Habitat suitability modeling growth and cover trends of staghorn coral (Acropora cervicornis) outplants in the lower Florida Keys

1 1 2 3 4 Habitat suitability modeling growth and cover trends of staghorn coral (Acropora cervicornis) 5 outplants in the lower Florida Keys 6 7 Glenna Dyson1,5*, Erich Bartels2, Easton R. White3, Ian R. Combs2, Kristen Mello-Rafter1, 8 Thomas C. Lippmann4,5, Jennifer A. Dijkstra1* 9 10 11 12 1 Center for Coastal and Ocean Mapping/UNH-NOAA Joint Hydrographic Center, University of 13 New Hampshire, Durham, New Hampshire, United States of America, 03824 14 2 Elizabeth Moore International Center for Coral Reef Research & Restoration, Mote Marine 15 Laboratory, Summerland Key, Florida, United States of America, 33042 16 3 Department of Biological Sciences, University of New Hampshire, Durham, New Hampshire, 17 United States of America, 03824 18 4 Center for Ocean Engineering, University of New Hampshire, Durham, New Hampshire, 19 United States of America, 03824 20 5 Department of Earth Science, University of New Hampshire, Durham, New Hampshire USA, 21 03824 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 15, 2025. ; https://doi.org/10.1101/2025.07.11.664295doi: bioRxiv preprint 2 22 23 * Corresponding authors 24 E-mail: [email protected] (GD); [email protected] (JD) 25 26 27 ABSTRACT 28 The decline of important reef building corals has motivated the development of habitat 29 suitability models used to identify optimal locations for coral restoration. In the Florida Keys 30 habitat suitability models incorporate coarse spatial data sampled over large areas, resulting in 31 recommended outplant sites at distant locations, making it logistically difficult and expensive to 32 access and regularly monitor. Restoration efforts to date show that outplanting success can vary 33 widely within a limited space, necessitating improved predictive abilities of coral outplant 34 success at high spatial resolutions within a restoration site. With the advent of Structure-from- 35 Motion image reconstruction, fine-scale, site specific, digital terrain models can be created to 36 support habitat suitability model development. In this study, generalized linear mixed models 37 used extracted seafloor terrain attributes and environmental variables to identify within site 38 locations of high Acropora cervicornis growth and healthy coral cover of long-term outplants. 39 Percent healthy coral cover significantly decreased after two years of outplantation. The 40 submodel of corals exclusively less than two years old was unable to identify environmental 41 conditions associated with higher healthy cover. For all corals, outplant recommendations for 42 higher healthy cover are in deeper waters, away from the coast, in less rough terrain, and closer 43 to the reef edge. Model results for growth support these recommended outplant sites, in addition .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 15, 2025. ; https://doi.org/10.1101/2025.07.11.664295doi: bioRxiv preprint 3 44 to concave locations near high slope relief. Finally, our results also indicate that marine heat 45 waves, but especially marine cold waves negatively correspond with coral growth, and high wind 46 events positively correspond with coral growth. These model results provide a basis for further 47 endeavors in modeling endangered organismal success, which are vulnerable to minute 48 differences in local environmental conditions. 49 Introduction 50 In the past half century, corals around the world have been decimated due to mounting 51 anthropogenic pressures including disease, overfishing, river run off, high intensity thermal and 52 wind events (1–4). These threats have resulted in increased efforts to promote resilient reefs, 53 capable of withstanding and recovering from disturbance through the creation of marine 54 protected areas, genetic engineering of corals towards elevated heat resistance, and outplanting 55 corals. Stony reef-building corals have been the primary target of outplantation. One such genus, 56 Acropora, has been targeted due to forming much of the essential reef structure throughout the 57 Florida Reef Tract (5) and its outplantation facilitates variation in benthic coral composition and 58 fish species (6). Despite being a good candidate for outplantation, due to asexual propagation and 59 a fast growth rate (7), the genus has shown to be especially vulnerable to environmental stress 60 (8,9) and disease (10,11). To support efforts towards enhancing reef resiliency, habitat suitability 61 models (HSMs) targeting Acropora restoration have been developed, including models of wild 62 Acropora population occurrence (12), recruitment (13), and global responses to ocean warming 63 and acidification (14). 64 These HSMs have the potential to identify optimal sites for Acropora outplantation, 65 however models for outplant success have been challenging for coral reef managers and 66 outplanters to implement. Current models often use coarsely resolved environmental data (9), .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 15, 2025. ; https://doi.org/10.1101/2025.07.11.664295doi: bioRxiv preprint 4 67 many collected through remote sensing efforts, resulting in recommended sites that are 68 financially and logistically difficult to access for outplantation and recurrent monitoring. 69 Predicted suitable habitat can be hundreds of square kilometers, although many restoration 70 activities occur at much finer spatial scales (10s of meters) where coral survival throughout the 71 same site is highly variable (15). Additionally, current models have primarily used wild 72 Acropora populations. Coral outplant’s growth and survivorship perform differently than wild 73 and transplanted coral populations (16,17), making these models imprecise in recommendations 74 for outplants. Lastly, when coral outplants are surveyed, long-term monitoring is often not 75 realized nor influential in informing outplanting techniques, and thus sustained recovery of 76 Acropora outplants is not clearly established (18,19). Consequently, there is a knowledge gap in 77 HSMs using fine-scale, within site, terrain attributes for long-term coral outplant locations. 78 This study aims to develop HSMs utilizing fine-scale terrain attributes which may 79 contribute to within site variations in growth (total and healthy) and healthy cover of a 80 commonly used stony coral for outplantation, Acropora cervicornis. A. cervicornis has been one 81 of the primary targeted Acropora species for restoration due to its ecological importance as a reef 82 builder and its physiological characteristics as a long-lived, fast growing species (20–22). HSMs 83 developed thus far have shown improved wild A. cervicornis populations in waters with 84 moderate sea surface temperatures and limited temperature ranges, moderate turbidity to mitigate 85 temperature fluctuations and UV radiation (9), and higher water flow to deliver food and oxygen 86 (9,23). From these results, restoration work has called for outplanting deeper, or where there is 87 higher water flow to decrease the negative impact of high intensity thermal events (24). These 88 HSMs show A. cervicornis has limited suitable habitats within the Florida Reef Tract, 89 constrained to the fore and back reef of the lower and upper Florida Keys, the Dry Tortugas, and .