Drone Imaging can Accurately Assess Coral Cover, Bleaching, and Growth Form for Shallow Coral Reefs

Drone Imaging can Accurately Assess Coral Cover, Bleaching, and Growth Form for Shallow Coral Reefs | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Drone Imaging can Accurately Assess Coral Cover, Bleaching, and Growth Form for Shallow Coral Reefs Lucian Himes, Theresa Rueger This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8565286/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 09 Apr, 2026 Read the published version in Coral Reefs → Version 1 posted 7 You are reading this latest preprint version Abstract Assessing the impacts of rapid environmental change on coral reefs is hindered by a major discrepancy between the regions with the greatest need and those that receive the most research funding. Remote sensing, particularly through the use of drones (or unmanned aerial vehicles (UAVs)), has the potential to significantly enhance the accessibility and efficiency of high-quality data collection in remote, biodiverse areas such as the Coral Triangle. To test this, we compared a UAV-based method, a common citizen science method (Reef Check), and a research-grade method (photo quadrats) to assess hard coral cover, coral bleaching, and coral growth forms across reef types (nearshore, fringing, midshelf) and depths (crest, 5 m, 10 m, 15 m) in Kimbe Bay, Papua New Guinea, during the fourth Global Mass Coral Bleaching Event. We found that, compared to in-water methods, the UAV delivered accurate reef health data for shallow reef crests of all reef types, including hard coral cover, bleached hard coral cover, and coral growth forms. Hard coral cover did not differ significantly between the crest and the other depths. However, bleaching was most significant at the reef crest and decreased significantly at deeper parts of the reef slope. Coral growth form composition varied significantly between reef types, but the scale of these differences decreased with depth. Our study demonstrates that UAVs can provide accurate health and community composition data for reefs with high biodiversity, significantly enhancing the availability of high-quality reef health data in the areas of highest need. unmanned aerial vehicle remote sensing coral triangle global mass bleaching depth gradient Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 1. Introduction For positive conservation outcomes, managers and policy makers need high-resolution, location-specific information on reef health and stressors with high spatial and temporal resolution (Montambault et al. 2015 ; Obura et al. 2019 ; Carriger et al. 2021 ). Unfortunately, there is a large discrepancy between the areas of highest coral reef biodiversity and conservation need, e.g, the Coral Triangle, and where research and management practices are most concentrated, e.g, Eastern Australia and the US Pacific, making high-resolution data less available in areas with the highest need (Fisher et al. 2011 ; Di Marco et al. 2017 ). Generalizable data is useful for predicting and implementing positive policy around the globe; however, projects that integrate localized data into their decision-making framework rather than relying on regional-scale measurements have been more successful (Gouezo et al. 2021 ; Ullah et al. 2023 ; Matthews et al. 2024 ). In a time of rapid environmental change, well-reasoned, timely, and location-specific management decisions are needed (Mumby and Harborne 2010 ; Danovaro et al. 2025 ). To increase the attainability of high-quality data, there is a need for low-cost monitoring methodologies. It is also important that these methods are scalable, as large amounts of relevant data are needed for informed management policies (Hughes et al. 2017 ; Lindenmayer et al. 2022 ). For coral reefs specifically, much of the in-situ data on disturbance events, such as marine heatwaves that lead to bleaching, comes from permanent transects or individual colonies (Depczynski et al. 2013 ; Banha et al. 2020 ; Byrne et al. 2025 ). While this is useful for particular research questions, exposure to stressors can vary in both time and spatial scales on a reef (Falter et al. 2014 ; Thomas et al. 2022 ). One way to achieve large-scale datasets at lower costs is through remote sensing, as it allows for the coverage of large areas in a highly efficient and cost-effective manner compared to ground sampling (Hedley et al. 2016 ; Turner et al.(Turner et al. 2003 ; Hedley et al. 2016 ). Potentially the most accessible form of remote sensing is through the use of drones, technically known as unmanned aerial vehicles (UAVs) (Tang and Shao 2015 ; Parsons et al. 2018 ). UAVs have several advantages compared to more intensive methods, such as satellites and manned aircraft. Satellites can provide data for large areas, but revisit times are fixed, and data quality is highly dependent on weather conditions, especially clouds (Manfreda et al. 2018 ; Kooistra et al. 2024 ). Manned aircraft have much larger startup and, in many cases, travel times to sampling areas than UAVs, as they are dependent on runway availability and locations. They also cost more to operate, and are more influenced by the environmental conditions, such as cloud cover, rain, and wind (Salamí et al. 2014 ). In contrast, UAVs can be easily and quickly deployed at relevant sites, and data acquisition can be done in minutes, depending on the needs of the project (Devoto et al. 2020 ). This flexibility is crucial in highly variable conditions. Combined with much lower operation costs and the variability in spatial resolution that can be achieved, UAVs can fulfill the need for high-quality, localized data from remote coral reefs in areas where research is often under-resourced. The overall aim of this study was to assess UAVs' potential and limitations as a tool for low-cost coral reef monitoring in remote areas. To achieve this, we compiled benthic survey data during the 4th global coral mass bleaching event in 2024 from nearshore, fringing, and midshelf reefs in Kimbe Bay, Papua New Guinea. We compared a survey method utilizing a low-cost, entry-level UAV (DJI Mini 2) to commonly used in-water methods: point-intercept transects and photo transects. Our specific objectives were to 1) compare the accuracy of data from UAV images to in-water methods for shallow depths of fringing, inshore, and midshelf reefs, and 2) determine whether the shallow depths surveyable with the UAV are representative of deeper depths across reef types. We assessed accuracy in three areas: total coral cover, coral health, and coral growth form. 2. Methods Sites : Field work was conducted in Kimbe Bay, on the northern side of West New Britain, Papua New Guinea (-5.166838, 150.499957). To test the effectiveness of UAVs in a range of conditions, we chose nine reefs in different areas that had varying amounts of coral cover and composition (Jones et al. 2004 ; Turak et al. 2021 ; Galbraith et al. 2022 ; Webber et al. 2022 ). One of the reefs, Gava Gava, is on the inside row of a set of near-shore reefs, and five reefs, Garbuna, Matane Walindi, Luba Luba, Lady Di, and Limuka, are on the outside row of the near-shore reefs. One reef, Restorf, is a fringing reef on a small island. The last two reefs, Vanessa’s and Christine’s, are midshelf reefs. Site locations are shown in Fig. 1 . Sites were chosen on each reef to have a consistent reef slope for at least 100 meters to a depth of 15 meters to avoid conducting transects on reef walls. In-water Sampling : Using SCUBA, we sampled the nine reefs at four different depths: the crest (approximately 1.5 m deep), 5 m, 10 m, and 15 m. At each depth, we deployed four 20-meter transects, each spaced from the other by at least 5 m. In-water sampling followed two methodologies along the same transect line. First, we took photos of each meter of the transect. The second in-water method, conducted on the same transect, followed the point-intercept transect (PIT) methods outlined by the Reef Check Foundation for monitoring coral reefs (2004). Substrate was classified among the eleven Reef Check categories, with a category for bleached coral added: hard coral, soft coral, recently killed coral, nutrient indicator algae, sponge, rock, rubble, sand, silt, other, bleached coral. The PIT method sampled the substrate directly beneath the transect every half meter for a total of 40 points per transect. With four transects conducted for each depth at each site, we collected 160 points per depth, in line with the methods that Reef Check recommends. Aerial Sampling : The UAV selected for this testing is the DJI Mini 2. This UAV was chosen due to its low cost, ease of use, and compactness. The UAV was flown over the crest of the reef in the section where the in-water transects were completed at a height of 10 m. Two photos were taken for each transect, for a total of eight photos per reef. The camera was pointed straight down when possible, or at the most vertical angle possible to avoid any sun glint on the water that obscured the reef. All photos included the entire visible reef crest, with the sides ending in the reef flat and in water too deep to identify substrate. At a height of 10 m, the UAV produced images of 10 m by 14.2 m of reef at a resolution of approximately 0.12 cm 2 per pixel. Table 1 Details of data gathered for each sampling method. Method Points Per Transect Categories Measured Depths sampled Hard Coral Cover Bleached Coral Cover Pale Coral Cover Coral Growth Form UAV 200 Yes Yes Yes Yes Crest Photo Transects 200 Yes Yes Yes Yes Crest, 5m, 10m, 15m Reef Check 40 Yes Yes No No Crest, 5m, 10m, 15m Photo analysis : Photos from the photo transects and the UAV were analyzed using Coral Point Count with Excel extensions (CPCe) (Kohler and Gill 2006 ). For the photo transects, ten photos were chosen. We started with the second photo of each transect and used every other photo from then on to avoid overlap between photos and eliminate the chance of counting points multiple times. The entire photo was included in the analysis, and random points could be generated anywhere on the image. Twenty points were randomly generated using the CPCe software on each photo and classified into 13 categories. The categories are “hard coral”, “pale hard coral”, “bleached hard coral”, “recently killed coral”, “algae”, “sponge”, “sand”, “silt/clay”, “rock”, “rubble”, “other”, and “tape, wand, or shadow”. Each of the coral categories, “hard coral”, “pale hard coral”, “bleached hard coral”, “recently killed coral”, was further differentiated into seven subcategories: “foliose coral”, “encrusting coral”, “branching coral”, “pillar corals”, “table corals”, “massive corals”, and “mushroom corals”. Algae were also differentiated: “nutrient indicator algae” and “turf”. Points classified as “tape, wand, or shadow” were not counted towards the percent cover. Across the ten photos, we classified a total of 200 points per transect for the photo transects. This resulted in a total of 800 points classified for each depth of each reef. The photos from the UAV were cropped to include what could be considered the reef crest. The reef flat and depths where substrate could not be identified were not included. Sections of reef that were in more than one photo were only included in the first photo. We generated 100 random points per photo, so each set of two photos from the UAV matched the sets of ten photos for each in-water transect with 200 total points. These points were assigned to the same categories as for the photo transects, and between the two UAV photos per transect and the four transects per reef, we obtained 200 points per transect and 800 points per reef. Statistical analysis : We performed all statistical analyses in R version 4.5.0 (R Core Team 2025 ). We used the glmmTMB package to fit beta regression models, the emmeans package to check for global fixed effects and perform pairwise analysis, and the performance package to check model assumptions for all models detailed below (Brooks et al. 2017 ; Ludecke et al. 2021 ; Lenth 2025 ). We used the vegan package to perform a PERMANOVA test (Oksanen et al. 2025 ). To determine whether there was variation between sites, we compared a null beta regression model with a beta regression model that included site as the only predictor variable. Both models used hard coral cover as the response variable. We found a significant difference between the models (Log-likelihood test: df = 10, Chi-sq = 53.2, p < 0.001), leading us to include site as a random factor in all other models. To compare hard coral cover from UAV images to in-water methods for shallow depths on fringing, nearshore, and midshelf reefs, we fit a beta regression model with hard coral cover as a proportion of benthic cover as the response variable, method (categorical; ‘UAV’, ‘Reef Check’, ‘Photo Transect’), reef type (categorical; ‘fringing’, ‘nearshore’, ‘midshelf’), and their interaction as predictors, and site as a random variable to account for non-independence. This model used data from UAV, photo transect, and reef check methods from the crest. To evaluate global effects among model parameters, we used the emmeans package. To evaluate the accuracy of UAV imagery to identify the scale of bleaching, we fit a beta regression model with bleached coral cover as a proportion of benthic cover as the response variable, method and reef type as the predictor variables, and site as the random variable to account for non-independence. This model used data from the UAV, photo transect, and reef check methods, with all samples from the reef crest. A second beta regression was fit with pale coral cover