coral population dynamics show that in FN (d), outbreaks occur at 50–80-year intervals and coral cover recovers between outbreaks (Table 4). In CN (e), outbreaks occur at 15-year intervals and corals only recover to 30–40% of potentially obtainable values. These data form the basis to model the transition (f) in chlorophyll, A. planci and coral cover in CN from pre-European (blue) to contemporary levels (red)

Fig. 7

Simulation results from the population model shows the averaged effects of varying levels of mean chlorophyll on relative population sizes of a larvae, b juveniles and c adults, and on d the frequency of seastar outbreaks, e the percent of coral eaten and f coral cover. Results are averages over model runs spanning 200 years and excluding a 20-years ‘burn-in’ period

  1. (1)

    The concentration of chlorophyll in the water column which determines the probability of A. planci larvae to survive until settlement and metamorphosis;

     
  2. (2)

    The availability of hard coral for consumption by the juvenile and adult A. planci.

     

The former is governed by the empirical relationship between larval survival and concentrations of chlorophyll in the water (Fig. 6a–c). At low levels of chlorophyll, the larval survival was low and thus A. planci populations remained low and coral cover was high. Conversely, at high levels of chlorophyll, the abundant larvae led to large populations of A. planci adults that can deplete the hard coral cover within a few years.

The model was used to compare the CN and FN regions using observed water quality data to drive the populations (Fig. 6d–f). Averaged over many generations, with distributions of chlorophyll concentrations reflecting those in the CN inshore region (Table 3, mean chlorophyll = 0.54 μg l−1), the model population formed outbreaks at 12–15 year intervals, consistent with present-day outbreak frequencies and intensities in CN (Fig. 6e; Table 4). Coral recovery between outbreaks remained incomplete, with coral cover averaging 20–28% of typical maximum values. At the chlorophyll distributions recorded in the FN (mean chlorophyll = 0.26 μg l−1), adult and juvenile seastar densities were 0.04–0.25 and 0.25–0.63 of CN densities respectively, outbreaks occurred only once in 50–80 years, and coral cover recovered to 75–90% of typical maximum values between outbreaks (Fig. 6d; Table 4). Taking the relatively pristine FN as reflective of conditions ~150 years ago, a potential transition from pristine to contemporary outbreak conditions in CN becomes apparent, demonstrating increasing outbreak frequencies and progressively declining coral cover (Fig. 6f).
Table 4

Results of simulation runs of the single-reef A. planci—coral simulation model (Fig. 6)

 

CN (Figs. 6b, e)

FN (Figs. 6b, d)

CN/FN

Adult A. planci (mean relative density)

40–63

16–25

1.8–3.2

Juvenile A. planci (mean relative density)

4,000–10,000

400–1,000

6–15

Coral Cover (mean % of typical maximum)

20–28

75–90

0.25–0.35

Interval between outbreaks (yrs)

12–15

50–80

0.13–0.22

The far northern region (FN, latitude 12.0°–15.0°S) has little agriculture and a low human population density, whereas the central/northern GBR (CN, latitude 15.1°–19.2°S) experiences elevated nutrient loads from rivers. Ratios between the two contrasting regions rather than absolute values and sensitivity analyses were used to overcome the effects of model assumptions

The model was also used to investigate the characteristics of A. planci, coral cover and the outbreak frequencies and intensities (Fig. 7a–f). In terms of the model, outbreaks were defined as events that reduced the coral cover to <2% and lead to mass mortality of seastars. The age structure of the A. planci populations changed with increasing chlorophyll levels (Fig. 7a–c). As chlorophyll increased, population sizes increased, and the age structure shifted to relatively more young seastars. Outbreaks only occurred at chlorophyll levels above ~0.25 μg l−1 and seastar populations were <10% of their maximum levels (Fig. 7d). At chlorophyll levels >0.5 μg l−1, coral cover declined by 75% of the initial cover and was dominated by young corals with slower growth rates, the rate of growth in the coral population slowed, and less coral was consumed (Fig. 7e–f). The maximum frequency of the outbreak waves was one cycle per 10–15 years, and was controlled by the rate of coral recovery, with slower rates resulting in lower outbreak frequencies.

