ArticlePDF Available

Limits to understanding and managing outbreaks of crown-of- thorns starfish (ACANTHASTER Spp.)



Outbreaks of crown-of-thorns starfish (Acanthaster spp.) remain a major cause of coral mortality in the Indo-Pacific, contributing to widespread and accelerating degradation of coral reef environments. This review examines the evidence for and against the principal hypotheses put forward to explain spatial and temporal patterns of outbreaks and also explores whether it is possible or feasible to intervene and limit ongoing degradation caused by crown-of-thorns starfish. The inherent biological characteristics of Acanthaster spp., such as exceptional fecundity, early maturation, and extreme flexibility in resource use, clearly contribute to extreme fluctuations in their abundance. Of the many hypotheses put forward to explain the occurrence of outbreaks, none has universal or unequivocal support. Clearly, however, the high incidence and severity of outbreaks at many reef locations cannot be sustained because anthropogenic changes to marine environments either have caused fundamental shifts in the population dynamics of Acanthaster spp. or have undermined the capacity of reef ecosystems to withstand these periodic disturbances. Reducing the incidence or severity of outbreaks of Acanthaster spp. is critical for reversing widespread and protracted declines in coral cover throughout the Indo-Pacific. Improved efficiency of direct controls provides opportunities to limit the progression and spread of outbreaks if detected early, but effective management of Acanthaster spp. really depends on definitive knowledge and appropriate action to address the ultimate causes of outbreaks. There are considerable, but not insurmountable, challenges to addressing important and persistent knowledge gaps relating to the biology of Acanthaster spp. This research is fundamental to ensure the persistence of coral reef ecosystems, especially given other emerging threats associated with global climate change.
Oceanography and Marine Biology: An Annual Review, 2014, 52, 133-200
© Roger N. Hughes, David J. Hughes, and I. Philip Smith, Editors
Taylor & Francis
1ARC Centre of Excellence for Coral Reef Studies,
James Cook University, Townsville, QLD 4811, Australia
E- mail:
2Australian Institute of Marine Science, PMB 3, Townsville MC, Queensland 4810, Australia
Outbreaks of crown- of- thorns starsh (Acanthaster spp.) remain a major cause of coral mortality
in the Indo- Pacic, contributing to widespread and accelerating degradation of coral reef environ-
ments. This review examines the evidence for and against the principal hypotheses put forward to
explain spatial and temporal patterns of outbreaks and also explores whether it is possible or fea-
sible to intervene and limit ongoing degradation caused by crown- of- thorns starsh. The inherent
biological characteristics of Acanthaster spp., such as exceptional fecundity, early maturation, and
extreme exibility in resource use, clearly contribute to extreme uctuations in their abundance.
Of the many hypotheses put forward to explain the occurrence of outbreaks, none has universal or
unequivocal support. Clearly, however, the high incidence and severity of outbreaks at many reef
locations cannot be sustained because anthropogenic changes to marine environments either have
caused fundamental shifts in the population dynamics of Acanthaster spp. or have undermined the
capacity of reef ecosystems to withstand these periodic disturbances. Reducing the incidence or
severity of outbreaks of Acanthaster spp. is critical for reversing widespread and protracted declines
in coral cover throughout the Indo- Pacic. Improved efciency of direct controls provides opportu-
nities to limit the progression and spread of outbreaks if detected early, but effective management of
Acanthaster spp. really depends on denitive knowledge and appropriate action to address the ulti-
mate causes of outbreaks. There are considerable, but not insurmountable, challenges to addressing
important and persistent knowledge gaps relating to the biology of Acanthaster spp. This research
is fundamental to ensure the persistence of coral reef ecosystems, especially given other emerging
threats associated with global climate change.
Coral reefs are increasingly regarded as one of the world’s most threatened ecosystems. Not only
have reef ecosystems suffered a long history of degradation (e.g., Pandol et al. 2005), but also
climate change is expected to have a greater effect on coral reefs than almost any other ecosystem
(Walther et al. 2002, Hoegh- Guldberg & Bruno 2010). Anthropogenic degradation of coral reef
ecosystems began centuries ago with extensive exploitation and harvesting of large vertebrate spe-
cies (Pandol et al. 2005). More recently, there have been sustained declines in the abundance of
corals (e.g., Gardner et al. 2003, Bellwood et al. 2004, De’ath et al. 2012) and associated shifts in
the biological and physical structure of benthic habitats (Hughes et al. 2010). Across the Caribbean,
average coral cover declined from approximately 50% in 1977 to less than 10% in 2001, represent-
ing an average annual loss of 1.67% (Gardner et al. 2003). In the Indo- Pacic, average annual coral
loss was 1.05% between 1982 and 2003 and is accelerating (Bruno & Selig 2007). Approximately
19% of the worlds coral reefs have been effectively destroyed, meaning that more than 90% of coral
has been lost, and there is little prospect of recovery (Wilkinson 2008). Moreover, 35% of reefs face
a similar fate within the next 10–40years unless there is effective management action to halt or
reverse ongoing coral loss (Wilkinson 2008).
Scleractinian corals are fundamental to the geomorphology, biodiversity, and productivity of
coral reef ecosystems (e.g., Hoegh- Guldberg 2004, Wild et al. 2004, Pratchett et al. 2008, Stella
et al. 2011). Most important, corals contribute to both biological and physical habitat structure
(Pratchett et al. 2008), providing essential resources (food and shelter) for many reef organisms
(Jones et al. 2004, Cole et al. 2008, Stella et al. 2011); high diversity of distinct microhabitats
(e.g., Messmer et al. 2011); and habitat structure that mediates important biological interactions,
such as competition (e.g., Munday 2001, Holbrook & Schmitt 2002) and predation (e.g., Caley &
St John 1996, Beukers & Jones 1998, Coker et al. 2014). There is considerable correlative and exper-
imental evidence showing that reef locations with high cover and diversity of scleractinian corals
support greater abundance and diversity of coral reef organisms, especially shes (e.g., Carpenter
et al. 1982, Jones 1988, Holbrook et al. 2000, Munday 2000, Jones et al. 2004, Messmer et al. 2011).
Accordingly, extensive loss of live corals leads to marked declines in the abundance and diversity of
coral reef shes (e.g., Wilson et al. 2006, Pratchett et al. 2008, 2011). Effects of coral loss are even
more pronounced and affect a greater diversity of reef organisms when combined with loss of struc-
tural complexity, either due to direct physical disturbances (e.g., severe tropical storms) that damage
coral skeletons or the gradual decomposition and erosion of corals killed by biological disturbances
(Pratchett et al. 2008).
Major causes of coral loss vary geographically. Most notably, the current status of coral
reefs throughout the world is strongly reective of the timing and extent of human colonisation
(e.g., Pandol et al. 2005). The most degraded reef environments (East Africa, South- East Asia,
and the Caribbean) are in areas with large human populations (Wilkinson 2004), which reects the
overarching effects of chronic (press) disturbances, such as overshing, pollution, sedimentation,
and eutrophication. There are also a range of acute (pulse) disturbances that strongly inuence the
structure and dynamics of coral reef assemblages (e.g., Gilmour et al. 2013). At the global scale,
the most pronounced acute disturbances are episodes of mass bleaching, linked to increasing ocean
temperatures (Hoegh- Guldberg 1999, Hughes et al. 2003); severe tropical storms (Gardner et al.
2003); or rapid and pronounced increases (variously called ‘plagues’, Vine 1973; ‘outbreaks’, Weber
& Woodhead 1970; or ‘infestations’, Endean 1977) in the abundance of crown- of- thorns starsh and
other coral predators.
In the Caribbean, high levels of coral mortality since 1977 are variously attributed to severe
tropical storms (e.g., Gardner et al. 2003), increasing incidence of coral disease (e.g., Aronson &
Precht 2001), or recent episodes of mass bleaching (Williams & Bunkley- Williams 2000). In reality,
it is the combination of different disturbances that is responsible for sustained and ongoing declines
in the cover of scleractinian corals across much of the Caribbean and associated shifts in the bio-
logical and physical structure of reef habitats. For example, phase shifts from coral- to macroalgae-
dominated systems in Jamaica can be traced back to overshing of herbivorous shes, which had
already occurred by the 1960s (Hughes 1994). However, marked changes in the biological and
physical structure of reef habitats were not apparent until severe storms (e.g., Hurricane Allen in
1981, Hurricane Gilbert in 1988) and the mass mortality of the sea urchin Diadema antillarum
in 1983. Each of these disturbances had an important and independent contribution to the resulting
degradation of coral reef systems (Hughes 1994).
In the Indo- Pacic, outbreaks of crown- of- thorns starsh (Acanthaster spp.) have long been
considered one of the major causes of coral loss (e.g., Pearson 1981, Bruno and Selig 2007, De’ath
et al. 2012). The rst well- documented outbreaks occurred in southern Japan in the late 1950s
(Yamazato 1969 in Yamaguchi 1986) and on Australia’s Great Barrier Reef (GBR) in the early
1960s (Pearson & Endean 1969). However, there are several earlier reports (as far back as the 1930s)
of high densities of Acanthaster spp. that probably represented outbreaks (Dana 1970, Vine 1973).
The severity and extent of coral loss caused by outbreaks of Acanthaster spp. in the 1960s and 1970s
generated considerable concern about the fate of coral reefs (e.g., Cornell & Surowiecki 1972). In
reviewing recovery of coral communities from different disturbances, Pearson (1981) reported that
“damage caused by Acanthaster infestations during the last 10 to 15years has been more extensive
and dramatic than that caused by any other natural or man- made disturbance” (p. 110). Since that
time, emerging threats associated with climate change, such as coral bleaching (Hoegh- Guldberg
1999) and disease (Bruno et al. 2007), have become the major focus of coral reef science and man-
agement (Hughes et al. 2003). However, outbreaks of crown- of- thorns starsh continue to occur
throughout the Indo- Pacic (e.g., Pratchett et al. 2011, De’ath et al. 2012, Baird et al. 2013), and
at many locations the effects of severe outbreaks have been far greater than combined effects of
all other major disturbances, including climate- induced coral bleaching (e.g., Trapon et al. 2011,
De’ath et al. 2012). Importantly, it is the combined effect of outbreaks of Acanthaster planci and
other diverse disturbances (e.g., sedimentation, cyclones, and bleaching) that have caused sustained
and accelerating degradation of coral reef ecosystems throughout the Indo- Pacic (e.g., Jones et al.
2004, Pratchett et al. 2006).
The purpose of this review is to synthesise established knowledge and insights on the causes
and consequences of outbreaks of crown- of- thorns starsh, focusing on the increased understand-
ing and research undertaken in the last two decades (mostly since 1990), given that there were
several substantive reviews of the research on the biology and ecology of Acanthaster spp. between
1960 and 1990 (e.g., Potts 1981, Moran 1986, Birkeland & Lucas 1990, Endean & Cameron 1990).
Since 1990, there have not been any major comprehensive reviews of the biology and population
dynamics of Acanthaster spp., but there have been numerous commentaries on specic issues
(e.g., Brodie 1992, Brodie et al. 2005) related to causes or consequences of population outbreaks.
In the last few years, there has also been renewed interest in outbreaks of Acanthaster spp., largely
attributable to fresh outbreaks of crown- of- thorns starsh at many locations throughout the Indo-
Pacic (e.g., Kayal et al. 2012, Baird et al. 2013). Moreover, there is an increasing realisation that
urgent action is needed to reverse sustained and ongoing declines in live coral cover that are occur-
ring throughout the world (e.g., Gardner et al. 2003, Bellwood et al. 2004, De’ath et al. 2012), and
of all the factors that are contributing to degradation of coral reef ecosystems in the Indo- Pacic,
outbreaks of Acanthaster spp. are considered to be the most amenable to direct and immediate
intervention (cf. climate- induced coral bleaching, increasing prevalence of coral disease, increas-
ing severity of tropical storms). Controlling outbreak populations of Acanthaster spp. is consid-
ered one of the most promising strategies to halt or reverse widespread declines in live coral cover
(e.g., De’ath et al. 2012) and thereby improve the capacity of reef systems to cope with inevitable
threats caused by sustained and ongoing climate change as well as other more direct anthropogenic
disturbances. A second purpose of this review is to highlight gaps in our knowledge of the biology
of Acanthaster spp. that persist, despite ve decades of research, and constrain both understanding
of the causes and effective management of outbreaks.
Biology of Acanthaster spp.
The crown- of- thorns starsh, specically Acanthaster planci (Linnaeus 1758), was rst described
based on the original description by Plancus and Gualtieri (Vine 1973). Crown- of- thorns starsh
have since been reported on coral reefs throughout the tropical Indo- Pacic from the Red Sea
(e.g., Goreau 1964) to Panama (e.g., Glynn 1973) but have never been recorded in the Caribbean
or Atlantic Ocean. Crown- of- thorns starsh are also found in a wide range of latitudes, from 34°N
on subtropical reefs in the Ryukyu Islands, Japan (Yamaguchi 1986), to 32°S at Lord Howe Island
(DeVantier & Deacon 1990). Marked geographic differences in appearance and allelic frequen-
cies within the broad geographic range suggest that there might be at least two species of crown-
of- thorns starsh (Benzie 1999), distributed within the Indian and Pacic oceans, respectively.
Recent molecular sampling (632 bp from the COI region) of crown- of- thorns starsh (nominally,
Acanthaster planci) from throughout the Indo- Pacic revealed that there are four strongly dif-
ferentiated clades from distinct regions of the Indo- Pacic that probably represent distinct spe-
cies: (1) Red Sea, (2) southern Indian Ocean, (3) northern Indian Ocean, and (4) Pacic (Vogler
et al. 2008). However, Vogler et al. (2013) found no genetic differentiation between the crown- of-
thorns starsh in the far eastern Pacic (sometimes considered to be a distinct species, Acanthaster
ellisii; e.g., Barham et al. 1973, but see Glynn 1974) from the remainder of the Pacic. For this
reason, Acanthaster spp. is used when referring to the entire species complex, whereas A. planci
is reserved for use when referring explicitly to crown- of- thorns starsh from the Pacic. Another
well- described species, A. brevispinus Fisher 1917, is known from deep- water habitats in the west-
ern Pacic, but this species is rarely found in coral reef habitats (Birkeland & Lucas 1990) and is
not considered in this review.
Moran (1986) described crown- of- thorns starsh as “one of the most well- known animals in
coral reef ecosystems” (p. 1) and went on to say that the biology of this animal has been particularly
well studied. It is true that there was considerable research on the basic biology of Acanthaster spp.
in the 1970s and 1980s, including studies on reproductive biology (e.g., Lucas 1973); diet (e.g., Brauer
et al. 1970, Branham et al. 1971, Ormond et al. 1973, Glynn 1974); and behaviour (e.g., Barnes et al.
1970, Moran et al. 1985). However, crown- of- thorns starsh remain something of an enigma, with
relatively little known about their demography and population dynamics. Moore (1990) suggested
that intermittent outbreaks might be attributable to fundamental switches in the inherent life- history
characteristics between endemic and epidemic characters, but this has never been explicitly tested.
Most of what is known about the reproductive biology and life cycle of crown- of- thorns starsh
(Figure 1) comes from detailed studies of Acanthaster planci in the western Pacic (e.g., Lucas
1973, Yamaguchi 1973, Nishihira & Yamazato 1974, Conand 1984, Babcock & Mundy 1992a,b).
While the reproductive biology and life history are likely to be broadly similar for other Acanthaster
spp. from the Indian Ocean, this needs to be veried as geographical (and taxonomic) differences
in their biology might account for marked geographical differences in the incidence and severity of
outbreaks (as discussed further in this review). Comprehensive and detailed information about well-
studied aspects of the biology of A. planci was provided by Moran (1986). Rather than repeat this
information, this review limits discussion to advances in the biological knowledge since 1986 and
also considers key aspects of the biology that are fundamental in understanding the proximal causes
of outbreaks, including inherent constraints of reproductive success, and the structure (age and size)
of normal versus outbreak populations. There is widespread recognition that the biology (especially
the reproduction and early life history) of crown- of- thorns starsh is key to understanding when and
why outbreaks occur (e.g., Birkeland 1982, 1989a).
One of the most important biological traits of Acanthaster spp., which is particularly relevant to
major population uctuations, is their enormous reproductive potential (Endean 1982, Conand
1984). Large female starsh can produce up to 65 million eggs per season (Conand 1984, Kettle &
Lucas 1987). It has long been recognised that Acanthaster spp. release millions of eggs each time
they spawn (e.g., Pearson & Endean 1969), but Conand (1983, 1984) provided the rst quantita-
tive analysis of size- based fecundity for Acanthaster planci, following detailed studies in Noumea.
These data correspond closely with similar research undertaken by Kettle and Lucas (1987), who
also found disproportionate increases in fecundity with increasing size, ranging from 0.5–2.5 million
eggs per year for individuals less than 30 cm to 46–65 million eggs per year for starsh that have a
40-cm diameter (Birkeland & Lucas 1990).