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 15, 2025. ; https://doi.org/10.1101/2025.07.11.664295doi: bioRxiv preprint 5 90 nearshore Broward-Miami reefs (15,23). This study’s use of fine-scale environmental data aims 91 to delineate optimal coral outplant sites within such regions by modeling outplant growth and 92 percent healthy cover within the forereef of the lower Florida Reef Tract. Growth (separated into 93 total and healthy) indicates where at fine-scales there is optimal mass transport (exchange of 94 nutrients and dissolved gasses) and minimal stress (25,26). Total growth indicates where the 95 most growth is occurring, whereas healthy growth indicates where the most growth is occurring 96 despite infection from disease. Percent healthy cover indicates where corals are least exposed to 97 disease, are more resilient to disease, or where corals are best able to recover from infection (27). 98 In this study, we use fine-scale three-dimensional terrain attributes derived from 99 Structure-from-Motion (SfM) photogrammetry, environmental data, and high intensity thermal 100 and wind events as predictor variables for modeling A. cervicornis growth and percent healthy 101 cover. The use of SfM is unique in these HSMs for its capacity to develop three-dimensional 102 terrain models which provide insight in terrain structure and complexity (28–30); the scale of 103 these models is also several orders of magnitude finer (mm) than spatial data often used in HSMs 104 (0.25km to 1 km). This study aims to address the utility of SfM and fine-scale data to predict 105 site-specific optimal coral outplant locations. Model results contribute towards further 106 understanding of A. cervicornis restoration and the applicability of fine-scale models for other 107 endangered species. 108 109 Methods 110 Outplanting .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 15, 2025. ; https://doi.org/10.1101/2025.07.11.664295doi: bioRxiv preprint 6 111 Percent cover and growth of A. cervicornis were determined at eight outplant sites in the 112 lower Florida Keys. Corals were outplanted by Mote Marine Laboratory at Sand Key, Eastern 113 Dry Rock, M32, ICC1, and ADAC1 (Fig 1). All sites reside within the Florida Keys National 114 Marine Sanctuary. Additionally, Eastern Dry Rock and Sand Key are sanctuary preservation 115 areas. 116 Coral outplant fragments ranged from 250 to 1000 per outplant site, and the corals’ initial 117 sizes ranged from 3cm to 50cm. Except for the singular 3cm coral plugs at ADAC1, all other 118 sites had outplanted clusters of five coral fragments. Corals were affixed to the seafloor using 119 either nails and cable ties; nails, cable ties, and epoxy; or affixed to plugs with epoxy. To 120 enhance genetic diversity, a variety of genotypes were outplanted. 121 122 Fig 1. Map of coral outplant sites. Eight survey sites within the lower Florida Reef Tract to 123 west of Key West (S4, S6, T3, T16, M32-2, M32-3, ICC1, and ADAC1). Shaded areas refer to 124 sanctuary preservation areas. 125 126 Image Collection 127 Eight coral outplant sites were surveyed in June of 2022, and one of the eight sites 128 (ICC1) was previously surveyed the year before in July of 2021, resulting in nine outplant 129 surveys (Table 1). Coral outplant ages ranged from 129 to 1,788 days. All sites, except ADAC1, 130 are on the forereef of the Florida Reef Tract. Images were collected in 10 x 30m plots at all sites. 131 Coded targets (which also functioned as scale bars to ensure accurately sized orthomosaics) were 132 systematically placed throughout each site in a grid formation to support co-registration. Images 133 were taken with two Canon 70D DSLR cameras housed in an Aquatica A70D. Planar images of 134 the seafloor were collected in a lawn-mower survey pattern to facilitate co-registration and limit .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 15, 2025. ; https://doi.org/10.1101/2025.07.11.664295doi: bioRxiv preprint 7 135 distortion, resulting in ~ 80% image overlap. Depth measurements taken from a dive computer 136 accounted for the four corners of each 10 x 30m site. 137 Table 1. Summary of coral outplant locations and time spent outplanted. Outplant ID Location Outplant Date Age of outplants when surveyed T3 Eastern Dry Rock 7/8/2017 1788 days S4 Sand Key 9/11/2018 1386 days S6 Sand Key 3/25/2019 1191 days M32-2 Marker 32 8/9/2019 1050 days M32-3 Marker 32 5/6/2020 779 days ICC1(21) ICC1(22) Site C 12/10/2020 237 days 564 days ADAC1 Acer 10/5/2021 261 days T16 Eastern Dry Rock 2/11/2022 129 days 138 139 Three-dimension Terrain Model 140 Orthomosaics and digital terrain models (DTMs) were constructed using Agisoft 141 Metashape Professional V1.2.6 software. Orthomosaics and DTMs followed similar protocol to 142 other researchers and colleagues (31,32), with slight modifications. Permanent control points, 143 galvanized nails hammered into the bedrock and epoxied at the base to the seafloor, were 144 georeferenced using a GARMIN system on-board the vessel for image reconstruction. A. 145 cervicornis cover was quantified from orthomosaics (Fig 2). 146 147 Fig 2. A 10 x 10m orthomosaic and a magnified section showing outplanted staghorn coral 148 in site ICC1. 149 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 15, 2025. ; https://doi.org/10.1101/2025.07.11.664295doi: bioRxiv preprint 8 150 Coral Cover 151 Feature class polygons of healthy coral cover and total coral cover were created. Healthy 152 tissue cover was identified as orange/beige tissue, without any unhealthy cover (coral tissue 153 affected by disease or covered with bacterial mats or macroalgae). Total coral cover included all 154 skeletal tissue cover, both healthy and unhealthy. Percent healthy cover was quantified as the 155 percentage of healthy coral cover from total coral cover. Coral cover was then classified by 156 fragment, wild (non-outplanted), or outplanted corals. Total and healthy growth was quantified 157 as the change in total skeletal cover and healthy tissue cover since outplantation, divided by time 158 since outplantation. Only one site, ADAC1, had wild coral populations within the survey plot 159 and were easily identifiable, as they were geographically distant from the outplants and much 160 larger than the recently outplanted 3mm coral plugs. ADAC1 outplants were all singular plugs; 161 other sites were composed of five fragment aggregates; older sites (S4, S6, T3) composed of five 162 fragment aggregates fused together to form a mass of thickets. Average coral fragment cover at 163 the time of outplantation was utilized to account for variations between initial size in calculating 164 growth. 165 Covariates 166 Bathymetric and Environmental Parameters 167 DTMs were imported into ArcGIS Pro and projected in UTM 17 zone projection with 168 WGS84 datum. DTM resolutions were standardized (utilizing the resample tool in ArcGIS Pro) 169 to 8mm and were georeferenced to align with the orthomosaics to ensure accurate sampling of 170 terrain attributes around corals. To ascribe seafloor terrain attributes to coral cover, a 0.1m buffer 171 around each coral polygon was created. Associating seafloor terrain and environmental attributes 172 - calculated using the Spatial Analyst toolbox and Analysis toolbox in ArcGIS Pro - with total .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 15, 2025. ; https://doi.org/10.1101/2025.07.11.664295doi: bioRxiv preprint 9 173 coral skeletal cover were extracted using ArcGIS Pro ModelBuilder for each coral outplant and 174 included as covariates in the HSM (Table 2; Fig S1). 