as the response variable, and only using UAV and photo transect images from the crest. To evaluate global fixed effects among both model parameters, we used the emmeans package. To evaluate the effectiveness of UAV imagery to determine growth forms of corals on reef crests, we fit a beta regression model with percent of total coral cover as the response variable, growth form, method, reef type, and all interactions as the predictor variables, and site as a random variable. We used UAV and photo transect data from the crest for this model. We used the emmeans package to evaluate global fixed effects and conduct pairwise comparisons among model parameters. To evaluate changes of coral cover with depth, we fit a beta regression model with hard coral cover as the response variable, depth (categorical; ‘crest’, ‘5m’, ‘10m’, ‘15m’), reef type, and their interaction as the predictor variable, and site as a random factor with random slopes for depth within site to evaluate whether depth trends were consistent across sites. We used photo transect data from all depths for this model. We used the emmeans package to evaluate fixed global effects. To evaluate changes with depth for coral health categories (‘bleached’, ‘pale’, ‘healthy’), we fit a beta regression model with percent of total coral as the response variable, health category, depth, reef-type, and all interactions as the predictor variable, and site as the response variable. Data from the photo transect methods and all depths were used for this model. We used the emmeans package to evaluate fixed global effects and perform pairwise analyses on model parameters. To evaluate changes in growth forms with depth, we fit a beta regression model with percent of total coral cover as the response variable, growth form, depth, reef type, and all interactions as the predictor variables, and site as a random factor. We used data from the photo transects and all depths for this model. Emmeans was used to evaluate fixed global effects and perform pairwise analysis. To evaluate the degree of difference between coral growth form frequencies with depth between reef types, we fit four PERMANOVA models, one for each depth, with growth form frequency as the response variable and reef type as the predictor variable. Significance was determined with Bray-Curtis dissimilarities and 999 permutations. 3. Results 3.1 Comparing UAV and in-water methods All three methods, reef check, photo transects, and UAV, reported similar total hard coral cover at the crest (global fixed effects test: F 2,Inf = 0.317, Chi-squared = 0.634, p = 0.728). Additionally, there was no effect of location, or interaction between method and reef-type (global fixed effects test: reef-type: F 2,Inf = 1.711, Chi-squared = 3.422, p = 0.181, method*reef-type: F 4,Inf = 1.152, Chi-squared = 4.608, p = 0.330). Fixed and random factors together explained 41% of the variance in the data (R 2 marginal = 0.185, R 2 conditional = 0.410). All three methods were consistent in estimating the amount of bleached coral cover at the crest (global fixed effects test: F 2,Inf = 0.195, Chi-squared = 0.390, p = 0.823). There was also no significant difference between locations for bleached coral on the crest (global fixed effects test: F 2,Inf = 1.433, Chi-squared = 2.866, p = 0.239). The model explained 47 percent of the variation in the data (R 2 marginal = 0.129, R 2 conditional = 0.470). There was a significant difference between the pale hard coral cover reported by the UAV and photo transect methods on the crest. Pale coral cover was overrepresented in UAV data (beta GLMM pairwise comparison: UAV vs photo transects, estimate = 0.764, standard error (s.e.) = 0.222, z = 3.446, p < 0.001). Reef check methods were not included in this model because the only health categories for hard coral in the reef check methods were healthy and bleached hard coral. The model explained 61 percent of the variation in the data (R 2 marginal = 0.479, R 2 conditional = 0.610). The percent cover of different coral growth forms had large variance on the crest, but both UAV and photo transect methods were able to capture the variations accurately (joint Wald test: growth form: F 6,Inf = 61.307, Chi-squared = 367.842, p < 0.001, growth form*method: F 6,Inf = 2.004, Chi-squared = 4.008, p = 0.135). Growth form varied between reef types, but both methods were able to report accurate information regardless of reef type (joint Wald test: growth form*location: F 12,Inf = 11.426, Chi-squared = 137.112, p < 0.001, growth form*method*location: F 12,Inf = 0.973, Chi-squared = 11.676, p = 0.472). The model explained 64 percent of the variation in the data (R 2 marginal = 0.637, R 2 conditional = 0.642). 3.2 Evaluating changes between depths Hard coral cover as a percent of total cover was not strongly influenced by depth or reef type (joint Wald tests: depth, F 3,Inf = 1.685, Chi-squared = 5.055, p = 0.168; reef type, F 2,Inf = 1.985, Chi-squared = 3.970, p = 0.137). There was stronger evidence for the interaction effect of depth and location, but the results were still not statistically significant (joint Wald Test: F 6,Inf = 1.939, Chi-squared = 11.634, p = 0.071). This model showed that high coral cover at the crest of a site did not correlate with high coral cover at other depths of the same site, with correlation values of -0.11 for 5 meters, -0.92 for 10 meters, and − 0.85 for 15 meters. There were strong positive correlations for all depths other than the crest (15m-10m = 0.49, 15m-5m = 0.60, 10m-5m = 0.99). This model explained 45 percent of the variance within the data (R 2 marginal = 0.450). The percentage of bleached and pale coral cover decreased with increasing depth, while healthy coral increased. For all health categories, there were significant differences associated with depth (joint Wald test: bleached: F 3,Inf = 5.558, Chi-squared = 16.674, p < 0.001; pale: F 3,Inf = 18.256, Chi-squared = 54.468, p < 0.001; healthy: F 3,Inf = 36.884, Chi-squared = 110.652, p < 0.001). There were significant differences between the crest and all other depths for all health categories (Supplementary Table 1). There were significant differences between locations, with nearshore sites having a higher percentage of pale coral and a lower percentage of healthy coral (Supplementary Table 2). This model explained 97 percent of the variance in the data (R 2 marginal = 0.973). The concentration of growth forms varied between depths and reef types. The model showed significant effects of depth, reef type, and their interaction on growth form (joint Wald tests: growth form*depth, F 18,Inf = 7.193, Chi-squared = 129.474, p < 0.001; growth form*reef type, F 12,Inf = 10.182, Chi-squared = 122.184, p < 0.001; growth form*depth*reef type, F 36,Inf = 2.131, Chi-squared = 76.716, p < 0.001). There were significant differences between depths for five of the seven growth forms, with foliose and encrusting coral generally increasing at depth, while massive coral decreased with depth (Supplementary Table 3). Branching, encrusting, and massive coral had significant differences between locations (Supplementary Table 4). This model explained 59 percent of the variation in the model (R 2 marginal = 0.591, R 2 conditional = 0.595). Massive coral had the largest discrepancy between locations, making up 66.8 percent of total coral cover at the crest of nearshore reefs, while only accounting for 23.8 and 24.3 percent of cover on the crest of fringing and midshelf reefs, respectively. Encrusting coral accounted for the majority of hard coral cover on the crest of the fringing reef, with 52.2 percent. Branching coral contributed the largest portion of coral cover on the crest of midshelf reefs, with 41.4 percent of cover. Of the three positions of reefs that we sampled (fringing, nearshore, and mid-shelf), the community composition, as assessed through growth form, varied most at shallow depths, while the differences decreased with increasing depth. When comparing four permanovas of the reef types at the four depth categories, there are decreases in Sum of Squares, F and R 2 values with increase in depth (crest: sum of squares = 1.86, R 2 = 0.425, F 2 = 12.2, p = 0.001, 5 meters: sum of squares = 1.78, R 2 = 0.327, F 2 = 8.03, p = 0.001, 10 meters: R 2 = 0.164, F 2 = 3.24, p = 0.007, 15 meters: sum of squares = 0.429, R 2 = 0.133, F 2 = 2.54, p = 0.042). The decreasing sum of squares, F, and R 2 values show that the degree of variation in community composition between positions declines with depth. Significant differences in the community composition exist between reefs in different locations, and they are most drastic at the crest. 4. Discussion Our study shows the potential of UAVs to increase the availability of high-resolution, scalable data for use in coral reef management. We show that UAV images have a high degree of overlap with common in-water methods for coral cover, bleaching extent, and coral growth form. We also demonstrated that there is no significant variation in total coral cover between depths; however, depth trends are not consistent across sites. Additionally, the most significant bleaching and growth form variations exist at the crest. Our study expands on previous literature, which evaluated the accuracy of UAV-derived data in comparison to in-water methods for a variety of metrics of coral reef health (Levy et al. 2018 ; Cornet and Joyce 2021 ; Casella et al. 2022 ), by looking specifically at the representative potential of UAV data across depths for coral reefs and by demonstrating the usefulness of such systems in a remote part of the Coral Triangle. The UAV provided accurate information for hard coral cover, bleached coral cover, and growth form when compared to in-water methods. There were differences in the pale coral cover reported, but these are most likely due to a difference in data interpretation between different lighting situations rather than an inherent inability of either method to accurately report data. There were no significant differences between data from points identified in-water using reef check methods, or from photo ID from photo transect and UAV methods, but the photo ID methods generally had lower standard errors due to larger sample sizes. The increased data sizes achievable with photo transects and UAVs give them an advantage over reef check methods ( Kuo et al. 2022 ). In addition, the ability to reprocess images to answer different questions is helpful for evaluating changing environments. The three main factors we evaluated, total coral cover, coral health, and growth form, are crucial for marine management decisions to identify the current health of the reef, the scale of contemporary impacts, and the long-term changes that are taking place ( Gardner et al. 2003; Darling et al. 2012 ). This information, combined with data on nutrient output, sedimentation, and sea surface temperature, can help researchers and managers understand the status of a reef and make informed decisions for management and conservation ( Heron et al. 2016 ). While there were no significant differences in total coral cover between depths, sites with higher coral cover at the crest did not necessarily have higher coral cover at depths. There were positive correlations between all depths other than the crest, which could indicate that the crest is experiencing a different set of stressors than corals at deeper depths. Kimbe Bay does not have strong currents (Steinberg et al. 2006 ), resulting in low water movement. Consequently, recreational and scientific divers frequently report distinct temperature stratifications, with warmer water at the surface and a stark shift to cooler water between 3 and 8 meters (Himes pers. comm. 2024). The reef crest experienced the highest amount of bleaching during our sampling period, with bleached coral decreasing at deeper depths. Similar trends were found on the Great Barrier Reef during the 2016 bleaching event, with shallower reefs experiencing more significant bleaching (Muir et al. 2017 ; Baird et al. 2018 ; Frade et al. 2018 ). In our study, there were no significant differences between reef types for bleached coral cover, but nearshore sites had a larger percentage of pale corals and a lower percentage of healthy corals compared to midshelf and fringing reefs. These patterns may be explained by increased turbidity and nutrient enrichment at nearshore sites. Sedimentation can negatively impact corals and increase their susceptibility to bleaching events (Rogers 1990 ; Tuttle and Donahue 2022 ). With low water movement in Kimbe Bay (Steinberg et al. 2006 ), the nearshore reefs we studied are likely experiencing significantly more sedimentation from runoff than the fringing and midshelf reefs we sampled, which could contribute to the lower rates of healthy corals at nearshore reefs. In Kimbe Bay, nearly 75 percent of catchment areas have been modified by development in some way (Brodie and Turak 2004 ; Bun et al. 2004 ). Assessing coral growth form along with total coral cover and bleaching extent is useful, as it can give an indication of habitat complexity and community composition important to assess wider reef health (Komyakova et al. 2013 ; McWilliam et al. 2018 ; Wong et al. 2018 ; Cornet and Joyce 2021 ; González-Barrios et al. 2025 ). We found that while there were significant differences in growth form frequencies between reef types at all depths, the greatest variation was found at the crest. The nearshore and fringing reefs had over 50 percent massive and encrusting coral, respectively, at the crest. The midshelf reefs had a more even spread along the crest, with the largest contributor being