Name: Measuring the Structure, Composition, and Change of Underwater Environments with Large-area Imaging (Video) | JoVE Journal
Uploaded: 2025-04-18T13:30:00.000+00:00
Duration: 9 min 19 s
Description: Learn a four-step large-area imaging protocol for coral reef ecology using underwater camera systems. This method enables detailed structural complexity, community composition, and demographic analyses of benthic marine ecosystems.

ai_agent_IP

ai_agent

maintenance

instantTrialDurationMinutes

isBiopharma

enableNewMyJoveSectionDesign

educationFeatures

seoMetaApi

ai_copilot_institution_blacklist

ai_chat_institution_blacklist

freeAccessWindow

ssoRoleCapture

scheduledMaintenanceBanner

joveLabsFeatures

disable_autoplay_for_bots

extended_trial_duration

5 minutes

This work is supported by the efforts of the 100 Island Challenge at Scripps Institution of Oceanography. We thank Schmidt Marine Technology Partners, Ed and Christy Scripps, and the Moore Family Foundation for their financial support of the associated research and large-area imaging training efforts which have helped to refine the methodology. In addition, we thank the crew of research vessels Hi&#39;ialiki, Hanse Explorer, and Plan B, who provided support for field efforts. We particularly thank the team at the Ecosystem Sciences Division of the Pacific Islands Fisheries Sciences Center at NOAA who assisted with field collection of the data presented here.

The authors have no conflicts of interest to disclose.

Large-area imaging is a tool enabling domain scientists to digitally visualize and analyze features of the environment at scales larger than that of the individual images collected. By capturing multiple images of the environment from multiple perspectives, LAI protocols help to create representations of relatively broad landscapes (relative to the spatial coverage of individual images) while maintaining the detail collected from the original imagery. The unique value of LAI lies in the ability to explore environmental data across scales, from the largest scale (defined by the areal extent of the survey) to the finest scale (defined by the realized resolution of the original imagery). However, to capitalize on this cross-scale strength, it is critical to assure regular and fluid access to all levels of the data captured, specifically in assuring facile access to both the original imagery and the derived 3D model. In each step of the protocol presented here, we highlight this unique strength of LAI by consistently assuring that the original imagery is accessible, usable, and securely archived alongside the derived LAI models.
The LAI method will deliver products that are linked directly to the original imagery collected. By shifting details of image acquisition, users can produce data products of differing quality and coverage. When surveying structurally complex coral reef environments, a user with limited survey time underwater (or a constant number of images available to be captured) may prioritize increasing the areal coverage of the survey area or increasing the level of detail of each section of the area sampled. There will necessarily be a tradeoff, with the large areal model having less detail (and perhaps more occlusions) per unit area and the detailed model covering less total area (with likely fewer occlusions). In this protocol, we include the use of two cameras, each with different lenses, which enables a user to sample a larger area (sufficient overlap with wide angle lens to fulfill SfM requirements) while simultaneously collecting more detailed original imagery (higher detail from zoom lens which has less photo-to-photo overlap). By including pose estimation of imagery from both cameras, the downstream visualization and analysis protocols include higher-resolution views from much of the sampled area. Although the protocol aims to expand the range of survey viability, users may find that the derived products lack sufficient areal coverage or sufficient detail of original images to complete preferred analytical routines. Users are encouraged to review the original imagery and the derived models to ensure that the environmental survey protocol meets program needs and to modify the in-water survey approach (e.g., shifting ground sampling distance, altering survey duration or number of images collected) to arrive at the preferred balance of areal coverage and resolution per unit area.