Finally, the data on reef connectivity and their source, sink and self-seeding levels were related to seastar population and outbreak intensities. We found that
  1. (1)

    Patterns of inter-reef connectivity had far less effect on the large-scale wave-like patterns of secondary outbreaks than differences in chlorophyll concentrations. Provided there is at least a low level of connectivity among reefs, further increases in the strength of connectivity had little effect on the outbreak patterns.

     
  2. (2)

    At the scale of individual reefs, the risk of a reef having a severe A. planci outbreak increased with its capacity to retain larvae through self-seeding and acting as a sink, and decreased with its capacity to reduce larvae by acting as a source. Thus, reefs that were the origin of outbreaks needed not outbreak themselves, and hence reefs where primary outbreaks are first observed may not be the source of the outbreak. Furthermore, reefs that were predominantly sources of larvae (i.e., had low larval retention) were 4 times less likely to outbreak than reefs that retained more of their larvae.

     

Discussion

Identifying the causes of ecological patterns and distinguishing anthropogenic changes from natural dynamics is exceedingly complex, but synthesis of information from various sources such as experiments, field surveys, long-term monitoring and simulation models can form a basis for attribution of causality (Fabricius and De’ath 2004). Using this approach, our study adds new and strong support to the hypothesis that food availability controls primary outbreaks of A. planci by enhancing the survival of larvae. It is important to differentiate between the different processes that govern population dynamics on the three reef ‘types’ (source reefs, primary outbreak reefs, and secondary outbreak reefs). Primary outbreaks can arise from small populations living on source reefs that encounter highly productive waters during spawning times. Hydrodynamics dictate that source reefs may be located either upstream of the primary outbreak reefs, or become primary outbreak themselves reefs through self-seeding. However, both source and primary outbreak reefs are likely to experience seasonal phytoplankton blooms as the larvae together with the phytoplankton move with the currents. Following these primary outbreaks, secondary outbreaks are then observed on individual reefs in downstream progressing waves (Moran 1986; Moran et al. 1992; Sweatman et al. 2008) that can be reconstructed using hydrodynamic models (Dight et al. 1990, our A. planci model). Primary outbreaks therefore develop from a small source population of starfish, with each mature starfish producing an extremely large number of offspring due to the release of larval food limitation. In contrast, secondary outbreaks comprise a large population of mature starfish and hence can sustain outbreaks in conditions where larval survival is relatively low.

Our larval feeding experiments showed a strong and non-linear (logistic) dose–response relationship between the availability of natural eukaryotic phytoplankton and larval development success. By using unfiltered natural seawater, the larvae in our experiments grew in conditions where both phytoplankton food and potential planktonic predator communities underwent natural successions in response to nutrient availability. We demonstrated larval food availability to be a strong driver, with low chlorophyll leading to a low rate of developmental completion, a prolonged pelagic phase of the larvae and small sizes of the post-metamorphosis juveniles, likely leading to higher pre- and post-settlement mortality (Allison 1994). The role of larval food limitation to echinoderm population dynamics in the field is also corroborated by the observation that only echinoderm groups with planktotrophic larvae have the propensity to exhibit boom and bust population dynamics, while echinoderms with non-feeding (lecitotrophic) larvae have more stable populations (Uthicke et al. 2009).