Initiation of gametogenesis is clearly related to both age and size of Acanthaster spp. Gonad
development in laboratory- reared starsh (Yamaguchi 1973, Lucas 1984) and in a eld popula-
tion in Fiji (Zann et al. 1987) started before the animals reached the age of 2years, and the largest
10 Hours
1 Day
2 Days
3 Days
5 Days
7 Days
11 Days
11+ DAYS
0.5–6 Months
0.5–2 Years
2–5 Years
fertilization cleavage
1.5 h 5–6 h 8–9 h
Gastrula larva
– Ovoid; ciliated; blastopore hindmost
Early bippinaria larva (0.5 mm long)
– Filter feeding
Advanced bippinaria larva (0.8 mm long)
– Axohydrocoels well-developed
Early brachiolaria larva (0.8 mm long)
– Actively swimming; brachioliar arms appear
Mid brachiolaria larva (1.0 mm long)
– Brachioliar arms well-developed; primordium developing
Late brachiolaria larva (1.0–1.5 mm long)
– Prominent brachioliar and other larval arms
– Negatively buoyant and testing substratum
absorption of larval
body into starsh
Newly settled juvenile (0.5 mm diameter)
– Five arms each with a red optic cushion,
a terminal tentacle and two pairs of tube feet
– Mouth develops a few days after metamorphosis
– Cream to pale yellow color
Coral-feeding juvenile to sub-adult
(10–200 mm diameter)
– Shift in diet results in high (von Bertalany) growth rate
– Gonads start to develop at later part of year 1
– High feeding rates
– Colours range from purplish blue to grey-green
– Arms and spines elongate
Coral-feeding adult
(200–350 mm diameter)
– High fecundity
– Major gonad development and spawning
– High feeding rates, but gradually decreases with age
– Growth rate also gradually decreases with age
– Observed aggregations mostly at this age/size range
Algal feeding juvenile
(1–10 mm diameter)
– Feeding on crustose coralline algae
– Low (exponential) growth rate
– Colour changes to pink and appears more spiny
– Arms relatively shorter compared to central disk
Senile adult
(>350 mm diameter)
– Decline and subsequent cessation of gametogenesis
– Low feeding rates
– Low growth rate and may undergo shrinkage
– Longevity in the eld still unconrmed
bsorption of larval
y into stars
Figure 1 Complete life cycle of Acanthaster planci adapted from laboratory rearing studies by Yamaguchi
(1973) and Lucas (1984) and compilations by Moran (1986) and Birkeland & Lucas (1990).
individuals in any given cohort were also the rst to exhibit gametogenesis (Lucas 1984, Zann et al.
1987). Gonads appear as aciniform rows along each side of the inner wall of the proximal part of
each arm (Figure2). When arms are dissected to expose internal digestive and reproductive organs,
male and female starsh are readily distinguishable as testes are cream or pale yellow in colour and
have smaller, more numerous lobes (Figure2A, 2B) compared to ovaries, which appear as larger,
spherical, yellow (sometimes almost orange) lobes (Figure2C, 2D). It is also apparent that the size
(weight or volume) and maturation stage of gonads are consistent among arms, such that gonad
somatic indices are generally based on subsampling of only one to three arms (Lucas 1973, Conand
1984, Ogura et al. 1985, but see Yokochi & Ogura 1987). Changes in the size of gonads parallels
changes in gonad index and size of oocytes (Yamazato & Kiyan 1973) and swelling in the proximal
region of arms.
(A) (B)
pc p
1 cm
1 cm
5 cm
5 cm
Figure 2 (See also colour gure in the insert) Aboral view of Acanthaster planci showing internal arrange-
ment of gonads (g), pyloric caeca (pc), podia (p), and cardiac stomach (cs) in (A) and (B) male and (C) and
(D) female specimens. (Photographs taken by C.F. Caballes.)
Larger starsh increasingly partition energy towards reproduction (ova production) at the
expense of the body wall and pyloric caeca (Kettle & Lucas 1987). Gonad indices in female
Acanthaster planci are usually higher compared to males, and this disparity becomes more pro-
nounced at the peak of the breeding season (Cheney 1974, Conand 1984, Yokochi & Ogura 1987,
Babcock & Mundy 1992a). Spermatogenic development in testes is mainly reected in the thickness
of the germinal layer (Yamazato & Kiyan 1973). In females, the stage of gametogenic cycle can be
assessed by looking at the size of oocytes, the presence or absence of layers of connective tissue that
bind oocytes together, and ovulation (Babcock & Mundy 1992a).
While the gametogenic cycle of asteroids may be regulated by endogenous (intrinsic) factors
such as age, size, and nutritional status or by exogenous (extrinsic) factors such as temperature,
photoperiod, and food availability (reviewed in Mercier & Hamel 2009), there are few studies of the
role of nutritional status on gametogenesis and fecundity in Acanthaster spp. Cheney (1974) found
that starving Acanthaster planci (by placing them in a cage for 1month without food) resulted in
reabsorption of gonads, atrophy of the pyloric caeca, and a decrease in the overall size (diameter
and weight) of individual starsh. Other environmental factors (e.g., extreme temperatures, reduced
salinity, and limited food availability) may also exert exogenous control on gametogenesis and
fecundity, but this has not been investigated. For the large part, scientists are still coming to terms
with spatial and temporal variation in occurrence of spawning, let alone understanding variation in
the fecundity of individual starsh and relating this to local environmental conditions.
Like most asteroids, Acanthaster planci is a gonochoristic species, whereby male and female indi-
viduals must be in close proximity and spawn simultaneously to effectively reproduce (Babcock et al.
1994). This is important because reproductive success may be greatly constrained when there is a
highly biased sex ratio (e.g., Stump 1994) or if densities of starsh are low and the distance between
individuals is large (Vine 1973). Initial studies on the sex ratio of A. planci, based on sampling of out-
break populations (e.g., Pearson & Endean 1969, Nishihira & Yamazato 1974), suggested that there
are generally an equal number of males and females. However, strongly male- biased sex ratios have
been recorded in several populations (e.g., Stump 1994, C. Caballes et al. unpublished data, 2013).
In September 2011, 93 large starsh (>30-cm diameter) were sampled in Guam (Hospital Point), of
which only 12 were female (C. Caballes et al. unpublished data, 2013). Similarly, at Lizard Island
(northern GBR) in March 2013, there were 115 starsh collected across a range of different locations
(with starsh ranging in size from 15 to 47 cm), of which only 2 individuals were female (C. Caballes
et al. unpublished data, 2013). Of the remaining individuals, 76 were male, and 37 were immature or
non- reproductive. Both these populations were sampled in the aftermath of peak outbreak densities,
and the strong male bias may reect generally lower survival of females, which invest much more
energy in reproduction (Stump 1994). However, if there is strong male sex bias in low- density popula-
tions generally, then this has the potential to greatly limit reproductive success.
Broadcast spawners (e.g., shes, corals, and many marine invertebrates) release copious quanti-
ties of gametes during spawning but typically achieve low fertilization rates unless (1) individuals
are highly aggregated, (2) spawning is synchronised, and (3) spawning occurs in low- to- moderate
ow conditions (e.g., Mercier & Hamel 2009). In Okinawa, Okaji (1991) found that dispersed popu-
lations spawned later and over a much longer period compared with aggregated populations. Okaji
(1991) suggested spawning is also more synchronised within aggregated populations, which results
in much higher reproductive success. Moreover, Cheney (1974) found that aggregated individuals
had consistently higher gonado- somatic indices (indicative of fecundity and reproductive potential)
compared with those from dispersed populations. More important, fertilization rates for Acanthaster
spp. and other free- spawning marine invertebrates decline precipitously with increasing distance
between male and female individuals (Levitan et al. 1992). For Acanthaster spp., fertilization suc-
cess is close to 100% when male starsh spawn adjacent to spawning females (Benzie et al. 1994)
but declines with increasing distance between spawning individuals (Figure3). However, fertiliza-
tion rates recorded when males and females are separated by relatively large distances (>20 m) are
signicantly greater than those of many other marine invertebrates (Yund 1990, Grosberg 1991,
Levitan et al. 1991; Figure3). This disparity could be caused by the greater size and fecundity of
Acanthaster spp. and the large quantity of gametes released during spawning (Babcock & Mundy
1992a, Babcock et al. 1994) or greater capacity for fertilization at low sperm concentrations (Benzie
& Dixon 1994). Even so, minimising the distance between spawning individuals by aggregating
will lead to a marked increase in reproductive success, suggesting that chance aggregation of adults
on reefs with very low overall densities of Acanthaster spp. may be sufcient to precipitate an out-
break (e.g., Vine 1973).
During spawning, gametes are shed from aboral rows of gonopores along the sides of each arm.
Exudates from spawning females appear as translucent spherical grains, while males exude milky
clouds of sperm. Despite the conspicuousness of spawning starsh, there have been relatively few
observations of natural spawning in the eld (Table1). It is still unclear whether Acanthaster spp.
spawns just once each year or whether there are multiple spawning events concentrated within a
particular spawning period. At Lodestone Reef in the GBR, Lucas (1973) showed that periods of
most active gametogenesis corresponded to periods of increasing temperature and marked changes
in the photoperiod. As for many other marine invertebrates (e.g., corals; Baird et al. 2009), tem-
perature appears to be the most important cue for seasonality in spawning. In warmer locations,
there is a tendency for starsh to reach maximum gonad maturity and begin spawning whenever
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
% Fertilization
Distance between male and female (m)
Benzie et al. 1994
(spawning early in
the breeding season)
Benzie et al. 1994
(spawning late in the
breeding season)
Babcock & Mundy
1992 (Davies Reef )
Babcock et al. 1994
(one male, Davies
Reef )
Babcock et al. 1994
(one male, Sesoko
Babcock et al. 1994
(ve males, Davies
Reef )
Pennington 1985 (one
male, uncontrolled #
of eggs)
Pennington 1985
(three males, ~1600
Figure 3 Fertilization rates for female Acanthaster planci (solid and dashed lines) versus Strongylocentrotus
droebachiensis (dotted lines) at varying distances downstream from spawning males.
Table1 Observations of spontaneous natural spawning in the eld
Region Location Date
Time of
day Depth Extent of spawning Reference
Japan Kii
17 Jul 1973 Evening Hayashi 1975
15 Aug 1973 Evening Hayashi 1975
9 Jun 1984 Afternoon 4–8 m 3 males and 1 female
Yokochi 1985
Okinawa 9 Jun 1988 Daytime 15 m Several individuals
Kishimoto, 1989
Okinawa 4–5 Jul 1988 1500 Cited in Babcock
& Mundy 1992a
Okinawa 13 Jun 1990 1500 3 male individuals
Cited in Babcock
& Mundy 1992a
Okinawa 17 Jun 1990 1425–1645 6 males and 1 female
Cited in Babcock
& Mundy 1992a
Okinawa 19 Jun 1990 Afternoon Single male Cited in Babcock
& Mundy 1992a
Red Sea N Gulf of
24 Jul 2004 1823, 1859 10 m,
21 m
Two individuals
spawned (water
temperature 24.5°C)
D. Zakai personal
Hawaii Kalohi
23–24 Apr 1970 Several starsh
Branham et al.
Guam Tanguisson
21 Apr 2006 1500 5 m Single female
(~35-cm diameter)
in a massive
spawned for 30 min
C. Caballes
Maldives North Malé
Apr 1991 Ciarapica &
Passeri 1993
Papua New
Yule Island 4 Jul 1989 1630 50–100 individuals
Cited in Babcock
& Mundy 1992a
Central GBR,
10 Jan 1968 1330 3 m Dense aggregation;
spawning lasted
30 min and involved
several males and
only one female
Pearson and
Endean 1969
20 Jan 1983 1500 Single female Cited in Babcock
& Mundy 1992a
Feb 1983 0900 20–30 males; no
Gladstone 1987
Rib Reef 21 Jan 1984 1615 2–3 males spawned
on top of boulder
Cited in Babcock
& Mundy 1992a
Rib Reef 13 Dec 1984 1600 4 m More than 50
clustered individuals
Cited in
Birkeland &
Lucas 1990
5 Jan 1987 1520–1800 3–8 m 40–100 starsh
spread over a
2400-m2 area
Cited in
Birkeland &
Lucas 1990
sea- surface temperatures exceed 27°C (Figure4). At higher latitudes, where temperatures never
reach this threshold (e.g., Hawaii, New Caledonia), gametogenesis and spawning are often con-
centrated in months when the temperature rst starts to rise (Figure4). Breeding and spawning
seasons of Acanthaster spp. at higher latitudes are mostly shorter, well- dened events compared
with more protracted gametogenesis and spawning at lower latitudes (e.g., Guam, Palau) (Table2).
Mature gonads have been reported year round in A. planci from Guam (e.g., Cheney 1972) but tend
to be found in only a few months of each year at most other locations (Table2).
Table1 (Continued) Observations of spontaneous natural spawning in the eld
Region Location Date
Time of
day Depth Extent of spawning Reference
5 Jan 1987 1600 Single female
(adjacent starsh did
not spawn)
Cited in Babcock
& Mundy 1992a
17 Jan 1988 3–4 starsh Cited in Babcock
& Mundy 1992a
18 Jan 1988 1430 1 starsh Cited in Babcock
& Mundy 1992a
5 Dec 1990 1534–1706 2.5 m 10 males; single
female was last to
begin spawning and
spawned for the
shortest time
Gladstone 1992
7 Dec 1990 2145 ≤7 m 68% of 129 starsh
(38 females and
50 males) spawned
over a 2-h period
Babcock &
Mundy 1992a
17 Dec 1990 1700 ≤7 m 3 male individuals
Babcock &
Mundy 1992a
11 Dec 1991 1630 1–4 m Single female initially
spawned, followed
by > 50 starsh
(mostly males)
Babcock &
Mundy 1992b
12 Dec 1991 1930 1–4 m 8 males and 1 female
Babcock &
Mundy 1992b
13 Dec 1991 2030 1–4 m Single male spawned Babcock &
Mundy 1992b
23 Jan 1992 2030 1–4 m 2 males released
gametes through
gonopores from only
a few arms
Babcock &
Mundy 1992b
18 Dec 2009 1630 1 m 12 males released
copious amounts of
sperm; most were
exposed, and
individuals were
well spaced from
each other (3–10 m
L. Vail & A.
Fiji Muaivuso 1 Feb 1970 1400–1600 0.5–4 m Dense clusters; many
individuals spawned
Owens 1971
Compilation of the limited observations of natural spawning of Acanthaster spp. (Table1) shows
that spawning occurs mostly in the late afternoon, although it has occasionally been observed in the
early morning (Gladstone 1987, Kishimoto 1989) and after dark (Babcock & Mundy 1992b). There is
no apparent link between the timing of spawning and lunar phases or tidal cycles (Babcock & Mundy
1992a), suggesting that spawning is not linked to environmental cues. It is known that spawning by
one individual will often instigate spawning by other individuals in the local proximity (e.g., Babcock
& Mundy 1992a) or within aquaria. However, this does not explain the synchronous behavioural
changes observed in many aggregations of Acanthaster spp., whereby individuals become partic-
ularly active, move to shallow promontories, and adopt the characteristic arched posture prior to
the release of gametes (Babcock & Munday 1992a). Laboratory experiments by Beach et al. (1975)
revealed that pheromones extracted from Acanthaster planci ovaries and testes synchronise spawn-
ing among neighbouring animals and also induce movement towards spawning individuals. Beach
et al. (1975) argued that gametogenic cycles are inuenced by local environmental conditions (mainly
temperature), but chemically mediated communication between ripe individuals is necessary to coor-
dinate gamete release. Consistent with this hypothesis, spawning by Acanthaster spp. is often syn-
chronised at localised scales, within populations, but not among populations (Babcock & Mundy
1992a, Yasuda et al. 2010). Babcock & Mundy (1992a) saw A. planci spawning in one area of Davies
Reef on the GBR, while A. planci in other parts of the same reef were not.
Spawning by Acanthaster spp. is concentrated in summer months at most locations, but the
evidence is contradictory regarding whether individual starsh spawn once annually (e.g., Babcock
& Mundy 1992a,b) or spawn multiple times (e.g., batch spawn) each year (e.g., Conand 1983). In
Noumea, Conand (1983) found that all female starsh examined contained oocytes at vastly dif-
ferent stages of maturity, suggesting that individual Acanthaster planci spawn sequentially over
several months. However, in the central GBR Babcock and Mundy (1992a) suggested that A. planci
Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar
Monthly mean sea surface temperature (°C)
New Caledonia
Figure 4 Seasonal variation in sea- surface temperatures and spawning times for Acanthaster planci at
different locations throughout the Pacic. Spawning (lled circles) tends to occur only when temperatures
are above 27°C, although at locations where temperatures never reach this level (e.g., Hawaii), spawning is
restricted to months when temperatures rst start to rise. Data sources: Palau, Idip 2003; Lizard Island, GBR,
L. Vail unpublished data; Guam, Cheney 1974; New Caledonia, Conand 1984; Hawaii, Cheney 1974; and
Okinawa, Japan, Yamazato & Kiyan 1973.
Table2 Peak seasons in annual reproductive cycle of Acanthaster planci from different locations based on gonad
development and spawning
Latitude Location
Month (April to March)
Reference04 05 06 07 08 09 10 11 12 01 02 03
33°N SW Honshua● ● Hayashi 1975
28°N Amami- Ohshimab○○●○ Yasuda et al. 2010
26.5°N Okinawac,d,e ○○○●●○ ○ ○○Yamazato & Kiyan 1973
Okinawac,d ○○●●○○○○○○○○Okaji 1989
Okinawab○○●○○ Yasuda et al. 2010
26°N Kerama Islandsb○ ● ● Yasuda et al. 2010
25°N Gulf of CaliforniadDana & Wolfson 1970
24.5°N Miyako Islandb○ ● ○ Yasuda et al. 2010
24°N Iriomotec○○●○○○○○○ ○Yokochi & Ogura 1987
Iriomotec○●●○○○○○○ ○Habe et al. 1989
24°N Sekisei Lagoonb●○○○ Yasuda et al. 2010
20–21°N Red Seac● ● Crump 1971
Red Seac● ● Moore 1985
21°N Hawaiia,b ● ○ ○ ○ Branham et al. 1971
13°N Guamf● ● Chesher 1969
Guamb●●● Cheney 1974
9°N PanamafGlynn 1974
7°N Palauc●●●●○●○○ ○●Idip 2003
7°N Philippinesc●○○○○○Bos et al. 2013
5–15°N Micronesiad● ● Eldredge 1970
MicronesiacCheney 1972
4°N Maldivesa Ciarapica & Passeri 1993
9°S Solomon IslandsdEldredge 1970
14°S W Samoad● ● Garlovsky & Bergquist 1970
17–19°S Central GBRa,c ○○○○●●○○Pearson & Endean 1969
Central GBRc,e ○ ○ ○○○○●● Lucas 1973
Central GBR a,c,d ○○●●● Babcock & Mundy 1992a
Central GBRa,c ○○○●●●○ Babcock & Mundy 1992b
Central GBRaGladstone 1992
18°S FijiaOwens 1971
19–23°S SE PolynesiadDevaney & Randall 1973
20°S NW Australiab ○●●●● Wilson & Marsh 1974
22°S New Caledonia b,c ○○ ○ ○●●●●○Conand 1984
27°S South Africad Schleyer 1998
32°S Lord Howe IslanddDeVantier & Andrews 1987
Note: , months surveyed/ sampled; , spawning/ breeding season; , mature gonads but only partial or no spawning; , not surveyed/ sampled but
spawning assumed.