branching corals, accounting for 41 percent of coral cover. This could indicate a different set of stressors affecting the different reef types. Nutrient enrichment and sedimentation can impact coral species and forms differently (Buckingham et al. 2022 ; Tuttle and Donahue 2022 ). However, in general, sedimentation causes a decline in coral cover, coral species, and net productivity of coral reefs (Rogers 1990 ). With massive and encrusting species seemingly more resistant to bleaching, an accumulation of stressors on nearshore reefs may have shifted community composition towards a more uniform, hardier makeup (McCowan et al. 2012 ). With increased development and disturbance events in the coral triangle, these trends towards less functionally diverse, more uniform coral communities are well documented (Hughes et al. 2018 ; Browne et al. 2019 ; Turak et al. 2021 ). The concentration of these effects on the reef crest indicates that shallower areas are being more affected by environmental disturbance, which matches global trends (De Bakker et al. 2016 ; Baird et al. 2018 ; Frade et al. 2018 ). Shifts in growth form composition and bleaching are greatest at the crest of the reefs we collected data from in Kimbe Bay, and while trends for coral cover on the crest were not predictive of those at depth, the major disturbances to coral reefs seem to be impacting the crest the most. UAVs can accurately and efficiently collect data on coral cover, bleaching, and growth form, and have the potential to greatly increase the spatial and temporal scales at which we can collect data, allowing for a better understanding of how coral reefs are changing. With coral bleaching events projected to increase in frequency and severity (Donner et al. 2005 ; Mellin et al. 2024 ), it is crucial to have accessible and reliable tools to evaluate change and make conservation decisions to protect what we have and facilitate recovery where possible. Declarations The authors declare that they have no conflict of interest. Author Contribution Conceptualization: L.H. and T.R. Data curation: L.H. Formal analysis: L.H. and T.R. Funding acquisition: L.H. Investigation: L.H. and T.R. Methodology: L.H. and T.R. Project administration: L.H. Resources: L.H. Supervision: L.H. and T.R. Validation: L.H. and T.R. Visualization: L.H. Writing—original draft: L.H. Writing—review and editing: L.H. and T.R. Acknowledgement We want to thank Mahonia Na Dari and its staff, especially S. Jonda, for providing facilities, knowledge, and support leading up to and throughout the completion of this project. We want to thank Mahonia Na Dari captains R. Martin and B. Mautu for their knowledge and support in navigating us to field sites, and the Tamare-Kilu communities for access to their reefs. We want to thank Walindi Plantation Resort and its staff for their logistical support and knowledge. We want to thank everyone who assisted with field work, including F. Noble, M. Giru, L. Yllan, and Y. Kobayashi. We acknowledge the use of AI-assisted technologies, specifically OpenAI’s ChatGPT, which assisted with coding and debugging during statistical analysis. Funding for this project was provided by the United States Fulbright Program. Data Availability We are in the process of making our data and R-script available; they will be made available by publication. References Baird AH, Madin JS, Álvarez-Noriega M, Fontoura L, Kerry JT, Kuo C-Y, Precoda K, Torres-Pulliza D, Woods RM, Zawada KJA, Hughes TP (2018) A decline in bleaching suggests that depth can provide a refuge from global warming in most coral taxa. Mar Ecol Prog Ser 603:257–264 Banha TNS, Capel KCC, Kitahara MV, Francini-Filho RB, Francini CLB, Sumida PYG, Mies M (2020) Low coral mortality during the most intense bleaching event ever recorded in subtropical Southwestern Atlantic reefs. Coral Reefs 39:515–521 Brodie J, Turak E (2004) Land use practices in the Stettin Bay catchment area and their relation to the status of the coral reefs in Kimbe Bay. Brooks ME, Kristensen K, Benthem JJ van, Magnusson A, Berg CW, Nielsen A, Skaug HJ, Maechler M, Bolker BM (2017) glmmTMB Balances Speed and Flexibility Among Packages for Zero-inflated Generalized Linear Mixed Modeling. R J 9:378–400 Browne N, Braoun C, McIlwain J, Nagarajan R, Zinke J (2019) Borneo coral reefs subject to high sediment loads show evidence of resilience to various environmental stressors. PeerJ 7:e7382 Buckingham MC, D’Angelo C, Chalk TB, Foster GL, Johnson KG, Connelly Z, Olla C, Saeed M, Wiedenmann J (2022) Impact of nitrogen (N) and phosphorus (P) enrichment and skewed N:P stoichiometry on the skeletal formation and microstructure of symbiotic reef corals. Coral Reefs Online 41:1147–1159 Bun YA, King T, Shearman P (2004) China’s impact on Papua New Guinea’s forestry industry. Forest Trends, Washington, D.C. Byrne M, Waller A, Clements M, Kelly AS, Kingsford MJ, Liu B, Reymond CE, Vila-Concejo A, Webb M, Whitton K, Foo SA (2025) Catastrophic bleaching in protected reefs of the Southern Great Barrier Reef. Limnol Oceanogr Lett 10:340–348 Carriger JF, Yee SH, Fisher WS (2021) Assessing Coral Reef Condition Indicators for Local and Global Stressors Using Bayesian Networks. Integr Environ Assess Manag 17:165–187 Casella E, Lewin P, Ghilardi M, Rovere A, Bejarano S (2022) Assessing the relative accuracy of coral heights reconstructed from drones and structure from motion photogrammetry on coral reefs. Coral Reefs 41:869–875 Cornet VJ, Joyce KE (2021) Assessing the Potential of Remotely-Sensed Drone Spectroscopy to Determine Live Coral Cover on Heron Reef. Drones 5:29 Danovaro R, Aronson J, Bianchelli S, Boström C, Chen W, Cimino R, Corinaldesi C, Cortina-Segarra J, D’Ambrosio P, Gambi C, Garrabou J, Giorgetti A, Grehan A, Hannachi A, Mangialajo L, Morato T, Orfanidis S, Papadopoulou N, Ramirez-Llodra E, Smith CJ, Snelgrove P, van de Koppel J, van Tatenhove J, Fraschetti S (2025) Assessing the success of marine ecosystem restoration using meta-analysis. Nat Commun 16:3062 De Bakker DM, Meesters EH, Bak RPM, Nieuwland G, Van Duyl FC (2016) Long-term Shifts in Coral Communities On Shallow to Deep Reef Slopes of Curaçao and Bonaire: Are There Any Winners? Front Mar Sci 3:247 Depczynski M, Gilmour JP, Ridgway T, Barnes H, Heyward AJ, Holmes TH, Moore J a. Y, Radford BT, Thomson DP, Tinkler P, Wilson SK (2013) Bleaching, coral mortality and subsequent survivorship on a West Australian fringing reef. Coral Reefs 32:233–238 Devoto S, Macovaz V, Mantovani M, Soldati M, Furlani S, Devoto S, Macovaz V, Mantovani M, Soldati M, Furlani S (2020) Advantages of Using UAV Digital Photogrammetry in the Study of Slow-Moving Coastal Landslides. Remote Sens 12:3566 Di Marco M, Chapman S, Althor G, Kearney S, Besancon C, Butt N, Maina JM, Possingham HP, Rogalla von Bieberstein K, Venter O, Watson JEM (2017) Changing trends and persisting biases in three decades of conservation science. Glob Ecol Conserv 10:32–42 Donner SD, Skirving WJ, Little CM, Oppenheimer M, Hoegh-Guldberg O (2005) Global assessment of coral bleaching and required rates of adaptation under climate change. Glob Change Biol 11:2251–2265 Falter JL, Zhang Z, Lowe RJ, McGregor F, Keesing J, McCulloch MT (2014) Assessing the drivers of spatial variation in thermal forcing across a nearshore reef system and implications for coral bleaching. Limnol Oceanogr 59:1241–1255 Fisher R, Radford BT, Knowlton N, Brainard RE, Michaelis FB, Caley MJ (2011) Global mismatch between research effort and conservation needs of tropical coral reefs. Conserv Lett 4:64–72 Frade PR, Bongaerts P, Englebert N, Rogers A, Gonzalez-Rivero M, Hoegh-Guldberg O (2018) Deep reefs of the Great Barrier Reef offer limited thermal refuge during mass coral bleaching. Nat Commun 9:3447 Galbraith GF, Cresswell BJ, McCormick MI, Bridge TC, Jones GP (2022) Contrasting hydrodynamic regimes of submerged pinnacle and emergent coral reefs. PLOS ONE 17:e0273092 González-Barrios FJ, Keith SA, Emslie MJ, Ceccarelli DM, Williams GJ, Graham NAJ (2025) Emergent patterns of reef fish diversity correlate with coral assemblage shifts along the Great Barrier Reef. Nat Commun 16:303 Gouezo M, Fabricius K, Harrison P, Golbuu Y, Doropoulos C (2021) Optimizing coral reef recovery with context-specific management actions at prioritized reefs. J Environ Manage 295:113209 Hedley JD, Roelfsema CM, Chollett I, Harborne AR, Heron SF, Weeks S, Skirving WJ, Strong AE, Eakin CM, Christensen TRL, Ticzon V, Bejarano S, Mumby PJ (2016) Remote Sensing of Coral Reefs for Monitoring and Management: A Review. Remote Sens 8:118 Hughes BB, Beas-Luna R, Barner AK, Brewitt K, Brumbaugh DR, Cerny-Chipman EB, Close SL, Coblentz KE, de Nesnera KL, Drobnitch ST, Figurski JD, Focht B, Friedman M, Freiwald J, Heady KK, Heady WN, Hettinger A, Johnson A, Karr KA, Mahoney B, Moritsch MM, Osterback A-MK, Reimer J, Robinson J, Rohrer T, Rose JM, Sabal M, Segui LM, Shen C, Sullivan J, Zuercher R, Raimondi PT, Menge BA, Grorud-Colvert K, Novak M, Carr MH (2017) Long-Term Studies Contribute Disproportionately to Ecology and Policy. BioScience 67:271–281 Hughes TP, Kerry JT, Baird AH, Connolly SR, Dietzel A, Eakin CM, Heron SF, Hoey AS, Hoogenboom MO, Liu G, McWilliam MJ, Pears RJ, Pratchett MS, Skirving WJ, Stella JS, Torda G (2018) Global warming transforms coral reef assemblages. Nature 556:492–496 Jones GP, McCormick MI, Srinivasan M, Eagle JV (2004) Coral decline threatens fish biodiversity in marine reserves. Proc Natl Acad Sci 101:8251–8253 Kohler KE, Gill SM (2006) Coral Point Count with Excel extensions (CPCe): A Visual Basic program for the determination of coral and substrate coverage using random point count methodology. Comput Geosci 32:1259–1269 Komyakova V, Munday PL, Jones GP (2013) Relative Importance of Coral Cover, Habitat Complexity and Diversity in Determining the Structure of Reef Fish Communities. PLOS ONE 8:e83178 Kooistra L, Berger K, Brede B, Graf LV, Aasen H, Roujean J-L, Machwitz M, Schlerf M, Atzberger C, Prikaziuk E, Ganeva D, Tomelleri E, Croft H, Reyes Muñoz P, Garcia Millan V, Darvishzadeh R, Koren G, Herrmann I, Rozenstein O, Belda S, Rautiainen M, Rune Karlsen S, Figueira Silva C, Cerasoli S, Pierre J, Tanır Kayıkçı E, Halabuk A, Tunc Gormus E, Fluit F, Cai Z, Kycko M, Udelhoven T, Verrelst J (2024) Reviews and syntheses: Remotely sensed optical time series for monitoring vegetation productivity. Biogeosciences 21:473–511 Lenth RV (2025) emmeans: Estimated Marginal Means, aka Least-Squares Means. https://CRAN.R-project.org/package=emmeans Levy J, Hunter C, Lukacazyk T, Franklin EC (2018) Assessing the spatial distribution of coral bleaching using small unmanned aerial systems. Coral Reefs 37:373–387 Lindenmayer DB, Lavery T, Scheele BC (2022) Why We Need to Invest in Large-Scale, Long-Term Monitoring Programs in Landscape Ecology and Conservation Biology. Curr Landsc Ecol Rep 7:137–146 Ludecke D, Ben-Shachar MS, Patil I, Waggoner P, Makowski D (2021) performance: An R Package for Assessment, Comparison and Testing of Statistical Models. J Open Source Softw 6:3139 Manfreda S, McCabe MF, Miller PE, Lucas R, Madrigal VP, Mallinis G, Dor EB, Helman D, Estes L, Ciraolo G, Müllerová J, Tauro F, Lima MID, Lima JLMPD, Maltese A, Frances F, Caylor K, Kohv M, Perks M, Ruiz-Pérez G, Su Z, Vico G, Toth B, Manfreda S, McCabe MF, Miller PE, Lucas R, Madrigal VP, Mallinis G, Dor EB, Helman D, Estes L, Ciraolo G, Müllerová J, Tauro F, Lima MID, Lima JLMPD, Maltese A, Frances F, Caylor K, Kohv M, Perks M, Ruiz-Pérez G, Su Z, Vico G, Toth B (2018) On the Use of Unmanned Aerial Systems for Environmental Monitoring. Remote Sens 10:641 Matthews SA, Williamson DH, Beeden R, Emslie MJ, Abom RTM, Beard D, Bonin M, Bray P, Campili AR, Ceccarelli DM, Fernandes L, Fletcher CS, Godoy D, Hemingson CR, Jonker MJ, Lang BJ, Morris S, Mosquera E, Phillips GL, Sinclair-Taylor TH, Taylor S, Tracey D, Wilmes JC, Quincey R (2024) Protecting Great Barrier Reef resilience through effective management of crown-of-thorns starfish outbreaks. PLOS ONE 19:e0298073 McCowan DM, Pratchett MS, Baird AH (2012) Bleaching susceptibility and mortality among corals with differing growth forms. McWilliam M, Hoogenboom MO, Baird AH, Kuo C-Y, Madin JS, Hughes TP (2018) Biogeographical disparity in the functional diversity and redundancy of corals. Proc Natl Acad Sci 115:3084–3089 Mellin C, Brown S, Cantin N, Klein-Salas E, Mouillot D, Heron SF, Fordham DA (2024) Cumulative risk of future bleaching for the world’s coral reefs. Sci Adv 10:eadn9660 Montambault JR, Wongbusarakum S, Leberer T, Joseph E, Andrew W, Castro F, Nevitt B, Golbuu Y, Oldiais NW, Groves CR, Kostka W, Houk P (2015) Use of monitoring data to support conservation management and policy decisions in Micronesia. Conserv Biol 29:1279–1289 Muir PR, Marshall PA, Abdulla A, Aguirre JD (2017) Species identity and depth predict bleaching severity in reef-building corals: shall the deep inherit the reef? Proc R Soc B Biol Sci 284:20171551 Mumby PJ, Harborne AR (2010) Marine Reserves Enhance the Recovery of Corals on Caribbean Reefs. PLoS ONE 5:e8657 Obura DO, Aeby G, Amornthammarong N, Appeltans W, Bax N, Bishop J, Brainard RE, Chan S, Fletcher P, Gordon TAC, Gramer L, Gudka M, Halas J, Hendee J, Hodgson G, Huang D, Jankulak M, Jones A, Kimura T, Levy J, Miloslavich P, Chou LM, Muller-Karger F, Osuka K, Samoilys M, Simpson SD, Tun K, Wongbusarakum S (2019) Coral Reef Monitoring, Reef Assessment Technologies, and Ecosystem-Based Management. Front Mar Sci 6:580 Oksanen J, Simpson GL, Blanchet FG, Kindt R, Legendre P, Minchin PR, O’Hara RB, Solymos P, Stevens MHH, Szoecs E, Wagner H, Barbour M, Bedward M, Bolker B, Borcard D, Borman T, Carvalho