Identifying the origin of A. planci primary outbreaks is key to successful management of the GBR. The first two outbreaks were first detected at Green Island off Cairns (Moran et al. 1992), a major tourist destination that was more frequently visited than many other reefs near-by. The third outbreak was first reported from Lizard Island in October 1993 and 10–11 months later from 7 other reefs in the Cairns and Lizard Island regions (16013C, Evening Reef, Swinger, Startle, Mackay, North Direction and Macgillivray Reefs), with a number of additional reefs having A. planci densities only slightly below outbreak threshold levels in 1994 (LTMP data, not shown). The multiple outbreak locations, and the multiple A. planci size classes (including juveniles) observed at Lizard Island in 1996 (Pratchett 2005) suggested a gradual population build-up in the whole region through several successful spawning events in prior years, indicating that not only the large 1991 flood but also conditions in following years provided conditions suitable for high recruitment success. Floods have reached or crossed this part of the shelf in 1991, 1994, 1995 and 1996 (Devlin and Brodie 2005), retention times of flood materials on the continental shelf may be long (Luick et al. 2007), and chlorophyll levels frequently exceed 0.5 and even 0.8 μg l−1 (Table 3). The weak and bidirectional currents may further increase the vulnerability of reefs in this region to develop outbreaks due to their relatively high rates of larval retention and self-seeding (James et al. 2002).

Long-term average chlorophyll values on the inner 25 km were twice as high in CN compared to FN. As offshore chlorophyll values were similar in both regions, it is unlikely that the high CN chlorophyll values were attributable to latitude or upwelling (Brodie et al. 2007). In CN, present river loads of nutrients and sediments are estimated to be 2–10 fold higher than before western colonization in ~1860, while river loads in the sparsely inhabited FN are considered largely unaltered (McKergow et al. 2005). As rivers are the main source of new nutrients to GBR inshore waters, regional differences in chlorophyll have been attributed to differences in river nutrient loads, reflecting past and present terrestrial runoff (Devlin and Brodie 2005; Brodie et al. 2007), although an explicit link between increasing river loads and changes to inshore water quality on the GBR has not been established.

The fact that primary A. planci outbreaks occurred on locations where floods intercepted large reefs on the GBR 3–5 years earlier shows that not only long-term chlorophyll concentrations but also large floods are a strong driver for A. planci primary outbreaks, in agreement with previous findings (Birkeland 1982; Brodie et al. 2005). Flood plume models (King et al. 2001) showed that the 1974 and 1991 floods reduced salinity for >60 days during the time when A. planci larvae are pelagic. Due to the high incidence of cloud cover during and after rain events, few satellite images are available to track the spread of flood plumes. However, Devlin and Brodie (2005) used aerial surveys to show a spreading of plumes into the main reef matrix, similar to the patterns shown in Fig. 5, after cyclones in 1996 and 1999, and to an even greater extent in 1994. These river floods are the largest source of new nutrients for the inshore GBR, and trigger phytoplankton blooms that average 2 μg l−1 chlorophyll and at times exceed 4 μg l−1 (Devlin and Brodie 2005). The application of experimental findings to field settings necessitates caution. Bearing this in mind, if we take the experimentally observed rates of change as indicative of the relative differences in developmental rates in the field, the odds of successful development of A. planci larvae could be up to ~60-fold higher during floods with 2.0 μg l−1 chlorophyll compared to the long-term average of 0.54 μg l−1 for the central/northern GBR.

In combination, the three components of this study, together with previous evidence (Birkeland 1982; Brodie et al. 2005), strongly support the assertion that removal of larval food limitation causes primary population outbreaks of A. planci on the GBR. In contrast to the parsimonious explanation of food limited control of A. planci populations, explanations of population control by predators rely on complex arguments (Birkeland and Lucas 1990). To date, there is no empirical evidence to support these arguments. Unlike water quality, predation pressure does not fluctuate widely on short time scales, hence it remains unclear how a chronic release from predation would occasionally lead to sudden increases in population densities. However, the reported correlation between reef protection status and A. planci outbreaks on a subset of GBR reefs (Sweatman 2008) suggests that both hypotheses may not necessarily be exclusive, and that predation may play some additive role in determining the propensity of individual reefs to be afflicted by A. planci outbreaks.