Method used to describe annual reproductive cycle of Acanthaster planci: aeld observation of spawning; bchanges in proportion of ripe and
spent gonads; cchanges in gonad index; dgonad condition or histology of ovaries and testes; echanges in oocyte size and frequency; fno method given.
spawn just once, mainly in early December. Babcock and Mundy (1992a,b) recorded changes in the
size and density of oocytes in A. planci at Davies Reef over much of the year, but concentrated sam-
pling over the austral summer (October to February). They saw a marked drop in the proportion of
females that were gravid, as well as marked reductions in mean size of gonads, in early December.
Accordingly, spawning was observed in the evening of 7 December 1990 at 2145, with 88 (of 129)
starsh seen spawning along a single, shallow (1- to 4-m) transect. Elsewhere on the GBR, spawn-
ing has been reported from December and January to February (Table2), but it is unclear whether
this reects geographic variation in the timing of annual spawning or whether A. planci on the GBR
can spawn multiple times in a year. It is clear that starsh at some locations (e.g., Guam) are gravid
year round (Cheney 1974), and reproductive behaviour is likely to vary geographically in response
to environmental regimes and food availability. More detailed studies of the gametogenic cycle,
maturation, and spawning of Acanthaster spp. are needed, combined with systematic sampling of
individually tagged starsh to establish spawning behaviour of individuals as well as populations.
This information is critical for potentially linking the precise timing of spawning behaviour and
associated reproductive success to spatial and temporal anomalies in local conditions, such as nutri-
ent dynamics (e.g., Fabricius et al. 2010).
Larval development
The life cycle of Acanthaster spp. is typical of most asteroids, with larval development divided into
two distinct bipinnaria stages (pelagic feeding larva characterised by bilateral arrangement of the
pre- and post- oral ciliated swimming and feeding bands) and three brachiolaria stages (feeding
larva characterised by the presence of brachiolar arms and attachment disk on the preoral lobe)
prior to metamorphosis and settlement (Figure1). Fertilized embryos develop to the blastula stage
after 8–9 h, then hatch after approximately 1day as free- swimming gastrula larvae (Lucas 1982).
After 2–4days, larvae have a completely formed alimentary canal and start lter- feeding on uni-
cellular algae and other suspended particulate matter (Yamaguchi 1973). Larvae then proceed to
the brachiolaria stage, developing brachiolar arms, which will eventually be used to locate favour-
able substrate prior to settlement (Henderson & Lucas 1971). The typical planktonic larval dura-
tion (PLD) for Acanthaster planci is 11days (Figure1), although, like many marine invertebrates
(Richmond 1987, Graham et al. 2008), these starsh can delay settlement and signicantly extend
their PLD (Yamaguchi 1973). The rate of development also varies with temperature (Henderson
& Lucas 1971) and food availability (Lucas 1982), such that the PLD can range from 9 to 42days.
Factors that limit the rate and success of larval development of Acanthaster spp. (mainly,
Acanthaster planci) have received a great deal of attention in the past years, as this was considered
key to understanding the initiation of population outbreaks (e.g., Lucas 1982, Olson & Olson 1989).
Much of this research has concentrated on larval nutrition and the extent to which larval survivor-
ship is constrained by abundance of phytoplankton, which in turn depend on the levels of nutrients.
Phytoplankton are generally considered to be the main food source for larval Acanthaster spp., as
both natural phytoplankton and cultured unicellular algae (single or mixed species) have been suc-
cessfully used to rear larvae under laboratory conditions (Henderson & Lucas 1971, Lucas 1975,
1982, Uchida & Nomura 1987, Keesing et al. 1996, Okaji 1996, Fabricius et al. 2010). However,
Lucas (1982) suggested that the amount of phytoplankton required to maintain cultured larvae was
much higher than generally occurs in near- reef waters (e.g., within the GBR lagoon), leading to sug-
gestions that larvae are severely food limited except during major phytoplankton blooms (‘larval
starvation hypothesis’; Lucas 1982). Similarly, Fabricius et al. (2010) reported minimal survival of
larval Acanthaster planci at chlorophyll concentrations below 0.25 µg.l−1, whereas larval survival
increased approximately 8-fold with each doubling of chlorophyll concentrations up to 3.0 µg.l−1.
These experiments were designed to answer specic questions about the larval nutrition of A. planci
and were limited in their representation of model ecosystems. For instance, any potential predators
on larval crown- of- thorns starsh (≥100 µm length) were explicitly excluded from the experimental
system by ltering (25-µm lters) incoming seawater. Even so, these data are used to argue that out-
breaks of A. planci may arise from pulses of recruits produced by temporary increases in productiv-
ity linked to high rainfall events and associated oods (Fabricius et al. 2010). It is not clear whether
predators or competitors would alter the survivorship of larval crown- of- thorns starsh sufciently
to affect the relationship with nutrient concentrations, but these experiments need to be repeated
and extended before any conclusions about limitations to larval survivorship can be made. It is also
controversial whether Acanthaster spp. are food limited, as several studies have successfully reared
larvae at normal (low) chlorophyll concentrations (Olson 1987, Johnson et al. 1991) though it is
increasingly clear that food limitation may constrain rates of development, if not survival.
Olson & Olson (1989) suggested that larval Acanthaster planci are capable of exploiting a diver-
sity of different sources of nutrition. Under normal conditions of low plankton abundance, high
levels of survivorship were facilitated by the ability of A. planci to utilise dissolved organic matter
(DOM) and bacteria (Olson & Olson 1989). Importantly, larvae reared in insitu culture chambers
showed no sign of food limitation and were able to develop at near- maximal rates despite low levels
of plankton (Olson 1987). Moreover, nutrient enrichment did not result in increased survivorship
but did result in a slight increase in the rate of development (Olson 1987). Accordingly, Hoegh-
Guldberg (1994) showed that dissolved free amino acids (DFAAs) can supply signicant amounts
of energy for developing larvae. Microscopic analyses by Ayukai (1994) showed that A. planci
larvae generally consume only large phytoplankton, as there was limited evidence of ultraplankton
(<5 m) or bacteria within stomach contents. However, A. planci may rely on alternative sources of
prey (including ultraplankton and bacteria) when preferred prey (large phytoplankton) are in limited
abundance. Given increased knowledge of patterns of resource use, food limitation experiments
with developing larvae should be repeated to test explicitly whether high concentrations of DFAA,
free- living bacteria, or ultraplankton may compensate for the limited abundance of phytoplankton
under eld conditions. Experimental work needs to be corroborated with eld assessment of the
nutritional condition of crown- of- thorns larvae before, during, and after major ood events to spe-
cically relate this to local phytoplankton concentrations. In the past, it was difcult to distinguish
the larvae of Acanthaster spp. from those of other echinoderms, but Roper (1997) developed a
technique to stain crown- of- thorns larvae with uorescently labelled monoclonal antibodies, which
readily distinguish Acanthaster spp. within plankton samples.
As for many marine organisms, predation on larval Acanthaster spp. is expected to be high,
especially during the late brachiolaria stage, when larvae come within the vicinity of reefs and
attempt to settle. However, logistical challenges to sampling Acanthaster spp. larvae in the eld
make it difcult to quantify natural rates of larval mortality and rates of predation. Unlike coral
eggs, which are heavily preyed on by planktivorous shes (e.g., Pratchett et al. 2001), Acanthaster
spp. gametes and larvae are often avoided by planktivorous shes and invertebrates (Yamaguchi
1974, 1975). Chemical analyses by Lucas et al. (1979) showed that eggs and larvae of Acanthaster
spp. contain saponins, which presumably make them less palatable. During a spawning event at
Blue Pearl Bay (Hayman Island, GBR), Gladstone (1992) observed that planktivorous sh feeding
nearby ignored gametes released by spawning Acanthaster planci. Reef shes within the vicinity
of a spawning A. planci at Arlington Reef, central GBR, also ignored gametes, except for one spe-
cies of damselsh, Abudefduf curacao, which was observed feeding on eggs shed by the spawn-
ing female (Pearson & Endean 1969). Butteryshes Chaetodon auripes and Chaetodon falcula
have also been observed to feed on A. planci gametes in Okinawa and the Maldives, respectively
(Keesing & Halford 1992, Ciarapica & Passeri 1993).
Keesing and Halford (1992) suggested that spawning usually occurs late in the afternoon or
at night, when the impact of visual predators would be minimised. However, little is known about
when or where developed larvae actually settle or whether natural rates of predation on larvae
that are trying to settle are high or low. Chesher (1969) suggested that predation by lter- feeders,
such as corals, would inict signicant mortality on settling larvae. Yamaguchi (1973) documented
that Pocillopora damicornis will feed on the larvae of A. planci and other coral reef asteroids.
However, predation by corals is likely to have limited inuence on overall survivorship for two rea-
sons. Firstly, Ormond & Campbell (1974) found that A. planci larvae could detect and readily avoid
live corals. Secondly, there are many areas of coral reef substrata that have relatively low cover of
live coral where starsh larvae could settle without being consumed by coral polyps (Reichelt et al.
1990b). Chesher (1969) suggested that larvae are attracted to aggregations of adult A. planci because
they have already removed corals that would otherwise prey on the larvae. More likely, however,
is that the feeding activities of adult A. planci increase the availability of microhabitats (dead but
intact coral colonies) that are conducive to settlement.
Abiotic factors play an important role in development and survivorship of larval Acanthaster
spp. (e.g., Henderson & Lucas 1971, Lucas 1973). Laboratory rearing experiments suggested that
Acanthaster planci have a narrow temperature tolerance. Optimal temperatures for larval develop-
ment appear to be between 26°C and 31°C (e.g., Lucas 1973, Lamare et al 2014). Temperatures of
32°C or greater appear lethal, while larvae simply did not complete development at temperatures
below 25°C (Henderson & Lucas 1971, Lucas 1973). In Guam, larvae were successfully reared at
27–29°C, while larvae reared at temperatures below 25°C did not advance to brachiolaria stage and
showed regression to earlier stages, even though they were observed to feed vigorously (Yamaguchi
1973). Similar temperature ranges were used to rear larvae through to settlement in the Red Sea
(28°C and 29°C; Ormond & Campbell 1974) and southern Japan (uctuating between 25°C and
30.3°C; Uchida & Nomura 1987). Importantly, temperature tolerance of A. planci varies with devel-
opmental stages (Johnson & Babcock 1994). The late brachiolaria stage appears to be the most
temperature sensitive (Habe et al. 1989), and this may constrain settlement to areas with relatively
warm temperatures (26–31°C). However, greater temperature tolerance in early larval stages allows
larvae to withstand exposure to cooler waters and slowly continue normal development during oce-
anic transport. Hatched gastrula larvae can tolerate temperatures between 13°C and 34°C, and
bipinnaria larvae can tolerate temperatures of 14.5–32°C (Habe et al. 1989). However, the rate of
development is greatly accelerated at higher temperatures: Habe et al. (1989) found that embryonic
development is completed in 31 h at 20°C but only 11 h at 32°C.
Although echinoderms are generally sensitive to changes in salinity (Diehl 1986), the early
larval stages of Acanthaster spp. appear to tolerate wide ranges in salinity. Gastrula larvae can
tolerate a salinity range of 21–45‰ and bipinnaria larvae can tolerate 21–50‰ salinity (Habe et al.
1989). Henderson (1969) also found that bipinnaria larvae can tolerate abrupt salinity changes from
36‰ down to 21‰, and they developed more rapidly at lower salinities. Despite the robust early
larval stages, late brachiolaria and metamorphosing stages are less tolerant to salinity and rupture
with 2‰ changes in salinity (Henderson & Lucas 1971). Larval development and metamorphosis
was completed at 26‰, but not at 22‰ (Lucas 1973). On the GBR, salinity in nearshore water is
inuenced by river discharge, and during times of ooding and heavy rainfall, salinity levels of
30‰ or less are often recorded (Brodie et al. 2005). Lucas (1973) showed that larval survival was
3-fold higher at 30‰ compared to ambient conditions, suggesting that temporary declines in salin-
ity bought about by periods of heavy rainfall and run- off may actually enhance larval survivorship
(Birkeland 1982). The tolerance of unfertilised gametes, especially sperm, to lowered salinity is
unknown. Sperm of Acanthaster planci have been shown to be sensitive to changes in temperature
and seawater chemistry (Uthicke et al. 2013) and may therefore be sensitive to changes in salinity
(but see Greenwood & Bennett 1981).
Towards the end of the brachiolaria stage, the brachiolar arms of Acanthaster spp. larvae elongate
to improve locomotion while supporting the weight of the starsh primordium (Olson et al. 1988).
At this time, larvae start to drift downward and ex the anterior body dorsally to orient the bra-
chiolar arms against the substratum to test its suitability for settlement (Yamaguchi 1973). Based on
laboratory experiments (Henderson & Lucas 1971, Ormond et al. 1973, Yamaguchi 1973, Ormond
& Campbell 1974, Lucas 1975, Johnson et al. 1991, Keesing & Halford 1992, Johnson & Sutton
1994) and eld observations (Yokochi & Ogura 1987, Zann et al. 1987), larval Acanthaster planci
are particular about where they settle. However, Lucas (1975) suggested that these strong settlement
preferences do not necessarily limit settlement success as long as larvae are transported over coral
reef habitats while they are still competent to settle. Larvae appear to settle preferentially in habitats
with ne- scale topographic complexity, so that the larvae are completely hidden within carbonate
matrix, or among coral rubble, prior to metamorphosis (Lucas 1975). This may be an adaptation
to minimise larval mortality at settlement and during metamorphosis, which is suggested to be in
excess of 85% (Keesing & Halford 1992). Natural rates of post- settlement mortality are extremely
difcult to measure, but as for many coral reef organisms (e.g., Trapon et al. 2013), there is increas-
ing realisation that biological interactions (competition and predation) are probably important in
determining patterns of settlement (Keesing & Halford 1992). Ormond and Campbell (1974) showed
that skeletons of dead colonies of Acropora hyacinthus were favoured over the other coral skeletons
they tested (Acropora diversa, Pocillopora verrucosa, and Stylophora pistillata), probably because
the branch spaces more closely match the size of the larvae. However, other work has suggested that
settlement cues for A. planci are more strongly inuenced by biolms (e.g., encrusting algae and
associated bacteria and detritus), rather than the microhabitat complexity. For example, A. planci
will not generally settle on glass or ceramic tiles (Henderson & Lucas 1971, Ormond & Campbell
1974, Johnson et al. 1991) unless these substrates are rst conditioned in natural environments to
generate ne growth of microalgae (Henderson & Lucas 1971).
Observations of recently settled Acanthaster planci in Suva Reef, Fiji (Zann et al. 1987), and
Ryukyu Islands, Japan (Yokochi & Ogura 1987), revealed a strong association with coralline
algae (e.g., Porolithon onkodes), which is expected given that newly settled starsh feed almost
exclusively on coralline algae (Yamaguchi 1973, Lucas 1984, Zann et al. 1987). Yamaguchi (1973)
observed A. planci larvae settling directly on dead coral encrusted with coralline algae (Porolithon
sp.) but found no settlement on bleached coralline algae or on pieces of beach rock covered with
lamentous algae (but see Henderson & Lucas 1971). Johnson et al. (1991) also found high rates
of settlement on coral rubble with high coverage of the coralline algae, Lithothamnium pseudoso-
rum, but reported signicantly lower settlement on tiles colonised by non- calcareous crustose red
algae (Peyssonellia sp.) and other species of coralline algae (Porolithon onkodes, Neogoniolithon
foslei). Techniques developed for large- scale culture of A. planci larvae have achieved high rates
of settlement on thalli of L. pseudosorum (Ayukai et al. 1996, Keesing et al. 1996). Treatment of
highly inductive shards of L. pseudosorum with antibiotics reduced settlement to low levels, signi-
fying that induction of settlement and metamorphosis of A. planci may be mediated by chemical
cues produced by epiphytic bacteria (Johnson et al. 1991). Settlement and metamorphosis were
inhibited in the absence of bacteria; larvae always settled on sections of thallus with high densities
of bacteria, but not in areas where epiphytic bacteria were sparse (Johnson et al. 1991, Johnson &
Sutton 1994). However, surface bacteria did not induce settlement when isolated from soluble algal
compounds, suggesting that bacteria require the algal substratum to produce inductive compounds
or that compounds from both the bacteria and coralline algae are required to induce settlement and
metamorphosis (Johnson & Sutton 1994).
Juvenile ecology
Following settlement, metamorphosis of Acanthaster spp. brachiolaria larvae occurs with the
absorption of the anterior part of the larval body into the starsh primordium (Yamaguchi 1973),
which emerges 2days later as a ve- arm juvenile starsh 0.3 to 0.7 mm in diameter with two pairs
of tube feet, a terminal tentacle, and a red optic cushion on each arm (Henderson & Lucas 1971,
Yamaguchi 1973, Lucas 1975). After 3weeks, these juvenile starsh start adding arms at 2-week
intervals, and the body turns pink, which camouages the juvenile starsh against the coralline
algae on which it is feeding (Yamaguchi 1973, Lucas 1975, Birkeland & Lucas 1990). During this
phase, juvenile starsh do not feed on coral tissue, possibly to avoid damage caused by mesenteric
laments when coming in contact with coral polyps (Yamaguchi 1973). While feeding on coralline
algae, growth rates of A. planci are slow (1.5–2.6 mm/ month) but increase rapidly when starsh
switch to feeding on scleractinian corals at around 6months (Yamaguchi 1974, Zann et al. 1987).