G, Chirico M, Caceres MD, Durand S, Evangelista HBA, FitzJohn R, Friendly M, Furneaux B, Hannigan G, Hill MO, Lahti L, Martino C, McGlinn D, Ouellette M-H, Cunha ER, Smith T, Stier A, Braak CJFT, Weedon J (2025) vegan: Community Ecology Package. https://cran.r-project.org/web/packages/vegan/index.html Parsons M, Bratanov D, Gaston KJ, Gonzalez F (2018) UAVs, Hyperspectral Remote Sensing, and Machine Learning Revolutionizing Reef Monitoring. Sensors 18:2026 R Core Team (2025) R: A Language and Environment for Statistical Computing. Rogers C (1990) Responses of coral reefs and reef organisms to sedimentation. Mar Ecol Prog Ser 62:185–202 Salamí E, Barrado C, Pastor E, Salamí E, Barrado C, Pastor E (2014) UAV Flight Experiments Applied to the Remote Sensing of Vegetated Areas. Remote Sens 6:11051–11081 Steinberg CR, Choukroun SM, Slivkoff MM, Mahoney MV, Brinkman RM (2006) Currents in the Bismarck Sea and Kimbe Bay. Tang L, Shao G (2015) Drone remote sensing for forestry research and practices. J For Res 26:791–797 Thomas L, Underwood JN, Rose NH, Fuller ZL, Richards ZT, Dugal L, Grimaldi CM, Cooke IR, Palumbi SR, Gilmour JP (2022) Spatially varying selection between habitats drives physiological shifts and local adaptation in a broadcast spawning coral on a remote atoll in Western Australia. Sci Adv 8:eabl9185 Turak E, DeVantier L, Szava-Kovats R, Brodie J (2021) Impacts of coastal land use change in the wet tropics on nearshore coral reefs: Case studies from Papua New Guinea. Mar Pollut Bull 168:112445 Turner W, Spector S, Gardiner N, Fladeland M, Sterling E, Steininger M (2003) Remote sensing for biodiversity science and conservation. Trends Ecol Evol 18:306–314 Tuttle LJ, Donahue MJ (2022) Effects of sediment exposure on corals: a systematic review of experimental studies. Environ Evid 11:4 Ullah H, Wahab MA, Rahman MJ, Al Mamun SN, Kumar U, Rahman MA, Souhardya SM, Kabir IE, Hussain M, Rahman MB, Chishty SMSUH (2023) Local ecological knowledge can support improved management of small-scale fisheries in the Bay of Bengal. Front Mar Sci 10:974591 Webber K, Srinivasan M, Coppock AG, Jones GP (2022) Spatial patterns in the cover and composition of macroalgal assemblages on fringing and nearshore coral reefs. Mar Freshw Res 73:1310–1322 Wong JSY, Chan YKS, Ng CSL, Tun KPP, Darling ES, Huang D (2018) Comparing patterns of taxonomic, functional and phylogenetic diversity in reef coral communities. Coral Reefs 37:737–750 (2004) Reef Check Instruction Manual: A Guide to Reef Check Coral Reef Monitoring. Reef Check, Inst. of the Environment, Los Angeles [Calif.] Additional Declarations No competing interests reported. Supplementary Files HimesandRuegerSupplementaryTables.docx Cite Share Download PDF Status: Published Journal Publication published 09 Apr, 2026 Read the published version in Coral Reefs → Version 1 posted Editorial decision: Revision requested 15 Feb, 2026 Reviews received at journal 11 Feb, 2026 Reviewers agreed at journal 29 Jan, 2026 Reviewers invited by journal 27 Jan, 2026 Editor assigned by journal 27 Jan, 2026 Submission checks completed at journal 14 Jan, 2026 First submitted to journal 09 Jan, 2026 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8565286","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":581425512,"identity":"95e1f0be-48fd-43ae-9853-f45b9b4c9fcd","order_by":0,"name":"Lucian Himes","email":"","orcid":"","institution":"Mahonia Na Dari – Guardian of the Sea","correspondingAuthor":false,"prefix":"","firstName":"Lucian","middleName":"","lastName":"Himes","suffix":""},{"id":581425517,"identity":"2315d75b-a792-4e5f-95bc-5d9e939064ca","order_by":1,"name":"Theresa Rueger","email":"data:image/png;base64,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","orcid":"","institution":"Newcastle University","correspondingAuthor":true,"prefix":"","firstName":"Theresa","middleName":"","lastName":"Rueger","suffix":""}],"badges":[],"createdAt":"2026-01-10 03:38:11","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8565286/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8565286/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1007/s00338-026-02864-5","type":"published","date":"2026-04-09T15:59:27+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":101406001,"identity":"34bc349d-16df-4e75-a720-9b6917902f1c","added_by":"auto","created_at":"2026-01-29 10:42:16","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":770415,"visible":true,"origin":"","legend":"\u003cp\u003e(A) Australia, Papua New Guinea, and Indonesia. (B)\u003cstrong\u003e \u003c/strong\u003eKimbe Bay. (C) West side of Kimbe Bay, dots represent field sites. (1) Restorf, (2) Vanessa’s, and (3) Christine’s. (D) Nearshore sampled reefs. (4) Garbuna, (5) Matane Walindi, (6) Luba Luba, ( 7) Limuka, (8) is Lady Di, and (9) Gava Gava. Imagery is used from Google Earth, Data SIO, NOAA, U.S. Navy, NGA, GEBCO, Image Landsat/Copernicus, Image © 2025 Maxar Technologies, Image © 2025 Airbus.\u003c/p\u003e","description":"","filename":"image1.png","url":"https://assets-eu.researchsquare.com/files/rs-8565286/v1/7d647d44974ff20007efc13d.png"},{"id":101406007,"identity":"3e97f60c-42db-4c9e-8029-343b1159abab","added_by":"auto","created_at":"2026-01-29 10:42:17","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":940140,"visible":true,"origin":"","legend":"\u003cp\u003e(A) Sample UAV image from Limuka Reef (B) Sample photo transect image from Limuka reef crest.\u003c/p\u003e","description":"","filename":"image2.png","url":"https://assets-eu.researchsquare.com/files/rs-8565286/v1/baa5cba873013aec9513367d.png"},{"id":101405942,"identity":"741a4e39-a6ad-4d97-a997-1cec6dcf0c15","added_by":"auto","created_at":"2026-01-29 10:42:08","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":75848,"visible":true,"origin":"","legend":"\u003cp\u003eFig 2. Hard coral cover and bleached coral cover on the crest for nearshore, fringing, and midshelf reefs by UAV, photo transects, and reef check methods.\u003c/p\u003e","description":"","filename":"image3.png","url":"https://assets-eu.researchsquare.com/files/rs-8565286/v1/8377df80af9791e39662133d.png"},{"id":101405860,"identity":"643d5910-15a8-435b-b07f-ae0bdd754724","added_by":"auto","created_at":"2026-01-29 10:41:51","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":50773,"visible":true,"origin":"","legend":"\u003cp\u003eFig 3. Percent of cover by reef type for each growth form for UAV and photo transects.\u003c/p\u003e","description":"","filename":"image4.png","url":"https://assets-eu.researchsquare.com/files/rs-8565286/v1/7229c7bec436557522f60641.png"},{"id":101405945,"identity":"bcc2161a-354f-400d-bf26-bc72ca4637eb","added_by":"auto","created_at":"2026-01-29 10:42:08","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":74678,"visible":true,"origin":"","legend":"\u003cp\u003eFig 4. Percent hard coral cover for all three depth categories grouped by site. Sites are grouped by reef type.\u003c/p\u003e","description":"","filename":"image5.png","url":"https://assets-eu.researchsquare.com/files/rs-8565286/v1/73fb932589b9c5eacba1f8f5.png"},{"id":101405940,"identity":"675ada84-d2fb-4f93-ba6d-0205d04767ce","added_by":"auto","created_at":"2026-01-29 10:42:07","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":92334,"visible":true,"origin":"","legend":"\u003cp\u003eFig 5. Percent of each health category from total coral cover by site and depth. Sites are grouped by reef type.\u003c/p\u003e","description":"","filename":"image6.png","url":"https://assets-eu.researchsquare.com/files/rs-8565286/v1/cb39d1cc625a6dc98dda0225.png"},{"id":101405987,"identity":"2791b03e-3c15-4601-ac09-cd62668ad6fd","added_by":"auto","created_at":"2026-01-29 10:42:10","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":101326,"visible":true,"origin":"","legend":"\u003cp\u003eFig 6. Percent of each growth form by site and depth. Sites are grouped by branching\u003c/p\u003e","description":"","filename":"image7.png","url":"https://assets-eu.researchsquare.com/files/rs-8565286/v1/c7a56b67caff9252618f022c.png"},{"id":106809110,"identity":"b631f41e-60bc-40ac-9ca4-e06dc33a04ce","added_by":"auto","created_at":"2026-04-13 16:07:03","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2931682,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8565286/v1/c644f465-a93a-403a-be56-6a99e65e2f70.pdf"},{"id":101405944,"identity":"d99829c9-d5d4-4cde-8483-b275e383b095","added_by":"auto","created_at":"2026-01-29 10:42:08","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":9864,"visible":true,"origin":"","legend":"","description":"","filename":"HimesandRuegerSupplementaryTables.docx","url":"https://assets-eu.researchsquare.com/files/rs-8565286/v1/85e58a4cfd37b530b1931879.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Drone Imaging can Accurately Assess Coral Cover, Bleaching, and Growth Form for Shallow Coral Reefs","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eFor positive conservation outcomes, managers and policy makers need high-resolution, location-specific information on reef health and stressors with high spatial and temporal resolution (Montambault et al. \u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e2015\u003c/span\u003e; Obura et al. \u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; Carriger et al. \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). Unfortunately, there is a large discrepancy between the areas of highest coral reef biodiversity and conservation need, e.g, the Coral Triangle, and where research and management practices are most concentrated, e.g, Eastern Australia and the US Pacific, making high-resolution data less available in areas with the highest need (Fisher et al. \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2011\u003c/span\u003e; Di Marco et al. \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2017\u003c/span\u003e). Generalizable data is useful for predicting and implementing positive policy around the globe; however, projects that integrate localized data into their decision-making framework rather than relying on regional-scale measurements have been more successful (Gouezo et al. \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Ullah et al. \u003cspan citationid=\"CR55\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; Matthews et al. \u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). In a time of rapid environmental change, well-reasoned, timely, and location-specific management decisions are needed (Mumby and Harborne \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2010\u003c/span\u003e; Danovaro et al. \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eTo increase the attainability of high-quality data, there is a need for low-cost monitoring methodologies. It is also important that these methods are scalable, as large amounts of relevant data are needed for informed management policies (Hughes et al. \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2017\u003c/span\u003e; Lindenmayer et al. \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). For coral reefs specifically, much of the \u003cem\u003ein-situ\u003c/em\u003e data on disturbance events, such as marine heatwaves that lead to bleaching, comes from permanent transects or individual colonies (Depczynski et al. \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2013\u003c/span\u003e; Banha et al. \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2020\u003c/span\u003e; Byrne et al. \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). While this is useful for particular research questions, exposure to stressors can vary in both time and spatial scales on a reef (Falter et al. \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2014\u003c/span\u003e; Thomas et al. \u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). One way to achieve large-scale datasets at lower costs is through remote sensing, as it allows for the coverage of large areas in a highly efficient and cost-effective manner compared to ground sampling (Hedley et al. \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e2016\u003c/span\u003e; Turner et al.(Turner et al. \u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e2003\u003c/span\u003e; Hedley et al. \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e2016\u003c/span\u003e).\u003c/p\u003e \u003cp\u003ePotentially the most accessible form of remote sensing is through the use of drones, technically known as unmanned aerial vehicles (UAVs) (Tang and Shao \u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e2015\u003c/span\u003e; Parsons et al. \u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). UAVs have several advantages compared to more intensive methods, such as satellites and manned aircraft. Satellites can provide data for large areas, but revisit times are fixed, and data quality is highly dependent on weather conditions, especially clouds (Manfreda et al. \u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Kooistra et al. \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). Manned aircraft have much larger startup and, in many cases, travel times to sampling areas than UAVs, as they are dependent on runway availability and locations. They also cost more to operate, and are more influenced by the environmental conditions, such as cloud cover, rain, and wind (Salam\u0026iacute; et al. \u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e2014\u003c/span\u003e). In contrast, UAVs can be easily and quickly deployed at relevant sites, and data acquisition can be done in minutes, depending on the needs of the project (Devoto et al. \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). This flexibility is crucial in highly variable conditions. Combined with much lower operation costs and the variability in spatial resolution that can be achieved, UAVs can fulfill the need for high-quality, localized data from remote coral reefs in areas where research is often under-resourced.