Biogeographical variation in diurnal behaviour of Acanthaster planci versus Acanthaster cf. solaris

Crown-of-thorns starfish (CoTS; Acanthaster spp.) are among the most extensively studied coral reef taxa, largely owing to their devastating impacts on live coral cover during population outbreaks. Much of this research has however, been conducted in the western Pacific, although it is now apparent that there are several distinct species of Acanthaster spp. across the Indo-Pacific. The purpose of this study was to test for biogeographical variation in behaviour, comparing between Acanthaster planci at Lankanfushi Island in the Maldives and Acanthaster cf. solaris at Rib Reef on Australia’s Great Barrier Reef. The extent to which CoTS were exposed (cf. concealed within or beneath coral substrates) was substantially higher (63.14%) for A. planci at Lankanfushi Island, compared to 28.55% for A. cf. solaris at Rib Reef, regardless of time of day. More importantly, only 52% of individuals were exposed at night at Rib Reef compared to >97% at reefs around Lankanfushi Island. Biogeographic variation in the behaviour of Acanthaster spp. was independent of differences in the size structure of starfish and coral cover at specific study sites, but may be attributable to other environmental factors such as habitat complexity or prey availability. This is the first study to explicitly test for biogeographical differences in the biology and behaviour of Acanthaster spp., potentially linked to species-specific differences in the causes and explanations of population outbreaks. However, we did not find evidence at this stage of differences in behavior among regions, rather behavioural differences observed were most likely products of different environments.


Introduction
Crown-of-thorns starfishes (CoTS; Acanthaster spp.) have gained considerable notoriety over the last few decades following outbreaks throughout the Indo-Pacific [1][2][3][4][5][6][7][8]. Along with anthropogenic climate change, outbreaks of CoTS are a major contributor to coral loss and reef degradation [9], causing extensive coral mortality [10] and shifts in the biological and physical structure of coral reef habitats [11][12][13][14]. Unsurprisingly, CoTS are among the most extensively studied organisms from coral reef environments [2], though the majority of this fieldwork was undertaken. The specific roles of these authors are articulated in the 'author contributions' section. Ocean Paradise Dive Centre provided field materials including SCUBA equipment in the Maldives. The funders did not have any additional role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript." Competing interests: DB and JFC were employed by Gili Lankanfushi Resort in the roles of 'Marine Biologist and Environmental Officer' and 'Marine Biologist', respectively, during fieldwork in the Maldives in March 2017. At the time, Gili Lankanfushi Resort provided salaries for DB and JFC but had no influence upon the study design, data collection and analysis, decision to publish, preparation of the manuscript or outcome of this study. This affiliation does not alter our adherence to PLOS ONE policies on sharing data and materials Gaafu Dhaalu Atoll in the south [38]. Elevated densities of A. planci were again recorded in the Maldives in 1999 to 2007 [39]. Most recently, severe and widespread outbreaks of A.planci have affected South Male and Ari Atolls [7,35], starting in 2014. This study was undertaken in March 2017, on the western facing fringing reef of the Lankanfinolhu-Lankanfushi-Himmafushi island group located on the eastern outer rim reef of North Male Atoll in central Maldives (Fig 1). The site was chosen due to the high density of A.planci in comparison with nearby reefs. Despite localised culling between October 2015 and June 2016, outbreaks of A. planci had caused extensive coral loss in the study area, which was further compounded by severe coral bleaching in May 2016 [35,40,41].
