A decadal analysis of bioeroding sponge cover on the inshore Great Barrier Reef

Decreasing coral cover on the Great Barrier Reef (GBR) may provide opportunities for rapid growth and expansion of other taxa. The bioeroding sponges Cliona spp. are strong competitors for space and may take advantage of coral bleaching, damage, and mortality. Benthic surveys of the inshore GBR (2005–2014) revealed that the percent cover of the most abundant bioeroding sponge species, Cliona orientalis, has not increased. However, considerable variation in C. orientalis cover, and change in cover over time, was evident between survey locations. We assessed whether biotic or environmental characteristics were associated with variation in C. orientalis distribution and abundance. The proportion of fine particles in the sediments was negatively associated with the presence-absence and the percent cover of C. orientalis, indicating that the sponge requires exposed habitat. The cover of corals and other sponges explained little variation in C. orientalis cover or distribution. The fastest increases in C. orientalis cover coincided with the lowest macroalgal cover and chlorophyll a concentration, highlighting the importance of macroalgal competition and local environmental conditions for this bioeroding sponge. Given the observed distribution and habitat preferences of C. orientalis, bioeroding sponges likely represent site-specific – rather than regional – threats to corals and reef accretion.

bioeroding sponges may be increasing over time, but the geographic extent and rate of these increases are largely unknown.
Several physiological and ecological hypotheses have been proposed to explain the observed increases in abundance of bioeroding sponges. Cliona is thought to be a robust sponge genus that is tolerant of disturbances and changing environmental conditions [23][24][25][26] as well as benefitting from the poor water quality that can adversely affect corals 21,27 . Based on the success of Cliona spp. in similar habitats, the inshore GBR was expected to be optimal habitat for bioeroding sponges where increases in cover may be occurring throughout the region 20 . However, poor water quality is also associated with low light conditions that may negatively impact growth of photo-symbiotic bioeroding sponges such as C. orientalis and C. varians 28,29 .
Increases in the abundance of bioeroding sponges will have implications for coral reefs in addition to the erosion of substratum 19 . Bioeroding sponges weaken reef substrata, produce carbonate sediments 11,30,31 , and are strong competitors against live corals 19,[32][33][34][35][36][37] , particularly following coral bleaching events 38 . However, the growth of Cliona spp. can be limited by macroalgae 32,39 , suggesting that the composition of the reef community may influence the success of Cliona.
Given that sponge erosion is expected to accelerate as oceans become more acidic 18,25,40,41 , there is a clear need to monitor bioeroding sponge populations 42,43 . The most conspicuous bioeroding sponge on the GBR is Cliona orientalis but percent cover has only been reported for a single GBR site 20,43,44 . Here, we quantified the abundance and trajectory of C. orientalis cover on the inshore GBR over a 10-year period (2005-2014) to resolve whether environmental conditions are drivers of change in sponge abundance. Our sampling covers a wide geographic area to assess whether previous reports of increasing Cliona abundance represent a GBR-wide trend or site-specific responses 20 .

Results and Discussion
C. orientalis was present in at least three survey years at 16 of the 35 inshore GBR locations. Where present, C. orientalis occupied as much surface substratum (0.73% ± 0.97 SD) as all other sponges combined (0.56% ± 1.11 SD). Havannah Island had the highest average cover at 3.6% (Fig. 1A), although C. orientalis cover reached as high as 5% at Fitzroy Island and High Island in certain years (Fig. 2). C. orientalis percent cover was lower than previously reported from Orpheus Island (>6%) 20 , possibly due to a greater area surveyed or the untargeted design in the current study. When absences are included (i.e., zero cover), the average percent cover of C. orientalis on the inshore GBR was 0.14% (±0.51 SD), which is comparable to the average cover of C. delitrix in the Florida Keys, USA (~0.1%) 45 and southeast Florida (~0.08%) 46 , but lower than C. delitrix cover in Colombia (~2%) 47 . Additional studies have assessed the abundance of bioeroding sponges by counting individual sponges 27,38,48,49 , although it is challenging to reliably compare these measures of abundance with percent cover.