Maximal growth rates (16.7–25.0 mm/ month) occur between 5 and 18months after A. planci switch
to feeding on live corals but slow substantially after this period, presumably because starsh begin
diverting energy from somatic growth as they become sexually mature (Birkeland & Lucas 1990).
Although juvenile Acanthaster spp. contain saponins, they are still likely to be extremely vul-
nerable to predation (Keesing & Halford 1992). Small juveniles are cryptic and are mostly active
at night (Zann et al. 1987), presumably to avoid visual predators like reef shes. This highly cryp-
tic and nocturnal behaviour continues until starsh reach at least 15-cm diameter, at an age of
approximately 20 months (Zann et al. 1987), after which starsh are much more active during
daylight hours. As a consequence, relatively few juvenile Acanthaster spp. have been found, despite
extensive eld sampling (Doherty & Davidson 1988, Johnson et al. 1992). Estimates of recruitment
(e.g., Doherty & Davidson 1988, Zann et al. 1990) are often based on the emergence of relatively
old (e.g., 1- to 2-year- old) individuals. The abundance and distribution of these older juveniles are
likely to differ greatly from patterns at settlement because of high rates of post- settlement mortal-
ity (e.g., Keesing & Halford 1992) and likely movement of larger juvenile starsh as they switch to
eating scleractinian corals (Endean & Cameron 1990).
Adult growth and longevity
The demography of Acanthaster spp. is extremely plastic in that adult growth and longevity seem
to be strongly dependent on local environmental conditions, such as food availability, tempera-
ture, and wave exposure (e.g., Ormond & Campbell 1971, Kenchington 1977, Lucas 1984). For this
reason, there has been considerable controversy surrounding even the most basic demographic ques-
tions, such as whether growth of Acanthaster spp. is determinate (or more precisely, asymptotic;
e.g., Yamaguchi 1974, Lucas 1984) or indeterminate (Kenchington 1977). This issue was discussed
at length by Moran (1986) but never resolved. Lucas (1984) argued that Acanthaster planci from
the GBR reached a maximum size of approximately 340-mm diameter at approximately 3 years
of age (Figure5), after which they entered a period of senescence and had a maximum longevity of
4–5years. These assertions are clearly at odds with extensive and increasing records of A. planci
that are up to 750 mm in diameter and 8 years of age (Endean & Cameron 1990, Stump 1996).
Based on their regenerative ability, as well as physical and chemical defences, Endean (1982) sug-
gested that adult crown- of- thorns starsh would have low mortality and should live for decades (see
also Ebert 1973). Large A. planci have been recorded, up to 750 mm in diameter (Stump 1996), but
mostly on the GBR and often outside active outbreaks. At high densities, Acanthaster spp. may have
highly constrained, nite growth and survivorship, which is possibly linked to strong intraspecic
competition and rapid depletion of prey resources during major outbreaks (Kettle 1990, Mills 2012).
However, recent studies (e.g., Pan et al. 2010) clearly showed that starsh in outbreak populations
can grow well beyond 350 mm and can live for more than 8years.
An important question for demographic studies (and improved understanding of the timing
and therefore possible causes of outbreaks) of Acanthaster spp. is whether variation in the sizes of
starsh within a given population reects distinct cohorts and therefore the range in ages of indi-
viduals (e.g., Pratchett 2005). The alternative is that substantial variation in growth rates among
individuals from a single cohort will obscure any relationship between size and age and that this
alone accounts for the range of sizes within outbreaking populations (Stump & Lucas 1990). To
test this, size- independent proxies of individual age have been explored, including spine length,
age pigments, and pigment bands (equivalent to growth rings) on spines (Birkeland & Lucas 1990).
The most reliable and widely adopted technique, developed by Stump and Lucas (1990), involves
estimating individual ages based on pigment banding on spines. Mark- recapture and tetracycline
staining of Acanthaster planci in the eld (Stump & Lucas 1990, Stump 1994) conrmed that
growth bands on the longest spines taken from the aboral surface of upper arms were laid down sea-
sonally; thus, banding couplets are reective of age in years. However, growth bands only become
apparent after sexual maturity, at around 2years of age (Stump 1996). Moreover, caution must be
taken in selecting the most appropriate spines (Stump & Lucas 1999) as marked differences in
putative ages (ranging from 1 to 17years) may be obtained from spines taken randomly across the
surface of individual starsh (Souter et al. 1997).
Initial studies of post- settlement growth of Acanthaster planci revealed marked changes in
growth rates at different life stages (e.g., Lucas 1984). The growth pattern is sigmoid (e.g., Lucas
1984, Stump 1996), with slow growth both when starsh rst settle and feed on calcareous algae
and when starsh attain sexual maturity at approximately 2+ years of age. However, when consider-
ing only the individuals 1+ years of age, growth can be described effectively using von Bertalanffy
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
Total diameter (mm)
Age (years)
Yamaguchi (1974) - Guam
Lucas (1984) - Aquaria
Zann et al. (1987) - Fiji
Habe et al. (1989) - Japan
Stump (1996) - GBR
Pan et al. (2010) - Philippines
Figure 5 Relationship between total diameter (millimetres) and age (years) for Acanthaster planci based on
data from throughout the western Pacic. There is a strong and consistent size- age relationship, despite initial
suggestions that marked plasticity in initial growth, combined with rapid attainment of asymptotic size, would
obscure any such relationship. The growth curve in the formative part of the life history (incorporating the
initially slow growth in the rst 6months) is best explained by a sigmoidal function. After 1year, however,
the generalised growth curve is largely consistent with von Bertalanffy growth equations. The dashed line
indicates the presumed maximum size and age of A. planci prior to 1990, based on the apparent asymptote at
340-mm diameter for laboratory- reared starsh (Lucas 1984). Field- based studies, however, suggest that there
is a much larger asymptotic size.
growth functions (Figure5), where the L (asymptotic size) is between 474 mm based on data from
Stump (1996) and 580 mm based on data from Pan et al. (2010). By combining size- at- age data
across all previous studies, including laboratory- based measurements of Lucas (1984), it is appar-
ent that there are distinct differences in the reported size (diameter) of starsh in consecutive age
classes; 30–100 mm for starsh with an estimated age between 1 and 2years, 100–250 mm for 2- to
3-year- old starsh, 250–300 mm for those 3–4years, 300400 mm for 4–5years, and greater than
380 mm for 5+ years (Figure5). Increased research and further validation of growth relationships
in different locations (especially outside the Pacic) may be needed to provide increased resolution
of ages necessary to establish the precise timing for the initiation of different outbreaks. However,
it is generally accepted that wide ranges in the size of individuals, especially within single popula-
tions, reect marked differences in ages rather than extreme variation in growth of starsh from the
same cohort (Stump 1996).
Lack of an effective method of tagging individuals has greatly impeded eld studies of the
demography of Acanthaster spp. (Glynn 1982). Early attempts to tag starsh involved embedded
tags, but starsh quickly ejected tags (within weeks) or dropped the arm to which tags were attached.
Some tags also lead to high rates of infection, or greatly modied individual behaviour, under-
mining efforts to document ‘natural’ rates of growth and survivorship (Moran 1986). Later tagging
efforts avoided damaging the dermal tissues, instead attaching harnesses around the oral disk or
coloured bands to individual spines (e.g., Keesing & Lucas 1992), but again tags had a limited life-
span. Stump (1996) marked Acanthaster planci using tetracycline injections to stain their spines. In
combination with counts of arms and records of the position of madreporites, this could be used to
conrm the identity of individual starsh. However, this technique did not allow easy recognition
of individuals in the eld. In 2012, J. Rivera- Posada (unpublished data) used loops of relatively inert
nylon monolament that were passed through predrilled holes in the skeletal elements of the aboral
disk and between the ambulacral ridge of the arms. Holes were drilled with Kirschner wires at low
speed to avoid thermal osteonecrosis or fracture that could lead to rapid ejection of tags. Needles
and cardiocatheters were then used to pass the nylon through the arm and skeletal elements, secur-
ing the two ends with small brass connector sleeves. These tags last at least 4weeks in aquaria but
are yet to be tested in the eld. By threading coloured plastic beads on to the nylon loop, it will be
possible to identify a large number of tagged Acanthaster spp., with wide application in studies of
demography and individual behaviour (e.g., movement).
Feeding behaviour
The crown- of- thorns starsh is just one of many different coral reef organisms that feed on scler-
actinian corals (Glynn 1988, Cole et al. 2008, Rotjan & Lewis 2008). However, the capacity of
Acanthaster spp. to deplete local cover of scleractinian corals is far greater than for any other coral-
livorous species (Glynn 1988, Birkeland 1996a, Carpenter 1996). Most corallivores are limited in
their rate of feeding because scleractinian corals have only a thin veneer of living tissue over the
surface of an indigestible calcareous skeleton (Keesing 1990). As a consequence, corallivores must
selectively pick live tissues from the surface of corals (which tends to limit the rate of feeding) or
else ingest large quantities of calcium carbonate, which is energetically costly (Motta 1988). In con-
trast, Acanthaster spp. (and other corallivorous starsh) are extremely well adapted to feed on scler-
actinian corals as they can digest tissue from a large area of coral surface at once. These asteroids
feed by everting their stomach through their oral opening and spreading it over the surface of live
corals or any other benthic prey (Jangoux 1982). Enzymes are then secreted through the gastric tis-
sues that digest coral tissues within 3–5 h (Goreau 1964, Brauer et al. 1970). Acanthaster planci has
a much larger stomach for its size than other corallivorous asteroids (e.g., Culcita novaeguineae),
enabling it to consume scleractinian corals two to ve times faster than other starsh of equivalent
size (Birkeland 1989a). Moreover, Benson (1975) suggested that Acanthaster spp. possess the most
specialised enzyme system for digestion of wax esters, which are a major component of coral tis-
sues. Even so, crown- of- thorns starsh consume a maximum of 150–250 cm2 of live coral per day,
depending on their body size (Chesher 1969, Glynn 1973).
A key facet of the feeding behaviour of Acanthaster spp., which has a major bearing on their
ecological impact, is their feeding preferences (Moran 1986). Moran (1986) identied many factors
that inuence feeding preferences of Acanthaster spp., including (1) the nutritional content and
growth form of corals; (2) coral defences (e.g., mesenterial laments, nematocysts, and secondary
metabolites); (3) coral defence by commensal infauna (mostly trapezid crabs); (4) the distribution
of corals; (5) local environmental conditions; and (6) prior conditioning of individual starsh (see
also Birkeland & Lucas 1990). This combination of different factors was expected to lead to com-
plex patterns of feeding preferences, which vary with geographical variation in the composition of
coral assemblages (Birkeland & Lucas 1990) and with size and abundance of the starsh (Moran
1986). To test this, we compiled published data on the proportional consumption of different coral
genera by Acanthaster spp. relative to their availability at different locations throughout the Pacic
(Figure6). These data do not represent feeding preferences per se because they do not strictly assess
the selection of one prey type over another (e.g., Keesing 1990, De’ath & Moran 1998, Pratchett
2007). Moreover, the relative consumption of different coral genera is generally inferred from
Average forage ratio
Figure 6 Variation in forage ratios (averaged across seven studies in the Pacic) for different coral genera,
indicative of general feeding preferences. Forage ratios compare proportional consumption of coral genera to
their relative abundance in the local area, following the work of Tokeshi & Daud (2011). Data sources: Glynn
1974, Bouchon 1985, Keesing 1992, Chess et al. 1997, Pratchett et al. 2009a, Pratchett 2010, Baird et al. 2013.
changes in the abundance or mortality of different coral genera during outbreaks, although many
factors (e.g., routine coral mortality, coral disease, predation by other corallivores, or bleaching)
may have contributed to coral loss (e.g., Pratchett et al. 2009a). However, studies that directly com-
pared mortality rates between sites with and without outbreak densities of A. planci (e.g., Pratchett
et al. 2009a, Pratchett 2010) found that coral mortality was negligible in sites where A. planci were
rare or had only recently been reported.
Variation in forage ratios, averaged over seven studies (Figure6) and six different geographi-
cal locations across the Pacic (Uva Island, eastern Pacic, Glynn 1974; Hawaii, Chess et al. 1997;
French Polynesia, Bouchon 1985; Papua New Guinea, Pratchett et al. 2009a; GBR, Keesing 1992,
Pratchett 2010; Indonesia, Baird et al. 2013), shows that Acropora and Montipora were the most
preferred coral genera. These two coral genera were consistently consumed in greater proportions
than would be expected from their availability. At the other extreme, there were four genera of
scleractinian corals (Hydonophora, Leptastrea, Oxypora, and Turbinaria) as well as Heliopora,
that did not decline in abundance during outbreaks of A. planci, presumably because they are non-
preferred coral prey. Average forage ratios for virtually all other corals (all except Acropora and
Montipora) were negative, although there was variation in apparent preference for individual coral
genera among studies and locations (Figure6). Reported forage ratios for Pachyseris, Astreopora,
Acanthastrea, Psammacora, Lobophyllia, and Diploastrea ranged from −1 to close to 1, indicat-
ing that abundance did not change during outbreaks of Acanthaster planci at some locations, while
there was extensive depletion at other locations. For example, there was a 63% decline in abundance
of Acanthastrea during an outbreak of A. planci at Moorea, French Polynesia, in 1980 (Bouchon
1985), but there was no apparent change in abundance of Acanthastrea during a recent outbreak of
A. planci in Sumatra, Indonesia (Baird et al. 2013).
Despite marked variation in forage ratios for some coral genera, the rank order of coral genera
revealed by comparing average forage ratios are in accord with more detailed studies of feeding
preferences conducted under laboratory conditions (e.g., Brauer et al. 1970, Collins 1975, Ormond
et al. 1976, Pratchett 2007) and eld- based studies of specic feeding behaviour (De’ath & Moran
1998). Prior studies of the feeding preferences of Acanthaster spp. have mostly compared among
Acropora, Pocillopora, and Porites, showing that Acropora is most preferred and Porites is least
preferred (reviewed by Moran 1986). Explanations for these overarching feeding preferences
(e.g., nutritional content, chemical deterrents, or colony defence) are lacking, and they are likely to
vary in their importance depending on the coral taxa (Pratchett 2007). Comparing Acropora spp.
with Pocillopora spp. (and other Pocilloporidae), Pratchett (2001) showed that differential feed-
ing preferences were largely attributable to differences in the effectiveness with which commensal
infauna (and especially trapezid crabs) defend their coral hosts (see also Glynn 1987). Notably,
A. planci did not distinguish between Acropora spp. and pocilloporid corals when commensals
were removed from all corals (Pratchett 2001). Porites spp. also contain symbiotic organisms
(Pedum spondyloideum and Spirobranchus giganteus); rather than preventing A. planci from eating
their host colony, these organisms enhance the survivorship of a few adjacent coral polyps, which
may enable subsequent regeneration of the colony (e.g., DeVantier & Endean 1988). The avoidance
of Porites spp. by Acanthaster spp. is generally ascribed to a combination of low nutritional value,
chemical deterrents to feeding (De’ath & Moran 1998), and the inability of starsh to attach to
large, smooth colonies, leading to high probability of dislodgement except in calm conditions.
The relative preference of Acanthaster spp. for certain corals (e.g., Acropora spp.) could result
in shifts in the structure of coral assemblages during outbreaks. However, even the least- preferred
corals are consumed during extremely severe outbreaks or when coral prey is scarce (e.g., Chesher
1969, Pearson & Endean 1969). This is particularly apparent when comparing effects of outbreaks
of A. planci at Moorea in 1980–1981 (Bouchon 1985) with those of subsequent outbreaks in 2007–
2012 (Kayal et al. 2012). In 1980–1981, Bouchon (1985) recorded proportional declines in the
abundance of the major coral genera Acropora, Pocillopora, and Porites of 45.5%, 0%, and 0%,
respectively. In contrast, Kayal et al. (2012) reported comprehensive extirpation of Acropora spp.
and Pocillopora spp. and greater than 95% decline in local cover of Porites spp. Kayal et al. (2012)
did report that there was sequential depletion of these major genera, with Acropora removed rst,
then Pocillopora, and then Porites. Clearly, selective effects of A. planci are most apparent during
outbreaks that cause only moderate loss of corals (Birkeland & Lucas 1990). The relative consump-
tion of different coral genera also appears to be different for low- density (non- outbreak) populations
of Acanthaster spp. (e.g., Tokeshi & Daud 2011), with feeding preferences appearing to be largely
dictated by the proximity of corals to appropriate shelter.
Outbreaks of crown- of- thorns starsh
Extreme variability in adult abundance is common among marine organisms, particularly those
with planktonic larvae (e.g., Roughgarden et al. 1988). However, few marine organisms show
changes in abundance of the magnitude, or rate, shown by crown- of- thorns starsh. The abun-
dance of Acanthaster spp. can increase by as much as six orders of magnitude within 1 to 2years
(reviewed by Birkeland & Lucas 1990). At Tutuila Island, American Samoa, the overall abundance
of A. planci increased from 1–2 starsh in 1976 to more than 200,000 starsh in late 1977 (Birkeland
& Randall 1979). Similarly, at Tanguisson Reef, Guam, densities of A. planci increased from less
than 0.1 starsh.ha−1 to more than 1,000 starsh.ha−1 during the course of 1967 (Chesher 1969).