To test for diurnal changes in the behaviour of CoTS at Lankanfushi Island, three permanent replicate 100m transects were established on both the reef crest (8m) and the reef slope (15m) on the inner atoll rim reef between Lankanfinolhu Island and Himmafushi Island (41 7'53.0"N 73˚33'13.7"E). During surveys, which were conducted randomly at various times of day and night from February 25 th to March 29 th , 2017, two divers carefully searched for CoTS within a 1m wide belt on either side of the transect line, along each of the six transects. Divers swam parallel to each other, whilst visually checking crevices and overhangs across the 2m belt to allow for different angles of visual perception, paying particular attention to areas where feeding scars were observed to ensure the likelihood of detection of both exposed and cryptic animals. In all, each of the six transects were surveyed at four different times of day. Times were grouped into categories including morning (surveys between 06:30-10:00), midday (11:30-13:00), afternoon (14:00-17:15) and night (18:00-22:00). Surveys were undertaken with permission from Gili Lankanfushi Resort, who manage the section of reef surveyed, negating the need for a permit.

Acanthaster cf. solaris on the Great Barrier Reef
Comparable research on A. cf. solaris was conducted at Rib Reef, in the central GBR, Australia (Fig 1) (18˚29'11.1"S 146˚52'33.8"E). The first documented outbreak of CoTS on the GBR occurred in 1962, with subsequent outbreaks starting in 1979, 1993 and 2009-10 [27]. In all instances, outbreaks have started in the northern GBR, between Cooktown and Cairns, before spreading north and south [21,42,43] affecting mid-shelf reefs along the entire length of the GBR. This study was conducted from December 2016 to May 2017, when outbreaks were concentrated in the area between Cairns and Townsville [34]. High densities of A. cf. solaris caused extensive coral depletion at Rib Reef during 2015 and 2016, though patches of high coral cover remained, mostly on distinct patch reefs on the leeward, north-west margin, as well as inside the lagoon.
Reef access was much more limited at Rib Reef, compared to Lankanfushi Island, in the Maldives. As such, two-four permanent replicate 50m transects were established during each of three distinct reef visits, with sampling conducted in a different location on each visit to account for prevailing conditions and changes in the distribution and abundance of CoTS. In all, five transects were surveyed on reef crest habitats (3m) and five on slope habitats (6m). CoTS were surveyed within 2.5m either side of transects on the reef crest, where there was high cover of tabular Acropora (live and dead). In deeper, less complex habitats, surveys were extended to 5m either side of the transect line to ensure enough CoTS were captured to allow for statistical analysis. As densities of CoTS were analysed, rather than abundance per site, this did not confound results. Each of the ten transects were surveyed at several different times of the day, including at least one survey during daylight hours and one survey after dark. Survey times were placed into the same time categories as the Maldives surveys for comparative analysis. Surveys at Rib Reef were undertaken under Marine Parks permit G15/37363.1 issued by the Great Barrier Reef Marine Park Authority.

Quantifying diurnal patterns of behaviour
All CoTS recorded within the area of belt transects at both Lankanfushi Island, Maldives and Rib Reef, central GBR, were measured (diameter from arm tip to opposite arm tip) in situ to the nearest cm. The extent to which individuals were visible from above (or at 90˚perpendicular to the substrate on particularly steep sections of reef) was then recorded to the nearest 10%. There were however, two distinct modes in these data, where the majority (85%) of starfish were either completely (100%) exposed or entirely concealed (0% exposed) beneath the reef substrate and only visible from an acute angle. As such, we categorised starfish as either cryptic, where �50% of the starfish was exposed, versus exposed where >50% of the starfish was visible from above. In addition to exposure, we also recorded whether each starfish was moving versus stationary, based on sustained movement of the body of the starfish. For starfish that were not moving, one or more arms were lifted away until it was possible to see whether the stomach was everted. Starfish with stomachs partially or fully everted were recorded as feeding, whereas individuals with stomachs completely retracted were considered to be resting.

Coral cover
Coral cover was determined along each of the transects, which were demarcated by a fibreglass tape, at both locations before CoTS surveys began. Benthic cover was determined to the lowest possible taxonomic level, with scleractinian corals identified to genus and growth form, using the point intercept method every 50cm.

