Temperature and Water Quality-Related Patterns in Sediment-Associated Symbiodinium Communities Impact Symbiont Uptake and Fitness of Juvenile Acroporid Corals

The majority of corals acquire their photo-endosymbiont Symbiodinium from environmental sources anew each generation. Despite the critical role that environmental availability of Symbiodinium plays in the potential for corals to acclimate and adapt to changing environments, little is known about the diversity of free-living Symbiodinium communities and how variation in these communities influences uptake and in hospite communities in juvenile corals. Here we characterize Symbiodinium community diversity in sediment samples collected from eight reefs representing latitudinal and cross-shelf variation in water quality and temperature regimes. Sediment-associated Symbiodinium communities were then compared to in hospite communities acquired by A. tenuis and A. millepora juveniles following 11 – 145 days of experimental exposure to sediments from each of the reefs. Communities associated with juveniles and sediments differed substantially, with sediments harbouring four times more unique OTUs than juveniles (1125 OTUs vs. 271). Moreover, only 10.6% of these OTUs were shared between juveniles and sediments, indicating selective uptake by acroporid juveniles. The diversity and abundance of Symbiodinium types differed among sediment samples from different temperature and water quality environments. Symbiodinium communities acquired by juveniles also differed among the sediment treatments, despite juveniles having similar parentage. Moreover, Symbiodinium communities displayed different rates of infection, mortality, and photochemical efficiencies. This study demonstrates that the biogeography of free-living Symbiodinium types found within sediment reservoirs follows patterns along latitudinal and water quality environmental gradients on the Great Barrier Reef. We also demonstrate a bipartite strategy for Symbiodinium uptake by juvenile corals of two horizontally-transmitting acroporid species, whereby uptake is selective within the constraints of environmental availability.

being the most abundant on Pacific reefs (Takabayashi, et al., 2012). Clade A was also 1 1 5 recovered from coral larvae exposed to sediments, and B and C from juveniles exposed 1 1 6 to reef waters in Japan (Adams et al., 2009). In addition, larvae infected using sediments 1 1 7 sourced from the northern and central GBR contained Symbiodinium C1, C2, and D,  were used to account for non-linear trends over time (Wood, 2006) using the package 3 0 2 'mgcv' (Wood, 2006(Wood, , 2008. Penalized regression spline smoothing functions were 3 0 3 applied to the interaction between sampling date and site, whilst site itself was also 3 0 4 included as a fixed effect. Plate was treated as a random effect and the variance 3 0 5 structure was allowed to vary through time using the varIdent weights argument. Temporal autocorrelation from sequential measurements was dealt with using first-order the 'anova' function from the 'nlme' package. A log-normal distribution was used, as  To examine if there were any relationships between F v /F m measurements and 3 1 4 normalized OTU abundances in A. tenuis juveniles exposed to sediments from central De'ath and Fabricius, 2008;De'ath, 2007;Furnas, 2003;Furnas et al., 2005). Values for 3 3 4 each covariate per site were extracted from interpolated modelled data using the R 3 3 5 packages 'dismo' (Hijmans et al., 2013) and 'raster' (Hijmans), and figures for each 3 3 6 were created using the package 'mapping' and a custom Queensland spatialPolygons  To address the highly correlated nature of many of the irradiance and water  (GAMS) across 7 of the 9 clades. GAMS and partial plots were constructed using the packages 'mgcv' (Wood, 2000(Wood, , 2006(Wood, , 2008 of normalized abundance data generated 3 5 7 from the full data set. Non-significant covariates and regression spline smoothing of OTUs identified to the same type were summed. A non-metric multidimensional 3 6 0 scaling plot (NMDS) of the full data set was constructed using the Bray-Curtis distance  Symbiodinium communities differed significantly between juvenile corals and 3 6 8 sediments when data were combined across all samples of sediments, juveniles (both 3 6 9 species) and time points (Figures 1, 2 OTUs were unique to juveniles. Only 10.6% (166) were shared between juveniles and 3 7 2 sediments ( Figure S2C). In hospite and ex hospite diversities were only moderately 3 7 3 correlated among sites (mean R 2 = 0.229 ± 0.04, Figure 3). OTUs shared between juveniles and sediments belonged predominantly to also present in both sample types, but at lower diversities (< 40 OTUs in total) ( Figure   3 7 7 S2). Overall, within-clade diversity was much lower in juveniles (< 100 OTUs) than in sediments also held a larger abundance of diverse types within clade A, including free-3 8 7 living types from A1, A3, and A4. Clades A and C were the most abundant Symbiodinium clades in the sediments 3 9 2 (A: 19.5% of total sediment reads, 177 of 1291 OTU sediment diversity, C: 11.5%, 148 3 9 3 of 1291, respectively), particularly a free-living type from clade A (OTU_10). A 3 9 4 majority of the unique OTU diversity retrieved from sediment samples was identified as 3 9 5 "uncultured" Symbiodinium (i.e. novel Symbiodinium without taxonomic descriptions, 3 9 6 >300 OTUs), which represented 82% of all uncultured Symbiodinium OTUs across the 3 9 7 whole dataset (i.e., sediment and juvenile samples combined; Figure S2).