More recently, Kayal et al. (2012) reported maximum densities of A. planci of 151,650 starsh/ km2
along the northern barrier reef of Moorea, French Polynesia, with densities increasing more than
10-fold over the course of just 1year. The combined feeding activities of high densities of crown- of-
thorns starsh also cause extensive coral depletion (e.g., Chesher 1969, Kayal et al. 2012), leading
to widespread concern that outbreaks of Acanthaster spp. are becoming more frequent and more
prevalent across the Indo- Pacic (e.g., Brodie et al. 2005).
Despite extensive and increasing reports of ‘outbreaks’ of Acanthaster spp. across the Indo-
Pacic, there are major inconsistencies and deciencies in published reports that constrain rigorous
and comprehensive analyses of the geographical extent and recurrence of outbreaks. These prob-
lems are partly caused by inherent complexities in dening and comparing the extent and severity
of outbreaks among different locations. On the GBR, for example, outbreaks are reported at the
scale of the entire reef system (Sweatman et al. 2008), recognising inherent coupling of outbreaks
among the well- connected reefs that make up this vast reef system (Reichelt et al. 1990a, Moran
et al. 1992). Elsewhere, however, outbreaks are often reported from individual locations, let alone
individual reefs (e.g., Koonjul et al. 2003). The incidence of outbreaks of crown- of- thorns starsh
on small, isolated, or unpopulated reefs is important for understanding the potential causes of (and
anthropogenic contribution to) outbreaks of Acanthaster spp., but there are likely to be many differ-
ent factors that inuence the initiation and impacts of outbreaks, which are rarely considered in pub-
lished reports of localised outbreaks. In the Maldives, for example, outbreaks of Acanthaster planci
were reported at many different atolls in the early 1990s (Ciarapica & Passeri 1993), which runs
counter to Birkeland’s (1982) assertions that outbreaks really only occur on high islands (discussed
in the section on geographical incidence of recent outbreaks). However, no attempt was made to
quantify temporal and spatial patterns in starsh densities, or to assess the population structure of
localised outbreaks, to understand how these outbreaks were initiated or spread among nearby atolls.
Dening outbreaks of crown- of- thorns starsh
Although there are conspicuous differences in the densities of Acanthaster spp. between outbreak-
ing and non- outbreaking populations, rigorous denitions of outbreaks are elusive. Potts (1981)
dened outbreaks of crown- of- thorns starsh as “any large aggregation of many hundreds or thou-
sands of individuals which persist at high densities for months or years and causes extensive mortal-
ity among coral over large areas of reef” (Potts 1981 p. 65). This denition encompasses aspects
of both the biological (pronounced and unexplained increases in the abundance of a species) and
ecological (rapid increases in the abundance of a species beyond that which can be sustained by
local resources) denitions of population outbreaks, but these qualitative denitions are not particu-
larly useful when attempting to rigorously account for the incidence and occurrence of outbreaks.
Normal background densities of Acanthaster planci may be spatially variable (e.g., Glynn 1990),
so it is important to distinguish between periodic outbreaks versus highly localised and chance
aggregations of starsh from populations with persistent moderate densities (Moore 1990). Various
quantitative denitions for outbreaks have thus been proposed (Table 3).
Moran & De’ath (1992) proposed an operational denition of 1,500 star−2 for outbreaks
of Acanthaster planci on Australia’s GBR by estimating the actual densities of starsh (after
accounting for sampling bias) that exceed sustainable limits. Substantial coral mortality was only
observed at reefs with more than 1,500 star−2 (equivalent to 15 starsh.ha−1 or 0.22 starsh
per 2-min manta tow), suggesting that this is the maximum density of starsh that can be sustained
by reefs with average coral cover (Moran & De’ath 1992). Similar estimates of the maximum sus-
tainable density of A. planci (1,000 star−2) were obtained by relating feeding rates of star-
sh to the average annual turnover in well- established coral assemblages (Keesing 1990). It was
recognised, however, that the density of Acanthaster spp. that can be sustained on any given reef
will vary enormously in time and space, depending on the abundance and composition of sclerac-
tinian corals, as well as the size and distribution of starsh. Keesing and Lucas (1992) estimated
that densities of 10–15 starsh per hectare could be sustained in areas with more than 20% coral
cover. However, sustained declines in coral cover on reefs across the Pacic (Bruno & Selig 2007)
mean that outbreaks of Acanthaster spp. that could once be sustained would now contribute to
accelerating coral loss. On Australia’s GBR, for example, average coral cover has declined from
28.0% in 1982 to 13.8% in 2012, caused in large part by outbreaks of Acanthaster planci (De’ath
et al. 2012). Sustained declines in mean coral cover could also increase vulnerability to current and
future outbreaks.
Whereas the density of Acanthaster spp. provides the most relevant measure to compare the
extent and severity of outbreaks (Moran 1986), it is not easy to obtain precise estimates of star-
sh abundance. Notably, density estimates for Acanthaster spp. depend greatly on the temporal
and spatial scales of sampling (Endean & Cameron 1990). Even during outbreaks, the density of
starsh varies within and among reefs (e.g., and among habitats) and may also vary on short tim-
escales within any given location. Changes in the behaviour of starsh after they attain adult size
Table3 Operational criteria used to distinguish outbreak densities
of Acanthaster spp. and non- outbreaking (normal) densities
Minimum threshold Reference
30–40 starsh per km2Clark & Weitzman 2006
14 starsh per 1000 m2Endean & Stablum 1975
40 starsh per 20-min swim Pearson & Endean 1969
100 starsh per 20-min swim or manta tow Chesher 1969
10 starsh per 1-min spot check Pearson & Garrett 1978
30–50 starsh in 20 min Faurea 1989
1 starsh per 24 m2Pearson & Endean 1969
100 starsh per hectare Dana et al.1972
260 starsh per hectare Glynn 1973
150 feeding scars per 250 m2Lison de Loma et al. 2006
(switching from generally cryptic and nocturnal to diurnally active) also signicantly affect the
ease with which starsh can be surveyed and resulting estimates of starsh densities (Endean &
Cameron 1990). Outbreak populations of Acanthaster spp. tend to be highly aggregated, such that
maximum recorded densities may be large, whereas there are few or no starsh at nearby loca-
tions (e.g., Pratchett 2005). It is possible, therefore, that densities of starsh on just one or a few
transects may exceed threshold densities indicative of an outbreak (e.g., 1,500 starsh/ km2; Moran
& De’ath 1992), whereas densities averaged across the entire reef or reef system remain well below
these levels. For this reason, some monitoring studies distinguish between ‘spot’ versus ‘reef- wide’
outbreaks (Engelhardt 1999) or ‘incipient’ versus ‘active’ outbreaks (e.g., Sweatman et al. 2008).
Moran (1986) called for standardised methods for quantifying the abundance of Acanthaster spp.,
recommending repeated and intensive surveys of starsh at individual reefs using area- based sur-
veys to quantify densities, combined with measurements of maximum diameter, to clearly establish
the ne- scale temporal and spatial dynamics of outbreak populations. The few studies that have
conducted intensive sampling at a single reef throughout the course of an outbreak (e.g., Pratchett
2005, Kayal et al. 2012) have provided signicant new insights into the causes and consequences of
outbreaks of Acanthaster spp.
Primary versus secondary outbreaks
Outbreaks of Acanthaster spp. are thought to arise in two different ways (e.g., Potts 1981, Johnson
1992), which will be reected in marked differences in population structure (Figure7). In many
cases, outbreak populations comprise individuals of similar size (<150 mm variation between the
smallest vs. largest individuals) with a unimodal size structure (Figure7B) suggestive of a single
massive inux of new recruits (Birkeland & Lucas 1990). These outbreak populations generally
consist of only a single cohort or year class (e.g., Dana et al. 1972, Glynn 1973, Zann et al. 1987)
and appear as dramatic increases in the abundance of starsh within weeks to months (e.g., Chesher
1969, Branham et al. 1971). However, there are many cases for which outbreak populations include
individuals of a great range of sizes with a multimodal size structure (Figure7A). In some instances,
outbreak populations have been shown to include starsh from at least ve or six different cohorts
(e.g., Stump 1996, Engelhardt 1999, Pratchett 2010), and there has been a gradual increase in the
density of Acanthaster spp. for many years, indicating the progressive accumulation of individuals
over several successive recruitment events (e.g., Zann et al. 1987, 1990, Stump 1996, Pratchett 2005).
Factors that contribute to instantaneous increases in the recruitment of Acanthaster spp.
(e.g., increased survivorship of starsh larvae; Lucas 1973, Birkeland 1982), leading to rapid and
dramatic increases in starsh densities, are likely to be different from those that cause slow, pro-
gressive increases in starsh densities (Johnson 1992). Importantly, the sustained and gradual accu-
mulation of crown- of- thorns starsh from multiple successive recruitment events may represent a
mechanism by which outbreaks are initiated (‘primary outbreaks’; Endean 1974, Johnson 1992,
Stump 1996), which then give rise to massive numbers offspring causing subsequent outbreaks on
nearby and downstream reefs (‘secondary outbreaks’; Endean 1974) and cause massive devastation
over large areas (Reichelt et al. 1990a). Birkeland and Lucas (1990) suggested that gradual increases
in the densities of Acanthaster spp. over several years (regardless of nal densities) do not really
t with the denition of an outbreak, arguing that ‘outbreaks’ must arise suddenly. The distinction
between primary versus secondary outbreaks, however, might be fundamental to understanding the
initiation of outbreaks of Acanthaster spp. (e.g., Pratchett 2005, Fabricius et al. 2010) as distinct from
waves of outbreaks that are a predictable consequence of large, established breeding populations.
Endean’s (1974) distinction between primary and secondary outbreaks, although logical, is not
easy to apply because ecologists rarely have sufcient information to identify the sources of out-
breaks with certainty, so the identication of primary outbreaks is largely conjectural. Intensive
monitoring of Acanthaster planci at Lizard Island, in the northern GBR, during the 1990s provided
a good example of a primary outbreak: The number of starsh increased gradually until it reached
outbreak densities in 1996 (Sweatman et al. 1998, Pratchett 2005). Moreover, at peak densities
of crown- of- thorns starsh, the local population comprised individuals ranging in size from 110
to 620 mm, and there was evidence of repeated annual recruitment from 1992 to 1997 (Pratchett
2005). The source of recruits settling at Lizard Island is not known, but they may represent the
progeny of the adult starsh that were already present on reefs around Lizard Island (albeit in low
densities) prior to the outbreak. With a high level of self- recruitment, these outbreaks may have
resulted from incremental increases in the reproductive output of the initial reproductive population
and all of their subsequent progeny, leading to exponential growth in population size. Detailed stud-
ies of this outbreak (e.g., Stump 1996, Pratchett 2005, 2010) did not provide any real clues regarding
Proportion of individuals
50 100 150 200 250 300 350 400 450 500
Proportion of individuals
Diameter (mm)
50 100 150 200 250 300 350 400 450 500
Diameter (mm)
Figure 7 Contrasting size structure of outbreak populations of Acanthaster planci in (A) July 1995 at Lizard
Island on Australia’s Great Barrier Reef (Stump 1996) and (B) September 1985 at Suva, Fiji (Zann et al. 1987).
At Lizard Island, the population comprised individuals ranging in size by more than 430 mm (n = 317) and
multiple age classes, suggestive of gradual accumulation of starsh over several years (i.e., primary out-
break; Endean 1977). In Fiji, the entire population comprised starsh ranging in size by less than 150 mm
(n = 158), dominated by a single year class and indicative of a single mass recruitment (i.e., ‘secondary out-
break; Endean 1977).
the cause(s). However, given the gradual accumulation of starsh over more than 5years, any factor
responsible for the initial onset of outbreaks is likely to be subtle and difcult to detect.
Secondary outbreaks of Acanthaster spp. are known mainly from the GBR and southern Japan
(Potts 1981), where waves of outbreaks spread among well- connected reef systems after the initia-
tion of one or more primary outbreaks (e.g., Kenchington 1977, Yasuda et al. 2009). However, out-
break populations with highly constrained size structure, characteristic of secondary outbreaks, have
also been reported at many small and isolated reef systems throughout the Pacic (e.g., Birkeland
1982), suggesting that the extensive dispersal of larvae from primary outbreaks may initiate suc-
cessive outbreaks in this area. To evaluate this, Timmers et al. (2012) explored genetic structure
and connectivity among outbreak populations of Acanthaster planci at 23 sites across the Pacic
Ocean, including the north- western (e.g., Guam), north- central (e.g., Hawaii), and south- central
Pacic (French Polynesia) regions. Strong regional, archipelagic, and interreef genetic structuring
of A. planci populations indicate that larval dispersal is highly constrained, and that high densities
of larvae do not spread across open ocean expanses to initiate secondary outbreaks at distant reefs
(see also Vogler et al. 2013). These data suggest either that primary outbreaks have gone undetected
within many small archipelagos or that, in some circumstances, primary outbreaks may result from
a single massive inux of larvae (Birkeland 1982).
Recent advances in genetic research, including a now- extensive library of microsatellite mark-
ers specic for Acanthaster planci (Yasuda et al. 2006, 2007, Wainwright et al. 2012), provide
greatly improved opportunities to study the initiation (e.g., source populations) and subsequent
spread of outbreaks. As such, future studies of outbreak populations not only should record the
density of Acanthaster spp. but also should measure maximum diameter and obtain genetic samples
(6–10 tube feet placed in 95% ethanol) for a representative sample of individual starsh.
Geographical incidence of recent outbreaks (1990–2013)
Since 1990, at least 246 outbreaks of Acanthaster spp. have been reported from across the Indo-
West Pacic (Figure8), which is three times the number of outbreaks reported (82) prior to 1990
(e.g., Moran 1986, Birkeland & Lucas 1990). Although mostly qualitative, these recent reports rep-
resent outbreaks or infestations that are distinct in space and time, at either a particular group of
reefs (mostly at the scale of small island nations in the Pacic) or discrete episodes of outbreaks
within a given geographic location (e.g., outbreaks of Acanthaster planci are reported in the Society
Islands, in French Polynesia in 1969–1970, 1979–1982, and 2006–2009). This apparent increase in
the number of reported outbreaks may or may not indicate a real increase in the spatial or temporal
incidence of outbreaks; it is surely attributable in part to an increase in surveys and monitoring of
reef environments, combined with increased awareness of Acanthaster spp. in this more recent
period (see also Baird et al. 2013).
One of the key arguments put forward as evidence of the role of humans in causing or exac-
erbating outbreaks of crown- of- thorns starsh is that the incidence of devastating outbreaks has
increased in recent times (e.g., Brodie et al. 2005). However, in locations where there have been
recurrent outbreaks since the 1960s (e.g., GBR, Japan, and Society Islands), there is no evidence of
increased frequency of outbreaks (or more specically, a decrease in the period between successive
outbreaks). On the GBR, it was 17years between the start of the rst recorded outbreak (1969)
and the emergence of the second major outbreak (1979), and it was the same period (17years)
between the start of the third (1993) and start of the most recent, fourth, major outbreak (2010).
In French Polynesia, it was only 10years between the rst two reported outbreaks (1969–1970 vs.
1979–1982), but then 27years until the devastating outbreak that occurred in 2006–2009 (Kayal
et al. 2012). Similarly, in Japan it was almost 20years between the most recent wave of outbreaks
and previous high- frequency, almost- chronic outbreaks that occurred throughout the 1960s and
Tanzania Seychelles
Red Sea
Bay of
40°E 60°E 80°E 100°E 120°E 140°E 160°E 180° 160°W 140°W 120°W 100°W 80°W 60°W
160°E 120°E 130°E 140°E
120°E 130°E 140°E
120°E 130°E 140°E
120°E 130°E 140°E150°E 160°E
0 400
0 400
0 400
0 3,000
Solomon Islands SEA OF JAPAN
Australia Fiji
New Caledonia
Papua New Guinea
20°N15°N10°N5°N5°S 30°N20°N10°N10°s20°s30°S
30°N 40°N20°N10°N0°N10°s20°s30°S
40°E 60°E 80°E 100°E 120°E 140°E 160°E 180° 160°W 140°W 120°W 100°W 80°W 60°W
(C) (D)
Figure 8 Geographical spread of reported outbreaks of Acanthaster spp. Each dot indicates a specic reef where outbreaks of crown- of- thorns starsh have been
reported since 1990. No account is made of the extent, severity, or recurrence of outbreaks at each location.
early 1970s (Yasuda et al. 2009). If anything, the period between successive outbreaks is becoming
longer, as might be predicted if the initiation of renewed outbreaks requires substantial recovery of
coral prey (e.g., Bradbury et al. 1985, Bradbury & Antonelli 1990, Fabricius et al. 2010).
Global degradation of coral reef ecosystems, especially sustained declines in the abundance or
cover of scleractinian corals, is widely attributed to the increasing frequency of acute disturbances
(e.g., Seymour & Bradbury 1999, Fabricius et al. 2010), such that major disturbances occur too often
to allow effective (although not necessarily complete) recovery of coral assemblages between succes-
sive disturbances. Moreover, the time required for coral assemblages to recover from periodic distur-
bances might be increasing (Seymour & Bradbury 1999). The minimum interval between outbreaks
of Acanthaster spp. at individual reefs will be set by local recovery of coral assemblages (Bradbury
et al. 1985, Fabricius et al. 2010) because low levels of coral cover would prevent outbreaks from
becoming established. This implies that the overall incidence of outbreaks of Acanthaster spp.
should decline with sustained and increasing degradation of coral reef ecosystems. In particular,
there should be longer periods between successive outbreaks on individual reefs. Alternatively, lim-
ited abundance of coral prey might constrain the overall abundance or persistence of Acanthaster
spp. while outbreaks continue to occur at equivalent or increasing frequency (Fabricius et al. 2010) or
might even become chronic (e.g., Zann et al. 1990, Mendonça et al. 2010), thereby causing accelerated
degradation of reef ecosystems. The current wave outbreaks on the GBR provide a good opportunity
to test whether outbreaks of Acanthaster planci can become established on reefs with low coral
cover, as many reefs in the central GBR were impacted by a severe tropical storm (category 5 cyclone
Yasi) that tracked across the GBR in early 2011 (e.g., Lukoschek et al. 2013). Coral cover remains at
or below 5% on many reefs exposed to this disturbance, and preferred coral prey of Acanthaster spp.