\u003c/p\u003e \u003cp\u003eThe overall aim of this study was to assess UAVs' potential and limitations as a tool for low-cost coral reef monitoring in remote areas. To achieve this, we compiled benthic survey data during the 4th global coral mass bleaching event in 2024 from nearshore, fringing, and midshelf reefs in Kimbe Bay, Papua New Guinea. We compared a survey method utilizing a low-cost, entry-level UAV (DJI Mini 2) to commonly used in-water methods: point-intercept transects and photo transects. Our specific objectives were to 1) compare the accuracy of data from UAV images to in-water methods for shallow depths of fringing, inshore, and midshelf reefs, and 2) determine whether the shallow depths surveyable with the UAV are representative of deeper depths across reef types. We assessed accuracy in three areas: total coral cover, coral health, and coral growth form.\u003c/p\u003e"},{"header":"2. Methods","content":"\u003cp\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003eSites\u003c/span\u003e:\u003c/p\u003e\n\u003cp\u003eField work was conducted in Kimbe Bay, on the northern side of West New Britain, Papua New Guinea (-5.166838, 150.499957). To test the effectiveness of UAVs in a range of conditions, we chose nine reefs in different areas that had varying amounts of coral cover and composition (Jones et al. \u003cspan class=\"CitationRef\"\u003e2004\u003c/span\u003e; Turak et al. \u003cspan class=\"CitationRef\"\u003e2021\u003c/span\u003e; Galbraith et al. \u003cspan class=\"CitationRef\"\u003e2022\u003c/span\u003e; Webber et al. \u003cspan class=\"CitationRef\"\u003e2022\u003c/span\u003e). One of the reefs, Gava Gava, is on the inside row of a set of near-shore reefs, and five reefs, Garbuna, Matane Walindi, Luba Luba, Lady Di, and Limuka, are on the outside row of the near-shore reefs. One reef, Restorf, is a fringing reef on a small island. The last two reefs, Vanessa\u0026rsquo;s and Christine\u0026rsquo;s, are midshelf reefs. Site locations are shown in Fig. \u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e. Sites were chosen on each reef to have a consistent reef slope for at least 100 meters to a depth of 15 meters to avoid conducting transects on reef walls.\u003c/p\u003e\n\u003cp\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003eIn-water Sampling\u003c/span\u003e:\u003c/p\u003e\n\u003cp\u003eUsing SCUBA, we sampled the nine reefs at four different depths: the crest (approximately 1.5 m deep), 5 m, 10 m, and 15 m. At each depth, we deployed four 20-meter transects, each spaced from the other by at least 5 m. In-water sampling followed two methodologies along the same transect line. First, we took photos of each meter of the transect. The second in-water method, conducted on the same transect, followed the point-intercept transect (PIT) methods outlined by the Reef Check Foundation for monitoring coral reefs (2004). Substrate was classified among the eleven Reef Check categories, with a category for bleached coral added: hard coral, soft coral, recently killed coral, nutrient indicator algae, sponge, rock, rubble, sand, silt, other, bleached coral. The PIT method sampled the substrate directly beneath the transect every half meter for a total of 40 points per transect. With four transects conducted for each depth at each site, we collected 160 points per depth, in line with the methods that Reef Check recommends.\u003c/p\u003e\n\u003cp\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003eAerial Sampling\u003c/span\u003e:\u003c/p\u003e\n\u003cp\u003eThe UAV selected for this testing is the DJI Mini 2. This UAV was chosen due to its low cost, ease of use, and compactness. The UAV was flown over the crest of the reef in the section where the in-water transects were completed at a height of 10 m. Two photos were taken for each transect, for a total of eight photos per reef. The camera was pointed straight down when possible, or at the most vertical angle possible to avoid any sun glint on the water that obscured the reef. All photos included the entire visible reef crest, with the sides ending in the reef flat and in water too deep to identify substrate. At a height of 10 m, the UAV produced images of 10 m by 14.2 m of reef at a resolution of approximately 0.12 cm\u003csup\u003e2\u003c/sup\u003eper pixel.\u0026nbsp;\u003c/p\u003e\u003ctable id=\"Tab1\" border=\"1\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e\n \u003cdiv class=\"CaptionContent\"\u003e\n \u003cp\u003eDetails of data gathered for each sampling method.\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003eMethod\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003ePoints Per Transect\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\" colspan=\"4\"\u003e\n \u003cp\u003eCategories Measured\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\" rowspan=\"2\"\u003e\n \u003cp\u003eDepths sampled\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eHard Coral Cover\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eBleached Coral Cover\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003ePale Coral Cover\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eCoral Growth Form\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eUAV\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e200\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCrest\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePhoto Transects\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e200\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCrest, 5m, 10m, 15m\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eReef Check\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e40\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCrest, 5m, 10m, 15m\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003c/p\u003e\n\u003cp\u003e\u003cspan type=\"Underline\" name=\"Emphasis\"\u003ePhoto analysis\u003c/span\u003e:\u003c/p\u003e\n\u003cp\u003ePhotos from the photo transects and the UAV were analyzed using Coral Point Count with Excel extensions (CPCe) (Kohler and Gill \u003cspan class=\"CitationRef\"\u003e2006\u003c/span\u003e). For the photo transects, ten photos were chosen. We started with the second photo of each transect and used every other photo from then on to avoid overlap between photos and eliminate the chance of counting points multiple times. The entire photo was included in the analysis, and random points could be generated anywhere on the image. Twenty points were randomly generated using the CPCe software on each photo and classified into 13 categories. The categories are \u0026ldquo;hard coral\u0026rdquo;, \u0026ldquo;pale hard coral\u0026rdquo;, \u0026ldquo;bleached hard coral\u0026rdquo;, \u0026ldquo;recently killed coral\u0026rdquo;, \u0026ldquo;algae\u0026rdquo;, \u0026ldquo;sponge\u0026rdquo;, \u0026ldquo;sand\u0026rdquo;, \u0026ldquo;silt/clay\u0026rdquo;, \u0026ldquo;rock\u0026rdquo;, \u0026ldquo;rubble\u0026rdquo;, \u0026ldquo;other\u0026rdquo;, and \u0026ldquo;tape, wand, or shadow\u0026rdquo;. Each of the coral categories, \u0026ldquo;hard coral\u0026rdquo;, \u0026ldquo;pale hard coral\u0026rdquo;, \u0026ldquo;bleached hard coral\u0026rdquo;, \u0026ldquo;recently killed coral\u0026rdquo;, was further differentiated into seven subcategories: \u0026ldquo;foliose coral\u0026rdquo;, \u0026ldquo;encrusting coral\u0026rdquo;, \u0026ldquo;branching coral\u0026rdquo;, \u0026ldquo;pillar corals\u0026rdquo;, \u0026ldquo;table corals\u0026rdquo;, \u0026ldquo;massive corals\u0026rdquo;, and \u0026ldquo;mushroom corals\u0026rdquo;. Algae were also differentiated: \u0026ldquo;nutrient indicator algae\u0026rdquo; and \u0026ldquo;turf\u0026rdquo;. Points classified as \u0026ldquo;tape, wand, or shadow\u0026rdquo; were not counted towards the percent cover. Across the ten photos, we classified a total of 200 points per transect for the photo transects. This resulted in a total of 800 points classified for each depth of each reef.\u003c/p\u003e\n\u003cp\u003eThe photos from the UAV were cropped to include what could be considered the reef crest. The reef flat and depths where substrate could not be identified were not included. Sections of reef that were in more than one photo were only included in the first photo. We generated 100 random points per photo, so each set of two photos from the UAV matched the sets of ten photos for each in-water transect with 200 total points. These points were assigned to the same categories as for the photo transects, and between the two UAV photos per transect and the four transects per reef, we obtained 200 points per transect and 800 points per reef.\u003c/p\u003e\n\u003cp\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003eStatistical analysis\u003c/span\u003e:\u003c/p\u003e\n\u003cp\u003eWe performed all statistical analyses in R version 4.5.0 (R Core Team \u003cspan class=\"CitationRef\"\u003e2025\u003c/span\u003e). We used the \u003cem\u003eglmmTMB\u003c/em\u003e package to fit beta regression models, the \u003cem\u003eemmeans\u003c/em\u003e package to check for global fixed effects and perform pairwise analysis, and the \u003cem\u003eperformance\u003c/em\u003e package to check model assumptions for all models detailed below (Brooks et al. \u003cspan class=\"CitationRef\"\u003e2017\u003c/span\u003e; Ludecke et al. \u003cspan class=\"CitationRef\"\u003e2021\u003c/span\u003e; Lenth \u003cspan class=\"CitationRef\"\u003e2025\u003c/span\u003e). We used the \u003cem\u003evegan\u003c/em\u003e package to perform a PERMANOVA test (Oksanen et al. \u003cspan class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e\n\u003cp\u003eTo determine whether there was variation between sites, we compared a null beta regression model with a beta regression model that included site as the only predictor variable. Both models used hard coral cover as the response variable. We found a significant difference between the models (Log-likelihood test: df\u0026thinsp;=\u0026thinsp;10, Chi-sq\u0026thinsp;=\u0026thinsp;53.2, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), leading us to include site as a random factor in all other models.\u003c/p\u003e\n\u003cp\u003eTo compare hard coral cover from UAV images to in-water methods for shallow depths on fringing, nearshore, and midshelf reefs, we fit a beta regression model with hard coral cover as a proportion of benthic cover as the response variable, method (categorical; \u0026lsquo;UAV\u0026rsquo;, \u0026lsquo;Reef Check\u0026rsquo;, \u0026lsquo;Photo Transect\u0026rsquo;), reef type (categorical; \u0026lsquo;fringing\u0026rsquo;, \u0026lsquo;nearshore\u0026rsquo;, \u0026lsquo;midshelf\u0026rsquo;), and their interaction as predictors, and site as a random variable to account for non-independence. This model used data from UAV, photo transect, and reef check methods from the crest. To evaluate global effects among model parameters, we used the \u003cem\u003eemmeans\u003c/em\u003e package.\u003c/p\u003e\n\u003cp\u003eTo evaluate the accuracy of UAV imagery to identify the scale of bleaching, we fit a beta regression model with bleached coral cover as a proportion of benthic cover as the response variable, method and reef type as the predictor variables, and site as the random variable to account for non-independence. This model used data from the UAV, photo transect, and reef check methods, with all samples from the reef crest. A second beta regression was fit with pale coral cover as the response variable, and only using UAV and photo transect images from the crest. To evaluate global fixed effects among both model parameters, we used the \u003cem\u003eemmeans\u003c/em\u003e package.\u003c/p\u003e\n\u003cp\u003eTo evaluate the effectiveness of UAV imagery to determine growth forms of corals on reef crests, we fit a beta regression model with percent of total coral cover as the response variable, growth form, method, reef type, and all interactions as the predictor variables, and site as a random variable. We used UAV and photo transect data from the crest for this model. We used the \u003cem\u003eemmeans\u003c/em\u003e package to evaluate global fixed effects and conduct pairwise comparisons among model parameters.\u003c/p\u003e\n\u003cp\u003eTo evaluate changes of coral cover with depth, we fit a beta regression model with hard coral cover as the response variable, depth (categorical; \u0026lsquo;crest\u0026rsquo;, \u0026lsquo;5m\u0026rsquo;, \u0026lsquo;10m\u0026rsquo;, \u0026lsquo;15m\u0026rsquo;), reef type, and their interaction as the predictor variable, and site as a random factor with random slopes for depth within site to evaluate whether depth trends were consistent across sites. We used photo transect data from all depths for this model. We used the emmeans package to evaluate fixed global effects.\u003c/p\u003e\n\u003cp\u003eTo evaluate changes with depth for coral health categories (\u0026lsquo;bleached\u0026rsquo;, \u0026lsquo;pale\u0026rsquo;, \u0026lsquo;healthy\u0026rsquo;), we fit a beta regression model with percent of total coral as the response variable, health category, depth, reef-type, and all interactions as the predictor variable, and site as the response variable. Data from the photo transect methods and all depths were used for this model. We used the emmeans package to evaluate fixed global effects and perform pairwise analyses on model parameters.