9 8
Approximately 100 unique OTUs were represented from each of clades A, C, D, E, and mostly due to strong differences in diversity across the inshore to offshore gradient. Symbiodinium communities in the sediments differed significantly between 4 0 7 inshore and offshore locations (p=0.001, Table 1; Figure 1). In total, the abundances of

Northern versus central sector sediment community comparisons
Symbiodinium communities within northern sector sediments also differed 4 1 6 significantly from those detected within central sector sediments (p=0.001, Table 1; abundances of 10 different A-types, including A1, A3, A13, and up to 3.1 times greater 4 2 0 abundances of C and C1 types, which were not found in sediments from central reefs 4 2 1 (B-H adjusted p-values < 0.05, Figure 5). There were also up to 3.6 times higher  Of the OTUs unique to juveniles, clade C had the greatest type diversity (99 4 3 1 OTUs), although each type was relatively rare in abundance, thus C types comprised 4 3 2 only 9.8% of juvenile reads overall. In terms of abundance, unique juvenile OTUs 4 3 3 belonged predominantly to B1, with this type representing 79.5% of all reads unique to 4 3 4 juveniles ( Figure S2). As found for sediments, in hospite Symbiodinium community 4 3 5 diversity was only moderately correlated among juveniles exposed to sediments from 4 3 6 the different sites (mean R 2 = 0.33 ± 0.14, Figure 3, Supplementary Results). In 4 3 7 addition, patterns in community composition for juveniles exposed to sediments from 4 3 8 inshore versus offshore sites differed between sectors (p=0.004, Table 1). The early Symbiodinium community in juveniles differed significantly when 4 4 2 juveniles were exposed to inshore versus offshore sediments (p=0.001, Table 1; Figure   4 4 3 1). The abundances of 12 OTUs differed significantly in juveniles exposed to inshore 4 4 4 versus offshore sediments ( Figure 6A). Juveniles exposed to inshore sediments were compared to juveniles exposed to offshore sediments (B-H adjusted p-values < 0.05). 4 4 7 Juveniles exposed to offshore sediments were dominated by A3, D1, and C15, and had 4 4 8 up to 2.9 times greater abundances of A3 and C15 types compared to juveniles exposed 4 4 9 to inshore sediments (B-H adjusted p-values < 0.05). The early Symbiodinium community in juveniles exposed to northern versus   Figure 6B). Juveniles exposed to northern sediments were dominated by A3 4 5 5 and D1 and had up to 3.9 times greater abundances of A1, A3, A7, and A4, D1a, D1, 4 5 6 C1, C90, and C (B-H adjusted p-values < 0.05). Juveniles exposed to sediments from 4 5 7 the central sector were dominated by B1, C15, C1, and C-types, and had up to 3.3 times 4 5 8 more B1, C, C15, and A4 compared to juveniles exposed to northern sediments. Symbiodinium communities in A. tenuis juveniles exposed to central sector and over time (B-H adjusted p-values < 0.05, Table S3). In particular, large community 4 7 0 shifts were seen 8 days later (day 19) at each site, with communities in juveniles 4 7 1 exposed to central offshore reef (Davies and Rib) sediments more closely resembling 4 7 2 each other, whereas communities in juveniles exposed to central inshore reef (Magnetic communities in juveniles exposed to offshore sediments were beginning to resemble 4 7 7 each other, particularly in their abundances of C15, A3 and C1. In contrast, it wasn't 4 7 8 1 1 until day 75 that Symbiodinium communities in juveniles exposed to inshore sediments 4 7 9 started to resemble each other, most notably in their abundances of B1, C1 and D1.