(e.g., Acropora) are particularly scarce (e.g., Lukoschek et al. 2013).
Despite increases in the overall number of reported outbreaks, the geographical distribution of
outbreaks reported in 1990–2012 is fairly consistent with the record prior to 1990 (e.g., Moran 1986,
Birkeland & Lucas 1990). Most notably, the vast majority (167/246) of outbreaks of Acanthaster
spp. are reported from the Pacic Ocean (Figure9), mainly from the western Pacic, including
Australia’s GBR, southern Japan, Micronesia, and Melanesia. There has been a slight increase in
the proportion of outbreaks reported from the Indian Ocean, increasing from 11.0% in 1956–1989 to
Red Sea Indian IAA Pacic
Total number of reported outbreaks
Pre-1990 (n = 86)
Post-1990 (n = 246)
Figure 9 Number of reported outbreaks in different regions of Indo- Pacic (Red Sea, Indian Ocean, Indo-
Australia Archipelago, and Pacic Ocean) pre-1990 versus post-1990. No account is made of the extent, sever-
ity, or recurrence of outbreaks at each location.
15.4% in 1990–2012, with unprecedented outbreaks of Acanthaster spp. recorded in some isolated
locations, such as Chagos (R. Roche unpublished data 2013). In most locations, outbreaks have now
been reported at least twice, and up to four times, since the 1960s within the same area, if not the
same reef, showing that outbreaks of Acanthaster spp. are not rare, singular events. The greatest
concentrations of outbreaks have occurred within extensive reef systems of the GBR and southern
Japan and within the Philippines (Figure8). However, there have also been recent outbreaks on
many small, isolated, and relatively unpopulated reef systems.
Analyses of all reported outbreaks since 1990 showed that the majority of outbreaks (139 of
246) have occurred on arrays of platform reefs along continental shelves, such as the GBR. An addi-
tional 72 (56%) outbreaks have been reported on high islands, such as Moorea in French Polynesia.
The remaining outbreaks (35 of 246) have been reported from low- lying islands, atolls, or com-
pletely submerged reef platforms. It is likely that the causes, especially anthropogenic contribu-
tions to the initiation of outbreaks, are likely to vary among these different reef systems (Birkeland
1982). Notably, continental and high island reefs will be much more subject to terrestrial run- off
(Birkeland 1982), which is one possible cause of outbreaks (see the section on the nutrient enrich-
ment hypothesis). The ip side to this is that human populations on low- lying islands and atolls have
reduced capacity for agriculture and are much more reliant on coral reef sheries (McClanahan
et al. 2002), and excessive exploitation of predatory shes has also been linked to increased severity
of outbreaks of A. planci (Dulvy et al. 2004). To assess whether outbreaks occur more or less than
expected on different types of reefs, the relative frequency of outbreaks needs to be compared to the
proportional area of continental and high island reefs versus low- lying atolls and submerged reefs,
but such data are not readily available.
Fine- scale patterns of outbreaks
Australia’s Great Barrier Reef, the GBR, extending 2,000 km along Australia’s eastern coast, is
the world’s largest reef system and has the dubious distinction of supporting both the most exten-
sive and the greatest frequency of outbreaks of Acanthaster spp. The rst documented outbreak
was detected in 1962 at Green Island (Pearson & Endean 1969), although high densities of star-
sh were reported on other GBR reefs in the 1950s and earlier (Vine 1973, Ganter 1987). Since
1962, there have been three additional outbreak episodes on the GBR, commencing in 1979, 1993,
and 2010. The initiation and progression of the rst two outbreaks, based on extensive but largely
uncoordinated sampling across the GBR, was discussed at length by Moran (1986). The publica-
tion of Moran’s review in 1986 roughly coincided with the advent of systematic monitoring on the
GBR to document patterns of outbreaks of Acanthaster planci in time and space (e.g., Sweatman
et al. 2008). When monitoring began in 1985, there were active outbreaks on many reefs near
Townsville (~18.5°S) and in the Swain reefs well off shore from Gladstone (~22°S) (Figure 10).
Subsequent monitoring documented the progressive spread in the southward extent of active out-
breaks (Figure10), consistent with patterns reported in previous outbreaks (Reichelt et al. 1990a).
However, outbreaks in the Swain region (southern GBR) appear to occur independently of the wave
of outbreaks that affect reefs from Cooktown to Mackay ( docs/ research/
biodiversity- ecology/ threats/ cots- animation.html).
The initiation and spread of outbreaks of Acanthaster planci on the GBR have been fairly con-
sistent in all four recorded outbreaks, including the current outbreak that started in 2010. Each of the
outbreaks appears to have been initiated on midshelf reefs in the north- central region of the GBR
(the ‘initiation box; Figure11) between Lizard Island (14.6°S) and Cairns (17°S) and mainly in the
northern portion of this area. During the initial waves of outbreaks (in 1962 and 1979), high densi-
ties of A. planci were rst detected (or at least reported) on reefs close to Cairns (e.g., Green Island),
although increases in densities of A. planci may have occurred even earlier on reefs to the north, as
recorded during the start of outbreaks in 1993 and 2010. Limited temporal and spatial resolution of
monitoring of GBR reefs (e.g., Sweatman et al. 1998) mean that it is still unclear whether outbreaks
start from a single reef within this area or arise simultaneously on a suite of closely positioned reefs
(Pratchett 2005). In 1994, outbreaks of Acanthaster spp. were rst recorded at Lizard Island but
also were reported soon after (within the same year) at several other reefs within the immediate
area, including Linnet, North Direction, and Rocky Islet (Sweatman et al. 1998). The exact timing
and location of primary outbreaks are critical for understanding and identifying potential causes
or triggers of outbreaks on the GBR, which suggests that this area should be monitored intensively
whenever initiation of a new wave of outbreaks seems possible. Alternatively, the pattern of initia-
tion and spread of outbreaks may be reconstructed using the size- frequency distribution of high-
density populations on disparate reefs (sensu Kenchington 1977), combined with newly developed
genetic markers (Yasuda et al. 2006, Wainwright et al. 2012) to potentially map the directionality of
dispersal among reefs within the initiation box.
The recurring pattern on the GBR is the southward progression of outbreaks from 17–20°S, at
a rate of 1° of latitude every 3years (Kenchington 1977, Reichelt et al. 1990a). This is attributed to
progressive colonisation of reefs by larvae spawned at the outbreak front, which disperse only 1–2°
of latitude (Kenchington 1977). Outbreaks of crown- of- thorns starsh also appear to spread north-
ward from the initiation area, but the northward progression is slower and less consistent (Moran
1992), although reefs in the northern region have not been surveyed as frequently as those in other
regions of the GBR. The progression of outbreaks northward and southward on the GBR has been
attributed to movement of adult starsh between reefs (e.g., Talbot & Talbot 1971), but differences
No A. planci
Few A. planci
Incipient outbreak
Active outbreak
Latitude (degrees S)
Figure 10 Occurrence of Acanthaster planci on reefs of the Great Barrier Reef (arranged by latitude)
between 1985 and 2013 based on manta tow surveys of reef perimeters. Note that the same reef may be
represented in several years. Reefs are divided into four categories: dots indicate reefs with low densities of
A. planci (≤3 per reef); smaller grey symbols indicate reefs with more than 3 starsh but less than incipient
outbreak densities (mean ≥ 0.22 A. planci per 2-min tow); larger grey symbols indicate reefs with incipient out-
break densities of A. planci (mean ≥ 0.22 A. planci per tow); large unlled symbols indicate reefs with active
outbreak densities (mean ≥ 1.0 A. planci per tow).
in the size- frequency distribution of disparate populations seem more likely to be the result of larval
dispersal (Kenchington 1977), with prevailing currents (Black & Moran 1991).
Disturbances caused by outbreaks of crown- of- thorns starsh
Acanthaster spp. have gained considerable notoriety, not only because of their tendency to undergo
rapid and dramatic increases in populations size but also because outbreak populations can cause
extensive and widespread depletion of scleractinian corals (e.g., Chesher 1969, Pearson & Endean
1969, Randall 1973, Colgan 1987). At low densities (<10 starsh.ha−1), Acanthaster spp. have a neg-
ligible impact on scleractinian coral cover (e.g., Glynn 1973, Zann et al. 1990). However, the com-
bined feeding activities of high densities of crown- of- thorns starsh (up to 151,650 starsh/ km2;
Kayal et al. 2012) can cause rapid and pronounced coral loss. In Guam, for example, high densi-
ties (>1000 starsh per hectare) of Acanthaster planci persisted for 30months in 1967–1969 and
killed virtually all scleractinian corals (>90% coral mortality) from 1 to 68 m depth along a 38-km
stretch of coastline (Chesher 1969). Similarly, in French Polynesia, large aggregations of A. planci
145°E 150°E
150 300 km
0.14 – 0.24
Chlorophyll (µg L–1)
0.25 – 0.32
0.33 – 0.56
0.57 – 0.96
Figure 11 (See also colour gure in the insert) Map of Australia’s Great Barrier Reef showing the varia-
tion in average (‘normal’) patterns of productivity (chlorophyll a concentrations). Data source: E- atlas (http://
maps.e- Also shown is the small area north of Cairns (the ‘initiation box’) where primary out-
breaks of Acanthaster planci are thought to be initiated and then spread (mostly southwards), causing progres-
sive waves of secondary outbreaks (e.g., Fabricius et al. 2010).
systematically moved around the entire circumference of Moorea and removed more than 96% of
the coral (Kayal et al. 2012). Whereas climate- related disturbances (e.g., bleaching and disease)
are a major concern for reef managers around the world, especially since 1998 (Knowlton 2001),
outbreaks of Acanthaster spp. continue to occur throughout the Indo- Pacic region (e.g., Pratchett
2005, Baine 2006) and have so far been responsible for far more coral mortality than has been
attributed to climate- related disturbances (Bruno & Selig 2007, Osborne et al. 2011, Trapon et al.
2011, Death et al. 2012).
Although outbreaks of Acanthaster spp. can cause massive and widespread coral depletion
(e.g., Chesher 1969, Pearson & Endean 1969, Randall 1973, Colgan, 1987), this is not always the
case. Outbreaks of Acanthaster spp. vary greatly, not only in the number, size, and density of star-
sh, but also in their effects (Moran 1986). For example, high densities of Acanthaster planci per-
sisted for more than 18months (1969–70) at Molokai Island, Hawaii, but had negligible impacts
on the abundance of scleractinian corals (Branham et al. 1971). Similarly, in Panama, high- density
populations of A. planci caused minimal reductions in the live cover of scleractinian corals (Glynn
1974, 1976). Large aggregations of many hundreds to thousands of crown- of- thorns starsh have
been reported on reefs throughout the Indian and Pacic oceans, including Panama (Glynn 1974,
1976); Samoa (Birkeland & Randall 1979); Micronesia (Chesher 1969, Colgan 1987); southern
Japan (Nishihira & Yamazato 1974, Keesing 1992); the GBR (e.g., Moran et al. 1988, Reichelt
et al. 1990a); Cocos- Keeling Islands (Colin 1977); and the Red Sea (Ormond & Campbell 1974).
However, incidences of large- scale destruction of scleractinian corals by outbreaks of Acanthaster
spp. have occurred primarily within the southern and western Pacic (Moran 1986, Birkeland &
Lucas 1990). More specically, devastating outbreaks of Acanthaster spp. have been most apparent
in southern Japan (Nishihira & Yamazato 1974, Keesing 1992), on Australia’s GBR (Pearson &
Endean 1969), in Micronesia (Chesher 1969, Colgan 1987), and in French Polynesia (Trapon et al.
2011, Kayal et al. 2012).
Variation in the effects of outbreaks of Acanthaster spp. in different regions of the Pacic might
be explained by the relative dominance of Acropora in local coral assemblages (Birkeland & Lucas
1990); Acropora tends to dominate in the southern and western Pacic and is consistently among
the corals that are rst and worst affected during outbreaks (e.g., Pratchett et al. 2009a, Pratchett
2010). In the northern and eastern Pacic (e.g., the main Hawaiian islands and Panama), Acropora
is relatively scarce and coral assemblages tend to be dominated by Pocillopora, which is much less
susceptible to crown- of- thorns attack (e.g., Glynn 1974, 1976), although it can be damaged exten-
sively by extremely severe outbreaks (e.g., Adam et al. 2011, Kayal et al. 2012). Geographic varia-
tion in the effects of Acanthaster spp. may also result from differences in the population dynamics
and behaviour among the four nominal sister species distributed in different parts of the Indo-
Pacic (Vogler et al. 2008). Acanthaster spp. from throughout the Indo- Pacic ostensibly look and
behave the same way, but devastating impacts of crown- of- thorns starsh appear to be conned to
the Pacic, which is the geographical range of Acanthaster planci. This warrants explicit compari-
sons of demographic rates, feeding rates, and feeding preferences among Acanthaster spp. from each
of the four distinct subpopulations identied by Vogler et al. (2008), extending the studies in the
Pacic to the Red Sea and both southern and northern Indian Ocean regions.
Effects of Acanthaster spp. can also vary at much smaller scales, within and among adjacent
coral reefs (e.g., Pratchett 2010). Highest densities of Acanthaster spp., and hence the greatest
depletion of live coral cover, tend to occur on the leeward side of reefs (Laxton 1974, Pratchett
2005), which is caused by either (1) high abundance of preferred coral prey (e.g., monospecic
stands of staghorn Acropora) in these habitats or (2) reduced water ow and turbulence in back
reef environments, which can potentially dislodge starsh while feeding in more exposed loca-
tions (Endean & Stablum 1973). However, crown- of- thorns starsh are rarely found within shallow,
semi- enclosed lagoons, which often support many suitable coral prey (e.g., Pratchett 2005, Pratchett
et al. 2011), suggesting that other environmental variables also inuence their ne- scale distribu-
tion. Acanthaster spp. have rarely been observed on mesophotic (>30 m deep) reefs of the GBR
(T. Bridge, unpublished data 2013), even though these habitats are extensive (Harris et al. 2013) and
can support a high cover of scleractinian corals (Bridge et al. 2013). These habitats could provide
refuges for coral species to reseed shallow water reef habitats in the aftermath of devastating out-
breaks of Acanthaster spp. (Bridge et al. 2013).
Directional shifts in coral composition
Disturbances inuence the structure of ecological communities either through selective mortal-
ity of particular species or by random, localised mass mortality across a wide range of different
species (often termed ‘catastrophic mortality’) that clears space for recolonisation (Petraitis et al.
1989). Small- scale or relatively discrete disturbance events (e.g., predation events) usually have a
disproportionate impact on certain individuals or species (Petraitis et al. 1989) and so exert a strong
structuring inuence on populations and communities. Such events may, for example, increase
diversity and promote coexistence of species by reducing the abundance of competitively dominant
species and allowing inferior competitors to persist (e.g., Porter 1972, 1974). However, disturbances
that differentially affect species may also reduce diversity by disproportionately affecting rare
species, thereby increasing the dominance of already- abundant species (e.g., Glynn 1974, 1976).
Catastrophic disturbances, meanwhile, eliminate most (if not all) of the species in an area and
may contribute to increased species diversity by preventing dominant species from monopolising
all available resources (Petraitis et al. 1989). Although effects of Acanthaster spp. on coral assem-
blages are essentially a predatory interaction, resulting loss of scleractinian corals can be cata-
strophic (e.g., Chesher 1969), and the specic effects will depend on the frequency and severity of
major outbreaks.
Endean and Cameron (1985) suggested that the severity of any given disturbance should be
judged not by how much coral is killed but by the type of coral killed. Notably, the removal of
long- lived and slow- growing species (e.g., Porites spp.) is likely to have longer- term impacts on
community structure compared with selective removal of short- lived, fast- growing species (Endean
& Cameron 1985). Like most disturbances (e.g., freshwater plumes, Jokiel et al. 1993; coral bleach-
ing, Marshall & Baird 2000), outbreaks of Acanthaster spp. tend to have a disproportionate impact
on fast- growing coral species (e.g., Acropora), which also recruit abundantly and often recover
rapidly in the aftermath of major disturbances (e.g., Linares et al. 2011). However, Acanthaster
planci do sometimes feed on large and old colonies of massive Porites (Done 1985). The specic
effects of crown- of- thorns outbreaks (and other recurrent disturbances) on the composition of coral
assemblages will depend on the frequency and severity of these disturbances. Severe but infrequent
disturbances are likely to have a disproportionate effect on slow- growing species that are inca-
pable of recovering between successive disturbances (Done 1985). However, frequent moderate
outbreaks (especially in isolated locations) are likely to cause the localised extirpation of corals that
are favoured by Acanthaster spp. (e.g., Berumen & Pratchett 2006, Pratchett et al. 2011).
A striking example of persistent shifts in the structure of coral assemblages has been docu-
mented in Moorea, French Polynesia, where a major outbreak of Acanthaster planci in 1980–1981
led to the disproportionate loss of Acropora corals (Bouchon 1985). Ongoing disturbances (including
cyclones, coral bleaching, and a further outbreak of A. planci) have since prevented recovery of
Acropora spp., such that the dominant coral genera in 2009 were Pocillopora and Porites (Berumen
& Pratchett 2006, Adjeroud et al. 2009, Pratchett et al. 2011, 2013), which were relatively unaffected
by outbreaks of A. planci. Given sufcient time between disturbances, Acropora might be expected
to recover eventually and regain its former dominance in Moorea. However, Pratchett et al. (2011)
measured recruitment rates for Acropora and other dominant coral genera (Porites and Pocillopora)
in Moorea and showed that the relative abundance of new recruits strongly reected the current
patterns of adult abundance. This is evidence of a positive- feedback mechanism, which is likely to
reinforce and sustain the altered community structure (Knowlton 1992, Nyström et al. 2008).