175 Table 2. Summary of spatial statistics utilized to characterize terrain and environmental 176 attributes. Terrain and Environmental Attribute Unit Definition ArcGIS tool Aggregate function Depth Meters Distance between water surface and benthos. Surface Information Max Slope Degree Benthic steepness. Surface Parameter Mean Aspect Northerness and Easterness Cosine(Degree) Sine(Degree) Direction the terrain is facing. Surface Parameter Mean Mean Curvature 1/Meters Second derivative of slope. Measures convexity/denudation or concavity/accumulation of terrain. Surface Parameter Mean Roughness Unitless Surface relief ratio measures a surface’s unevenness. Focal Statistics and Raster Calculator N/A Distance from Reef Edge Meters Apart from ADAC1, all coral outplant sites were along the edges of the reef, near drop- offs. The Euclidean distance of the A. cervicornis from the reef drop-off Near N/A Distance from Coast Meters The Euclidean distance of the A. cervicornis from the nearest coast Near N/A Distance from other A. cervicornis Meters The Euclidean distance of the A. cervicornis from the nearest A. cervicornis Near N/A 177 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 15, 2025. ; https://doi.org/10.1101/2025.07.11.664295doi: bioRxiv preprint 10 178 High intensity events 179 Water temperature and wind data spanning 30 years were cumulatively collected from 180 three NOAA buoys: Sand Key (January 1991 – September 2005), Sombrero Key (February 1998 181 – February 2008), and Key West (February 2005-January 2021). Marine heat waves (MHW) and 182 marine cold waves (MCW) are defined herein as five consecutive days or more where water 183 temperatures exceed the 90th percentile (heat wave) or fall below the 10th percentile (cold wave), 184 based on a 30 year historical base-line (33). This day-specific definition of marine heat and cold 185 waves accounts for extreme temperatures within a given season, which can disrupt seasonal 186 biological processes (like reproduction). Consequently, cold waves occurring in summer months 187 are reflected and vice versa. The duration and intensity of these events were defined by the 188 number of days within a heat/cold wave event and the number of degrees exceeding or falling 189 below the 90th/10th percentile, respectively. To avoid collinearity with time, the annual average 190 intensity of the event per duration of time throughout the corals lifetime was incorporated into 191 the model. 192 High wind events were defined as any day(s) when the average daily wind speed 193 exceeded 12.78 knots, equivalent to the average sustained wind of a tropical storm (34). Like 194 MHW and MCS, high wind events were similarly transformed to avoid collinearity with time. 195 Statistical and Model Analysis 196 Coral fragments were unable to characterize growth and were excluded from the analysis. 197 All statistical analysis and models were run in R (RStudio Team, 2020). We used a single sample 198 two-tailed Student’s t-test to determine if there was a significant difference between growth 199 (both total and healthy) and the null hypothesis of zero growth. To evaluate differences in growth 200 and percentage of healthy cover across sites and time, a one-way Analysis of Variance .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 15, 2025. ; https://doi.org/10.1101/2025.07.11.664295doi: bioRxiv preprint 11 201 (ANOVA) was performed, with the null hypothesis of no differences among sites. If differences 202 were observed a post-hoc Tukey’s HSD test (p < 0.05) was used to determine which sites 203 differed from one another. 204 Five generalized linear mixed models (GLMM), operating within a frequentist statistical 205 framework, were implemented using the glmmTMB package in R (35,36). Three models were 206 applied to the entire dataset of observed coral outplants: one for percent healthy coral cover, one 207 for total growth, and one for healthy growth. Dramatically different trends in percent healthy 208 cover were observed for corals less than and greater than 2 years old, resulting in submodels 209 being run based on these age groups (young and old coral models). Model selection took the 210 form of excluding highly correlating covariates. No covariates were excluded in the models for 211 total growth, healthy growth, and percent healthy cover due to adequate sample size, but 212 submodels with reduced sample size excluded highly correlating variables. All covariates, except 213 time, were standardized to z-score values, resulting in model coefficients indicating change in 214 growth (cm2/yr) for every change in standard deviation of the environmental and terrain 215 covariate. The percentage composition of a coral outplant with healthy tissue was doubly bound 216 between 0% and 100% using a beta-distributed GLMM using with a logit link function. Due to 217 logit link limitations, healthy cover equal to 0% and 100% were transformed to 0.001 or 0.999, 218 to fall within values of 0<y<1, due to logit link limitations. Healthy and total cover models 219 utilized a Gaussian distribution. For all models, sites were run as random effects. Model 220 diagnostics were visually inspected and affirmed from the “Performance” package and the 221 Akaike Information Criterion (AIC) (37). Model convergence was affirmed from the glmmTMB 222 hessian statistic and residuals were checked for model assumptions. 223 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 15, 2025. ; https://doi.org/10.1101/2025.07.11.664295doi: bioRxiv preprint 12 224 Results 225 After wild and fragmented corals were excluded, a total of 588 observed corals were 226 included in the niche model. M32-3 had the highest number of coral outplants (105), and site T3 227 had the fewest (27) observed coral outplants. The deepest coral outplant locations were in T16 228 (M = 8.35m) and shallowest at T3 (M = 6.44m). The steepest coral outplant location was M32-2 229 (M = 37.74°) and the least steep was T3 (M = 23.99°). The least rough coral outplant locations 230 were at site S6 (M = 0.471) and ICC1(22) had the highest roughness (M = 0.493). Qualitative 231 observations of high surrounding roughness seem indicative of surrounding rocks and soft/other 232 hard corals. The most concave coral outplant locations were S6 (M = -2.148 m-1), and the most 233 convex locations at ADAC1 (M = 3.119 m-1). Covariates exhibited minimal collinearity, except 234 for a notable correlation between distance from the reef edge and high wind events (R2 = -0.80). 235 The models of young and old corals displayed high collinearity among high intensity events, as 236 well as with distance from the reef edge and depth. 237 Percent Healthy Cover Model 238 Temporal trends indicate healthy tissue declined with time (GLMM, z = -2.63, p = 239 0.008). Sites that had outplants less than 2 years old (T16, ADAC1, ICC1(21), ICC1(22)) had 240 overall greater percentages of healthy coral cover. Among sites with coral outplants greater than 241 2 years old (M32-2, M32-3, S6, S4 and T3), the percentage of healthy tissue cover was lower 242 with the lowest mean percentage of healthy tissue cover observed at T3, the oldest outplant site 243 (Fig 3). 244 245 Fig 3. Percentage of healthy coral cover at each site. Site and associating time spent 246 outplanted increase from left to right. Boxplots show the median, and upper and lower quartiles. .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 15, 2025. ; https://doi.org/10.1101/2025.07.11.664295doi: bioRxiv preprint 13 247 Sites on the left side of the red dotted line are less than 2 years outplanted and corals on the right 248 side are more than 2 years outplanted. 249 250 Overall, percent healthy cover was variable across and within sites. Of the 588 corals, 251 261 exhibited no disease or turf algae coverage and 41 exhibited no healthy cover. M32-2 had 252 the most corals with no healthy cover. Coral outplant sites within Sand Key exhibited higher 253 percentages of healthy cover trends than other sites with older corals (Fig 3). All coral outplant 254 sites, apart from the most recent outplants at site ADAC1, had some corals that displayed some 255 sign of unhealthy tissue cover. Percent of healthy cover decreased over time (GLMM, z = -2.63, 256 p = 0.008). Sites with outplants less than 2 years old (T16, ADAC1, ICC1(21), ICC1(22)) did not 257 differ in percent healthy coral cover and had greater percentages of healthy coral cover 258 (ANOVA, F8,579= 30.89, p < 0.001). Post-two years of outplantation some corals observed were 259 completely diseased and/or covered with turf algae, with no discernable healthy tissue coverage 260 (0% healthy coral cover), and therefore assumed to be dead (Fig 3). 