C. orientalis occurred less frequently at locations with high accumulation of fine sediments. The model predicted a 50% probability of C. orientalis occurrence at 17% fine sediments, suggesting that even moderate accumulation of silt and clay sized particles prevents the establishment of C. orientalis (Fig. 3A). Furthermore, sites with large accumulations of fine sediments had low percent cover of C. orientalis (Fig. 3B). The amount of fine sediments distinguishes exposed and sheltered locations, as waves and currents resuspend fine particles and prevent accumulation 50 . Both suspended and deposited sediments can influence the composition of sponge communities 51 and have negative physiological effects on sponges 52 , including reduced reproductive output 53 and increased respiration 54 . The deposition of fine sediment may hinder filter-feeding or reduce the light available for photosynthesis 55 . The negative correlations observed between fine sediments and the distribution and abundance of C. orientalis suggest that sediments have negative physiological effects on C. orientalis, although these effects have not been demonstrated experimentally.
As coral cover declines on the GBR 4 , changes in the cover of bioeroding taxa may dictate future reef growth 2,13,18 . In this study, the average change in C. orientalis percent cover was 0.03% yr −1 (±0.08 SD). Cover increased at 10 out of 16 locations (Fig. 1B), although only one trend was statistically significant (0.2% yr −1 at Fitzroy Island (East); t = 2.8, p < 0.05). C. orientalis cover exhibited non-linear patterns at some sites, possibly due to disturbances such as cyclones or outbreaks of crown of thorns starfish 56 , which altered community composition and potentially increased the detectability of C. orientalis. The rate of change in C. orientalis cover was similar to the rate of change in sponge cover at the same locations (0.03% yr −1 ± 0.10 SD), but slower than the changes in other benthic groups (Fig. 4). These time series indicate that cover of C. orientalis and other sponges has remained largely stable over the past decade on the inshore GBR despite changes to the reef community, such as a decline in octocoral cover (Fig. 4).
Few studies have reported the rate of change in percent cover of bioeroding sponges. Therefore, we estimated rates of change in cover of other Cliona spp. to provide context for the rates of change in C. orientalis cover measured in this study. The fastest estimated rate of change was for C. orientalis cover from 1998 to 2004 at Orpheus Island on the GBR (~0.9% yr −1 ) 20 . Slower rates of increase were reported from the Caribbean, where C. caribbaea cover increased ~0.14% yr −1 from 1979 to 1998 in Belize 19 , bioeroding sponge cover increased ~0.05% yr −1 from 2005 to 2009 in southwest Florida 57 and C. delitrix cover changed <0.01% yr −1 from 2003 to 2009 in southeast Florida 46 . In contrast, C. delitrix cover decreased (−0.03% yr −1 ) in the Florida Keys 45 . The rate of change reported here (0.03% yr −1 for C. orientalis) is relatively low in the context of these estimates, but also encompassed a comparatively large number of survey locations. It is worth noting that many of the observations of increased cover of bioeroding sponges were initiated prior to 2001 [19][20][21] and that subsequent studies have not observed increased cover 38,45,46,57 .
Changes in C. orientalis cover are best explained by the abundance of macroalgae (Fig. 5, Table 1). Increases in C. orientalis cover occurred at locations with low macroalgal cover (t = −3.0, P = 0.01). However, these locations  also had low average chlorophyll a concentration in the water (Fig. 5), which also significantly affected the change in C. orientalis cover (t = −2.4, P = 0.03). Therefore, the fastest increases in C. orientalis cover occurred at locations with a combination of low macroalgal cover and low chlorophyll concentrations, which were clustered near     (Fig. 6). When analysed together, neither macroalgal cover nor chlorophyll concentration was significantly associated with change in C. orientalis cover (P > 0.05), likely due to the positive correlation between macroalgal cover and chlorophyll concentration (r = 0.56). While macroalgal cover explained 39% of the variation in change in C. orientalis cover (Table 1), macroalgal cover (or chlorophyll a) did not predict the distribution or abundance of C. orientalis (Table 1, Supplementary Figure 1). These results suggest that macroalgae outcompete bioeroding sponges for space: all but one of the locations with increased C. orientalis cover had less than 10% macroalgal cover and all had less than 0.45 µg/L chlorophyll a (Fig. 5), a water quality threshold that separates reefs with low and high macroalgal abundance 58 . Previous work observed that macroalgal cover was negatively correlated with C. orientalis cover 39 and macroalgae have also been reported to outcompete C. tenuis for substratum in the Caribbean 32 . In addition, several studies have observed that large colonies of bioeroding sponges occur where macroalgal cover is low 32,59 . By extension, controls on macroalgal growth, such as fish and urchin herbivory 39 as well as dissolved nutrient levels 58 , may indirectly affect the growth of bioeroding sponges.