Statistical analysis
To assess the variation in CoTS density between reef locations and habitats, a linear model with gaussian error distribution was constructed, to which the time of observation (Time: Morning, Midday, Afternoon, Night) and interaction terms for location:Habitat and location: Time was also included. Similarly, a linear model was constructed to determine variations in the size of CoTS individuals between reef locations, habitats and time of day, using the same interaction terms as above. Analysis of variance and post-hoc comparisons (using R package 'lsmeans') of factor levels was also conducted to determine the significance of these variations in density and size. Potential interactions were tested and retained if they improved model fit on the basis of the corrected Akaike information criterion [44]. To test for differences between the proportions of CoTS observed exhibiting different behaviours between reefs, Chi-Squared two sample tests for proportions were conducted. Additionally, to assess the differences between the coral communities at the two reef locations, non-metric multidimensional scaling and an analysis of similarity (ANOSIM) were conducted using R 3.5.1 [45] using the 'vegan' (v2.5.6, [46]) package, using a Bray-Curtis dissimilarity matrix and Wisconsin double standardization.
To test for variation in the cryptic behaviour of CoTS, we constructed a Bayesian generalised linear model to investigate the effects of 7 different predictor variables; location (Lankanfushi Island versus Rib Reef), habitat (crest and slope), body size, coral cover, behaviour (moving or stationary), time of day (morning, midday, afternoon, night) and CoTS density (per 100m 2 ). Exposure was recorded as a binary variable (either cryptic or exposed) and thus modelled using a Bernoulli distribution using the R package 'brms' [47]. All potential interactions were tested and retained if they improved model fit on the basis of the widely applicable information criterion [48]. Effects plots with 95% credible intervals were produced to show the magnitude and direction of the effect for each variable reported as log odds ratios. Marginal effects plots were also produced to show how CoTS exposure varied with respect to each of the predictor variables individually.
Overall, CoTS were exposed (where >50% of the animal was visible from directly above) in only 37.14% (410/1104) of records. The proportion of CoTS exposed was however, Diurnal behaviour of Acanthaster planci versus Acanthaster cf. solaris substantially higher at Lankanfushi Island (63.14%) compared to Rib Reef (28.55%), regardless of time of day. The proportion of exposed CoTS increased through the day at both locations, with lowest levels recorded in the morning (33.97% at Lankanfushi Island and 13.78% at Rib Reef) and highest levels of exposure recorded at night (97.47% at Lankanfushi Island and 52.09% at Rib Reef).
The best model (Bayesian generalised linear model) to account for variation in cryptic behaviour of CoTS across both locations included an interaction between reef (Lankanfushi Island versus Rib Reef) and time of day (morning, midday, afternoon, night), percent Acropora Diurnal behaviour of Acanthaster planci versus Acanthaster cf. solaris at each site, as well as the size (total diameter), and the behaviour of starfish (whether starfish were resting, feeding or moving) (Fig 4). At both locations, the probability that CoTS were exposed rather than cryptic was significantly greater at night as opposed to during the day (2.58[1.46,3.97], p<0.05); log odds posterior mean and 95% highest posterior density intervals calculated by contrasting night observations to the next most exposed time; afternoon (P value is inferred as the HPD intervals do not overlap 0). While exposure was not significantly higher at Lankanfushi Island across all time periods (0.54[1.59, 0.49], p>0.05), there was a significant interaction between reef location and time of day with the differences between night and afternoon being much more pronounced at Lankanfushi Island (4.19 [2.04, 7.42], p<0.05) compared to Rib Reef (0.96 [0.39, 1.52]). Furthermore, post hoc analyses revealed that the differences in exposure between Rib Reef and Lankanfushi Island were only significant during night time observations (Fig 4c). The probability of exposure increased with increasing size of CoTS (0.086[0.059, 0.11], p<0.05), in a manner that was consistent across locations (Fig 4b).