261 All models converged according to the hessian statistic and had AIC values of -1646.86 262 (all corals), -1118.50 (young corals), and -646.52 (old corals). Model results of young and old 263 corals show increased distance from coast (GLMM, z = 2.10, p = 0.035), decreased distance 264 from the reef edge (GLMM, z = -2.02, p = 0.043), increased depth (GLMM, z = 2.58, p = 0.009), 265 and decreased roughness (GLMM, z = -6.28, p < 0.001) correlated with higher healthy coral 266 cover (Fig 4). Submodels of young and old corals attribute the correlating relations come from 267 older corals; no environmental variables could predict percent healthy cover in the first two years 268 of outplantation. In contrast, increased depth (GLMM, z = 2.47, p = 0.013) and decreased 269 roughness (GLMM, z = -6.20, p < 0.001) correlated with healthy coral cover in corals ≥ 2 yrs old 270 (Fig 5). .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 15, 2025. ; https://doi.org/10.1101/2025.07.11.664295doi: bioRxiv preprint 14 271 272 Fig 4. Percent coral healthy tissue model covariate estimates for each covariate for all 273 corals. Lines depict the covariate estimates confidence intervals. Asterisks indicate the 274 significance level of estimates. 275 276 Fig 5. Percent coral healthy tissue model covariate estimates for each covariate for the 277 subset models of young (2yro) corals. Lines depict the covariate estimates 278 confidence intervals. Asterisks indicate the significance level of estimates. 279 280 281 Healthy and Total Growth Model 282 Outplanted corals showed total and healthy growth (t-test, tt = 22.72, pt < 0.001; t-test, th 283 = 14.56, ph < 0.001) (Fig 6). Total and healthy growth rates varied across sites, but did not 284 change over time (GLMM, zt = 0.97, pt = 0.33; GLMM, zh = -1.46, ph = 0.145) (Fig 6). Since the 285 time corals were outplanted, total growth across sites averaged 26.03 cm2/yr (CI: 24.19-28.77 286 cm2/yr) with an average healthy growth of 16.19cm2/yr (CI: 14.12-18.53 cm2/yr). 287 288 Fig 6. Healthy and total coral growth for all sites. The red dotted line at 0cm2/yr depicts where 289 the growth rate is positive (above) or negative (below). Total and healthy growth rates could be 290 negative due to fragmentation, reducing the coral size to smaller than its initial size. Healthy 291 growth could be negative due to disease. 292 293 The total growth model and healthy growth model converged according to the hessian 294 statistic and had AIC values of 4928.68 and 4955.47, respectively. Several variables correlated 295 with higher total and healthy coral growth. This included distance from shoreline (GLMM, zt = 296 2.23, pt = 0.026; GLMM, zh = 2.45, ph = 0.014), increased slope (GLMM, zt = 2.27, pt = 0.023; 297 GLMM, zh = 2.57, ph = 0.010), decreased roughness (GLMM, zt = -2.81, pt = 0.004; GLMM, zh = 298 -4.22, ph < 0.001), decreased curvature (concavity) (GLMM, zt = -3.84, pt < 0.001; GLMM, zh = .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 15, 2025. ; https://doi.org/10.1101/2025.07.11.664295doi: bioRxiv preprint 15 299 -3.29, ph = 0.001), increased high wind events (GLMM, zt = 2.03, pt = 0.042; GLMM, zh = 2.59, 300 ph = 0.009), and decreased marine cold waves (GLMM, zt = -3.73, pt < 0.001; GLMM, zh = - 301 3.16, ph = 0.0015). Increased marine heat waves correlated with lower total growth only 302 (GLMM, z = -2.23, p = 0.025) (Fig 7). 303 Fig 7. Healthy and total growth model covariate estimates for all corals. Lines depict the 304 covariate estimates confidence intervals. Asterisks indicate the significance level of estimates. 305 Discussion 306 Percent healthy cover, unlike growth, decreased after two years from outplantation. A 307 possible reason for the decline in healthy coral cover after two years of outplanting may be due 308 to the onset of spawning around 2 years of age, which likely diverts energetic resources from 309 fighting infection to reproduction and growth (17,38,39). In contrast to results for percent healthy 310 cover for all corals, the model for corals outplanted for less than two years was unable to identify 311 correlating terrain attributes that positively related to percent healthy cover, emphasizing the 312 importance of using long-term coral outplants when developing suitability models. The best 313 predictors for high healthy coral cover over two years of age were greater depth and less 314 roughness, indicating these environmental conditions are especially important when informing 315 outplant locations for A. cervicornis reefs. 316 Overall, our fine-scale models suggest that depth, distance from the reef edge and the 317 coast, and seafloor roughness correlated to percent healthy coral cover. For depth, healthy coral 318 cover increased with depth. Scleractinian corals are sensitive to UV radiation and high 319 temperatures, with greater temperatures and UV radiation occurring in the upper 10m (24,40,41). 320 All our study sites were shallower than 10m. Our results revealed depth as a factor for percent 321 cover of healthy coral, suggesting that even small differences in depth in the upper 10m of the 322 water column may influence coral outplant success. Corals that were outplanted farther from the .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 15, 2025. ; https://doi.org/10.1101/2025.07.11.664295doi: bioRxiv preprint 16 323 coastline were predicted to have greater healthy coral cover than those outplanted closer to the 324 coastline, which supports results found by others (9). This is likely the result of the greater 325 concentration of anthropogenic factors, such as greater thermal stress, sedimentation, and higher 326 concentration of pollution, all of which are known to negatively impact coral health (1). Within 327 the reefscape, corals outplanted closer to the reef edge had a greater percentage of healthy coral 328 tissue. Reef edges are likely to have greater water flow which reduces bleaching, the build-up of 329 sediment on the reef, and increases in mass transport (42,43), all factors that mitigate disease 330 (12,44–46). 331 Interestingly and in contrast to larger scale studies and our hypothesis, healthy coral 332 cover positively associated with decreased roughness (i.e., smoother areas). There are a number 333 of potential reasons for these unexpected results. First, increased surrounding roughness is 334 indicative of higher coral abundance which may facilitate contact-spread of disease (27,28,34). 335 Second, complex environmental terrain may support a higher number of corallivores due to the 336 greater number of available shelters. Corallivores create lesions in the coral tissues, which 337 provides greater opportunities for disease to colonize scleractinian tissue, including A. 338 cervicornis (27,47–50). Lastly, high terrain roughness and surrounding corals with rough tissue 339 results in higher mass transfer due to an increase in micro-turbulence (27,51,52), which may lead 340 to an increase in nutrient and gas exchange and thus a greater capacity to uptake disease-causing 341 microbes. 342 Total and healthy coral growth was hypothesized to decrease over time and both growth 343 models were assumed to perform differently. These assumptions were made due to evidence that 344 energetic resources are diverted away from growing when approaching reproductive years or 345 when fighting off disease (17,38,39), neither of which were evident in this study. Corals in the .