Coastal GBR water quality is considered amenable to benefit from improved land management (Haynes et al. 2007) due to the dominant role of the rivers in providing new nutrients to the inshore GBR, and the potentially long residency times of these newly imported materials (Luick et al. 2007). Large and early river floods from the Burdekin and Wet Tropics rivers occurred again in January 2008 and 2009 (Fig. 4), potentially providing conditions to trigger a new A. planci primary outbreak wave that may again kill a significant proportion of GBR corals. Our study suggests that reductions in phytoplankton biomass to summer values of <0.5 μg l−1 (De’ath and Fabricius 2010) through better land management could reduce the frequency of primary A. planci outbreaks. Legislation and incentives have now been put in place to reduce river discharges of nutrients, sediments and pesticides from agricultural areas. However until a reduction in nutrient levels is achieved, two additional precautionary management measures should aim to maintain very low A. planci densities in the high-risk area (the midshelf reefs and hard bottom inter-reefal areas that are directly intercepted by floods). These are: (a) large permanent fishing closures in the high-risk area, allowing fish populations to reach carrying capacity to safeguard against cascading changes in food webs, and (b) potentially some targeted efforts by divers, especially in the years following large floods, to remove some of the A. planci before they start to aggregate and spawn, This three-pronged approach constitutes the best presently available strategy to slow or reverse the loss of coral cover throughout the whole central and southern GBR, at a time where rising seawater temperatures exert increasing pressure on coral reefs, and increasing climatic instability may increase the frequency of extreme floods.

Notes

Acknowledgments

Many thanks to the GBR Long-Term Chlorophyll Monitoring Program for the chlorophyll data, to the AIMS Long-Term Monitoring Program for coral cover and A. planci field distribution data, to J. Scandol for compilation of the A. planci life history data, and to M. Slivkoff for the processing of the Modis satellite image. KO conducted the laboratory experiments, and GD developed the coral—A. planci simulation model. We gratefully acknowledge support for the experimental study by T. Ayukai, J. Lucas and J. Keesing. We thank J. Caley, B. Schaffelke, S. Uthicke and H. Sweatman for constructive comments on earlier versions of the manuscript, and J. Brodie, K. Day and E. Wolanski for sharing ideas. The study was funded by the Marine and Tropical Sciences Research Facility (MTSRF) and the Australian Institute of Marine Science, with the experimental study being funded by the Great Barrier Reef Marine Park Authority and James Cook University.