The proportion of A. planci that were observed feeding (as opposed to moving or resting) was higher at Lankanfushi Island (44.03%) compared to A. cf. solaris at Rib Reef (35.00%: χ 2 = 4.31, df = 1, p = 0.019). Conversely, a significantly higher proportion of CoTS were recorded resting at Rib Reef (56.99%) compared to Lankanfushi Island (46.54%; χ 2 = 5.48, df = 1, p = 0.0097), whereas the proportion of starfish actively moving was similar between locations (9.43% at Lankanfushi Island and 8.02% at Rib Reef;; χ 2 = 0.19, df = 1, p = 0.67). It was apparent that the proportion of individuals observed feeding at Lankanfushi Island was more consistent throughout the day (ranging from 38-64%) and was significantly higher during the midday surveys (64.5%) compared to night time (38.2%; χ 2 = 3.49, df = 1, p = 0.031), whereas the incidence of feeding by A. cf. solaris at Rib Reef increased throughout the day. The proportion of A. cf. solaris recorded feeding at night (65.4%) was more than twice that recorded at any time during the day (13.4-26.7%; χ 2 = 58.27, df = 1, p<0.0001, compared to next highest proportion, afternoon). To test whether this result was attributable to the large number of smaller (<25cm total diameter) starfish at Rib Reef, whereby smaller starfish may be more cryptic and tend to feed nocturnally [23,3,28], we compared behaviour of starfish �25 cm total diameter, but this did not alter the overall pattern; even for among A. cf. solaris �25 cm total diameter, the proportion of individuals feeding increased from 13.02% in the morning to 62.92% at night.

Discussion
Given their important contribution to degradation and disturbances on coral reefs, there has been extensive and increasing research into the biology and ecology of Acanthaster spp. over the last three decades (reviewed by [15,27]), with a particular focus on reproductive biology, outbreak dynamics and population control. However, most of this research has been conducted in the western Pacific, and mainly on the GBR (e.g., [49]). Despite marked genetic differences between the Pacific and Indian Ocean populations of CoTS, which were first revealed 20 years ago [50], there has been comparatively limited research on the biology or behaviour of CoTS from the Indian Ocean (but see [15,35]). It is implicitly assumed that all putative species (at least four different species; [25]) from different bioregions behave in much the same way. Whilst this study has revealed potentially important differences in the diurnal behaviour between A. planci from Lankanfushi Island in the Maldives and A. cf. solaris from Rib Reef on the GBR, this is likely attributable to varying environmental conditions, rather than speciesspecific traits.
The most pronounced difference observed between A. planci from the Maldives and A. cf. solaris on the GBR were higher levels of cryptic behaviour for CoTS on the GBR, especially at night. The proportion of starfish exposed at night was higher than during the day at both locations, but for A. cf. solaris on the GBR only 52% of individuals were found out in the open during the night, while a similar proportion remained largely concealed within and/ or beneath live or dead corals. In contrast, at Lankanfushi Island in the Maldives, virtually all individuals of A. planci (>97%) were fully exposed at night. Similarly, Ormond and Campbell [32] reported marked increases in activity and exposure at night for Acanthaster spp. from the Red Sea, where CoTS were reported to feed almost exclusively at night. Increased nocturnal exposure in the Maldives did not translate into higher feeding incidence, though movement was greatest at night. At Rib Reef on the GBR however, greater exposure at night was accompanied by pronounced nocturnal feeding. In some locations, Acanthaster spp. are reported to switch from feeding mainly at night to feeding day and night with the onset of population outbreaks (e.g., [51]). These shifts in diurnal behaviour may occur in response to increased food competition or food limitation, though we'd expect to see increased diurnal feeding at higher CoTS densities. In our study, diurnal feeding was more apparent at Lankanfushi Island, despite lower CoTS densities compared to Rib Reef. Coral cover was however, much lower at Lankanfushi Island (21.4%) than at Rib Reef (43.4%), and a pronounced difference in preferred prey existed between the two locations as live coral cover was dominated by Acropora at Rib Reef and Porites at Lankanfushi Island. To fully explain behavioural changes that may occur due to changes in CoTS densities and depletion of prey sources, we would need to observe the behaviour of A. planci versus A. cf. solaris through the entire course of population outbreaks at multiple reefs.