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 15, 2025. ; https://doi.org/10.1101/2025.07.11.664295doi: bioRxiv preprint 17 346 present study did not exhibit any decrease in growth trends, but it is important to acknowledge 347 that this study did not include monitoring the same corals over time to establish growth trends. 348 Further studies should monitor annual changes in coral growth or decline since outplantation. 349 Other than MHW, covariates that significantly correlate to total growth are also significantly 350 correlated to healthy growth in the same manner and proportion, indicating optimal conditions 351 for growth did not change after infections. 352 Total and healthy growth exhibited similar correlations as percent healthy cover with 353 distance from the coast and roughness. Additionally slope, curvature, MCW, and high wind 354 events correlated to total and healthy growth. Less terrain roughness and increased distance from 355 shore positively correlated with higher total and healthy growth and likely occurs for similar 356 reasons. Steeper slopes positively correlated with healthy and total growth. Steeper slopes may 357 indicate areas of high-water flow (53,54), suggesting that while healthy coral cover benefits from 358 water flow, growth is dependent upon the strength of such water flow providing ample nutrients 359 and dissolved oxygen for mass transport. Higher curvature (convex terrain) correlated with lower 360 growth and concave terrain correlated with higher growth. Large-scale concave geomorphology 361 has been shown to negatively correlate to coral growth, likely as a result of less water flow (55). 362 When zooming in to observe coral biogenic morphology, concavity in coral structure causes 363 sediment accumulation (56) and an increase ratio in polyp to surface area, causing faster food 364 depletion (57). Intermediate of these spectrums, at fine spatial scales, convex terrain represents 365 coral outplant locations next to encroaching soft corals and rocks, limiting water flow and 366 competition for resources, thereby reducing mass transport (58). Convex terrain was also highly 367 rough and therefore likely correlated negatively with growth for similar reasons. .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 15, 2025. ; https://doi.org/10.1101/2025.07.11.664295doi: bioRxiv preprint 18 368 It is not surprising that MHW and MCW negatively correlated to total and healthy coral 369 growth. MHWs have a long-standing evidential history of adversely influencing coral success 370 (21,40,59–62), and climate projections show MHWs are the main drivers inhibiting coral growth 371 (63) and causing coral reef decline (2). While attention has focused on the effect of MHW over 372 MCW, due to the expected increase in intensity and duration of MHW and decreased prevalence 373 of MCW due to climate change (2), model results in this study revealed MCW had a greater 374 negative impact on A. cervicornis total and healthy growth than MHW. One of the most severe 375 cold-water events in the Florida Keys on record occurred in in the winter of 1976-1977 (64) and 376 in 2010, causing mass mortalities throughout the Florida Reef Tract (8,65,66). A. cervicornis and 377 other corals’ high vulnerability to cold water events raises some concerns over selective breeding 378 and outplanting of heat tolerant corals to combat bleaching (67,68). This form of selection may 379 leave reefs more vulnerable to the rare occurrence of cold wave events. Future selective breeding 380 and genetic modifications should take this into account to ensure resilience under varying 381 environmental conditions. 382 In contrast to other studies, our model predicts that high wind events may enhance coral 383 growth. These events have been associated (1–3) with increased lesions (34), reduced growth 384 (69), and increase coastal run-off (13). However, the benefits of high winds include increased 385 wave action that can lead to higher incidences of asexual propagation through fragmentation 386 (7,38) and increased water flow which may mitigate bleaching by bringing in cool deeper waters 387 (70–72). Our study sites were located at depths of less than 10 m, making it likely that high wind 388 events had a direct effect on the coral reefs. Although the impact of high wind events on corals 389 requires further study, our findings indicate they are less of a concern than high intensity thermal 390 and cold wave events. .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 15, 2025. ; https://doi.org/10.1101/2025.07.11.664295doi: bioRxiv preprint 19 391 Considerations for future outplanting 392 Model results of curvature and roughness highlight recurring challenges with coral 393 restoration; often coral outplanting results in unfavorable environmental conditions for optimal 394 coral growth and survivorship. Large reef-building coral thickets can increase coral vulnerability 395 to density induced disease spread (27,73), corallivore presence (48,74), and decreased water flow 396 penetration into the thickets (42,75). In less rough and convex terrain A. cervicornis had higher 397 growth and healthy cover trends, yet the coral itself has rough and convex morphology, resulting 398 in the coral body building unfavorable terrain. Possible courses of actions include outplanting 399 corals as soon as they are sexually reproductive so outplants may immediately start recruiting to 400 the coral reef and to increase the adaptive potential of outplants to be more resilient to disease 401 and optimize mass transport in the long-term. 402 Coral outplanting and monitoring capabilities can often stand at odds with maximizing 403 restoration success. Model results within this study and others have demonstrated that the 404 predicted suitable habitats are in deeper, cooler, waters (at least below 15m deep) that act as 405 refuge from MHW (24,64,76), and are located farther away from the coast and its associated 406 anthropogenic impacts (1). However, these locations are especially hard for restoration groups to 407 regularly access and monitor coral success. Consequently, greater effort could be made to work 408 collaboratively with restoration groups to develop habitat suitability models which identify 409 suitable sites useful to coastal managers and practitioners. 410 The fine-scale digital terrain models developed from SfM can be used to inform several 411 factors that may be useful in coral restoration areas. These include, but are not limited to, water 412 flow (which is often used as a surrogate for the relative amount of food and oxygen delivered to 413 species), presence of corallivores, disease, and areas that may be subject to upwelling and .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 15, 2025. ; https://doi.org/10.1101/2025.07.11.664295doi: bioRxiv preprint 20 414 therefore mitigate temperature (47,77,78); all factors that help to define ecosystem health and 415 distribution/abundance of biota (79). Model results of roughness and curvature within this study 416 support this, where areas of high terrain roughness and convexity are primarily attributed to 417 encroaching soft and other hard corals and rocks. The insights gained from SfM shows how 418 higher resolution DTMs are able to bridge the gap from broader environmental conditions to 419 more localized factors that influence organismal success (42), factors that are otherwise 420 unavailable from satellite and sonar derived bathymetry (28,29,80,81). 421 Due to the insulating capacity that deeper and distant waters provide corals from climate 422 change and anthropogenic influence, auxiliary efforts towards technological developments for 423 easier monitoring are necessary for long-term restoration work. Current reef restoration has 424 focused on the shallow reef crest, reef crest, spur and groove terrain, forereef terrace, and deep 425 reef (82). These sites, apart from the deep reef, range in depth from 0.7m to 10m deep and 426 constitute over 56,000m2 of restorable area (82). These areas, while accessible, are the most 427 vulnerable to climate change, necessitating outplanting deeper and farther from the coast. 