References

  1. Allison GW (1994) Effects of temporary starvation on larvae of the sea star Asterina miniata. Mar Biol 118:255–261CrossRefGoogle Scholar
  2. Babcock BC, Mundy CN (1992) Reproductive biology, spawning and field fertilisation rates of Acanthaster planci. Austr J Mar Freshw Res 43:525–534CrossRefGoogle Scholar
  3. Birkeland C (1982) Terrestrial runoff as a cause of outbreaks of Acanthaster planci (Echinodermata: Asteroidea). Mar Biol 69:175–185CrossRefGoogle Scholar
  4. Birkeland C, Lucas JS (1990) Acanthaster planci: major management problem of coral reefs. CRC Press, Boca RatonGoogle Scholar
  5. Brodie J, Fabricius K, De’ath G, Okaji K (2005) Are increased nutrient inputs responsible for more outbreaks of crown-of-thorns starfish? An appraisal of the evidence. Mar Pollut Bull 51:266–278CrossRefPubMedGoogle Scholar
  6. Brodie J, De’ath G, Devlin M, Furnas M, Wright M (2007) Spatial and temporal patterns of near-surface chlorophyll a in the Great Barrier Reef lagoon. Mar Freshw Res 58:342–353CrossRefGoogle Scholar
  7. Bruno J, Selig E (2007) Regional decline of coral cover in the Indo-Pacific: timing, extent, and subregional comparisons. PLoS ONE 2:e711. doi:710.1371/journal.pone.0000711 CrossRefPubMedGoogle Scholar
  8. Charpy L, Blanchot J (1999) Picophytoplankton biomass, community structure and productivity in the Great Astrolabe Lagoon, Fiji. Coral Reefs 18:255–262CrossRefGoogle Scholar
  9. Crosbie ND, Furnas MJ (2001) Abundance, distribution and flow-cytometric characterization of picophytoprokaryote populations in central (17 degree S) and southern (20 degree S) shelf waters of the Great Barrier Reef. J Plankton Res 23:809–828CrossRefGoogle Scholar
  10. De’ath G, Fabricius KE (2010) Water quality as a regional driver of coral biodiversity and macroalgae on the Great Barrier Reef. Ecol Appl (in press)Google Scholar
  11. Devlin MJ, Brodie J (2005) Terrestrial discharge into the Great Barrier Reef Lagoon: nutrient behavior in coastal waters. Mar Pollut Bull 51:9–21CrossRefPubMedGoogle Scholar
  12. Dight IJ, Bode L, James MK (1990) Modelling the larval dispersal of Acanthaster planci. 1. Large scale hydrodynamics, Cairns Section, Great Barrier Reef Marine Park. Coral Reefs 9:115–123CrossRefGoogle Scholar
  13. Dulvy N, Freckleton R, Polunin N (2004) Coral reef cascades and the indirect effects of predator removal by exploitation. Ecol Lett 7:410–416CrossRefGoogle Scholar
  14. Fabricius KE, De’ath G (2004) Identifying ecological change and its causes: a case study on coral reefs. Ecol Appl 14:1448–1465CrossRefGoogle Scholar
  15. Furnas MJ (2003) Catchments and corals: terrestrial runoff to the Great Barrier Reef. Australian Institute of Marine Science, CRC Reef. Townsville, AustraliaGoogle Scholar
  16. Guillard RRL (1975) Culture of phytoplankton for feeding marine invertebrates. In: Smith WL, Chanley MH (eds) Culture of marine invertebrate animals. Plenum Press, New York, pp 26–60Google Scholar
  17. Haynes D, Brodie J, Waterhouse J, Bainbridge Z, Bass D, Hart B (2007) Assessment of the water quality and ecosystem health of the Great Barrier Reef (Australia): conceptual models. Environ Manage 40:993–1003CrossRefPubMedGoogle Scholar
  18. James MK, Armsworth PR, Mason LB, Bode L (2002) The structure of reef fish metapopulations: modelling larval dispersal and retention patterns. Proc R Soc Lond B Biol Sci 269:2079–2086CrossRefGoogle Scholar
  19. Keesing JK, Wiedermeyer WL, Okaji K, Halford AR, Hall KC, Cartwright CM (1996) Mortality rates of juvenile starfish Acanthaster planci and Nardoa spp. measured on the Great Barrier Reef, Australia and in Okinawa, Japan. Oceanol Acta 19:441–448Google Scholar
  20. King B, McAllister F, Wolanski E, Done T, Spagnol S (2001) River plume dynamics in the central Great Barrier Reef. In: Wolanski E (ed) Oceanographic processes of coral reefs: physical and biological links in the Great Barrier Reef. CRC Press, Boca Raton, pp 145–160Google Scholar
  21. Lourey MJ, Ryan DAJ, Miller IR (2000) Rates of decline and recovery of coral cover on reefs impacted by, recovering from and unaffected by crown-of-thorns starfish Acanthaster planci: a regional perspective of the Great Barrier Reef. Mar Ecol Progr Ser 196:179–186CrossRefGoogle Scholar