LAI methods provide value to underwater science by capturing information-rich, wide-coverage &#39;snapshots&#39; of benthic environments that can be time-efficient and require limited domain-specific expertise for collection. The value of these data products can be considered with reference to existing data streams as well as for novel and accelerated domain-specific applications. Considering the comparison with existing data streams, analysis products from LAI can provide ecological data that are directly comparable to data collected in situ by underwater observers84,85,86,87. We provide here a quantitative analysis of ecological data outputs derived from each, classical in situ monitoring activities and from standardized analysis of LAI products, following this protocol. By focusing on four common metrics in coral reef monitoring efforts (structural complexity, benthic community composition, density of mobile invertebrates, and coral size structure; Figure 5), we demonstrate strong quantitative concordance in data outputs. Notably, for the data streams that require fine-scale observations (e.g., taxonomic identifications, precise definitions of biological borders), LAI workflows that include regular and reliable access to original imagery provide unique strength to parallel the observational opportunities that have historically been limited only to immersive, in situ sampling. Advances in data handling and visualization provided by Viscore and described in this protocol offer unique value in assuring the comparability of LAI-derived ecological data and in situ monitoring products, allowing for unbiased maintenance of long-term data streams with the incorporation of digitally enhanced workflows of LAI.
For underwater scientists, LAI offers the opportunity to leverage novel and accelerated workflows in data acquisition and exploration. LAI has unique strengths as a high-resolution mapping tool that contains information not just describing the relative composition of the benthic environment but also spatial characteristics. Explicit to the creation of a 3D model from the SfM workflow,&#160;LAI products contain information about the structural complexity which can be explored across multiple scales56. As tools capturing larger-area seascapes, LAI products can provide an opportunity to consider patterns of spatial distribution and neighborhood characteristics for benthic organisms45,66. Moreover, by being able to visualize benthic landscapes at larger scales, it becomes possible to detect features that are not readily visible in underwater habitats due to the limitations of long-distance visibility underwater, for example large-scale (3-4 m) polygonal patterning of a common macroalga on a Pacific coral reef64.
While LAI provides opportunities for large-scale analysis, there have been stated concerns regarding the challenges linked with efficient in-water collection and image post-processing. Expanding the spatial extent of image acquisition underwater will demand technological advances away from diver-deployed imaging, to the utilization of assisted mapping systems88, and ultimately the use of ROVs38 and AUVs67,89. Robust survey methodologies in regard to the acquisition pattern and camera systems will ensure a smooth transition and consistency in the data generated across these different platforms. Given the computational demands as well as the large data format of LAI products, some marine scientists have expressed concerns regarding technological accessibility of the workflow90 and the large time demands linked with ecological data extraction84,86. A growing number of tools are being presented, however, that leverage creative solutions to the technical hurdles of data extraction80,91,92,93. Importantly, the use of AI-enhanced workflows of LAI analysis is limited by the quality of the input signal provided. As such, there remains a consistent demand in maintaining standards and quality of image acquisition and data management in LAI protocols regardless of whether the data extraction is conducted by a human observer, by a trained AI algorithm, or (ideally) by an AI-accelerated, human-in-the-loop workflow. By maintaining a consistent focus on the primary importance of the original imagery in the LAI protocols, as described here, unique opportunities emerge to explore underwater habitats robustly, transparently, and consistently.