428 Opportunities to survey coral reef growth and health more efficiently using a variety of remote 429 sensing techniques should be explored as it can increase monitoring capabilities. Potential 430 applications include multibeam sonar, ICESat-2 satellite monitoring, and unmanned aerial 431 surveillance. Complementing these survey methods with HSMs and deep learning coral detection 432 methods (83) can provide more accessible monitoring of corals in environments which may be 433 difficult to monitor recurringly in-situ. 434 Summary and Conclusion 435 As climate change progresses and environmental conditions continue to change, the 436 available habitat for endangered species will continue to shrink. Conservation efforts to restore .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 15, 2025. ; https://doi.org/10.1101/2025.07.11.664295doi: bioRxiv preprint 21 437 sensitive species populations requires attentive and rigorous research for what supports 438 organismal success. Within modeling research, this necessitates finer-scale modeling, at the scale 439 of meters to centimeters. While large-scale modeling is helpful towards understanding suitable 440 habitat shifts under climate change, site-specific models are necessary to inform active 441 restoration strategies. Structure-from-motion photogrammetry is unique from other survey 442 methods in capturing continuous spatial data at the scale of millimeters. While SfM 443 photogrammetry data collection is more time intensive compared to MBES, satellite, or UAS 444 imagery data collection, the resulting data resolution is imperative to informing environmental 445 influences on vulnerable species. 446 Cumulatively, our results indicate that the variance of success within the fine-scale can be 447 accounted for and incorporated into outplant decisions. A. cervicornis habitat suitability models 448 of coral growth and healthy cover inform recommendations to outplant 449 1) on steeper parts of the reefscape, along the reef edge 450 2) within marine sanctuaries, preferably in as distant locations as possible from the coast 451 3) in deeper waters, preferably deeper than 10m 452 4) where there is minimal surrounding corals and rocks 453 These outplant recommendations are developed with restoration group capabilities in mind, for 454 outplant sites they are already frequenting. The creation of these models for A. cervicornis 455 supports the applicability of fine-scale modeling of other endangered and vulnerable species, to 456 support conservation efforts. 457 Acknowledgement .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 15, 2025. ; https://doi.org/10.1101/2025.07.11.664295doi: bioRxiv preprint 22 458 Thank you to the managers and staff at Mote Marine Laboratory for their support with field work 459 and commitment to restoring Florida’s reefscape. This study was supported by NOAA grant # 460 NA20NOS4000196. 461 462 References 463 1. Golbuu Y, van Woesik R, Richmond RH, Harrison P, Fabricius KE. River discharge reduces reef 464 coral diversity in Palau. Mar Pollut Bull. 2011 Apr 1;62(4):824–31. 465 2. IPCC. Climate Change 2023: Synthesis Report. Contribution of Working Groups I, II and III to the 466 Sixth Assessment Report of the Intergovernmental Panel on Climate Change [Core Writing Team, H. 467 Lee and J. Romero (eds.)] [Internet]. First. Geneva, Switzerland: IPCC; 2023 Jul [cited 2024 Mar 4]. 468 Available from: https://www.ipcc.ch/report/ar6/syr/ 469 3. Valentine JF, Heck KL. Perspective review of the impacts of overfishing on coral reef food web 470 linkages. Coral Reefs. 2005 Jun 1;24(2):209–13. 471 4. Williams D, Miller M. Coral disease outbreak: pattern, prevalence and transmission in Acropora 472 cervicornis. Mar Ecol Prog Ser. 2005;301:119–28. 473 5. Acropora Biological Review Team. Atlantic Acropora Status Review Document [Internet]. National 474 Marine Fisheries Service, Southeast Regional Office; 2005 Mar p. 152+. Available from: 475 https://citeseerx.ist.psu.edu/document?repid=rep1&type=pdf&doi=61b96fc5c6b30c2c3b23e4b7af395 476 1bd7365a982 477 6. Calle-Triviño J, Muñiz-Castillo AI, Cortés-Useche C, Morikawa M, Sellares-Blasco R, Arias- 478 González JE. Approach to the Functional Importance of Acropora cervicornis in Outplanting Sites in 479 the Dominican Republic. Front Mar Sci [Internet]. 2021 [cited 2024 Jan 23];8. Available from: 480 https://www.frontiersin.org/articles/10.3389/fmars.2021.668325 481 7. Lirman D, Schopmeyer S, Galvan V, Drury C, Baker AC, Baums IB. Growth Dynamics of the 482 Threatened Caribbean Staghorn Coral Acropora cervicornis: Influence of Host Genotype, Symbiont 483 Identity, Colony Size, and Environmental Setting. PLOS ONE. 2014 Sep 30;9(9):e107253. 484 8. Kemp DW, Colella MA, Bartlett LA, Ruzicka RR, Porter JW, Fitt WK. Life after cold death: reef 485 coral and coral reef responses to the 2010 cold water anomaly in the Florida Keys. Ecosphere. 486 2016;7(6):e01373. 487 9. van Woesik R, Roth LM, Brown EJ, McCaffrey KR, Roth JR. Niche space of corals along the Florida 488 reef tract. PLoS ONE. 2020 Apr 7;15(4):e0231104. 489 10. Precht WF, Gintert BE, Robbart ML, Fura R, van Woesik R. Unprecedented Disease-Related Coral 490 Mortality in Southeastern Florida. Sci Rep. 2016 Aug 10;6(1):31374. 491 11. Aronson RB, Precht WF. White-band disease and the changing face of Caribbean coral reefs. 492 Hydrobiologia. 2001 Sep 1;460(1):25–38. .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 15, 2025. ; https://doi.org/10.1101/2025.07.11.664295doi: bioRxiv preprint 23 493 12. van Woesik R, Banister RB, Bartels E, Gilliam DS, Goergen EA, Lustic C, et al. Differential survival 494 of nursery-reared Acropora cervicornis outplants along the Florida reef tract. Restor Ecol. 495 2021;29(1):e13302. 496 13. Riegl B, Purkis SJ, Keck J, Rowlands GP. Monitored and modeled coral population dynamics and the 497 refuge concept. Mar Pollut Bull. 2009 Jan 1;58(1):24–38. 498 14. Couce E, Ridgwell A, Hendy EJ. Future habitat suitability for coral reef ecosystems under global 499 warming and ocean acidification. Glob Change Biol. 2013;19(12):3592–606. 500 15. Ware M, Garfield EN, Nedimyer K, Levy J, Kaufman L, Precht W, et al. Survivorship and growth in 501 staghorn coral (Acropora cervicornis) outplanting projects in the Florida Keys National Marine 502 Sanctuary. PLOS ONE. 2020 May 6;15(5):e0231817. 503 16. Lirman D, Thyberg T, Herlan J, Hill C, Young-Lahiff C, Schopmeyer S, et al. Propagation of the 504 threatened staghorn coral Acropora cervicornis: methods to minimize the impacts of fragment 505 collection and maximize production. Coral Reefs. 2010 Sep 1;29(3):729–35. 506 17. Omori M. Coral restoration research and technical developments: what we have learned so far. Mar 507 Biol Res. 2019 Oct 14;15:1–33. 508 18. Boström-Einarsson L, Babcock RC, Bayraktarov E, Ceccarelli D, Cook N, Ferse SCA, et al. Coral 509 restoration – A systematic review of current methods, successes, failures and future directions. PLOS 510 ONE. 2020 Jan 30;15(1):e0226631. 511 19. Fox HE, Harris JL, Darling ES, Ahmadia GN, Estradivari, Razak TB. Rebuilding coral reefs: success 512 (and failure) 16 years after low-cost, low-tech restoration. Restor Ecol. 2019;27(4):862–9. 513 20. Devlin-Durante MK, Miller MW, Group CAR, Precht WF, Baums IB. How old are you? Genet age 514 estimates in a clonal animal. Mol Ecol. 2016;25(22):5628–46. 515 21. Hughes TP. Life histories and population dynamics of early successional corals. In: Gabrie C, Salvat 516 B, editors. Moorea, French Polynesia: Antenne Museum-EPHE; 1985 [cited 2023 Aug 4]. p. 101–6. 