Widespread variation in heat tolerance of the coral Acropora hyacinthus spanning variable thermal regimes across Palau

Climate change is poised to dramatically change ecosystem composition and productivity, leading scientists to consider the best approaches to fostering population resilience and diversity in the face of these changes. Here we present results of a large-scale experimental assessment of bleaching resistance, a critical trait for coral population persistence as oceans warm, in 293 colonies of the coral Acropora hyacinthus across 39 reefs in Palau. We find bleaching resistant individuals originate significantly more often from warmer reefs, although they inhabit almost every reef regardless of temperature at low frequency. High levels of variation within reefs, where colonies experience similar temperatures, suggests that bleaching resistance is not solely due to phenotypic plasticity, but also involves adaptive alleles and host-symbiont interactions. To the extent that it is heritable, bleaching resistance could be used in promoting nursery growth, habitat restoration, or breeding, while employing large numbers of resistant colonies to preserve genetic variation.

included evolutionary responses to climate change stress could lead to much higher levels 68 of stable persistence than ecological models without evolution. The key parameter in 69 models that predict population persistence is increased heritable phenotypic variation, 70 reefs near each of the northern and southern patch reef regions (although we could not re-150 sample one of the northern fore reefs to run the bleaching assay), for a total of 39 reefs. 151

152
To identify 10 colonies on each reef, we haphazardly swam transects from the 153 exterior to the interior of each patch reef to capture the full range of abiotic and biotic 154 conditions at each location. We selected fore reef colonies by swimming parallel to the 155 reef crest, just inside the edge of each reef. We tagged each colony with a unique numeric 156 identifier (PA1-400), and recorded its location using a handheld GPS (Garmin, Kansas). 157 Additionally, to characterize the thermal environment, we deployed a HOBO temperature 158 logger (OnSet Computing, Massachusetts) recording at ten-minute intervals to all odd-159 numbered colonies (five per reef). Temperature data were collected at all locations from 160 We experimentally stressed fragments from each of the 400 corals that we tagged 165 and monitored for this study. We clipped medium sized fragments (approximately 8cm 166 width) from the edge of each colony using garden clippers, loosely packaged the 167 fragments in bubble wrap and stored them in a cooler to be transported to the Palau 168 International Coral Reef Center (PICRC). Upon return, fragments were placed in a 169 flowing seawater system at ambient temperatures where they recovered from transport 170 overnight. The following day, we clipped the larger fragments into five smaller fragments 171 and placed them in our experimental warming system. 172 heater and 2 chillers (Nova Tec, Maryland), and had fresh seawater inflow at all times 175 (ca. one volume change in 2-3 hours) in addition to an aquarium pump (~ 280 L hour -1 ) to 176 increase flow around the fragments. Tanks were illuminated on a 12 hour light:dark cycle 177 using LED light fixtures (ca. 22-66 μmol m -2 s -1 ). Three experimental treatments ramped a 178 fragment from each colony to temperatures of 34°C, 34.5°C, or 35°C, with two additional 179 control tanks that did not ramp but were maintained at 30°C for the duration of the 180 experiment. Each experiment consisted of two ramping cycles over the course of two 181 days. Ramps began at 10:00 A.M. and increased temperature for three hours until they 182 reached their target temperature at 1:00 P.M. Heating lasted for three hours (until 4:00 183 P.M.) at which point chillers cooled each tank back to 30°C, typically within 30 minutes. 184 When tanks were not ramping, they maintained a constant temperature of 30°C. We 185 repeated this cycle over two days, after which corals were held at 30°C overnight before 186 preservation. We preserved coral tissue by removing tissue from each fragment using an 187 airbrush loaded with seawater. We then centrifuged the tissue for five minutes at 5,000 g, 188 removed the supernatant and resuspended the slurry in 2 ml of RNALater. We stored 189 these samples at 4°C for approximately 24 hours, after which time we transferred them to 190 -20°C until shipping to Hopkins Marine Station. 191