The differences in behaviour of CoTS between the two locations, particularly the increased levels of exposure at Lankanfushi Island, is most likely linked to the lower levels of complex branching corals (e.g., branching and tabular Acropora) at this location. Reef habitats with high cover of complex corals are likely to offer more opportunities for Acanthaster spp. to remain concealed, even while continuing to feed. While coral composition and habitat (slope vs crest) had no effect on the probability of exposure for CoTS that were recorded, the effect of percent Acropora was likely masked in our model by the inclusion of reef location. Indeed, our results showed percent Acropora performing the same function as reef location, when the latter was removed from the model. Despite this, it is also likely that detectability itself varies with coral cover and habitat complexity [36]. It was very apparent, for example, that a greater number of CoTS were detected during nocturnal surveys on the shallow reef crest compared to diurnal surveys at the same location. While it is possible that these starfish are moving among habitats, it is also possible that the starfish are so cryptic during the day that they completely avoid detection. Study sites at Lankanfushi Island were surveyed in the aftermath of the 2016 bleaching event [40], which combined with sustained high densities of A. planci since 2015, have caused extensive depletion of acroporid and pocilloporid corals, which are generally preferred prey of Acanthaster spp. [27]. The lack of nutritional value afforded by the remaining non-preferred coral species (predominantly massive Porites) [52], combined with declines in structural complexity following the inevitable degradation of branching coral skeletons, may explain the generally higher levels of exposure and consistently high levels of feeding behaviour of CoTS at Lankanfushi Island (cf. Rib Reef) regardless of time of day. Conversely, the higher abundance of preferred prey species (mostly Acropora) at Rib Reef likely explains higher densities and wider size ranges of CoTS recorded at this location. Despite being subject to moderate bleaching in 2016 and 2017 [53] and sustained outbreaks of A. cf. solaris, there was still moderate cover of Acropora spp. at Rib Reef, allowing prey to not only sustain large densities of adult starfish, but also facilitate secondary waves of recruitment and replenishment.
The most likely explanation for the diurnal variation in exposure, and specifically increases in cryptic behaviour of CoTS during daylight hours, is that predation risk is substantially greater during the day compared to night [33]. The role of predator evasion in explaining probability of exposure is further reinforced by the relationship between body size and probability of exposure across both species and study locations. Therefore, as well as clear differences in topography, regional differences in predation risk may also account for differences in behaviour between locations, rather than it being a reflection of inherent differences between species of Acanthaster. Predation risk will vary with size, abundance and composition of potential predators which will, in turn, vary due to inherent biogeographical patterns, as well as overarching influences of anthropogenic extraction and reef degradation [54]. In the Philippines, for example, the proportion of A. cf. solaris with sublethal injuries, presumed to be caused by sublethal predation, was higher inside a marine reserve compared to nearby areas that were open to fishing [54]. If Acanthaster spp. respond to increased predation risk inside marine reserves by being more cryptic, this might lead to differences in detectability, which will need to be considered when comparing density estimates inside versus outside of marine reserves (e.g., [19]). There remains a definite need to quantify predation rates on CoTS at different times of day and in different habitats [15,55].
This study has revealed differences in behaviour between A. planci at Lankanfushi Island in the Maldives and A. cf. solaris at Rib Reef on Australia's GBR. These preliminary results from just two locations suggest these differences are a product of variation in extrinsic factors, in particular coral prey availability and structural complexity afforded by differences in coral composition, along with possible differences in predation risk. It is important, however, given recognition of species boundaries for Acanthaster spp. across the Indo-Pacific [24,25], that we expect species-specific differences in their biology and behaviour which may, in turn, mean that there are different causes and explanations for population outbreaks in different biogeographic regions. This study is the first to present behavioural data from a population of A. planci in the Maldives, which is useful for the management and control of CoTS throughout this region, but more work is required to establish whether any behavioural differences are reflective of intrinsic biological differences between species of Acanthaster.