517 Available from: http://www.reefbase.org/resource_center/publication/icrs.aspx 518 22. Irwin A, Greer L, Humston R, Devlin-Durante M, Cabe P, Lescinsky H, et al. Age and intraspecific 519 diversity of resilient Acropora communities in Belize. Coral Reefs. 2017 Dec 1;36(4):1111–20. 520 23. Banister RB, van Woesik R. Ranking 67 Florida Reefs for Survival of Acropora cervicornis 521 Outplants. Front Mar Sci [Internet]. 2021 [cited 2023 Jul 19];8. Available from: 522 https://www.frontiersin.org/articles/10.3389/fmars.2021.672574 523 24. Giraldo-Ospina A, Kendrick GA, Hovey RK. Depth moderates loss of marine foundation species 524 after an extreme marine heatwave: could deep temperate reefs act as a refuge? Proc R Soc B Biol Sci. 525 2020 Jun 10;287(1928):20200709. 526 25. Fabricius KE. Factors Determining the Resilience of Coral Reefs to Eutrophication: A Review and 527 Conceptual Model. In: Dubinsky Z, Stambler N, editors. Coral Reefs: An Ecosystem in Transition 528 [Internet]. Dordrecht: Springer Netherlands; 2011 [cited 2023 Aug 1]. p. 493–505. Available from: 529 https://doi.org/10.1007/978-94-007-0114-4_28 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 15, 2025. ; https://doi.org/10.1101/2025.07.11.664295doi: bioRxiv preprint 24 530 26. Manzello DP. Coral growth with thermal stress and ocean acidification: lessons from the eastern 531 tropical Pacific. Coral Reefs. 2010 Sep 1;29(3):749–58. 532 27. Shore A, Caldwell JM. Modes of coral disease transmission: how do diseases spread between 533 individuals and among populations? Mar Biol. 2019 Feb 26;166(4):45. 534 28. Burns JHR, Delparte D, Gates RD, Takabayashi M. UTILIZING UNDERWATER THREE- 535 DIMENSIONAL MODELING TO ENHANCE ECOLOGICAL AND BIOLOGICAL STUDIES OF 536 CORAL REEFS. Int Arch Photogramm Remote Sens Spat Inf Sci. 2015 Apr 9;XL-5/W5:61–6. 537 29. Fukunaga A, Burns JHR, Craig BK, Kosaki RK. Integrating Three-Dimensional Benthic Habitat 538 Characterization Techniques into Ecological Monitoring of Coral Reefs. J Mar Sci Eng. 2019 539 Feb;7(2):27. 540 30. McDowall P, Lynch HJ. Ultra-Fine Scale Spatially-Integrated Mapping of Habitat and Occupancy 541 Using Structure-From-Motion. PLOS ONE. 2017 Jan 11;12(1):e0166773. 542 31. Young GC, Dey S, Rogers AD, Exton D. Cost and time-effective method for multi-scale measures of 543 rugosity, fractal dimension, and vector dispersion from coral reef 3D models. PLOS ONE. 2017 Apr 544 13;12(4):e0175341. 545 32. Pierce J, Butler MJ, Rzhanov Y, Lowell K, Dijkstra JA. Classifying 3-D Models of Coral Reefs 546 Using Structure-From-Motion and Multi-View Semantic Segmentation. Front Mar Sci [Internet]. 547 2021 Oct 29 [cited 2025 Apr 28];8. Available from: 548 https://www.frontiersin.orghttps://www.frontiersin.org/journals/marine- 549 science/articles/10.3389/fmars.2021.706674/full 550 33. Hobday AJ, Alexander LV, Perkins SE, Smale D, Straub S, Oliver ECJ, et al. A hierarchical 551 approach to defining marine heatwaves. Prog Oceanogr. 2016 Feb;141:227–38. 552 34. Goergen EA, Moulding AL, Walker BK, Gilliam DS. Identifying Causes of Temporal Changes in 553 Acropora cervicornis Populations and the Potential for Recovery. Front Mar Sci [Internet]. 2019 554 [cited 2022 Jun 20];6. Available from: https://www.frontiersin.org/article/10.3389/fmars.2019.00036 555 35. Brooks ME, Kristensen K, Benthem K J ,van, Magnusson A, Berg C W, Nielsen A, et al. glmmTMB 556 Balances Speed and Flexibility Among Packages for Zero-inflated Generalized Linear Mixed 557 Modeling. R J. 2017;9(2):378. 558 36. Brooks ME, Kristensen K, Benthem KJ van, Magnusson A, Berg CW, Nielsen A, et al. Modeling 559 zero-inflated count data with glmmTMB [Internet]. bioRxiv; 2017 [cited 2023 Apr 4]. p. 132753. 560 Available from: https://www.biorxiv.org/content/10.1101/132753v1 561 37. Lüdecke D, Ben-Shachar M, Patil I, Waggoner P, Makowski D. performance: An R Package for 562 Assessment, Comparison and Testing of Statistical Models. J Open Source Softw. 2021 Apr 563 21;6(60):3139. 564 38. Lirman D. Fragmentation in the branching coral Acropora palmata (Lamarck): growth, survivorship, 565 and reproduction of colonies and fragments. J Exp Mar Biol Ecol. 2000 Aug 23;251(1):41–57. 566 39. REEFocus. CRF. 2020 [cited 2023 Apr 5]. NEW RESEARCH TRACKS TIMING OF CORAL 567 SPAWNING ON FLORIDA’S CORAL REEF. Available from: .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 15, 2025. ; https://doi.org/10.1101/2025.07.11.664295doi: bioRxiv preprint 25 568 https://www.coralrestoration.org/post/new-research-tracks-timing-of-coral-spawning-on-floridas- 569 coral-reef 570 40. Asner GP, Vaughn NR, Martin RE, Foo SA, Heckler J, Neilson BJ, et al. Mapped coral mortality and 571 refugia in an archipelago-scale marine heat wave. Proc Natl Acad Sci. 2022 May 572 10;119(19):e2123331119. 573 41. Buma AGJ, De Boer MK, Boelen P. Depth Distributions of Dna Damage in Antarctic Marine Phyto- 574 and Bacterioplankton Exposed to Summertime Uv Radiation. J Phycol. 2001;37(2):200–8. 575 42. Stocking JB, Rippe JP, Reidenbach MA. Structure and dynamics of turbulent boundary layer flow 576 over healthy and algae-covered corals. Coral Reefs. 2016 Sep 1;35(3):1047–59. 577 43. Stocking JB, Laforsch C, Sigl R, Reidenbach MA. The role of turbulent hydrodynamics and surface 578 morphology on heat and mass transfer in corals. J R Soc Interface. 2018 Dec 19;15(149):20180448. 579 44. Brodnicke OB, Bourne DG, Heron SF, Pears RJ, Stella JS, Smith HA, et al. Unravelling the links 580 between heat stress, bleaching and disease: fate of tabular corals following a combined disease and 581 bleaching event. Coral Reefs. 2019 Aug 1;38(4):591–603. 582 45. Nelson HR, Altieri AH. Oxygen: the universal currency on coral reefs. Coral Reefs. 2019 Apr 583 1;38(2):177–98. 584 46. Pollock FJ, Lamb JB, Field SN, Heron SF, Schaffelke B, Shedrawi G, et al. Sediment and Turbidity 585 Associated with Offshore Dredging Increase Coral Disease Prevalence on Nearby Reefs. PLOS ONE. 586 2014 Jul 16;9(7):e102498. 587 47. Borland HP, Gilby BL, Henderson CJ, Leon JX, Schlacher TA, Connolly RM, et al. The influence of 588 seafloor terrain on fish and fisheries: A global synthesis. Fish Fish. 2021;22(4):707–34. 589 48. Montalbetti E, Fallati L, Casartelli M, Maggioni D, Montano S, Galli P, et al. Reef complexity 590 influences distribution and habitat choice of the corallivorous seastar Culcita schmideliana in the 591 Maldives. Coral Reefs. 2022 Apr 1;41(2):253–64. 592 49. Nicolet KJ, Chong-Seng KM, Pratchett MS, Willis BL, Hoogenboom MO. Predation scars may 593 influence host susceptibility to pathogens: evaluating the role of corallivores as vectors of coral 594 disease. Sci Rep. 2018 Mar 27;8(1):5258. 595 50. Williams GJ, Aeby GS, Cowie ROM, Davy SK. Predictive Modeling of Coral Disease Distribution 596 within a Reef System. PLOS ONE. 2010 Feb 17;5(2):e9264. 597 51. Hearn C, Atkinson M, Falter J. A physical derivation of nutrient-uptake rates in coral reefs: effects of 598 roughness and waves. Coral Reefs. 2001 Dec 1;20(4):347–56. 599 52. Morrow KM, Pankey MS, Lesser MP. Community structure of coral microbiomes is dependent on 600 host morphology. Microbiome. 2022 Jul 28;10(1):113. 601 53. Eugene Shinn. Spur and Groove Formation on the Florida Reef Tract. SEPM J Sediment Res 602 [Internet]. 1963 [cited 2023 Apr 5];Vol. 33. Available from: 603 https://pubs.geoscienceworld.org/jsedres/article/33/2/291-303/112994 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 15, 2025. ; https://doi.org/10.1101/2025.07.11.664295doi: bioRxiv preprint 26 604 54. Hernández-Molina FJ, Serra N, Stow DAV, Llave E, Ercilla G, Van Rooij D. Along-slope 605 oceanographic processes and sedimentary products around the Iberian margin. Geo-Mar Lett. 2011 606 Dec 1;31(5):315–41. 