The gradual increases in C. orientalis cover observed at multiple locations suggest that broader ecological changes may be responsible for increases in C. orientalis cover. Water quality is declining across the inshore GBR, driven by inputs of terrestrial nutrients that are delivered during seasonal flood events 60,61 . Dissolved nutrient levels increase during floods 61 , which can lead to phytoplankton blooms and higher concentrations of organic material in the water 62 , which is a primary food for some Cliona species 63 . Nutrient levels likely increased over the survey period, as river flows were high, particularly during the middle of the study 56,64 . At locations with high nutrient levels, additional nutrients would likely have benefited the already high macroalgal cover 58 . However, at locations with low nutrient levels and little cover of macroalgae, additional nutrients may have contributed to increases in C. orientalis cover (Fig. 5). Thus, increases in C. orientalis cover may reflect additional nutrient loads entering the GBR lagoon, but are restricted to locations where nutrient concentrations are insufficient to support high macroalgal cover.
While the response of C. orientalis to high nutrient levels has not been investigated experimentally, several other Cliona species exhibit positive associations with elevated nutrients, including C. delitrix and C. vastifica 21,27,47,65 . However, not all Cliona species respond the same way, as several exhibited either positive or negative responses to a chlorophyll a gradient in Mexico 48 . On the GBR, observation of higher abundance of bioeroding sponges on inshore versus offshore reefs suggests that bioeroding sponges benefit from high nutrient conditions 66 . The correlations reported here suggest that C. orientalis is affected by local environmental conditions, specifically fine sediments, dissolved nutrients (chlorophyll a), and macroalgal cover, but experimental evidence of how these conditions affect Cliona species is lacking. Figure 6. The spatial distribution of changes in C. orientalis cover on the inshore GBR. Circles represent locations where change over time was measured and an * represent locations where C. orientalis was absent (not detected in at least 3 survey years). Blue circles indicate increases in cover; white circles indicate zero change in cover; and red circles indicate decreases in cover. The map was created using R statistical software (version 3.3.1; https://www.r-project.org) and the packages "ggplot2" 69 , "mapdata" 70 , and "oz" 71 .
Scientific RepoRts | 7: 2706 | DOI:10.1038/s41598-017-02196-z Factors other than fine sediments, macroalgal cover, and chlorophyll a explained little variation in the cover or distribution of C. orientalis. The cover by other taxa (scleractinian corals, soft corals, sponges, macroalgae) did not influence the distribution or abundance of C. orientalis (Table 1; Supplementary Figure 1), suggesting that competition with these groups does not exclude C. orientalis from its habitat. Total carbon in the sediment explained some variation in C. orientalis abundance, but the effect was not significant in a model that included both total carbon and fine sediments (Table 1). Latitude explained little variation in C. orientalis abundance or distribution (Table 1) or in the environmental predictors (Supplementary Figure 2). Importantly however, processes affecting C. orientalis at small spatial scales were not accounted for. For example, whilst the presence-absence of C. orientalis varied between nearby locations (i.e., kilometres; Fig. 6), presence-absence also varied within locations (i.e., 250 m). Much of the unexplained variation in the distribution and cover of C. orientalis may be due to small-scale factors, such as the availability of hard substratum 44 .