Across terrestrial environments, researchers have taken advantage of standardized large-area sampling of ecological communities, particularly in the context of long-term study sites, including Barro Colorado Island1, the Hubbard Brook experimental forest2, and others3. Through the collection of spatially explicit and taxonomically resolved distributional data, such sampling has been used to explore fundamental ecological dynamics, such as dispersion and recruitment patterns3,4,5, habitat preference and availability, dispersal kernels, resource limitation3,5,6,7,8, and space use9,10. However, to date, most spatial studies of marine communities have relied on metrics of relative cover, reported as percent cover occupied by taxon or group11,12,13,14,15. Aggregated estimates of relative cover, however, are insufficient to resolve details of population-level demography as well as community-level dynamics. Studies that have provided detailed analyses of benthic communities have relied upon laborious in-water monitoring protocols16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32, but the scale (including taxonomic, spatial, and chronological scales) of these studies is notably limited due to the operational demands of in-water methodology.
Large-area imaging (LAI) is an approach that combines information from numerous images through computationally intensive workflows to create photo-realistic representations of environments at scales much larger than that of the constituent images33. The LAI workflow is particularly well-suited to applications in underwater habitats given the limited visibility due to light absorption and scattering in water. Because of the limited visibility, imagery that captures fine details of the benthos must be acquired near to the subject; to capture a landscape (or seascape) view of a broad swath of benthic habitat while retaining fine detail of individual benthic subjects thus requires composite imaging. Further, in structurally complex environments, it is essential to account for the three-dimensional (3D) structure in reconstructing the composite imaging to produce faithful representations of the position and relative proximity of benthic organisms. The Structure-from-Motion (SfM) photogrammetric method has been applied to environments with relatively immobile benthic organisms, including coral reefs34,35,36, Antarctic benthic ecosystems37, cold-water coral reefs38, cold seeps39, and seagrass habitat40, generating composite imaging without stereoscopy used to reconstruct a landscape scene with follow-on ortho-map generation and point cloud estimation.
In coral reef science, LAI has offered the potential to visualize reefscapes at increasingly large spatial scales and to share these visualizations across digital media. LAI can be used to estimate the coverage of reef organisms, the density and distribution of coral colonies, as well as the shape and condition of individual organisms41,42,43,44,45,46,47. Further, when LAI products are collected from the same location at different points in time, it is possible to record changes in the size and condition of individual organisms48,49,50,51. Given that most scleractinian coral colonies grow on the order of millimeters to centimeters radially per year, time-series LAI collected across years can provide an invaluable data stream for reporting on the biology and ecology of these species52. Repeated and co-registered LAI data offer unique insights with which to study coral reefs in a format that can be shared, archived, and used as a basis for collaboration worldwide.
As the use of LAI has expanded among coral reef ecologists53, so has the diversity of camera systems and survey methodologies52. A chosen LAI protocol should target the resolution and scope of desired ecological metrics while staying bounded within available resources. The quality of any photogrammetric reconstruction will ultimately be dependent on the resolution of the source imagery and the spatial coverage of the survey area. Image quality is determined by an influence of camera parameters including sensor resolution and focal length, as well as the collection procedure, principally the distance from the benthos54, all of which contribute to the effective Ground Sampling Distance (GSD) of a particular set of images. Additionally, fast shutter speeds, small apertures, and low ISO values will produce images that are sharp, in focus, and have low electronic noise, respectively. Keeping each of these settings at thresholds that produce sufficient quality imagery can be a challenge in underwater environments where light levels are low. Larger sensors, like those found on digital single-lens reflex (DSLR) style and mirrorless cameras, generate better image quality and, in turn, more accurate reconstructions in comparison to smaller, more mobile solutions like action cameras55. Additional features that should not be overlooked when considering an appropriate camera model include a built-in intervalometer and sufficient storage and battery capacity to support extended-duration image collection efforts in the field.
Survey design should be driven by the ecological hypothesis, with the candidate metrics determining necessary resolution and spatial coverage. Within coral reef ecology, LAI has been used to characterize structural complexity35,36,56,57,58,59, community composition and assemblage60,61,62, spatial distribution45,63,64,65,66, and community trajectories48,49,50,67,68,69. The resolution of image quality should be appropriate to ecological data needs, with finer scale resolution on the sub-mm detail necessary to support polyp scale observations of competition along colony borders70 or surveys of small juvenile corals66,71. In contrast, extracting broad-scale habitat and structural metrics for coastal mapping72,73,74 requires a greater spatial extent with a reduced need for resolution at the cm-m scale. The demand for resolution must be balanced with the spatial extent required to obtain sufficient sampling and operational limits of the time needed to complete an LAI survey33.
Described here is an end-to-end protocol for conducting an LAI survey, which focuses on maximizing the quality, utility, and value of the source imagery, breaking down the protocol into four major steps: image collection, model construction, ecological analyses, and data curation33. The collection of approximately 3,500 LAI image surveys of over 2,000 unique reef sites over the past decade has contributed to the refinement of the methodology for each step presented here (https://doi.org/10.6075/J0T43RN1). The resulting protocol is a method for robust data collection and accurate and precise model reconstructions, which enable the collection of detailed ecological data across a broad range of applications, including structural complexity, community composition, and population demographics (e.g., density and size structure). We additionally include metadata standards for archiving LAI data, the establishment of which is essential to assure the preservation, transparency, and collaborative potential of these digital twins.