Selective depletion of certain coral species, linked to marked feeding preferences of Acanthaster
spp. (e.g., Brauer et al. 1970, Collins 1975, Ormond et al. 1976, Colgan 1987, Keesing 1990, De’ath
& Moran 1998), has the capacity to greatly alter coral diversity (e.g., Porter 1972, 1974, Glynn
1974, Colgan 1987) and therefore habitat heterogeneity. In the eastern Pacic, Acanthaster planci
tend to avoid the most abundant coral, Pocillopora (Glynn 1974, 1976, 1980). By feeding mostly on
rare coral species, this further increases the dominance of Pocillopora, leading to declines in coral
diversity (see also Branham et al. 1971). In the Indo- West Pacic, however, A. planci feeds predomi-
nantly on Acropora spp. and Montipora spp. (e.g., Ormond et al. 1976, Colgan 1987, Keesing 1990,
De’ath & Moran 1998, Pratchett 2010), which are relatively abundant and often competitively domi-
nant corals. This presumably increases the prevalence of other less- abundant coral species, which
Porter (1972, 1974) suggested might lead to increases of coral diversity. Pratchett (2010) explicitly
tested for changes in coral diversity during a moderate outbreak of Acanthaster planci at Lizard
Island, northern GBR, and showed that despite disproportionate effects on preferred corals, coral
diversity declined (rather than increased) with declines in coral cover. Contrary to Porter’s (1972)
suggestion, it seems unlikely that outbreaks of A. planci (even relatively moderate outbreaks) would
ever cause increases in coral diversity because crown- of- thorns starsh are not sufciently averse
to rare corals. In the absence of preferred corals, Acanthaster spp. will certainly feed on other less-
preferred corals (Moran 1986, Keesing 1990, Endean & Cameron 1990), so further outbreaks on
highly degraded reefs are likely to lead to even more extensive depletion of corals, with collateral
effects on the local diversity of other reef- associated organisms.
Indirect effects of outbreaks of crown- of- thorns starsh
Extensive coral depletion caused by outbreaks of Acanthaster spp., as well as directional shifts in
the composition of coral assemblages, have broad impacts on a wide variety of coral reef organ-
isms. Outbreaks of crown- of- thorns starsh have been linked to increased abundance of soft corals
(e.g., Endean 1971, Chou & Yamazato 1990), algae (Larkum 1988), urchins (Belk & Belk 1975),
and herbivorous sh species (Endean & Stablum 1973, Wass 1987), while causing declines in abun-
dance of coral- dependent shes (e.g., Sano et al. 1984, 1987, Williams 1986, Munday et al. 1997)
and motile invertebrates (Garlovsky & Bergquist 1970). Changes in the abundances of these reef
organisms are the indirect result of massive reductions in the abundance of scleractinian corals. For
example, increases in the abundance of urchins (specically, Echinometra mathaei and Diadema
spp.) following outbreaks of Acanthaster planci have been related to increased food availability, as
algae colonise skeletons of dead, but intact, corals (e.g., Belk & Belk 1975, Larkum 1988). Given
the potential severity of starsh outbreaks, it is not surprising that many coral reef organisms are
indirectly affected.
Most studies that have considered secondary effects of outbreaks of Acanthaster spp. have mea-
sured changes in the abundance of coral reef shes associated with localised coral loss (e.g., Williams
1986, Sano et al. 1987, Hart et al. 1996, Munday et al. 1997, Adam et al. 2011, Pratchett et al. 2012).
Aside from herbivorous sh species (e.g., Endean & Stablum 1973, Wass 1987, Adam et al. 2011),
most shes tend to decline in abundance in the aftermath of major outbreaks of Acanthaster spp.
that cause extensive coral loss (e.g., Bouchon- Navaro et al. 1985, Williams 1986, Sano et al. 1987,
Munday et al. 1997). These effects are most pronounced for specialist shes that depend on corals
for food (e.g., butteryshes; Williams 1986) or habitat (e.g., coral- dwelling gobies, Munday et al.
1997; coral- dwelling damselshes, Pratchett et al. 2012). At least 133 species (and 11 different
families) of coral reef shes feed on scleractinian corals (Cole et al. 2008), the majority (69 species)
of which are butteryshes (family Chaetodontidae). Also, 320 species (from 39 different families)
have been shown to use live corals as habitat (Coker et al. 2014). These data suggest that 8–10%
of coral reef shes will be directly and adversely affected by extensive coral depletion (see also
Munday et al. 2008). Localised depletion of preferred cover may ultimately lead to local or global
extinction of shes that are directly reliant on corals (e.g., Kokita & Nakazono 2001, Munday 2004).
This is particularly so for highly specialised shes that rely on only a limited suite of coral species,
although it is important to look at other biological traits that may offset extinction risk in these spe-
cies (Lawton et al. 2011).
The effects of severe outbreaks of Acanthaster planci also extend well beyond the few shes
that are directly dependent on live corals (e.g., Sano et al. 1987), especially where coral depletion
is associated with loss of habitat structure. Extensive coral depletion caused by large or persistent
outbreaks of A. planci (e.g., Chesher 1969, Pearson and Endean 1969, Colgan 1987, Sano et al.
1987) almost invariably leads to marked declines in habitat and topographical complexity, which
are critical for sustaining a high diversity of reef shes and other reef- associated organisms (Wilson
et al. 2006, Pratchett et al. 2009b). Once dead, the exposed skeletons of scleractinian corals are sus-
ceptible to biological and physical erosion (Hutchings 2011). Over time, skeletons of erect branch-
ing corals (e.g., Acropora and Pocillopora) break down into coral rubble (Sheppard et al. 2002),
whereas more robust skeletons of massive corals (e.g., Porites) may become dislodged or gradually
eroded insitu (Sheppard et al. 2002). The structural collapse of dead coral skeletons takes 4–7years
(Pratchett et al. 2008), and if there is no substantial recovery of corals in the meantime, then the cor-
responding loss of habitat structure and topographic complexity can have far- reaching effects on the
abundance and diversity of shes (Sano et al. 1984, 1987, Pratchett et al. 2009b). In southern Japan,
for example, Sano et al. (1987) reported more than 65% fewer individuals and species of shes at a
reef that had been devastated by localised outbreaks of Acanthaster planci, compared with nearby
reefs with extensive growth of staghorn Acropora. On the reef that had been devastated by A.
planci, extensive stands of Acropora corals were rapidly eroded, converting once highly complex
3-dimensional habitats into at, homogeneous rubble elds (Sano et al. 1987).
Causes of Acanthaster outbreaks
Unifying theories for population outbreaks were proposed by scientists working in terrestrial
environments long before outbreaks of the crown- of- thorns starsh were even known to occur
(e.g., MacArthur 1955, Elton 1958). Both MacArthur (1955) and Elton (1958) argued that popula-
tion outbreaks are manifestations of inherent instability within certain systems, attributed to either
(1) particular life- history characteristics (e.g., high fecundity, short generation times, high mortal-
ity during their early life history, and generalised patterns of prey and habitat use) that predis-
pose an organism to major uctuations in population size or (2) major changes in the physical
or biological environment that release the outbreaking population from usual regulating factors
(e.g., Andrewartha & Birch 1984, Berryman 1987). Numerous hypotheses have been put forward
to explain the occurrence of population outbreaks of Acanthaster spp. (reviewed by Moran 1986,
Birkeland & Lucas 1990). These hypotheses generally fall into two groups that place importance
either on factors affecting recruitment rates (i.e., ‘natural causes hypothesis’, Vine 1973; ‘larval
recruitment hypothesis’, Lucas 1973; ‘terrestrial run- off hypothesis’, Birkeland 1982) or on changes
in the behaviour or survivorship of post- settlement individuals (i.e., ‘predator removal hypothesis’,
Endean 1969; ‘adult aggregation hypothesis’, Dana et al. 1972; ‘prey- threshold hypothesis’, Antonelli
& Kazarinoff 1984). Whereas several of these hypotheses have been considered biologically improb-
able (e.g., Potts 1981, Birkeland & Lucas 1990), no single hypothesis has universal support. Sudden
and dramatic increases in the abundance of starsh must involve successful recruitment (Birkeland
& Lucas 1990), but both pre- and post- recruitment processes are likely to contribute to the dynamic
nature of Acanthaster populations (Bradbury & Antonelli 1990), as has also been shown for many
other marine organisms (e.g., Jones 1987, 1991, Hughes 1990). Many biologists and theoretical ecol-
ogists concur that single- factor hypotheses that seek to explain the occurrence of crown- of- thorns
outbreaks in all locations and at all times are likely to oversimplify the population dynamics for this
organism (reviewed by Birkeland & Lucas 1990, Bradbury & Antonelli 1990). It is also important
to recognise that Acanthaster spp., probably more so than any other coral reef organism, are predis-
posed to major uctuations in population abundance (Birkeland 1989b).
Given the life- history characteristics of Acanthaster spp., it is almost harder to explain the per-
sistence of low- density populations than it is to explain outbreaks (Endean & Cameron 1990). On
any given reef, it is likely that outbreaks will occur periodically through the effects of random envi-
ronmental variation on reproductive success or larval survival. As discussed previously, Acanthaster
spp. have extremely high fecundity (e.g., Conand 1983, 1984), and fertilisation rates, as well as
developmental rates and survivorship of larvae, are highly subject to the vagaries of local environ-
mental conditions. Moore (1990) examined the characteristics of key locations where Acanthaster
spp. occur, but never in outbreak densities, and suggested that a combination of (1) low and frag-
mented coral cover (causing individuals to be dispersed), (2) hydrodynamic conditions that cause
larvae to be retained rather than exported, and (3) relatively high populations of predators (reducing
starsh numbers and causing individuals to disperse) that prevent outbreaks from occurring.
Understanding the causes of crown- of- thorns outbreaks has been greatly hindered by a lack
of data on ne- scale temporal and spatial changes in the population structure and dynamics of
Acanthaster spp. (Moran 1986). In particular, there are few data on changes in the distribution, den-
sity, and spawning behaviour of Acanthaster spp. in the period immediately preceding an outbreak.
This is because most studies of outbreak populations (e.g., Chesher 1969, Pearson & Endean 1969,
Branham et al. 1971, Sakai 1985) are initiated after starsh densities have already increased to out-
break levels. Also, few studies have continually monitored changes in the structure and dynamics
of Acanthaster populations at regular intervals over an extended period, encompassing an entire
outbreak cycle (Moran 1986). On the GBR, for example, long- term and extensive monitoring of
the distribution and abundance of Acanthaster planci is undertaken by the Australian Institute of
Marine Science (Sweatman et al. 2011), but the methods developed to sample over vast reef areas
prohibits the collection of detailed information on population structure and reproductive condition.
Inherent trade- offs in the collection of ne- scale biological information versus broad- scale surveys
to detect changes in the abundance of Acanthaster spp. across extensive reef areas represent one
of the greatest challenges to understanding the processes that contribute to the initiation of new
and distinct outbreaks. On the GBR, it may be possible to focus detailed surveys in the area where
primary outbreaks are known to occur. Elsewhere, however, the distribution of primary versus sec-
ondary outbreaks is largely unknown.
Natural versus anthropogenic drivers
Reviews of the effects of disturbances on coral reefs (e.g., Pearson 1981) invariably distinguish
between natural (e.g., storms and other weather events) versus anthropogenic (e.g., overshing and
pollution) disturbances. The connotations of this are obvious, in that it is the anthropogenic dis-
turbances that are considered responsible for the recent (anthropocene) degradation of coral reef
ecosystems (e.g., Hughes et al. 2003) and need to be managed. However, the distinction between
natural and anthropogenic disturbances is not always clear (e.g., Potts 1981). Severe tropical storms
(cyclones, hurricanes, and typhoons), for example, are recurrent disturbances that have impacted
coral reefs throughout their evolution and development. However, Webster et al. (2005) reported
that anthropogenic climate change is increasing the severity, if not the frequency, of severe tropi-
cal storms (but see Klotzbach 2006, Landsea et al. 2006), with obvious ramications for coral reef
ecosystems. The role of anthropogenic activities in causing or exacerbating outbreaks of crown- of-
thorns starsh is also highly controversial (e.g., Potts 1981).
When extensive outbreaks of Acanthaster spp. were documented in the late 1960s (Chesher
1969, Pearson & Endean 1969), it was immediately assumed that these were new and unprecedented
phenomena linked to human activity, such as coastal development (Chesher 1969), pesticides,
and pollutants (Randall 1972), or excessive harvesting or predatory organisms (Endean 1977).
In support of this, Endean (1982) pointed out that the rst affected reefs (e.g., Green Island) on
the GBR were those that had greatest human visitations. Several scientists (e.g., Dana 1970, Vine
1970, 1973, Weber & Woodhead 1970) put forward the contrary view that outbreaks of Acanthaster
spp. were a natural occurrence that had simply gone unnoticed prior to the 1960s. Rapid increases
in the number of reports of ‘infestations’ and ‘plagues’ of Acanthaster spp. from throughout the
Indo- Pacic following initial awareness of the issue were taken as evidence for this (Vine 1973).
Some argued that outbreaks had occurred in both the recent (e.g., Vine 1973) and distant past
(e.g., Walbran et al. 1989a,b).
There are many anecdotal accounts of crown- of- thorns starsh occurring in high densities at
locations across the Indo- Pacic well before outbreaks were reported by scientists (Vine 1970,
1973). Fishers in Santa Ysabel in the Solomon Islands recalled a time in the 1930s when night sh-
ing was hazardous because of the abundance of Acanthaster planci (Vine 1970). Former trochus
and pearl shell divers on the GBR recalled seeing large numbers of the starsh on individual reefs
going back to the 1930s (Ganter 1987). In 1913, H.L. Clark collected three A. planci from the
reef at of Mer Island in the eastern Torres Strait (far northern GBR) without diving (Clark 1921),
suggesting that starsh must have been common. These anecdotal records suggest that localised
outbreaks have occurred in the past (Vine 1973). However, the previous occurrence of waves of out-
breaks as seen on the GBR in recent decades cannot be conrmed in the absence of systematic,
broad- scale monitoring.
Further evidence of past outbreaks (over geological time) has been sought in the form of
mesodermal skeletal elements of Acanthaster spp. in the sediment record. Skeletal elements from
A. planci have been found in numerous sediment cores from GBR reefs (Frankel 1977, 1978, Walbran
et al.1989a,b), but reconstructing a history of A. planci numbers and drawing conclusions about the
existence of past outbreaks is not simple because of disturbance by burrowing organisms, varying
sedimentation rates, and differential compaction of the sediments (Fabricius & Fabricius 1992,
Keesing et al. 1992, Pandol 1992). Only a few cores from a limited number of reefs have been
studied intensively, and the resulting reconstructions are relatively variable. Even if outbreaks did
occur prior to the 1950s (which appears likely), the key question is whether outbreaks are occurring
more frequently in recent times, and if this reects increasing anthropogenic changes to marine and
coastal environments (Brodie 1992).
While debate continues about the role of anthropogenic activities in causing or exacerbating
outbreaks of Acanthaster spp., it is clear that the current regime of disturbances to which most
coral reefs are subject cannot be sustained (Gardner et al. 2003, Bruno & Selig 2007, De’ath et al.
2012, Fabricius 2013). On the GBR, for example, De’ath et al. (2012) reported a 50.7% decline in
mean coral cover across 214 reefs that have been repeatedly and comprehensively surveyed (using
reef- wide manta tows) from 1985 to 2012. The timing and rates of coral loss varied spatially (see
also Sweatman et al. 2011), but these data are evidence of signicant reef- wide habitat degradation,
largely attributable to recurrent outbreaks of Acanthaster planci that compound other large- scale
and persistent disturbances (Osborne et al. 2011, De’ath et al. 2012). Over this 27-year period, out-
breaks of A. planci affected 49% of reefs, and hind casting showed that coral cover would have
increased at 0.89% per year (as opposed to annual declines of 0.53%) were it not for impacts of
crown- of- thorns starsh (De’ath et al. 2012). Similarly, in the central Pacic, the reefs surrounding
Moorea Island in French Polynesia have been subject to a high frequency of different disturbances,
including seven distinct episodes of mass coral bleaching, two major cyclones, and two outbreaks
of A. planci since 1979 (Trapon et al. 2011). Despite this frequency and diversity of disturbances in
Moorea, signicant long- term coral loss and degradation of reef environments are clearly attribut-
able to the devastating effects of A. planci outbreaks in 1980–1981 and 2007–2011 (Adam et al. 2011,
Trapon et al. 2011, Kayal et al. 2012). It has been suggested that if disturbances of this frequency and
magnitude had occurred throughout the geological period (Holocene) during which contemporary
coral assemblages evolved and coral reefs developed, then the biological and physical structure
would be vastly different (Randall 1972). Most notably, large colonies of slow- growing massive cor-
als (mostly Porites) could not withstand extremely severe outbreaks of Acanthaster spp. that cause
high rates of mortality across all preferred and non- preferred corals (e.g., Endean & Cameron 1985,
Done et al. 1988, Done 1992).
A signicant and increasing effect of Acanthaster spp. on coral assemblages and reef eco-
systems is not in itself evidence that outbreaks are unnatural (Birkeland & Lucas 1990). Rather,
other anthropogenic disturbances (e.g., shing and harvesting, sedimentation, eutrophication, and
pollutants) might have undermined the capacity of reef ecosystems to withstand these periodic
disturbances, eroding their resilience and leading changes in the ecosystem responses to persistent
and ongoing disturbances. Even more likely is that the pervasive effects of humans on coastal eco-
systems have fundamentally altered the structure and function of both crown- of- thorns populations
and reef ecosystems, forever altering any semblance of a natural system. Managing ongoing effects
of Acanthaster spp. is conditional on identifying the specic factor(s) that cause or exacerbate con-
temporary outbreaks, and much of the current discussion is centred around one of two alterna-
tive hypotheses: (1) nutrient enrichment and (2) predatory release (e.g., Birkeland & Lucas 1990,
McClanahan et al. 2002, Brodie et al. 2005, Mendonça et al. 2010, Fabricius 2013).