8 0
After approximately three months of exposure to sediments (day 90), the abundances of 4 8 1 C1 and C15 had increased to become dominant in juveniles exposed to sediments from 4 8 2 all sites, following steady increases between days 41 and 90 (Figure 7, B -H adjusted p-4 8 3 values < 0.05, Table S3). Infection was significantly more rapid in juveniles exposed to offshore 4 8 7 sediments (19.1 days ± 0.09) than in juveniles exposed to inshore sediments (24.3 days 4 8 8 ±2.13) (negative binomial generalized linear mixed model: p = 0.014), with no effect of 4 8 9 location (p = 0.45) ( Figure 8A). The cross-shelf (shore) effect was predominantly 4 9 0 driven by significantly lower mean times to infection for juveniles exposed to offshore inshore Pandora sediments (26.2 days ±4.37, p = 0.0145). Juveniles exposed to inshore  Survival of juveniles did not differ when juveniles were exposed to offshore 4 9 6 (68.9±5.7 days) versus inshore (71.95±1.2) sediments (Poisson GLMM: p = 0.179), 4 9 7 although significant differences did exist at the site level (p = 0.002) ( Figure 8B). 4 9 8 Juveniles exposed to Rib sediments survived significantly fewer days (51 ± 0 days) 4 9 9 compared to those exposed to Davies (80.8 ± 7.9, TPH: p = 0.003) and Magnetic (75 ± 5 0 0 0, TPH: p = 0.03) sediments. However, survival did not differ significantly between 5 0 1 juveniles exposed to Rib versus Pandora sediments (69.1 ± 2.1, TPH: p = 0.1). Survival 5 0 2 of juveniles exposed to sediments from Davies, Magnetic and Pandora did not differ  3.6 Photo-physiology of juveniles exposed to sediment treatments 5 0 6 3.6.1 Temporal patterns in photochemical efficiency compared among juveniles 5 0 7 exposed to central sector sediments 5 0 8 The photochemical efficiency (F v /F m ) of juveniles varied significantly among 5 0 9 sediment treatments and over time ( Figure 8C; GAM: p = 0.0283, Table S7). F v /F m 5 1 0 values for juveniles exposed to Davies sediments were initially ~0.6, but dropped sharply by day 48, and then generally increased until day 145. In comparison to 5 1 2 juveniles exposed to Davies sediments, temporal patterns in F v /F m values differed 5 1 3 significantly for juveniles exposed to sediments from the other three locations: = 0.0467). Differences in F v /F m yields among juveniles exposed to sediments from 5 1 6 different locations were also correlated with distinct shifts in Symbiodinium 5 1 7 communities in these juveniles, for example, the increase in A3 coincided with  In northern sediment treatments, juveniles of both A. millepora and A. tenuis were dominated by Symbiodinium A3 and D1, and to a lesser extent by D1a in A. millepora. Overall, Symbiodinium communities did not differ between A. tenuis and A. 5 2 5 millepora juveniles exposed to northern sediments after 35 days (permutational abundance of nine OTUs varied between species (B-H adjusted p-values < 0.05, Table   5 2 8 1 2 S4). In contrast, overall Symbiodinium communities hosted by juveniles of these two 5 2 9 species did differ significantly after 27 -30 days of exposure to central sediments (Df 5 3 0 1,41 , F = 4.83, R 2 = 0.107, p = 0.001). A. tenuis juveniles exposed to central sediments 5 3 1 were more diverse and had significantly greater abundances of six A types, B1, and C1 5 3 2 (B-H adjusted p-values < 0.05, Table S5) compared to A. millepora, which was 5 3 3 characterized by additional diversity within C15, D1 and D1a types. abundances of Symbiodinium types from clades A, B, C and D, but not types within 5 3 8 clades E, F and G (Figure 9, Figure S5, Table S7). Responses of free-living 5 3 9 Symbiodinium to these environmental covariates were generally linear, in contrast to the decreasing SST, and decreasing in abundance with lower carbonate and mud content.