<table><tbody><tr><td>1&quot; x 4 1/4&quot; x 3 1/4&quot; custom machined acetal</td><td>N/A</td><td>N/A</td><td>1.1 Gear Preparation; For contruction of camera slides and mounts</td></tr><tr><td>1/2&quot; marine grade high density polyethylene</td><td>King Starboard</td><td>N/A</td><td>1.1 Gear Preparation; For contruction of camera outer frame and camera mounting panels</td></tr><tr><td>18-8 Flathead Stainless Steel Phillips Flat Head Screws, 3/8&amp;rdquo;-16 Thread Size, 1-1/2&amp;rdquo; Long</td><td>McMaster-Carr</td><td>91771A628</td><td>1.1 Gear Preparation; For camera frame assembly</td></tr><tr><td>18-8 Stainless Steel Socket Head Screw, 10-24 Thread Size, 1-1/8&quot; Long</td><td>McMaster-Carr</td><td>92196A248</td><td>1.1 Gear Preparation; Used to secure mounting plate to handles</td></tr><tr><td>1 lb dive weights</td><td>House of Scuba</td><td>WBELT24 1LB</td><td>1.2 Plot Setup; Used for reference floats</td></tr><tr><td>200DL Underwater Housing for Nikon D780 DSLR Camera</td><td>Ikelite</td><td>71019</td><td>1.1 Gear Preparation; Underwater housing for digital camera</td></tr><tr><td>24mm fixed lens (AF-S NIKKOR 24mm f/1.8G ED)</td><td>Nikkor</td><td>20057</td><td>1.1 Gear Preparation; Wide-angle lens for greater image overlap</td></tr><tr><td>250# gray longline poly soft</td><td>Continental Western</td><td>503086</td><td>1.2 Plot Setup; To build scale bars. Ground control markers for determining model scale</td></tr><tr><td>3 lb drilling hammer</td><td>Estwing</td><td>B3-3LB</td><td>1.2 Plot Setup; Used to install stainless steel stakes at survey site</td></tr><tr><td>3/8-16 X 18&quot; THRD ROD W/60 DEGREE POINT 316 S/S</td><td>Ababa Bolt</td><td>37C1800ROD6/60DEG</td><td>1.2 Plot Setup; For permanent installation to mark survey site</td></tr><tr><td>316 Stainless Steel Nylon-Insert Locknut Super-Corrosion-Resistant, 1/4&quot;-20 Thread Size</td><td>McMaster-Carr</td><td>90715a125</td><td>1.2 Plot Setup; For scale bars and frame assembly.</td></tr><tr><td>316 Stainless Steel Nylon-Insert Locknut Super-Corrosion-Resistant, 3/8&quot;-16 Thread Size</td><td>McMaster-Carr</td><td>90715A145</td><td>1.2 Plot Setup; Affixed to left stainless steel stake for orientation of the plot</td></tr><tr><td>316 Stainless Steel Phillips Flat Head Screws, 10-32 Thread Size, 3/8&quot; Long</td><td>McMaster-Carr</td><td>91500a827</td><td>1.1 Gear Preparation; For camera frame assembly</td></tr><tr><td>4&quot;x4&quot; Agisoft marker printed on waterproof paper</td><td>Agisoft</td><td>N/A</td><td>1.2 Plot Setup; To build corner tiles. Ground control markers for determining model orientation</td></tr><tr><td>4&quot;x4&quot;x1/4&quot; Stainless steel tile</td><td>N/A</td><td>N/A</td><td>1.2 Plot Setup; To build corner tiles. Ground control markers for determining model orientation</td></tr><tr><td>4&quot;x4&quot;x3/4&quot; custom printed plastic agisoft marker high density polyethylene color core</td><td>N/A</td><td>N/A</td><td>1.2 Plot Setup; To build scale bars. Ground control markers for determining model scale</td></tr><tr><td>512 GB Extreme PRO SDXC UHS-I Card - C10, U3, V30, 4K UHD, SD Card</td><td>SanDisk</td><td>SDSDXXY-512G-GN4IN</td><td>1.1 Gear Preparation; High speed, large capacity storage card. Up to 