192
We visually evaluated coral fragments at 8:00 A.M. each day of the experiment 193 (one observation before each ramp and a final observation the day following the second 194 ramp for a total of three observations). We visually assessed coral fragments using a five point scale that consisted of the following categories: 1no bleaching, 2visually 196 discolored, 3moderately discolored, clearly bleaching, 4severely discolored, nearly 197 complete bleaching, 5no color, completely bleached. We evaluated each fragment 198 using two scorers who were required to agree on a score for each fragment during each 199 assessment. We also took photographs of each tank immediately following the visual 200 bleaching score assessment for confirmation in cases where the visual score disagreed 201 with symbiont quantification by flow cytometry (described below). In order to more precisely quantify the symbiont density in each fragment after 206 experimental heating, we used a Guava EasyCyte HT (Millepore, Massachusetts) flow 207 cytometer to count symbiont cells. To prepare each sample, we centrifuged 500 μl of 208 each cell suspension sample plus 500 μl filtered seawater at 15,000 g for 5 minutes, 209 removed the supernatant and resuspended each pellet in 300 μl 0.01% SDS solution. We 210 homogenized each sample using a PowerGen rotostat (Fisher Thermosciences, 211 Massachusetts) set at the highest setting for 5 seconds. We then needle sheared each 212 sample through a 25-gauge needle 10 times to ensure cells were not clumped together. 213 Each sample was then diluted by 1:200 in 0.01% SDS and run in triplicate on the flow 214 cytometer. We gated sample counts on the forward scatter channel, counting all events 215 exceeding 10 2 fluorescent units. Additionally, we counted all events that exceeded 10 4 216 fluorescence units on the 690 nm detector (which can detect the autofluorescence of 217 chlorophyll) as a symbiont count. After subtracting events we detected in the negative control (0.01% SDS), we calculated the proportion of symbionts as the total number of 219 symbiont events divided by the total number of events above the forward scatter gate. For 220 a subset of samples, we also quantified the total protein content of the homogenized 221 (undiluted) sample using a Pierce BCA protein assay (Thermo Scientific, Massachusetts) 222 as per protocols in Krediet et al. (2015). We measured the absorbance of these samples in 223 triplicate on a Tecan Plate Reader at 562 nm. We note that while cell count values may 224 not represent absolute numbers of cells within tested colonies (e.g. due to cell 225 fragmentation during tissue airbrushing and preservation, or Guava cell counting error), 226 values do confidently represent relative differences between samples. Previous work found that oceanic waters routinely flush through the interior of 233 northern fore reefs during tidal cycles (Skirving et al. 2005), which might make these 234 environments colder than southern interior waters. Similarly, we also expected fore reef 235 environments that are exposed to cooler upwelling conditions to have fewer extreme 236 temperature events than protected patch reef environments. The thermal profiles that we 237 recovered from loggers deployed on the same reef were highly correlated, we therefore 238 used the average thermal profile for each reef for these analyses. After trimming the 239 temperature measurements to begin and end at the same time points, we tested these 240 predictions by comparing the mean number of time events above 31°C in patch reefs compared to fore reefs using a one-tailed t-test. Other temperature thresholds resulted in 242 similar rankings. We then conducted a comparison between northern and southern patch 243 reefs using a one-tailed test with the prediction that northern reefs should experience 244 fewer high temperature events than southern reefs. We examined whether the rate of bleaching resistance over the course of the two-249 day experiment was dependent on geographic region (northern vs. southern reefs). First, 250 we calculated the average change in visual bleaching score (VBS) from day zero (no heat 251 stress) to day one (one cycle of heat stress) and from day zero to day two (two cycles of 252 heat stress). We then used a t-test to detect differences in bleaching rates between those 253 two regions and within regions across days of the experiment. 254

255
We normalized the reduction in symbiont density for each temperature treatment 256 by dividing the density of symbionts remaining after heat stress by the density of the 257 control treatments. We then used these standardized values to assess the degree to which 258 corals bleached during heat stress. We determined the performance of each coral as the 259 fraction of the average proportion of symbionts that were present after the heat treatments 260 compared to the average of the two control fragments. We also estimated the temperature 261 at which each colony lost 50% of its original symbionts. To do this we assumed a linear 262 decline in symbiont density from the initial symbiont proportions in the control treatment results in a subset of individuals that are adapted to that environment or individuals can 287 acclimate to local conditions. One prediction if the thermal environment selects for heat 288 resistant colonies, or results in acclimation to increased temperature, is that bleaching 289 resistant colonies should reside in warmer environments. We therefore assessed whether 290 bleaching responses were correlated with the number of extreme temperature events on 291 each reef. Because data loggers deployed to each reef were highly correlated (see above), 292 we used the average number of time points above 31°C for all loggers on a reef to 293 represent their exposure to thermally challenging conditions. We also conducted these 294 analyses using increasing thresholds of 32°C, 33°C and 34°C, none of which 295 substantially changed the outcome (not shown). 296