607 55. Sebens KP, Helmuth B, Carrington E, Agius B. Effects of water flow on growth and energetics of the 608 scleractinian coral Agaricia tenuifolia in Belize. Coral Reefs. 2003 Apr 1;22(1):35–47. 609 56. Philipp E, Fabricius K. Photophysiological stress in scleractinian corals in response to short-term 610 sedimentation. J Exp Mar Biol Ecol. 2003 Feb 26;287(1):57–78. 611 57. Merks RMH, Hoekstra AG, Kaandorp JA, Sloot PMA. Polyp oriented modelling of coral growth. J 612 Theor Biol. 2004 Jun 21;228(4):559–76. 613 58. Chadwick NE, Morrow KM. Competition Among Sessile Organisms on Coral Reefs. In: Dubinsky Z, 614 Stambler N, editors. Coral Reefs: An Ecosystem in Transition [Internet]. Dordrecht: Springer 615 Netherlands; 2011 [cited 2024 Mar 6]. p. 347–71. Available from: https://doi.org/10.1007/978-94- 616 007-0114-4_20 617 59. Ainsworth TD, Hurd CL, Gates RD, Boyd PW. How do we overcome abrupt degradation of marine 618 ecosystems and meet the challenge of heat waves and climate extremes? Glob Change Biol. 619 2020;26(2):343–54. 620 60. Donovan MK, Burkepile DE, Kratochwill C, Shlesinger T, Sully S, Oliver TA, et al. Local conditions 621 magnify coral loss after marine heatwaves. Science. 2021 May 28;372(6545):977–80. 622 61. Foo SA, Asner GP. Sea surface temperature in coral reef restoration outcomes. Environ Res Lett. 623 2020 Jul 1;15(7):074045. 624 62. Oliver ECJ, Benthuysen JA, Darmaraki S, Donat MG, Hobday AJ, Holbrook NJ, et al. Marine 625 Heatwaves. Annu Rev Mar Sci. 2021;13(1):313–42. 626 63. Cantin NE, Cohen AL, Karnauskas KB, Tarrant AM, McCorkle DC. Ocean warming slows coral 627 growth in the central Red Sea. Science. 2010 Jul 16;329(5989):322–5. 628 64. Porter JW, Battey JF, Smith GJ. Perturbation and change in coral reef communities. Popul Biol. 1981 629 Nov 2;79:1678–81. 630 65. Colella MA, Ruzicka RR, Kidney JA, Morrison JM, Brinkhuis VB. Cold-water event of January 631 2010 results in catastrophic benthic mortality on patch reefs in the Florida Keys. Coral Reefs. 2012 632 Jun 1;31(2):621–32. 633 66. Lirman D, Schopmeyer S, Manzello D, Gramer LJ, Precht WF, Muller-Karger F, et al. Severe 2010 634 Cold-Water Event Caused Unprecedented Mortality to Corals of the Florida Reef Tract and Reversed 635 Previous Survivorship Patterns. PLOS ONE. 2011 Aug 10;6(8):e23047. 636 67. Buerger P, Alvarez-Roa C, Coppin CW, Pearce SL, Chakravarti LJ, Oakeshott JG, et al. Heat- 637 evolved microalgal symbionts increase coral bleaching tolerance. Sci Adv. 2020 May 638 13;6(20):eaba2498. 639 68. Howells EJ, Abrego D, Liew YJ, Burt JA, Meyer E, Aranda M. Enhancing the heat tolerance of reef- 640 building corals to future warming. Sci Adv. 2021 Aug 20;7(34):eabg6070. .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 15, 2025. ; https://doi.org/10.1101/2025.07.11.664295doi: bioRxiv preprint 27 641 69. Muko S, Arakaki S, Nagao M, Sakai K. Growth form-dependent response to physical disturbance and 642 thermal stress in Acropora corals. Coral Reefs. 2013 Mar 1;32(1):269–80. 643 70. Carrigan AD, Puotinen M. Tropical cyclone cooling combats region-wide coral bleaching. Glob 644 Change Biol. 2014;20(5):1604–13. 645 71. Manzello DP, Brandt M, Smith TB, Lirman D, Hendee JC, Nemeth RS. Hurricanes benefit bleached 646 corals. Proc Natl Acad Sci. 2007 Jul 17;104(29):12035–9. 647 72. Page CE, Leggat W, Heron SF, Fordyce AJ, Ainsworth TD. High flow conditions mediate damaging 648 impacts of sub-lethal thermal stress on corals’ endosymbiotic algae. Conserv Physiol. 2021 Jan 649 1;9(1):coab046. 650 73. Anderson RM, May RM. Population biology of infectious diseases: Part I. Nature. 1979 651 Aug;280(5721):361–7. 652 74. Koval G, Rivas N, D’Alessandro M, Hesley D, Santos R, Lirman D. Fish predation hinders the 653 success of coral restoration efforts using fragmented massive corals. PeerJ. 2020 Oct 2;8:e9978. 654 75. Reidenbach MA, Koseff JR, Monismith SG. Laboratory experiments of fine-scale mixing and mass 655 transport within a coral canopy. Phys Fluids. 2007 Jul;19(7):075107. 656 76. Harrison HB, Álvarez-Noriega M, Baird AH, Heron SF, MacDonald C, Hughes TP. Back-to-back 657 coral bleaching events on isolated atolls in the Coral Sea. Coral Reefs. 2019 Aug 1;38(4):713–9. 658 77. Barott KL, Rohwer FL. Unseen players shape benthic competition on coral reefs. Trends Microbiol. 659 2012 Dec 1;20(12):621–8. 660 78. Chong-Seng KM, Cole AJ, Pratchett MS, Willis BL. Selective feeding by coral reef fishes on coral 661 lesions associated with brown band and black band disease. Coral Reefs. 2011 Jun 1;30(2):473–81. 662 79. Pygas DR, Ferrari R, Figueira WF. Review and meta-analysis of the importance of remotely sensed 663 habitat structural complexity in marine ecology. Estuar Coast Shelf Sci. 2020 Apr 5;235:106468. 664 80. Burns JHR, Fukunaga A, Pascoe KH, Runyan A, Craig BK, Talbot J, et al. 3D HABITAT 665 COMPLEXITY OF CORAL REEFS IN THE NORTHWESTERN HAWAIIAN ISLANDS IS 666 DRIVEN BY CORAL ASSEMBLAGE STRUCTURE. Int Arch Photogramm Remote Sens Spat Inf 667 Sci. 2019 Apr 17;XLII-2/W10:61–7. 668 81. Combs IR. Characterizing the Impacts of Stony Coral Tissue Loss Disease on Coral in Southeast 669 Florida Using 3D Photogrammetry [Internet] [M.S.]. ProQuest Dissertations and Theses. [United 670 States -- Florida]: Florida Atlantic University; 2019 [cited 2023 Mar 28]. Available from: 671 https://www.proquest.com/docview/2333948527/abstract/5040EC761EAC4D06PQ/1 672 82. Restoring Seven Iconic Reefs: A Mission to Recover the Coral Reefs of the Florida Keys. 673 83. Trudeau GA, Lowell K, Dijkstra JA. Coral reef detection using ICESat-2 and machine learning. Ecol 674 Inform. 2025 Jul 1;87:103099. .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 15, 2025. ; https://doi.org/10.1101/2025.07.11.664295doi: bioRxiv preprint 28 675 84. Kopp S. New Surface Analysis Capabilities in ArcGIS Pro 2.7 [Internet]. ArcGIS Blog. 2/21 [cited 676 2023 Apr 10]. Available from: https://www.esri.com/arcgis-blog/products/arcgis-pro/analytics/new- 677 slope-aspect-curvature/ 678 85. Evans IS. General geomorphometry, derivatives of altitude, and descriptive statistics. In: Spatial 679 Analysis in Geomorphology. Routledge; 1972. 680 86. Day M, Chenoweth S. 6.14 Surface Roughness of Karst Landscapes. In: Shroder JF, editor. Treatise 681 on Geomorphology [Internet]. San Diego: Academic Press; 2013 [cited 2023 Jul 13]. p. 157–63. 682 Available from: https://www.sciencedirect.com/science/article/pii/B9780123747396001081 683 87. Maxwell AE, Shobe CM. Land-surface parameters for spatial predictive mapping and modeling. 684 Earth-Sci Rev. 2022 Mar 1;226:103944. 685 88. Restoring Seven Iconic Reefs: A Mission to Recover the Coral Reefs of the Florida Keys. Mission 686 Iconic Reefs; 687 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 15, 2025. ; https://doi.org/10.1101/2025.07.11.664295doi: bioRxiv preprint .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 15, 2025. ; https://doi.org/10.1101/2025.07.11.664295doi: bioRxiv preprint .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 15, 2025. ; https://doi.org/10.1101/2025.07.11.664295doi: bioRxiv preprint .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 15, 2025. ; https://doi.org/10.1101/2025.07.11.664295doi: bioRxiv preprint .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 15, 2025. ; https://doi.org/10.1101/2025.07.11.664295doi: bioRxiv preprint .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 15, 2025. ; https://doi.org/10.1101/2025.07.11.664295doi: bioRxiv preprint .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 15, 2025. ; https://doi.org/10.1101/2025.07.11.664295doi: bioRxiv preprint .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted July 15, 2025. ; https://doi.org/10.1101/2025.07.11.664295doi: bioRxiv preprint