\u003c/p\u003e\n\u003cp\u003eTo evaluate changes in growth forms with depth, we fit a beta regression model with percent of total coral cover as the response variable, growth form, depth, reef type, and all interactions as the predictor variables, and site as a random factor. We used data from the photo transects and all depths for this model. Emmeans was used to evaluate fixed global effects and perform pairwise analysis.\u003c/p\u003e\n\u003cp\u003eTo evaluate the degree of difference between coral growth form frequencies with depth between reef types, we fit four PERMANOVA models, one for each depth, with growth form frequency as the response variable and reef type as the predictor variable. Significance was determined with Bray-Curtis dissimilarities and 999 permutations.\u003c/p\u003e"},{"header":"3. Results","content":"\u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003e3.1 Comparing UAV and in-water methods\u003c/h2\u003e \u003cp\u003eAll three methods, reef check, photo transects, and UAV, reported similar total hard coral cover at the crest (global fixed effects test: F\u003csub\u003e2,Inf\u003c/sub\u003e = 0.317, Chi-squared\u0026thinsp;=\u0026thinsp;0.634, p\u0026thinsp;=\u0026thinsp;0.728). Additionally, there was no effect of location, or interaction between method and reef-type (global fixed effects test: reef-type: F\u003csub\u003e2,Inf\u003c/sub\u003e = 1.711, Chi-squared\u0026thinsp;=\u0026thinsp;3.422, p\u0026thinsp;=\u0026thinsp;0.181, method*reef-type: F\u003csub\u003e4,Inf\u003c/sub\u003e = 1.152, Chi-squared\u0026thinsp;=\u0026thinsp;4.608, p\u0026thinsp;=\u0026thinsp;0.330). Fixed and random factors together explained 41% of the variance in the data (R\u003csup\u003e2\u003c/sup\u003e\u003csub\u003emarginal\u003c/sub\u003e\u0026thinsp;=\u0026thinsp;0.185, R\u003csup\u003e2\u003c/sup\u003e\u003csub\u003econditional\u003c/sub\u003e\u0026thinsp;=\u0026thinsp;0.410).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eAll three methods were consistent in estimating the amount of bleached coral cover at the crest (global fixed effects test: F\u003csub\u003e2,Inf\u003c/sub\u003e = 0.195, Chi-squared\u0026thinsp;=\u0026thinsp;0.390, p\u0026thinsp;=\u0026thinsp;0.823). There was also no significant difference between locations for bleached coral on the crest (global fixed effects test: F\u003csub\u003e2,Inf\u003c/sub\u003e = 1.433, Chi-squared\u0026thinsp;=\u0026thinsp;2.866, p\u0026thinsp;=\u0026thinsp;0.239). The model explained 47 percent of the variation in the data (R\u003csup\u003e2\u003c/sup\u003e\u003csub\u003emarginal\u003c/sub\u003e\u0026thinsp;=\u0026thinsp;0.129, R\u003csup\u003e2\u003c/sup\u003e\u003csub\u003econditional\u003c/sub\u003e\u0026thinsp;=\u0026thinsp;0.470).\u003c/p\u003e \u003cp\u003eThere was a significant difference between the pale hard coral cover reported by the UAV and photo transect methods on the crest. Pale coral cover was overrepresented in UAV data (beta GLMM pairwise comparison: UAV vs photo transects, estimate\u0026thinsp;=\u0026thinsp;0.764, standard error (s.e.)\u0026thinsp;=\u0026thinsp;0.222, z\u0026thinsp;=\u0026thinsp;3.446, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). Reef check methods were not included in this model because the only health categories for hard coral in the reef check methods were healthy and bleached hard coral. The model explained 61 percent of the variation in the data (R\u003csup\u003e2\u003c/sup\u003e\u003csub\u003emarginal\u003c/sub\u003e\u0026thinsp;=\u0026thinsp;0.479, R\u003csup\u003e2\u003c/sup\u003e\u003csub\u003econditional\u003c/sub\u003e\u0026thinsp;=\u0026thinsp;0.610).\u003c/p\u003e \u003cp\u003eThe percent cover of different coral growth forms had large variance on the crest, but both UAV and photo transect methods were able to capture the variations accurately (joint Wald test: growth form: F\u003csub\u003e6,Inf\u003c/sub\u003e = 61.307, Chi-squared\u0026thinsp;=\u0026thinsp;367.842, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001, growth form*method: F\u003csub\u003e6,Inf\u003c/sub\u003e = 2.004, Chi-squared\u0026thinsp;=\u0026thinsp;4.008, p\u0026thinsp;=\u0026thinsp;0.135). Growth form varied between reef types, but both methods were able to report accurate information regardless of reef type (joint Wald test: growth form*location: F\u003csub\u003e12,Inf\u003c/sub\u003e = 11.426, Chi-squared\u0026thinsp;=\u0026thinsp;137.112, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001, growth form*method*location: F\u003csub\u003e12,Inf\u003c/sub\u003e = 0.973, Chi-squared\u0026thinsp;=\u0026thinsp;11.676, p\u0026thinsp;=\u0026thinsp;0.472). The model explained 64 percent of the variation in the data (R\u003csup\u003e2\u003c/sup\u003e\u003csub\u003emarginal\u003c/sub\u003e\u0026thinsp;=\u0026thinsp;0.637, R\u003csup\u003e2\u003c/sup\u003e\u003csub\u003econditional\u003c/sub\u003e\u0026thinsp;=\u0026thinsp;0.642).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003e3.2 Evaluating changes between depths\u003c/h2\u003e \u003cp\u003eHard coral cover as a percent of total cover was not strongly influenced by depth or reef type (joint Wald tests: depth, F\u003csub\u003e3,Inf\u003c/sub\u003e = 1.685, Chi-squared\u0026thinsp;=\u0026thinsp;5.055, p\u0026thinsp;=\u0026thinsp;0.168; reef type, F\u003csub\u003e2,Inf\u003c/sub\u003e = 1.985, Chi-squared\u0026thinsp;=\u0026thinsp;3.970, p\u0026thinsp;=\u0026thinsp;0.137). There was stronger evidence for the interaction effect of depth and location, but the results were still not statistically significant (joint Wald Test: F\u003csub\u003e6,Inf\u003c/sub\u003e = 1.939, Chi-squared\u0026thinsp;=\u0026thinsp;11.634, p\u0026thinsp;=\u0026thinsp;0.071). This model showed that high coral cover at the crest of a site did not correlate with high coral cover at other depths of the same site, with correlation values of -0.11 for 5 meters, -0.92 for 10 meters, and \u0026minus;\u0026thinsp;0.85 for 15 meters. There were strong positive correlations for all depths other than the crest (15m-10m\u0026thinsp;=\u0026thinsp;0.49, 15m-5m\u0026thinsp;=\u0026thinsp;0.60, 10m-5m\u0026thinsp;=\u0026thinsp;0.99). This model explained 45 percent of the variance within the data (R\u003csup\u003e2\u003c/sup\u003e\u003csub\u003emarginal\u003c/sub\u003e\u0026thinsp;=\u0026thinsp;0.450).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThe percentage of bleached and pale coral cover decreased with increasing depth, while healthy coral increased. For all health categories, there were significant differences associated with depth (joint Wald test: bleached: F\u003csub\u003e3,Inf\u003c/sub\u003e = 5.558, Chi-squared\u0026thinsp;=\u0026thinsp;16.674, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001; pale: F\u003csub\u003e3,Inf\u003c/sub\u003e = 18.256, Chi-squared\u0026thinsp;=\u0026thinsp;54.468, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001; healthy: F\u003csub\u003e3,Inf\u003c/sub\u003e = 36.884, Chi-squared\u0026thinsp;=\u0026thinsp;110.652, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). There were significant differences between the crest and all other depths for all health categories (Supplementary Table\u0026nbsp;1). There were significant differences between locations, with nearshore sites having a higher percentage of pale coral and a lower percentage of healthy coral (Supplementary Table\u0026nbsp;2). This model explained 97 percent of the variance in the data (R\u003csup\u003e2\u003c/sup\u003e\u003csub\u003emarginal\u003c/sub\u003e\u0026thinsp;=\u0026thinsp;0.973).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThe concentration of growth forms varied between depths and reef types. The model showed significant effects of depth, reef type, and their interaction on growth form (joint Wald tests: growth form*depth, F\u003csub\u003e18,Inf\u003c/sub\u003e = 7.193, Chi-squared\u0026thinsp;=\u0026thinsp;129.474, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001; growth form*reef type, F\u003csub\u003e12,Inf\u003c/sub\u003e = 10.182, Chi-squared\u0026thinsp;=\u0026thinsp;122.184, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001; growth form*depth*reef type, F\u003csub\u003e36,Inf\u003c/sub\u003e = 2.131, Chi-squared\u0026thinsp;=\u0026thinsp;76.716, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). There were significant differences between depths for five of the seven growth forms, with foliose and encrusting coral generally increasing at depth, while massive coral decreased with depth (Supplementary Table\u0026nbsp;3). Branching, encrusting, and massive coral had significant differences between locations (Supplementary Table\u0026nbsp;4). This model explained 59 percent of the variation in the model (R\u003csup\u003e2\u003c/sup\u003e\u003csub\u003emarginal\u003c/sub\u003e\u0026thinsp;=\u0026thinsp;0.591, R\u003csup\u003e2\u003c/sup\u003e\u003csub\u003econditional\u003c/sub\u003e\u0026thinsp;=\u0026thinsp;0.595). Massive coral had the largest discrepancy between locations, making up 66.8 percent of total coral cover at the crest of nearshore reefs, while only accounting for 23.8 and 24.3 percent of cover on the crest of fringing and midshelf reefs, respectively. Encrusting coral accounted for the majority of hard coral cover on the crest of the fringing reef, with 52.2 percent. Branching coral contributed the largest portion of coral cover on the crest of midshelf reefs, with 41.4 percent of cover.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eOf the three positions of reefs that we sampled (fringing, nearshore, and mid-shelf), the community composition, as assessed through growth form, varied most at shallow depths, while the differences decreased with increasing depth. When comparing four permanovas of the reef types at the four depth categories, there are decreases in Sum of Squares, F and R\u003csup\u003e2\u003c/sup\u003e values with increase in depth (crest: sum of squares\u0026thinsp;=\u0026thinsp;1.86, R\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;=\u0026thinsp;0.425, F\u003csub\u003e2\u003c/sub\u003e\u0026thinsp;=\u0026thinsp;12.2, p\u0026thinsp;=\u0026thinsp;0.001, 5 meters: sum of squares\u0026thinsp;=\u0026thinsp;1.78, R\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;=\u0026thinsp;0.327, F\u003csub\u003e2\u003c/sub\u003e\u0026thinsp;=\u0026thinsp;8.03, p\u0026thinsp;=\u0026thinsp;0.001, 10 meters: R\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;=\u0026thinsp;0.164, F\u003csub\u003e2\u003c/sub\u003e\u0026thinsp;=\u0026thinsp;3.24, p\u0026thinsp;=\u0026thinsp;0.007, 15 meters: sum of squares\u0026thinsp;=\u0026thinsp;0.429, R\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;=\u0026thinsp;0.133, F\u003csub\u003e2\u003c/sub\u003e\u0026thinsp;=\u0026thinsp;2.54, p\u0026thinsp;=\u0026thinsp;0.042). The decreasing sum of squares, F, and R\u003csup\u003e2\u003c/sup\u003e values show that the degree of variation in community composition between positions declines with depth. Significant differences in the community composition exist between reefs in different locations, and they are most drastic at the crest.\u003c/p\u003e \u003c/div\u003e"},{"header":"4. Discussion","content":"\u003cp\u003eOur study shows the potential of UAVs to increase the availability of high-resolution, scalable data for use in coral reef management. We show that UAV images have a high degree of overlap with common in-water methods for coral cover, bleaching extent, and coral growth form. We also demonstrated that there is no significant variation in total coral cover between depths; however, depth trends are not consistent across sites. Additionally, the most significant bleaching and growth form variations exist at the crest. Our study expands on previous literature, which evaluated the accuracy of UAV-derived data in comparison to in-water methods for a variety of metrics of coral reef health (Levy et al. \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Cornet and Joyce \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Casella et al. \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2022\u003c/span\u003e), by looking specifically at the representative potential of UAV data across depths for coral reefs and by demonstrating the usefulness of such systems in a remote part of the Coral Triangle.\u003c/p\u003e \u003cp\u003eThe UAV provided accurate information for hard coral cover, bleached coral cover, and growth form when compared to in-water methods. There were differences in the pale coral cover reported, but these are most likely due to a difference in data interpretation between different lighting situations rather than an inherent inability of either method to accurately report data. There were no significant differences between data from points identified in-water using reef check methods, or from photo ID from photo transect and UAV methods, but the photo ID methods generally had lower standard errors due to larger sample sizes. The increased data sizes achievable with photo transects and UAVs give them an advantage over reef check methods (\u003cem\u003eKuo et al. 2022\u003c/em\u003e). In addition, the ability to reprocess images to answer different questions is helpful for evaluating changing environments. The three main factors we evaluated, total coral cover, coral health, and growth form, are crucial for marine management decisions to identify the current health of the reef, the scale of contemporary impacts, and the long-term changes that are taking place (\u003cem\u003eGardner et al. 2003; Darling et al. 2012\u003c/em\u003e). This information, combined with data on nutrient output, sedimentation, and sea surface temperature, can help researchers and managers understand the status of a reef and make informed decisions for management and conservation (\u003cem\u003eHeron et al. 2016\u003c/em\u003e).