297
To test the relation of local heating to heat resistance, we used a fully factorial 298 linear model where the average fraction of the symbiont population that was lost in the 299 heated treatments (measured by flow cytometry) was the dependent variable and the 300 number of extreme temperature events on a reef, the area of the reef (excluding fore reef 301 sites), and the average depth of the colonies on each reef were the independent variables. 302 303 Finally, we characterized the proportion of each reef that is comprised of heat 304 resistant and sensitive corals (defined as any coral meeting 2 out of 3 criteria described 305 above, a total of 74 bleaching resistant and 42 bleaching prone corals were included). 306

309
Cell count and visual score data 310 We recorded control and heated symbiont density data from 361 colonies across 312 39 reefs. We removed 35 colonies from the dataset for which we had only one control 313 value (n=30) or only one heat treatment value (n=5). We also removed an additional 33 314 colonies for which the heated branches showed anomalously higher density of symbionts 315 than did the controls (ratio range 4.4 to 1.2). Most of these colonies (n=20) showed low On average colonies retained 53% of their original symbionts after heating for 354 two daily cycles as measured by flow cytometry. However, different colonies showed 355 wide variation in symbiont retention after standardized heating, ranging from 100% retention to less than 5% ( Fig. 2A). We saw >90% retention of symbionts after heating in 357 31 corals from 18 reefs. By contrast, 45 corals from 24 reefs retained less than 25% of 358 their symbionts under the same experimental conditions. Bleaching-resistant corals (in 359 the top 25% of retention) averaged 75% symbiont retention whereas bleaching-prone 360 corals (bottom 25%) averaged 37% retention ( Fig. 2A). 361

362
To test if our characterization of heat resistance based on the retention of 363 symbionts was robust, we also scored heat resistance based on the drop in Visual 364 Bleaching Score compared to controls. Increasing VBS were correlated with decreasing 365 symbiont proportions (r 2 =0.3054, p<2 x 10 -16 , Fig. 2supplemental 1). The difference 366 (delta) in VBS and symbiont cell counts between control treatments and heated 367 treatments was also correlated but did not explain a large portion of the variance 368 (r 2 =0.1287, p<2x10 -16 , Fig. 2supplemental 1). Because these analyses suggest that VBS 369 and cell counts capture slightly different aspects of bleaching, we incorporated metrics of 370 bleaching resistance based on the visual bleaching scores alone. Colonies ranged from no 371 change to a VBS shift of four categories, with a change of one category or less (e.g. 372 category 1 to category 2) occurring in 42 corals across 19 reefs. By contrast 22 colonies 373 showed a VBS change of 3-4 (Fig. 2B). 374 375 Last, we also used the visual bleaching scores for colonies after two days at the 376 highest temperature, 35°C, to find colonies with the highest heat resistance. Most 377 colonies (171) showed total bleaching at this temperature but about 24% showed severe 378 bleaching and 10% showed only moderate bleaching or less (Fig. 2C). 379 We used these three measures of heat resistance (proportion symbiont retained, 381 drop in visual bleaching score, and absolute visual bleaching score at the highest 382 temperature) to characterize each colony we tested. We considered a colony to be highly 383 heat resistant for a criterion if it was in the upper 25% -34% of colonies. For symbiont 384 retention, this included all colonies that retained more than 73% of their symbionts after 385 heating (average 89.4%). For visual bleaching scores, our highly resistant category 386 included all colonies that dropped less than one visual bleaching value (85 colonies or 387 30%), and all colonies that showed less than total bleaching at the highest temperature Corals that meet 2-3 heat resistance criteria start with slightly but insignificantly 393 fewer symbiont cells compared to corals that meet none of those criteria (9% vs 9.7%, 394 t=1.4576, df=148.19, p=0.1471), but show much higher levels of symbionts after heating 395 (average 6.3% vs 3.3%, p=3.743x10 -9 ), resulting in retention rates of 75% versus 37% 396 (Table 2). Likewise, Visual Bleaching Scores were the same in controls for both 397 categories (2.4 vs 2.3, t = 1.8309, df = 141.66, p = 0.0692), but differed strongly when 398 corals meeting 2-3 criteria versus those meeting none were heated (VBS of 3.7 versus 5.0 399 at 35°C, Table 2). 400