Conclusion
Here, we present a large-scale monitoring effort to assess temporal changes in the abundance of the bioeroding sponge Cliona orientalis on the inshore GBR. Whilst Cliona abundance increased at 11 of 16 locations, increases in macroalgal cover and decreases in scleractinian and octocoral cover all outpaced changes in Cliona abundance. Low deposition of fine sediments was strongly associated with both the presence and abundance of C. orientalis, suggesting that the sponge requires exposed habitat. Increased cover of C. orientalis was only observed where mean chlorophyll a concentration was less than 0.45 µg/L and macroalgal cover was low, suggesting that C. orientalis can only increase in habitats where macroalgae are nutrient-limited. Experimental work that identifies the limiting environmental conditions (light, suspended sediment, nutrients) for C. orientalis is clearly warranted. Given the clumped distribution and strong association with local environmental conditions (e.g., sediment, macroalgae), bioeroding sponges such as C. orientalis likely represent site-specific -rather than regional -threats to coral health and reef accretion on the GBR.

Methods
Benthic surveys. Benthic cover was surveyed at 35 locations on the inshore GBR between 2005 and 2014 as part of the Inshore Water Quality and Coral Reef Monitoring program at the Australian Institute of Marine Science 56 . Briefly, at each location, two sites and two depths (2 and 5 m) were surveyed using five, fixed, 20 m transects. Every 0.5 m along each transect, photographs were taken of the benthos, which were used to determine presence-absence, percent cover, and change in percent cover. Survey data were pooled across sites and depths to relate to environmental variables measured at each location.
Percent cover was measured from digital photographs of the benthos. Five markers were overlaid onto each photograph and percent cover was calculated as the proportion of points occupied by each taxon. Percent cover of C. orientalis (encrusting ß form), other sponges, scleractinian corals, octocorals, and macroalgae was calculated for each of the four within-location survey sites. The influence of other benthic taxa on C. orientalis cover was explored using biplots of cover at each within-location site.
Trends in cover were analysed for each within-location site where C. orientalis was detected in at least three survey years. Trends were estimated for each location separately as the locations were surveyed at different frequencies over the course of the study. Change in percent cover was estimated for each within-location site using linear regression. Thus, change in percent cover represents the average of the within-location sites (1-4) where C. orientalis was detected. Analysis of presence-absence of C. orientalis at each location followed the same criterion as change in percent cover, whereby C. orientalis was considered present if it occurred in at least three survey years at any of the sites.
Environmental variables. Survey data were related to environmental variables collected at the location scale (not sites or transects). Water quality was assessed using satellite-derived data from the eReefs Marine Water Quality Dashboard (http://ereefs.org.au/ereefs), including chlorophyll a concentration, coloured dissolved organic matter, and non-algal particulates (1 km resolution). The data nearest each survey location were analysed for each survey year. Sediment was collected from the 5 m survey sites and the proportion of fine particles, carbon content, and nitrogen content in the sediment were measured as described in 56 , with average values compared to C. orientalis cover. Fine particles in the sediments were defined as all particles smaller than 63 µm and expressed as a proportion of the total sediment 67 . Data analysis. Exploratory plots were prepared to identify correlations amongst the environmental predictors and to compare the effects of different benthic taxa on C. orientalis cover. Note that only fine sediment, chlorophyll a, and total C in sediment were included in the model, as other environmental variables were strongly correlated with either the proportion of fine sediments or chlorophyll a (all r > 0.7). Uncorrelated environmental variables were used to predict the presence-absence of C. orientalis (generalized linear model (GLM) with binomial errors and logit link), the percent cover of C. orientalis (GLM with negative binomial errors and log link), and changes in C. orientalis cover per year (linear model). Latitude was used to account for the spatial relationships among locations. Model fit was evaluated by plotting residual and fitted values. For generalized linear models, model fit was also evaluated using the chi-square probability of the residual deviance and residual degrees of freedom and by comparing observed and simulated residuals from each model. Three models were used to assess whether other taxa, the environment, or geography explained patterns in C. orientalis distribution. Models were compared using AIC and R 2 values. For the GLM, R 2 was calculated as the deviance ratio of models with and without predictors. The most parsimonious model was identified as the model that maximized explanatory power with the fewest predictors.