Nutrient enrichment hypothesis
The notion that outbreaks of Acanthaster spp. may arise due to enhancement of larval survivor-
ship through nutrient enrichment has been proposed several times (e.g., Pearson & Endean 1969,
Lucas 1973, Nishihira & Yamazato 1974, Birkeland 1982, Brodie 1992, Brodie et al. 2005, Fabricius
et al. 2010). Birkeland (1982) suggested that outbreaks of Acanthaster planci at several locations
in Micronesia and Polynesia tended to occur 3years after extremely heavy rainfall events, often
preceded by extended droughts. Birkeland (1982) argued that such events provide a pulse of nutri-
ents that stimulate phytoplankton blooms, which supplement otherwise- limited food for crown-
of- thorns larvae (Lucas 1973). However, enhanced survival of larval crown- of- thorns starsh may
also be related to specic environmental conditions (low salinity and high temperatures) at times
and in locations affected by river run- off (Henderson 1969, Lucas 1973). Fundamental to this ter-
restrial run- off hypothesis is the notion that outbreaks occur suddenly, and 3years following heavy
rainfall periods (Birkeland 1982), which accounts for the time required for larval starsh to settle
on the reef, metamorphose into the adult form, begin feeding on corals, and attain sufcient size
(200–300 mm) to emerge from the reef matrix and become readily apparent (Figure1).
However, some of Birkeland’s (1982) ndings have since been questioned, based on inconsisten-
cies in either the initiation of outbreaks or the timing of severe tropical storms and peak rainfall
events (Endean & Cameron 1990). In Guam, for example, major outbreaks were rst reported in
1967 (Chesher 1969), meaning that the larval recruitment event that led to this outbreak would
have preceded the drought- breaking rains in July 1965 by 18months (Endean & Cameron 1990).
Similarly for the GBR, Fabricius et al. (2010) argued that each of the major episodes of outbreaks
was initiated by a major ooding event (in 1958, 1974, 1991, and 2008) that contributed to increased
survival of larvae (see also Day 2000). However, the time between ooding events and subsequent
outbreaks of A. planci ranges from 2 to 5years (Table4). The purported years of major ooding
events given in Fabricius et al. (2010) also do not correspond to the documented incidence of major
drought- breaking oods, based on barium/ calcium ratios in long- lived colonies of Porites from
Havanah and Pandora reefs (McCulloch et al. 2003). Some of the biggest oods (1968 and 1981)
recorded using this method (McCulloch et al. 2003) did not appear to initiate outbreaks (Table4),
although Fabricius et al. (2010) did stress that oods must occur in November to January to benet
crown- of- thorns larvae. Also, oods may not cause outbreaks of Acanthaster spp. if they occur too
soon after the preceding outbreak so that the coral cover has not had time to recover sufciently to
sustain a new outbreak (Fabricius 2013).
Extended delays (>3years) between ood events and reported outbreaks on the GBR in 1962
and 1979 (Table4) may be attributed to limitations in the detection of outbreaks prior to the imple-
mentation of reef- wide monitoring in 1985. For more recent outbreaks, however, high densities of
A. planci would have already recruited on reefs around Lizard Island when ooding events occurred
in 1991 and 2008. Moreover, these recent outbreaks almost certainly developed through several
consecutive years of high recruitment from 1994 to 1998 (e.g., Pratchett 2005), making links to
individual ooding events somewhat tenuous. In addressing these observations, Fabricius et al.
(2010) reported that “oods have reached or crossed this part of the shelf in 1991, 1994, 1995 and
1996” (p. 603), inferring that this was an unusual period with a high frequency of ooding events. It
is possible that initial outbreaks that occurred on the GBR in 1958, 1971, and 2008 also comprised
multiple cohorts and consecutive years of high recruitment, but there are no data on population
structure to assess this explicitly.
Further complexities associated with linking outbreaks of Acanthaster planci on the GBR
to periodic major ooding events relate to the spatial patterns of outbreaks (Brodie et al. 2005).
Notably, outbreaks of A. planci tend to occur predominantly on midshelf reefs (Moran 1986), rather
than inshore reefs where the inuence of terrestrial run- off is greatest (Brodie et al. 2005), or on off-
shore reefs. Also, outbreaks are initiated north of Cairns (close to either Cairns at 17.0°S or Lizard
Island at 14.5°S), rather than on reefs in immediate proximity to major river systems (Brodie 1992).
Fabricius et al. (2010) argued that the conuence of high nutrients and midshelf reefs is limited to
northern latitudes, between 14.5°S and 17.0°S, and it is only here that nutrient concentrations exceed
minimal thresholds (>0.25–0.5 µg.l−1) necessary for survivorship of crown- of- thorns larvae. They
suggested that ood plumes from major rivers (particularly the Burdekin River) travel northward,
staying close to the coast, but are deected off shore by Cape Grafton, a promontory just south of
Cairns. There are, however, large areas of the GBR (e.g., in the far northern section and Swains)
that have long- term average chlorophyll concentrations greater than 0.5 µg.l−1 (Figure11). Also,
chlorophyll concentrations in summer months (November– May), which is when A. planci spawn,
are generally greater than 0.5 µg.l−1 throughout the Wet Tropics (16°S to 19°S), independent of any
major ooding events (Brodie et al. 2005).
If the productivity of waters on the GBR is consistently below levels (0.25–0.5 µg.l−1) needed
to sustain larval growth and survivorship (e.g., Fabricius et al. 2010), it makes it hard to explain the
southward propagation of waves of outbreaks that cause widespread devastation. The conventional
Table4 Timing of major ood events on Australia’s Great Barrier Reef (specically, peak ow
events from the Burdekin River, located 19.6°S) relative to the agreed start of each of the four
major waves of outbreaks
Peak ood events Start of corresponding
Time between ood
and outbreak LocationMcCulloch et al. 2003 Fabricius et al. 2010
1958 1958 1962 4years Green Island (16.7°S)
1968 —
1970 —
1974 1979 5years Green Island (16.7°S)
1981 —
1988 1993 5years Lizard Island (14.6°S)
1991 1991 1993 2years Lizard Island (14.6°S)
1998 —
NA 2008 2010 2years Lizard Island (14.6°S)
wisdom (Brodie 1992, Brodie et al. 2005, Fabricius et al. 2010) is that once primary outbreaks have
become established, the immense numbers of larvae produced by high densities of well- fed starsh
will generate enough late- stage larvae to settle successfully on downstream reefs (to the south)
regardless of low rates of larval survival. It is certainly true that three waves of outbreaks have
propagated effectively through the midshelf reefs (>25 km off shore) north and south of Townsville
(~19°S), where chlorophyll concentrations rarely exceed 0.25 µg.l−1. However, it is unclear whether
this is because of the sheer volume of larvae spawned on reefs to the north (of which only a small
proportion actually survive) or evidence of the capacity of larvae to successfully develop and settle
despite relatively oligotrophic conditions (e.g., Olson 1987). It does seem illogical that low nutri-
ent levels would prevent the formation of primary outbreaks on reefs south of 16°S and yet allow
for the extensive formation of devastating secondary outbreaks. Moreover, simulation models do
not reproduce the southward propagation of waves of outbreaks when using high levels of larval
mortality expected to occur when chlorophyll concentrations are less than 0.25 µg.l−1 (Fabricius
et al. 2010). A more parsimonious explanation might be that primary outbreaks are established over
several years and independent of any ood events, but the subsequent spread of outbreaks might
be conditional on years of high larval survivorship, which is facilitated by major ood events that
enhance food availability.
A logical extension of the terrestrial run- off hypothesis is that outbreaks would be expected to
occur more frequently on high islands compared to atolls (thehigh island hypothesis’; Birkeland
1982) because of localised elevated nutrient concentrations due to runoff. Tsuda (1971) and Pearson
(1975) noted that outbreaks occurred predominantly on reefs near high islands or along continental
shelves. Birkeland (1982) showed that ‘large populations’ of Acanthaster planci were reported on
19 (of 23) high islands compared to 2 (of 22) atolls across Micronesia and Polynesia based on data
presented by Marsh & Tsuda (1973). Analyses of all reported outbreaks since 1990 showed that
outbreaks have occurred on at least 35 low islands or atolls across the Indo- Pacic (e.g., Eniwetok
Atoll and Majuro Atoll in the Marshall Islands), although the majority of outbreaks (29% and 56%,
respectively) were reported from high islands (e.g., Moorea, French Polynesia) and continental
shelves (e.g., the GBR). Outbreaks on low islands and atolls cannot be readily linked to terres-
trial run- off. However, there might be other sources of nutrients that cause plankton blooms and
thereby enhance larval survival away from high islands or major rivers, including (1) upwellings
(e.g., Sweatman et al. 2008, Mendonça et al. 2010), (2) bioturbation and resuspension of sediments
by severe tropical storms, and (3) oceanographic features that create high- productivity fronts (Houk
et al. 2007). On the GBR, it is evident that here have been outbreaks on reefs in the Swains region
that occur almost independently and asynchronously with waves of outbreaks propagating from
north of Cairns (Sweatman et al. 1998). Since the Swain Reefs are more than 100 km off shore, they
are rarely if ever exposed to ood plumes, although there is some evidence that upwelling occurs in
the region (Kuchler & Jupp 1988).
Houk et al. (2007) described interannual uctuations in ocean productivity in the northern
Pacic associated with the Transition Zone Chlorophyll Front (TZCF), which may explain the irreg-
ular occurrence of outbreaks across this region. However, these periods of peak productivity coin-
cide with the ‘emergence’ (presumably from deeper water) of high densities of adult Acanthaster
planci (Houk et al. 2007, Houk & Raubani 2010) rather than settlement of larvae, which must have
occurred about 2years earlier; it is unclear how this phenomenon relates to secondary outbreaks.
Despite the growing enthusiasm for the nutrient enrichment hypothesis, several authors have cau-
tioned against its broad applicability (e.g., Potts 1981, Olson & Olson 1989, Endean & Cameron
1990, Lane 2012). Lane (2012) pointed out that outbreaks of Acanthaster spp. have been occurring
despite overarching declines in global ocean productivity. Also, there is no evidence of increased
incidence of crown- of- thorns outbreaks in areas with enhanced nutrient concentrations (because of
either periodic high precipitation and high erosion rates or areas with upwellings) within the Indo-
Australia archipelago or ‘Coral Triangle’ (Lane 2012). It is also important to remember that much
of the potential importance of nutrient enrichment depends on larvae of Acanthaster spp. being
generally food limited (Lucas 1982, Okaji et al. 1997, Fabricius et al. 2010).
Olson (1985, 1987) set out to test Lucass suggestion that larval A. planci rarely complete devel-
opment at phytoplankton concentrations that are typical of GBR waters in the absence of ood
plumes. He developed an apparatus that allowed larvae to be supplied with seawater with con-
trolled concentrations of phytoplankton while being held in chambers underwater in the eld. He
found that A. planci larvae developed at near- maximal rates at ambient chlorophyll levels in the
absence of phytoplankton blooms (Olson 1987). He therefore suggested that uctuations in larval
food resources have a limited role in explaining interannual variation in larval recruitment, which
must underlie sudden increases in the abundance of Acanthaster spp. (Olson 1987).
Okaji (1966) tried to continue this line of research using the same apparatus, as well as a modi-
ed form with inline lters, to further manipulate phytoplankton densities. In spite of changing the
chambers every 2 days, he found that chlorophyll concentrations within the chambers increased
over the course of his experiment to well above ambient levels, presumably through contamination
and retention of phytoplankton. He concluded that the apparatus was unreliable and abandoned the
approach in favour of the laboratory experiments described later by Fabricius et al. (2010).
Olson (1987) did not measure chlorophyll concentrations in the chambers in the course of his
main experiments, but he did specically test for such an effect in a pilot experiment (Olson 1985)
and found that chlorophyll concentrations in the experimental chambers remained at or below ambi-
ent levels after 2days in shallow water and bright sunlight. Based on this observation, the lar-
val chambers were changed every 2days during the main experiment (Olson 1987) specically to
prevent any accumulation of phytoplankton. These divergent results reinforce the importance of
repeating and extending the limited studies relating growth and survival of Acanthaster larvae to
phytoplankton concentration; they also emphasise the need to conrm the ndings in the eld.
Despite discrepancies and inconsistencies in the spatial and temporal occurrence of outbreaks
(e.g., Table4), high rainfall, terrestrial run- off, and elevated nutrients are likely to increase the like-
lihood that outbreaks of Acanthaster spp. will actually occur, but they do not necessarily account
for all recorded outbreaks. Importantly, the nutrient enrichment hypothesis provides one of the
only plausible mechanisms by which anthropogenic activities may have exacerbated outbreaks of
Acanthaster spp. (increasing their severity or frequency) over recent decades (Brodie 1992, Brodie
et al. 2005). When proposing the terrestrial run- off and high island hypotheses, Birkeland (1982)
maintained that outbreaks were essentially a natural phenomenon triggered by irregular rainfall
events. He did acknowledge that land clearing may increase nutrient concentrations in coastal envi-
ronments following terrestrial run- off (Birkeland 1982). If so, there might also be a signal whereby
outbreaks of Acanthaster spp. are greatest in areas closest to heavily populated coastlines (but
see Lane 2012) or coinciding with periods of extensive settlement and clearing of coastal land.
However, outbreaks have been reported on isolated and unpopulated reefs (e.g., Chagos), but they
might be less frequent or less severe than on reefs subjected to increased anthropogenic inuences.
Unfortunately, inconsistencies in monitoring and reporting of nominal outbreaks make it virtually
impossible to test these ideas explicitly. This should be a priority for future monitoring studies.
Predator removal hypothesis
One of the earliest hypotheses to account for outbreaks of Acanthaster spp. was the predator removal
hypothesis (Endean 1969), which assumed that populations of crown- of- thorns starsh are normally
regulated by high levels of post- settlement or adult predation (Endean 1969, McCallum 1987, 1990).
This hypothesis was given increased credibility by two recent studies (Dulvy et al. 2004, Sweatman
2008) that reported increased incidence or severity of outbreaks of crown- of- thorns starsh in areas
subject to sheries exploitation (see also Ormond et al. 1990). Sweatman (2008) compared the rates
of occurrence of starsh outbreaks on GBR reefs that were open to shing and on reefs where
shing was prohibited. Because outbreaks come in waves, only those reefs that were close to known
outbreaks were considered; 75% of reefs that were open to shing suffered outbreaks, compared
to 20% for reefs that had been closed to shing for a minimum of 5years (Sweatman 2008). These
studies (Ormond et al. 1990, Dulvy et al. 2004, Sweatman 2008) did not reveal the mechanistic
basis for the observed results but suggested that harvesting of coral reef shes may increase the
likelihood that outbreaks of Acanthaster spp. will occur by removing one of the key regulatory
mechanisms that prevent extreme population uctuations.
When the predator removal hypothesis was initially proposed, the principal known predator of
adult Acanthaster spp. was the giant triton, Charonia tritonis (Pearson & Endean 1969). Harvesting
of tritons for sale as curios was suggested to have reduced predator numbers in the decades leading
up to the rst recorded outbreak in the 1960s (Endean 1969). However, the little that is known of
the ecology of C. tritonis suggests that even at preharvest densities they would not be effective in
controlling outbreak populations of crown- of- thorns starsh. For example, when large C. tritonis
were placed in a cage and supplied with abundant adult A. planci, average consumption was only 0.7
starsh per triton per week (Pearson & Endean 1969).
More recently, attention has focused on predation by shes, particularly emperors (family
Lethrinidae) because these are relatively large, generalist benthic carnivores that feed on and
around areas of coral (e.g., Sweatman 1997, Mendonça et al. 2010). They are also widely shed, so
there is the possibility that their numbers have been reduced through overexploitation, which could
be related to the apparent increase in frequency of outbreaks on the GBR. Other known predators
of Acanthaster spp. are triggershes and puffershes (Campbell & Ormond 1970, Owens 1971),
although numbers of these shes are unlikely to have changed through exploitation, at least on the
GBR. Direct evidence that any sheries target species are major predators of Acanthaster spp. is
meagre (Sweatman 1997). Remains of Acanthaster spp. have been found in the gut contents of some
sheries target species (Randall et al. 1978, Birdsey 1988; Table5), but only rarely. Ormond et al.
(1990) pointed out that prey- switching behaviour, which is a critical component of the population
models, means that a lack of observations of predation or the absence of Acanthaster spp. in preda-
tors’ gut contents when starsh densities are low does not necessarily mean that predation by shes
is not important in regulating starsh populations.
Few studies have sampled the gut contents of potential predators in locations where Acanthaster
spp. are known to be present. However, Sweatman (1997) examined the gut contents of 98 lethrinids
that were caught close to an area with outbreak densities of adult Acanthaster planci on the GBR
and did not nd any starsh remains. Similarly, Mendonça et al. (2010) examined gut contents
for more than 20 large (~50 cm total length) individuals of potential starsh predators, including
snappers Lutjanus bohar and Lutjanus johni, emperors Lethrinus spp., and Cheilinus lunulatus,
at reefs infested with Acanthaster spp. but failed to detect starsh remains in guts of any of the
shes. It is possible that these shes rarely feed on adult Acanthaster spp. but instead target juve-
nile starsh (Endean 1976). Endean (1976) recorded remains of juvenile A. planci in the guts of
a Queensland grouper, Epinephelus lanceolatus, although predation rates