  22. Lucas JS (1973) Reproductive and larval biology and Acanthaster planci (L.) in Great Barrier Reef. Micronesica 9:197–203Google Scholar
  23. Lucas JS (1982) Quantitative studies of feeding and nutrition during larval development of the coral reef asteroid Acanthaster planci (L.). J Exp Mar Biol Ecol 65:173–194CrossRefGoogle Scholar
  24. Luick JL, Mason L, Hardy T, Furnas MJ (2007) Circulation in the Great Barrier Reef Lagoon using numerical tracers and in situ data. Cont Shelf Res 27:757–778CrossRefGoogle Scholar
  25. McCallum HI (1987) Predator regulation of Acanthaster planci. J Theor Biol 127:207–220CrossRefGoogle Scholar
  26. McKergow L, Prosser I, Hughes A, Brodie J (2005) Sources of sediment to the Great Barrier Reef World Heritage Area. Mar Pollut Bull 51:200–211CrossRefPubMedGoogle Scholar
  27. Miller I (2002) Historical patterns and current trends in the broadscale distribution of crown-of-thorns starfish in the northern and central sections of the Great Barrier Reef. In: Proceedings of the 10th international Coral Reef Symposium, pp 1273–1279Google Scholar
  28. Moran PJ (1986) The Acanthaster phenomenon. Oceanogr Mar Biol Ann Rev 24:379–480Google Scholar
  29. Moran PJ, De’ath G, Baker VJ, Bass DK, Christie CA, Miller IR, Miller-Smith BA, Thompson AA (1992) Pattern of outbreaks of crown-of-thorns starfish (Acanthaster planci (L.)) along the Great Barrier Reef since 1996. Aust J Mar Freshw Res 42:555–568CrossRefGoogle Scholar
  30. Okaji K (1996) Feeding ecology in the early life stages of the crown-of-thorns starfish, Acanthaster planci (L.). PhD Thesis, James Cook University, p 121Google Scholar
  31. Okaji K, Ayukai T, Lucas JS (1997) Selective feeding by larvae of the crown-of-thorns starfish, Acanthaster planci (L.). Coral Reefs 16:47–50CrossRefGoogle Scholar
  32. Pratchett MS (2005) Dynamics of an outbreak population of Acanthaster planci at Lizard Island, northern Great Barrier Reef (1995–1999). Coral Reefs 24:453–462CrossRefGoogle Scholar
  33. R Development Core Team (2009) R: a language and environment for statistical computing. R Foundation for Statistical Computing, ViennaGoogle Scholar
  34. Scandol JP (1993) CotSim—scientific visualisation and gaming-simulation for the Acanthaster phenomenon. Great Barrier Reef Marine Park Authority, Townsville (Australia)Google Scholar
  35. Scandol JP (1999) CotSim—an interactive Acanthaster planci metapopulation model for the central Great Barrier Reef. Mar Models Online 1:1–4Google Scholar
  36. Seymour RM, Bradbury RH (1999) Lengthening reef recovery times from crown-of-thorns outbreaks signal systemic degradation of the Great Barrier Reef. Mar Ecol Prog Ser 176:1–10CrossRefGoogle Scholar
  37. Sweatman HPA (1995) A field study of fish predation on juvenile crown-of-thorns starfish. Coral Reefs 14:47–53CrossRefGoogle Scholar
  38. Sweatman H (2008) No-take reserves protect coral reefs from predatory starfish. Curr Biol 18:R598–R599CrossRefPubMedGoogle Scholar
  39. Sweatman H, Cheal A, Coleman G, Delean S, Fitzpatrick B, Miller I, Ninio R, Osborne K, Page C, Thompson A (2001) Long-term monitoring of the Great Barrier Reef: status report number 5. Australian Institute of Marine Science, Townsville 106 ppGoogle Scholar
  40. Sweatman H, Cheal A, Coleman N, Emslie M, Johns K, Jonker M, Miller I, Osborne K (2008) Long-term monitoring of the Great Barrier Reef. Australian Institute of Marine Science, Townsville 379 ppGoogle Scholar
  41. Uthicke S, Schaffelke B, Byrne M (2009) A boom–bust phylum? Ecological and evolutionary consequences of density variations in echinoderms. Ecol Monogr 79:3–24CrossRefGoogle Scholar
  42. Wolanski E, De’ath G (2005) Predicting the impact of present and future human land-use on the Great Barrier Reef. Estuar Coast Shelf Sci 64:504–508CrossRefGoogle Scholar

Copyright information

© Springer-Verlag 2010

Authors and Affiliations

  1. 1.Australian Institute of Marine ScienceTownsville MCAustralia
  2. 2.Coralquest Inc.Atsugi, KanagawaJapan

Personalised recommendations