4 4
This contrasted with A3 abundances, which decreased with increasing WQI. The distributions of types C15, C3 or C90. Interestingly, temperature did not significantly 5 5 0 explain D1 or D1a abundance, which significantly increased as carbonate decreased in 5 5 1 contrast to in hospite patterns (i.e., higher abundance in offshore locations Figure 6a). S.   Figure S4A and B, Table S6). predisposition for acquiring certain Symbiodinium types is supported by the acquisition of 6 1 0 clade C symbionts in offshore sediment treatments when their comparative availability in 6 1 1 offshore sediments was low, potentially signifying the presence of recognition mechanisms, 6 1 2 consistent with findings that clade C ultimately dominates adult communities of these corals contributions to the structuring of symbiont communities, has also been observed in closely- environmental availability of symbionts, in conjunction with an understanding of the genetic 6 1 7 mechanisms regulating symbiont uptake by the host, is therefore required to predict the 6 1 8 potential for Symbiodinium communities in acroporid juveniles to affect coral health through 6 1 9 differential selection of symbionts in response to environmental change.
The in hospite Symbiodinium communities in juveniles also varied through time, availability within the sediments from the start to the end of the experiment indicates that 6 2 7 lack of symbionts available for infection was not the underlying reason, but does suggest that 6 2 8 active selection by the host or competition among symbionts was driving temporal changes in 6 2 9 community structure. Although developmental changes associated with metamorphosis in 6 3 0 other invertebrate systems may be more extreme than changes associated with growth in A. in Symbiodinium communities associated with A. tenuis juveniles may be tied to changing 6 3 5 requirements associated with the onset of specific physiological processes (e.g., calcification) by these Acroporid species. Additionally, A. millepora juveniles in both the northern and and D1a (generally heat tolerant types), than A. tenuis. If species-specific preferences for D 6 5 2 types exist, then by not hosting thermally tolerant types, A. tenuis juveniles may be at a  variation in water quality, temperature, and sediment characteristics. by the coral Acropora millepora on the Great Barrier Reef. PLoS One 6, e25536.     Variability of Symbiodinium communities in waters, sediments, and corals of thermally  horizontally-transmitting corals. bioRxiv. Great Barrier Reef, and indicators of water quality and mapping risk. Townsv. Aust. Inst.     Furnas, M., Mitchell, A., Skuza, M., and Brodie, J. (2005). In the other 90%: phytoplankton    Stress Response Corals Symbiodinium a rapidly Chang. Environ., 29.  Huang, H., Zhou, G., Yang, J., Liu, S., You, F., and Lei, X. (2013). Diversity of free-living Res. 9, 117-128.  Karim, W., Nakaema, S., and Hidaka, M. (2015). Temperature effects on the growth rates    of Symbiodinium OTU's using Bray-Curtis distances. Symbols and their sizes represent: them; blue, green and grey colours denote central sediments and juveniles exposed to them. were found in significantly greater abundances in: A) offshore than in inshore sediments, and 1 0 5 0 B) in northern sector sediments than central sediments.  Analyses were restricted to 7 time points between day 11 and day 90 of sediment exposure have y-axes that are scaled to best show abundance dynamics for each Symbiodinium type. Colours represent nine key Symbiodinium types, whose abundances differed significantly  sediment treatment at day 117, highlighting the level of variation in symbiont proliferation 1 0 7 0 within juveniles exposed to sediments from different sites.