401
The geography of heat resistant corals 402 Corals that are heat resistant on the basis of at least two of our three criteria are 404 found at many locations: at least one heat resistant colony (out of the 10 we surveyed) 405 inhabited 31 of 39 reefs (Fig. 3 upper). However, heat resistant corals tend to be more number of corals that met 2 of 3 heat resistance criteria on a reef is also correlated with the number of warm temperature events that reefs experienced during this study (Fig. 4 -449 Supp.1). However, even most cooler reefs in our study had one or two heat resistant 450 colonies, suggesting that factors other than temperature contribute to the distribution of 451 bleaching resistance. Across 39 reefs that we sampled in Palau, we found wide variation in bleaching 510 susceptibility resulting from heat stress of the table top coral A. hyacinthus. In a simple 511 two-day standardized heat stress experiment, colonies ranged from retaining virtually all 512 of their original symbiont load at 35°C (ca. 5°C above ambient temperatures) to less than 513 30% at this temperature (Fig. 2). Overall, 10% of the 293 colonies we sampled met all 514 three of our criteria for high bleaching resistance, and another 15% met two of three 515 criteria; if this fraction is representative of the species as a whole, then the absolute 516 number of bleaching resistant colonies across the Palau archipelago is large and they are 517 distributed across a wide range of environments. 518 519 If heritable, the natural variation we identified in these populations is the 520 foundation on which selection can act due to differential mortality from high 521 temperatures. Given the current pace of climate change, selective mortality of bleaching-522 prone individuals due to heat stress is likely to be a strong force shaping the evolution of For example, Acropora on fore reefs bleached less than their patch reef counterparts and 552 Pocillopora showed the opposite, although these differences were not large. It should be 553 noted that small-scale variation in thermal regimes (e.g. PR 17 and to a lesser extent FR 554 60 and 61 are particularly warm relative to nearby reefs) could be contributing to 555 bleaching resistance, making environmental designations based on reef morphology alone 556 difficult to interpret in the context of identifying locations likely to harbor bleaching 557 resistant corals. Overall, whether coral assemblages are well adapted to their local 558 temperature regimes is an area for continued research. 559 560 Nevertheless, we found substantial numbers of heat resistant colonies on reefs that 561 experienced very few extreme temperature events. For example, cooler patch reefs in Palau's northern lagoon had 17 heat resistant colonies. Although this was a low 563 percentage of the colonies we surveyed, these individuals added substantially to our 564 inventory of heat resistant colonies. This is also true of shallow fore reef locations, which 565 harbored 17 heat resistant colonies in our final data set ( Fig. 3 and Fig. 4B). 566

567
These results show that bleaching resistance due to heat stress is not always 568 restricted to the warmest environments. Although we found patch reefs that had a high 569 prevalence of bleaching resistant colonies were also associated with a large number of 570 extreme temperature events, notably the western edge of the southern lagoon, fore reef 571 settings are the most extensive habitat for A. hyacinthus  2000), and they may paradoxically house the majority of heat resistant colonies in this 573 species in Palau. Finding these colonies will be more difficult because they only occur at 574 a frequency of 18% in our survey. However, high census population sizes on fore reefs 575 means that they could contain the highest numbers of bleaching resistant individuals. 576 577 Symbiont and environmental contributions to heat resistance 578 579 Overall, the mechanisms that create a mosaic pattern of bleaching resistance 580 among colonies on a reef can fall into several broad categories with very different 581 underlying causes: symbiont type, environmental history, and host genetics. 582

583
In particular, the symbionts themselves could be contributing significantly to 584 variable resistance among coral colonies. Coral holobiont heat tolerance is known to vary 585 with the genotypes and physiologies of the single-celled dinoflagellate endosymbionts 586 that carry out photosynthesis within host tissues (e.g. Sogin et al. 2017). In particular, 587 Durusdinium symbionts tend to confer higher heat resistance to the holobiont 588 pathogens from distant locations into an immunologically naïve population, which is a 688 risk that has long been known when transporting individuals for other purposes such as 689 commercial fisheries (Sindermann 1992 controller with reasonable accuracy, a heater, and a source of cooler fresh seawater. Our 709 basic data collection protocol in the field uses a simple five-point visual bleaching scale. 710 Although there was considerable variation in symbiont density for each visual bleaching 711 score level, the agreement between the visual scores and symbiont proportions measured 712 using flow cytometry, when combined with the ease of collecting visual scores relative to 713 flow cytometry measurements makes this approach especially effective at rapidly and 714 inexpensively assaying coral populations. Visual bleaching scores miss some important 715 details (e.g. that symbiont load is negatively correlated with bleaching severity), 716 however, the ease of documenting strong differences in bleaching results with visual 717 bleaching scoressevere versus visible bleaching for exampleopens up this tool to 718 wider adoption. This will be of use not only to scientists, but importantly also to local 719 managers with limited project budgets and manpower.  Supplemental methods and results 1107