\u003c/p\u003e \u003cp\u003eWhile there were no significant differences in total coral cover between depths, sites with higher coral cover at the crest did not necessarily have higher coral cover at depths. There were positive correlations between all depths other than the crest, which could indicate that the crest is experiencing a different set of stressors than corals at deeper depths. Kimbe Bay does not have strong currents (Steinberg et al. \u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e2006\u003c/span\u003e), resulting in low water movement. Consequently, recreational and scientific divers frequently report distinct temperature stratifications, with warmer water at the surface and a stark shift to cooler water between 3 and 8 meters (Himes pers. comm. 2024).\u003c/p\u003e \u003cp\u003eThe reef crest experienced the highest amount of bleaching during our sampling period, with bleached coral decreasing at deeper depths. Similar trends were found on the Great Barrier Reef during the 2016 bleaching event, with shallower reefs experiencing more significant bleaching (Muir et al. \u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e2017\u003c/span\u003e; Baird et al. \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Frade et al. \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). In our study, there were no significant differences between reef types for bleached coral cover, but nearshore sites had a larger percentage of pale corals and a lower percentage of healthy corals compared to midshelf and fringing reefs. These patterns may be explained by increased turbidity and nutrient enrichment at nearshore sites. Sedimentation can negatively impact corals and increase their susceptibility to bleaching events (Rogers \u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e1990\u003c/span\u003e; Tuttle and Donahue \u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). With low water movement in Kimbe Bay (Steinberg et al. \u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e2006\u003c/span\u003e), the nearshore reefs we studied are likely experiencing significantly more sedimentation from runoff than the fringing and midshelf reefs we sampled, which could contribute to the lower rates of healthy corals at nearshore reefs. In Kimbe Bay, nearly 75 percent of catchment areas have been modified by development in some way (Brodie and Turak \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2004\u003c/span\u003e; Bun et al. \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e2004\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eAssessing coral growth form along with total coral cover and bleaching extent is useful, as it can give an indication of habitat complexity and community composition important to assess wider reef health (Komyakova et al. \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e2013\u003c/span\u003e; McWilliam et al. \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Wong et al. \u003cspan citationid=\"CR57\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Cornet and Joyce \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Gonz\u0026aacute;lez-Barrios et al. \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). We found that while there were significant differences in growth form frequencies between reef types at all depths, the greatest variation was found at the crest. The nearshore and fringing reefs had over 50 percent massive and encrusting coral, respectively, at the crest. The midshelf reefs had a more even spread along the crest, with the largest contributor being branching corals, accounting for 41 percent of coral cover. This could indicate a different set of stressors affecting the different reef types. Nutrient enrichment and sedimentation can impact coral species and forms differently (Buckingham et al. \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; Tuttle and Donahue \u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). However, in general, sedimentation causes a decline in coral cover, coral species, and net productivity of coral reefs (Rogers \u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e1990\u003c/span\u003e). With massive and encrusting species seemingly more resistant to bleaching, an accumulation of stressors on nearshore reefs may have shifted community composition towards a more uniform, hardier makeup (McCowan et al. \u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e2012\u003c/span\u003e). With increased development and disturbance events in the coral triangle, these trends towards less functionally diverse, more uniform coral communities are well documented (Hughes et al. \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Browne et al. \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; Turak et al. \u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). The concentration of these effects on the reef crest indicates that shallower areas are being more affected by environmental disturbance, which matches global trends (De Bakker et al. \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2016\u003c/span\u003e; Baird et al. \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Frade et al. \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e2018\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eShifts in growth form composition and bleaching are greatest at the crest of the reefs we collected data from in Kimbe Bay, and while trends for coral cover on the crest were not predictive of those at depth, the major disturbances to coral reefs seem to be impacting the crest the most. UAVs can accurately and efficiently collect data on coral cover, bleaching, and growth form, and have the potential to greatly increase the spatial and temporal scales at which we can collect data, allowing for a better understanding of how coral reefs are changing. With coral bleaching events projected to increase in frequency and severity (Donner et al. \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2005\u003c/span\u003e; Mellin et al. \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e2024\u003c/span\u003e), it is crucial to have accessible and reliable tools to evaluate change and make conservation decisions to protect what we have and facilitate recovery where possible.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003eThe authors declare that they have no conflict of interest.\u003c/p\u003e\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eConceptualization: L.H. and T.R. Data curation: L.H. Formal analysis: L.H. and T.R. Funding acquisition: L.H. Investigation: L.H. and T.R. Methodology: L.H. and T.R. Project administration: L.H. Resources: L.H. Supervision: L.H. and T.R. Validation: L.H. and T.R. Visualization: L.H. Writing\u0026mdash;original draft: L.H. Writing\u0026mdash;review and editing: L.H. and T.R.\u003c/p\u003e\u003ch2\u003eAcknowledgement\u003c/h2\u003e\u003cp\u003eWe want to thank Mahonia Na Dari and its staff, especially S. Jonda, for providing facilities, knowledge, and support leading up to and throughout the completion of this project. We want to thank Mahonia Na Dari captains R. Martin and B. Mautu for their knowledge and support in navigating us to field sites, and the Tamare-Kilu communities for access to their reefs. We want to thank Walindi Plantation Resort and its staff for their logistical support and knowledge. We want to thank everyone who assisted with field work, including F. Noble, M. Giru, L. Yllan, and Y. Kobayashi. We acknowledge the use of AI-assisted technologies, specifically OpenAI\u0026rsquo;s ChatGPT, which assisted with coding and debugging during statistical analysis. Funding for this project was provided by the United States Fulbright Program.\u003c/p\u003e\u003ch2\u003eData Availability\u003c/h2\u003e\u003cp\u003eWe are in the process of making our data and R-script available; they will be made available by publication.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n \u003cli\u003eBaird AH, Madin JS, \u0026Aacute;lvarez-Noriega M, Fontoura L, Kerry JT, Kuo C-Y, Precoda K, Torres-Pulliza D, Woods RM, Zawada KJA, Hughes TP (2018) A decline in bleaching suggests that depth can provide a refuge from global warming in most coral taxa. Mar Ecol Prog Ser 603:257\u0026ndash;264\u003c/li\u003e\n \u003cli\u003eBanha TNS, Capel KCC, Kitahara MV, Francini-Filho RB, Francini CLB, Sumida PYG, Mies M (2020) Low coral mortality during the most intense bleaching event ever recorded in subtropical Southwestern Atlantic reefs. Coral Reefs 39:515\u0026ndash;521\u003c/li\u003e\n \u003cli\u003eBrodie J, Turak E (2004) Land use practices in the Stettin Bay catchment area and their relation to the status of the coral reefs in Kimbe Bay.\u003c/li\u003e\n \u003cli\u003eBrooks ME, Kristensen K, Benthem JJ van, Magnusson A, Berg CW, Nielsen A, Skaug HJ, Maechler M, Bolker BM (2017) glmmTMB Balances Speed and Flexibility Among Packages for Zero-inflated Generalized Linear Mixed Modeling. R J 9:378\u0026ndash;400\u003c/li\u003e\n \u003cli\u003eBrowne N, Braoun C, McIlwain J, Nagarajan R, Zinke J (2019) Borneo coral reefs subject to high sediment loads show evidence of resilience to various environmental stressors. PeerJ 7:e7382\u003c/li\u003e\n \u003cli\u003eBuckingham MC, D\u0026rsquo;Angelo C, Chalk TB, Foster GL, Johnson KG, Connelly Z, Olla C, Saeed M, Wiedenmann J (2022) Impact of nitrogen (N) and phosphorus (P) enrichment and skewed N:P stoichiometry on the skeletal formation and microstructure of symbiotic reef corals. Coral Reefs Online 41:1147\u0026ndash;1159\u003c/li\u003e\n \u003cli\u003eBun YA, King T, Shearman P (2004) China\u0026rsquo;s impact on Papua New Guinea\u0026rsquo;s forestry industry. Forest Trends, Washington, D.C.\u003c/li\u003e\n \u003cli\u003eByrne M, Waller A, Clements M, Kelly AS, Kingsford MJ, Liu B, Reymond CE, Vila-Concejo A, Webb M, Whitton K, Foo SA (2025) Catastrophic bleaching in protected reefs of the Southern Great Barrier Reef. Limnol Oceanogr Lett 10:340\u0026ndash;348\u003c/li\u003e\n \u003cli\u003eCarriger JF, Yee SH, Fisher WS (2021) Assessing Coral Reef Condition Indicators for Local and Global Stressors Using Bayesian Networks. Integr Environ Assess Manag 17:165\u0026ndash;187\u003c/li\u003e\n \u003cli\u003eCasella E, Lewin P, Ghilardi M, Rovere A, Bejarano S (2022) Assessing the relative accuracy of coral heights reconstructed from drones and structure from motion photogrammetry on coral reefs. Coral Reefs 41:869\u0026ndash;875\u003c/li\u003e\n \u003cli\u003eCornet VJ, Joyce KE (2021) Assessing the Potential of Remotely-Sensed Drone Spectroscopy to Determine Live Coral Cover on Heron Reef. Drones 5:29\u003c/li\u003e\n \u003cli\u003eDanovaro R, Aronson J, Bianchelli S, Bostr\u0026ouml;m C, Chen W, Cimino R, Corinaldesi C, Cortina-Segarra J, D\u0026rsquo;Ambrosio P, Gambi C, Garrabou J, Giorgetti A, Grehan A, Hannachi A, Mangialajo L, Morato T, Orfanidis S, Papadopoulou N, Ramirez-Llodra E, Smith CJ, Snelgrove P, van de Koppel J, van Tatenhove J, Fraschetti S (2025) Assessing the success of marine ecosystem restoration using meta-analysis. Nat Commun 16:3062\u003c/li\u003e\n \u003cli\u003eDe Bakker DM, Meesters EH, Bak RPM, Nieuwland G, Van Duyl FC (2016) Long-term Shifts in Coral Communities On Shallow to Deep Reef Slopes of Cura\u0026ccedil;ao and Bonaire: Are There Any Winners? Front Mar Sci 3:247\u003c/li\u003e\n \u003cli\u003eDepczynski M, Gilmour JP, Ridgway T, Barnes H, Heyward AJ, Holmes TH, Moore J a. Y, Radford BT, Thomson DP, Tinkler P, Wilson SK (2013) Bleaching, coral mortality and subsequent survivorship on a West Australian fringing reef. Coral Reefs 32:233\u0026ndash;238\u003c/li\u003e\n \u003cli\u003eDevoto S, Macovaz V, Mantovani M, Soldati M, Furlani S, Devoto S, Macovaz V, Mantovani M, Soldati M, Furlani S (2020) Advantages of Using UAV Digital Photogrammetry in the Study of Slow-Moving Coastal Landslides. Remote Sens 12:3566\u003c/li\u003e\n \u003cli\u003eDi Marco M, Chapman S, Althor G, Kearney S, Besancon C, Butt N, Maina JM, Possingham HP, Rogalla von Bieberstein K, Venter O, Watson JEM (2017) Changing trends and persisting biases in three decades of conservation science. Glob Ecol Conserv 10:32\u0026ndash;42\u003c/li\u003e\n \u003cli\u003eDonner SD, Skirving WJ, Little CM, Oppenheimer M, Hoegh-Guldberg O (2005) Global assessment of coral bleaching and required rates of adaptation under climate change. Glob Change Biol 11:2251\u0026ndash;2265\u003c/li\u003e\n \u003cli\u003eFalter JL, Zhang Z, Lowe RJ, McGregor F, Keesing J, McCulloch MT (2014) Assessing the drivers of spatial variation in thermal forcing across a nearshore reef system and implications for coral bleaching. Limnol Oceanogr 59:1241\u0026ndash;1255\u003c/li\u003e\n \u003cli\u003eFisher R, Radford BT, Knowlton N, Brainard RE, Michaelis FB, Caley MJ (2011) Global mismatch between research effort and conservation needs of tropical coral reefs. Conserv Lett 4:64\u0026ndash;72\u003c/li\u003e\n \u003cli\u003eFrade PR, Bongaerts P, Englebert N, Rogers A, Gonzalez-Rivero M, Hoegh-Guldberg O (2018) Deep reefs of the Great Barrier Reef offer limited thermal refuge during mass coral bleaching. Nat Commun 9:3447\u003c/li\u003e\n \u003cli\u003eGalbraith GF, Cresswell BJ, McCormick MI, Bridge TC, Jones GP (2022) Contrasting hydrodynamic regimes of submerged pinnacle and emergent coral reefs. PLOS ONE 17:e0273092\u003c/li\u003e\n \u003cli\u003eGonz\u0026aacute;lez-Barrios FJ, Keith SA, Emslie MJ, Ceccarelli DM, Williams GJ, Graham NAJ (2025) Emergent patterns of reef fish diversity correlate with coral assemblage shifts along the Great Barrier Reef. Nat Commun 