2 used per camera for image storage</td></tr><tr><td>5TB Elements Portable External Hard Drive HDD, USB 3.0</td><td>Western Digital</td><td>WDBU6Y0050BBK-WESN</td><td>1.3 Image Capture; Large volume external hard drive for image storage and image backup</td></tr><tr><td>60 mm fixed lens (AF-S Micro NIKKOR 60mm F2.8G ED)</td><td>Nikkor</td><td>2177</td><td>1.1 Gear Preparation; Macro zoom lens, optional for dual-camera setup</td></tr><tr><td>Acetal machined and tapped for 1&quot;x12&quot; 3/8&quot; 16 thread support braces</td><td>N/A</td><td>N/A</td><td>1.1 Gear Preparation; Camera frame support columns</td></tr><tr><td>AquaMend Epoxy Putty Stick</td><td>JD Industrial Supply</td><td>470550</td><td>1.2 Plot Setup; Used to install stainless steel stakes at survey site</td></tr><tr><td>Architectural 6063 Aluminum U-Channel, 1/8&quot; Wall Thickness, 1/2&quot; High x 3/4&quot; Wide Outside</td><td>McMaster-Carr</td><td>9001k46</td><td>1.2 Plot Setup; To build scale bars. Ground control markers for determining model scale</td></tr><tr><td>Black-Oxide 18-8 Stainless Steel Pan Head Phillips Screws, 1/4&quot;-20 Thread, 1/2&quot; Long</td><td>McMaster-Carr</td><td>91249a537</td><td>1.1 Gear Preparation; To attach ikelite handle to housing</td></tr><tr><td>Black-Oxide 18-8 Stainless Steel Pan Head Phillips Screws, 1/4&quot;-20 Thread, 5/8&quot; Long</td><td>McMaster-Carr</td><td>91249A539</td><td>1.2 Plot Setup; To build scale bars. Ground control markers for determining model scale</td></tr><tr><td>Blue Steel Rope</td><td>Continental Western</td><td>402020</td><td>1.2 Plot Setup; Used to secure dive float to the benthos during surveys</td></tr><tr><td>D780 camera body</td><td>Nikon</td><td>1618</td><td>1.1 Gear Preparation; Camera body model</td></tr><tr><td>DGX Tech Compass w/Bungee Mount and Cord</td><td>Dive Gear Express</td><td>DX-9050x</td><td>1.2 Plot Setup; For collection of plot bearings and as an addition to the camera frame as a navigational aid</td></tr><tr><td>Dive computer - Suunto Zoop Novo</td><td>Suunto&amp;nbsp;</td><td>N/A</td><td>1.2 Plot Setup; To record depth at reference tiles</td></tr><tr><td>Dive slate</td><td>TexWipe</td><td>TX5835</td><td>1.2 Plot Setup; Used to record plot metadata such as tile depth, and coded target numbers</td></tr><tr><td>DL 8 inch Dome Port</td><td>Ikelite</td><td>75340</td><td>1.1 Gear Preparation; Dome port for underwater housing</td></tr><tr><td>FLOAT, PVC SPONGE, 5-3/4&quot; DIA. BY 3/4&quot;, RUST</td><td>Memphis Net &amp;amp; Twine</td><td>SB1</td><td>1.2 Plot Setup; Used as a visual reference to determine plot boundaries and swim height of camera operator</td></tr><tr><td>Garmin 78s GPS</td><td>Garmin</td><td>010-00864-01</td><td>1.2 Plot Setup; Used to record location of survey site</td></tr><tr><td>High performance computer</td><td>N/A</td><td>N/A</td><td>2.0 Model Construction; For 3D dense point cloud processing, recommended specifications to include a high speed 10+ core CPU, 128GB RAM (64 GB minimum), 1TB solid state drive, and a dedicated NVIDIA or AMD GPU.&amp;nbsp;</td></tr><tr><td>Inflatable surface dive float</td><td>Omer</td><td>Atol 6246</td><td>1.2 Plot Setup; Dual purpose surface marker buoy and&amp;nbsp;</td></tr><tr><td>JOHNSON Cross Check Level: Nonmagnetic, 2 1/4 in x 1 7/16 in x 3/16 in, Plastic, Hanging Hole, 1mm/m</td><td>Grainger</td><td>6C225</td><td>1.1 Gear Preparation; Optional addition to the camera frame as a navigational aid</td></tr><tr><td>Long Tape Measure,1/2 In x 30m,Pumpkin</td><td>Grainger</td><td>3LJW9</td><td>1.2 Plot Setup; Used to set up plot area</td></tr><tr><td>Manta reel SR. Reel</td><td>Manta Industries</td><td>N/A</td><td>1.2 Plot Setup; Attached to dive float for use during surveys</td></tr><tr><td>Metashape Professional License</td><td>Agisoft</td><td>N/A</td><td>2.0 Model Construction; Software for dense point cloud processing</td></tr><tr><td>Non-glare clear acrylic</td><td>N/A</td><td>N/A</td><td>1.2 Plot Setup; To build corner tiles. Ground control markers for determining model orientation</td></tr><tr><td>O-Ring 0132.45 for DL Port System, ULTRAcompact Housings</td><td>Ikelite</td><td>132.45</td><td>1.1 Gear Preparation; O-ring for underwater housing</td></tr><tr><td>O-Ring 0132.59 for DSLR &amp;amp; Mirrorless Housings (Type 1)</td><td>Ikelite</td><td>132.59</td><td>1.1 Gear Preparation; O-ring for underwater housing</td></tr><tr><td>Paracord or Dacron Line</td><td>N/A</td><td>N/A</td><td>1.2 Plot Setup; Used to attach referene floats to dive weights</td></tr><tr><td>Passivated 18-8 Stainless Steel Phillips Flat Head Screw, 82 Degree Countersink, 1/4&quot;-20 Thread, 1&quot; Long</td><td>McMaster-Carr</td><td>91771a542</td><td>1.1 Gear Preparation; Frame slider panel</td></tr><tr><td>Passivated 18-8 Stainless Steel Phillips Flat Head Screw, 82 Degree Countersink, 1/4&quot;-20 Thread, 1-3/4&quot; Long</td><td>McMaster-Carr</td><td>91771a548</td><td>1.1 Gear Preparation; Frame slider panel</td></tr><tr><td>Passivated 18-8 Stainless Steel Phillips Flat Head Screw, 82 Degree Countersink, 10-32 Thread, 3/8&quot; Long</td><td>McMaster-Carr</td><td>91771A827</td><td>1.2 Plot Setup; To build corner tiles. Ground control markers for determining model orientation</td></tr><tr><td>Pelican 1060 micro case</td><td>Pelican</td><td>1060-025-100</td><td>1.2 Plot Setup; Housing for GPS unit that is affixed to the inflatable dive float</td></tr><tr><td>Plastic-Head Thumb screw 1/4&quot;-20 x 2&quot;</td><td>McMaster-Carr</td><td>91185A819</td><td>1.1 Gear Preparation; Use to secure camera housing to frame</td></tr><tr><td>Right Hand Quick Release Handle with Extension</td><td>Ikelite</td><td>4077.02</td><td>1.1 Gear Preparation; Handle for underwater housing</td></tr><tr><td>R</td><td>N/A</td><td>N/A</td><td>3.0 Ecological Analysis; Software for running structural complexity scripts</td></tr><tr><td>Taglab</td><td>N/A</td><td>N/A</td><td>3.0 Ecological Analysis; Software for segmentation analysis</td></tr><tr><td>Trigger Extension v2 for Shutter or Back Button Focus</td><td>Ikelite</td><td>4077.95</td><td>1.1 Gear Preparation; Trigger extenstion for underwater housing shutter button control</td></tr><tr><td>Vacuum pump with gauge</td><td>Ikelite</td><td>47011</td><td>1.1 Gear Preparation; To test integrity of o-ring seals for underwater camera housings</td></tr><tr><td>Viscore</td><td>N/A</td><td>N/A</td><td>2.0 Model Construction; 3.0 Ecological Analysis; Software for ecological analysis</td></tr></tbody></table>