16:303\u003c/li\u003e\n \u003cli\u003eGouezo M, Fabricius K, Harrison P, Golbuu Y, Doropoulos C (2021) Optimizing coral reef recovery with context-specific management actions at prioritized reefs. J Environ Manage 295:113209\u003c/li\u003e\n \u003cli\u003eHedley JD, Roelfsema CM, Chollett I, Harborne AR, Heron SF, Weeks S, Skirving WJ, Strong AE, Eakin CM, Christensen TRL, Ticzon V, Bejarano S, Mumby PJ (2016) Remote Sensing of Coral Reefs for Monitoring and Management: A Review. Remote Sens 8:118\u003c/li\u003e\n \u003cli\u003eHughes BB, Beas-Luna R, Barner AK, Brewitt K, Brumbaugh DR, Cerny-Chipman EB, Close SL, Coblentz KE, de Nesnera KL, Drobnitch ST, Figurski JD, Focht B, Friedman M, Freiwald J, Heady KK, Heady WN, Hettinger A, Johnson A, Karr KA, Mahoney B, Moritsch MM, Osterback A-MK, Reimer J, Robinson J, Rohrer T, Rose JM, Sabal M, Segui LM, Shen C, Sullivan J, Zuercher R, Raimondi PT, Menge BA, Grorud-Colvert K, Novak M, Carr MH (2017) Long-Term Studies Contribute Disproportionately to Ecology and Policy. BioScience 67:271\u0026ndash;281\u003c/li\u003e\n \u003cli\u003eHughes TP, Kerry JT, Baird AH, Connolly SR, Dietzel A, Eakin CM, Heron SF, Hoey AS, Hoogenboom MO, Liu G, McWilliam MJ, Pears RJ, Pratchett MS, Skirving WJ, Stella JS, Torda G (2018) Global warming transforms coral reef assemblages. Nature 556:492\u0026ndash;496\u003c/li\u003e\n \u003cli\u003eJones GP, McCormick MI, Srinivasan M, Eagle JV (2004) Coral decline threatens fish biodiversity in marine reserves. Proc Natl Acad Sci 101:8251\u0026ndash;8253\u003c/li\u003e\n \u003cli\u003eKohler KE, Gill SM (2006) Coral Point Count with Excel extensions (CPCe): A Visual Basic program for the determination of coral and substrate coverage using random point count methodology. Comput Geosci 32:1259\u0026ndash;1269\u003c/li\u003e\n \u003cli\u003eKomyakova V, Munday PL, Jones GP (2013) Relative Importance of Coral Cover, Habitat Complexity and Diversity in Determining the Structure of Reef Fish Communities. PLOS ONE 8:e83178\u003c/li\u003e\n \u003cli\u003eKooistra L, Berger K, Brede B, Graf LV, Aasen H, Roujean J-L, Machwitz M, Schlerf M, Atzberger C, Prikaziuk E, Ganeva D, Tomelleri E, Croft H, Reyes Mu\u0026ntilde;oz P, Garcia Millan V, Darvishzadeh R, Koren G, Herrmann I, Rozenstein O, Belda S, Rautiainen M, Rune Karlsen S, Figueira Silva C, Cerasoli S, Pierre J, Tanır Kayık\u0026ccedil;ı E, Halabuk A, Tunc Gormus E, Fluit F, Cai Z, Kycko M, Udelhoven T, Verrelst J (2024) Reviews and syntheses: Remotely sensed optical time series for monitoring vegetation productivity. Biogeosciences 21:473\u0026ndash;511\u003c/li\u003e\n \u003cli\u003eLenth RV (2025) emmeans: Estimated Marginal Means, aka Least-Squares Means. https://CRAN.R-project.org/package=emmeans\u003c/li\u003e\n \u003cli\u003eLevy J, Hunter C, Lukacazyk T, Franklin EC (2018) Assessing the spatial distribution of coral bleaching using small unmanned aerial systems. Coral Reefs 37:373\u0026ndash;387\u003c/li\u003e\n \u003cli\u003eLindenmayer DB, Lavery T, Scheele BC (2022) Why We Need to Invest in Large-Scale, Long-Term Monitoring Programs in Landscape Ecology and Conservation Biology. Curr Landsc Ecol Rep 7:137\u0026ndash;146\u003c/li\u003e\n \u003cli\u003eLudecke D, Ben-Shachar MS, Patil I, Waggoner P, Makowski D (2021) performance: An R Package for Assessment, Comparison and Testing of Statistical Models. J Open Source Softw 6:3139\u003c/li\u003e\n \u003cli\u003eManfreda S, McCabe MF, Miller PE, Lucas R, Madrigal VP, Mallinis G, Dor EB, Helman D, Estes L, Ciraolo G, M\u0026uuml;llerov\u0026aacute; J, Tauro F, Lima MID, Lima JLMPD, Maltese A, Frances F, Caylor K, Kohv M, Perks M, Ruiz-P\u0026eacute;rez G, Su Z, Vico G, Toth B, Manfreda S, McCabe MF, Miller PE, Lucas R, Madrigal VP, Mallinis G, Dor EB, Helman D, Estes L, Ciraolo G, M\u0026uuml;llerov\u0026aacute; J, Tauro F, Lima MID, Lima JLMPD, Maltese A, Frances F, Caylor K, Kohv M, Perks M, Ruiz-P\u0026eacute;rez G, Su Z, Vico G, Toth B (2018) On the Use of Unmanned Aerial Systems for Environmental Monitoring. Remote Sens 10:641\u003c/li\u003e\n \u003cli\u003eMatthews SA, Williamson DH, Beeden R, Emslie MJ, Abom RTM, Beard D, Bonin M, Bray P, Campili AR, Ceccarelli DM, Fernandes L, Fletcher CS, Godoy D, Hemingson CR, Jonker MJ, Lang BJ, Morris S, Mosquera E, Phillips GL, Sinclair-Taylor TH, Taylor S, Tracey D, Wilmes JC, Quincey R (2024) Protecting Great Barrier Reef resilience through effective management of crown-of-thorns starfish outbreaks. PLOS ONE 19:e0298073\u003c/li\u003e\n \u003cli\u003eMcCowan DM, Pratchett MS, Baird AH (2012) Bleaching susceptibility and mortality among corals with differing growth forms.\u003c/li\u003e\n \u003cli\u003eMcWilliam M, Hoogenboom MO, Baird AH, Kuo C-Y, Madin JS, Hughes TP (2018) Biogeographical disparity in the functional diversity and redundancy of corals. Proc Natl Acad Sci 115:3084\u0026ndash;3089\u003c/li\u003e\n \u003cli\u003eMellin C, Brown S, Cantin N, Klein-Salas E, Mouillot D, Heron SF, Fordham DA (2024) Cumulative risk of future bleaching for the world\u0026rsquo;s coral reefs. Sci Adv 10:eadn9660\u003c/li\u003e\n \u003cli\u003eMontambault JR, Wongbusarakum S, Leberer T, Joseph E, Andrew W, Castro F, Nevitt B, Golbuu Y, Oldiais NW, Groves CR, Kostka W, Houk P (2015) Use of monitoring data to support conservation management and policy decisions in Micronesia. Conserv Biol 29:1279\u0026ndash;1289\u003c/li\u003e\n \u003cli\u003eMuir PR, Marshall PA, Abdulla A, Aguirre JD (2017) Species identity and depth predict bleaching severity in reef-building corals: shall the deep inherit the reef? Proc R Soc B Biol Sci 284:20171551\u003c/li\u003e\n \u003cli\u003eMumby PJ, Harborne AR (2010) Marine Reserves Enhance the Recovery of Corals on Caribbean Reefs. PLoS ONE 5:e8657\u003c/li\u003e\n \u003cli\u003eObura DO, Aeby G, Amornthammarong N, Appeltans W, Bax N, Bishop J, Brainard RE, Chan S, Fletcher P, Gordon TAC, Gramer L, Gudka M, Halas J, Hendee J, Hodgson G, Huang D, Jankulak M, Jones A, Kimura T, Levy J, Miloslavich P, Chou LM, Muller-Karger F, Osuka K, Samoilys M, Simpson SD, Tun K, Wongbusarakum S (2019) Coral Reef Monitoring, Reef Assessment Technologies, and Ecosystem-Based Management. Front Mar Sci 6:580\u003c/li\u003e\n \u003cli\u003eOksanen J, Simpson GL, Blanchet FG, Kindt R, Legendre P, Minchin PR, O\u0026rsquo;Hara RB, Solymos P, Stevens MHH, Szoecs E, Wagner H, Barbour M, Bedward M, Bolker B, Borcard D, Borman T, Carvalho G, Chirico M, Caceres MD, Durand S, Evangelista HBA, FitzJohn R, Friendly M, Furneaux B, Hannigan G, Hill MO, Lahti L, Martino C, McGlinn D, Ouellette M-H, Cunha ER, Smith T, Stier A, Braak CJFT, Weedon J (2025) vegan: Community Ecology Package. https://cran.r-project.org/web/packages/vegan/index.html\u003c/li\u003e\n \u003cli\u003eParsons M, Bratanov D, Gaston KJ, Gonzalez F (2018) UAVs, Hyperspectral Remote Sensing, and Machine Learning Revolutionizing Reef Monitoring. Sensors 18:2026\u003c/li\u003e\n \u003cli\u003eR Core Team (2025) R: A Language and Environment for Statistical Computing.\u003c/li\u003e\n \u003cli\u003eRogers C (1990) Responses of coral reefs and reef organisms to sedimentation. Mar Ecol Prog Ser 62:185\u0026ndash;202\u003c/li\u003e\n \u003cli\u003eSalam\u0026iacute; E, Barrado C, Pastor E, Salam\u0026iacute; E, Barrado C, Pastor E (2014) UAV Flight Experiments Applied to the Remote Sensing of Vegetated Areas. Remote Sens 6:11051\u0026ndash;11081\u003c/li\u003e\n \u003cli\u003eSteinberg CR, Choukroun SM, Slivkoff MM, Mahoney MV, Brinkman RM (2006) Currents in the Bismarck Sea and Kimbe Bay.\u003c/li\u003e\n \u003cli\u003eTang L, Shao G (2015) Drone remote sensing for forestry research and practices. J For Res 26:791\u0026ndash;797\u003c/li\u003e\n \u003cli\u003eThomas L, Underwood JN, Rose NH, Fuller ZL, Richards ZT, Dugal L, Grimaldi CM, Cooke IR, Palumbi SR, Gilmour JP (2022) Spatially varying selection between habitats drives physiological shifts and local adaptation in a broadcast spawning coral on a remote atoll in Western Australia. Sci Adv 8:eabl9185\u003c/li\u003e\n \u003cli\u003eTurak E, DeVantier L, Szava-Kovats R, Brodie J (2021) Impacts of coastal land use change in the wet tropics on nearshore coral reefs: Case studies from Papua New Guinea. Mar Pollut Bull 168:112445\u003c/li\u003e\n \u003cli\u003eTurner W, Spector S, Gardiner N, Fladeland M, Sterling E, Steininger M (2003) Remote sensing for biodiversity science and conservation. Trends Ecol Evol 18:306\u0026ndash;314\u003c/li\u003e\n \u003cli\u003eTuttle LJ, Donahue MJ (2022) Effects of sediment exposure on corals: a systematic review of experimental studies. Environ Evid 11:4\u003c/li\u003e\n \u003cli\u003eUllah H, Wahab MA, Rahman MJ, Al Mamun SN, Kumar U, Rahman MA, Souhardya SM, Kabir IE, Hussain M, Rahman MB, Chishty SMSUH (2023) Local ecological knowledge can support improved management of small-scale fisheries in the Bay of Bengal. Front Mar Sci 10:974591\u003c/li\u003e\n \u003cli\u003eWebber K, Srinivasan M, Coppock AG, Jones GP (2022) Spatial patterns in the cover and composition of macroalgal assemblages on fringing and nearshore coral reefs. Mar Freshw Res 73:1310\u0026ndash;1322\u003c/li\u003e\n \u003cli\u003eWong JSY, Chan YKS, Ng CSL, Tun KPP, Darling ES, Huang D (2018) Comparing patterns of taxonomic, functional and phylogenetic diversity in reef coral communities. Coral Reefs 37:737\u0026ndash;750 (2004) Reef Check Instruction Manual: A Guide to Reef Check Coral Reef Monitoring. Reef Check, Inst. of the Environment, Los Angeles [Calif.]\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"coral-reefs","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"core","sideBox":"Learn more about [Coral Reefs](http://link.springer.com/journal/338)","snPcode":"338","submissionUrl":"https://submission.nature.com/new-submission/338/3","title":"Coral Reefs","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"unmanned aerial vehicle, remote sensing, coral triangle, global mass bleaching, depth gradient","lastPublishedDoi":"10.21203/rs.3.rs-8565286/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8565286/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eAssessing the impacts of rapid environmental change on coral reefs is hindered by a major discrepancy between the regions with the greatest need and those that receive the most research funding. Remote sensing, particularly through the use of drones (or unmanned aerial vehicles (UAVs)), has the potential to significantly enhance the accessibility and efficiency of high-quality data collection in remote, biodiverse areas such as the Coral Triangle. To test this, we compared a UAV-based method, a common citizen science method (Reef Check), and a research-grade method (photo quadrats) to assess hard coral cover, coral bleaching, and coral growth forms across reef types (nearshore, fringing, midshelf) and depths (crest, 5 m, 10 m, 15 m) in Kimbe Bay, Papua New Guinea, during the fourth Global Mass Coral Bleaching Event. We found that, compared to in-water methods, the UAV delivered accurate reef health data for shallow reef crests of all reef types, including hard coral cover, bleached hard coral cover, and coral growth forms. Hard coral cover did not differ significantly between the crest and the other depths. However, bleaching was most significant at the reef crest and decreased significantly at deeper parts of the reef slope. Coral growth form composition varied significantly between reef types, but the scale of these differences decreased with depth. Our study demonstrates that UAVs can provide accurate health and community composition data for reefs with high biodiversity, significantly enhancing the availability of high-quality reef health data in the areas of highest need.\u003c/p\u003e","manuscriptTitle":"Drone Imaging can Accurately Assess Coral Cover, Bleaching, and Growth Form for Shallow Coral Reefs","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-01-29 10:39:26","doi":"10.21203/rs.3.rs-8565286/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2026-02-16T01:52:28+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-02-11T06:19:07+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"190171362462000996281361406666891532459","date":"2026-01-29T23:28:00+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2026-01-27T23:25:48+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-01-27T23:22:34+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-01-14T08:16:10+00:00","index":"","fulltext":""},{"type":"submitted","content":"Coral Reefs","date":"2026-01-10T03:27:19+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"coral-reefs","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"core","sideBox":"Learn more about [Coral Reefs](http://link.springer.com/journal/338)","snPcode":"338","submissionUrl":"https://submission.nature.com/new-submission/338/3","title":"Coral Reefs","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"9fe99443-6b65-4949-9ae7-df01684d60e8","owner":[],"postedDate":"January 29th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2026-04-13T16:03:30+00:00","versionOfRecord":{"articleIdentity":"rs-8565286","link":"https://doi.org/10.1007/s00338-026-02864-5","journal":{"identity":"coral-reefs","isVorOnly":false,"title":"Coral Reefs"},"publishedOn":"2026-04-09 15:59:27","publishedOnDateReadable":"April 9th, 2026"},"versionCreatedAt":"2026-01-29 10:39:26","video":"","vorDoi":"10.1007/s00338-026-02864-5","vorDoiUrl":"https://doi.org/10.1007/s00338-026-02864-5","workflowStages":[]},"version":"v1","identity":"rs-8565286","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8565286","identity":"rs-8565286","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}