measuring the structure, composition, and change of underwater environments with large-area imaging

Measuring the Structure, Composition, and Change of Underwater Environments with Large-area Imaging

Digital imaging and processing technology have evolved to facilitate the expansion of large-area imaging surveys, which increase our capacity to study the status, trends, and dynamics of organisms living in subtidal habitats. By creating photo-realistic digital twins for ex situ analyses, these approaches allow small field teams to collect substantially more data than was previously possible. Here we present a four-step, large-area imaging survey pipeline and analysis methodology, including image collection, model construction, ecological analysis, and data curation, that has been developed and refined through experimentation over the past decade. Each step described has a consistent focus on the unique value of the original source imagery. While types of data extracted from large-area image surveys are vast, we include here workflows to extract ecological data for structural complexity, community composition, and demographic analyses valuable for monitoring and hypothesis-driven efforts. We additionally include recommendations for metadata standards, which complement the collection of large-area imaging data and support archival efforts facilitating transparency and collaboration between research groups.

Successful large-area image collection in the field should result in the creation of a dense point cloud reconstruction with full top-down coverage of the survey area, while inadequate redundancy in coverage may result in gaps or full degradation of the point cloud (Figure 4). For a set of 43 large-area image surveys conducted across the Hawaiian archipelago in 2016, an average of 99.6% of images were aligned per dataset, with 66% of image sets having 100% of images successfully aligned. Images collected from the wide-angle lens camera had an average GSD of 0.52 mm px-1 while images from the zoom lens camera had an average GSD of 0.18 mm px-1. These surveys generated dense point clouds with an average size of 557.7 million points (15 GB).
The ecological workflows described here were designed to generate data comparable to existing methodologies for coral reef monitoring78. Ecological data extracted from the LAI surveys across Hawaii show, apart from a few outliers, measures of linear rugosity resulting from the structural complexity workflow which align well with relative values compared to in situ measures of complexity across sites81 (Figure 5A,B). Additional community composition analyses from LAI to measure the percent coverage of major benthic functional groups show similar alignment compared with traditional photoquadrat surveys82 (Figure 5C,D). Quadrat sampling here was used to measure the density of sessile invertebrates, the most common being sea urchins, which were summarized as categorical measures of relative abundance. LAI methods regularly recorded higher levels of abundance relative to in situ methods81 (Figure 5E,F), which may be due to the ability to comprehensively search and survey all individuals within a given area compared with a rapid visual census. Segmenting coral colonies using the 2D orthoprojection also revealed similar size distributions of common coral taxa to in situ methods83 (Figure 5G,H).
A major advantage of LAI surveys is the ability to archive and track changes in reef areas across time through coregistration of the dense point cloud. Coral reefs are living substrates, which for LAI surveys means it can be challenging to identify permanent surfaces, either natural or installed, that can be used to reliably coregister dense point clouds from different time points. The example from Millennium atoll (Figure 6) shows an example with both high growth and structural loss where the assisted coregistration workflow was used for dense point cloud coregistration despite little to no stability of the reef surface across time.
<img alt="Camera housing design schematics and prototypes; Ikelite housing setups; top, side views; DSLR fit." class="xfigimg" src="/files/ftp_upload/67700/67700fig01.jpg" data-altupdatedat="1788304762000" data-alt="Figure 1"> 
Figure 1: Diagram of assembled camera frames. Example views of the (A,B) dual and (C,D) single camera frame setups. Ikelite camera housings are attached to the frame on the camera mounting panel using a slider plate attached to the housing handles. Optional instruments to aid in navigation such as a level, compass, and dive computer can be attached to the frame as shown in B. <a href="https://www.jove.com/files/ftp_upload/67700/67700fig01large.jpg" target="_blank">Please click here to view a larger version of this figure.</a>
<img alt="Mosaic plot setup diagram with GPS points, depth tiles, scale bars, and stainless stakes." class="xfigimg" src="/files/ftp_upload/67700/67700fig02.jpg" data-altupdatedat="1788304762000" data-alt="Figure 2"> 
Figure 2: Diagram of set up 100 m2 large-area image plot. Schematic of a fully set up large-area image plot 100 m2 in area. Temporary plot markers include six boundary tile markers, four scale bars, and four reference floats. Permanent plot markers include two stainless steel stakes with the left stake, when looking onshore, including a lock nut. The GPS reference for the plot should be taken above the left-center tile or stake. <a href="https://www.jove.com/files/ftp_upload/67700/67700fig02large.jpg" target="_blank">Please click here to view a larger version of this figure.</a>
<img alt="Seafloor mapping diagram with grid and diver setup, illustrating underwater survey method." class="xfigimg" src="/files/ftp_upload/67700/67700fig03.jpg" data-altupdatedat="1788304762000" data-alt="Figure 3"> 
Figure 3: Diver collection pattern. Example of (A) swim path of diver to ensure sufficient coverage and overlap of the plot area with a buffer included and (B) arrangement of the diver with the camera system with the reference floats providing a guide for the swim height. <a href="https://www.jove.com/files/ftp_upload/67700/67700fig03large.jpg" target="_blank">Please click here to view a larger version of this figure.</a>
<img alt="Coral bleaching stages; sequential diagram showing coral degradation process in oceanic environment." class="xfigimg" src="/files/ftp_upload/67700/67700fig04.jpg" data-altupdatedat="1788304762000" data-alt="Figure 4"> 
Figure 4: Reduction in image overlap. The degradation of the dense point cloud as the number of overlapping images is reduced through random sampling. Each panel shows the dense point cloud generated from (A) all images, (B) ½, (C) 1/3, (D) 1/5, and (E) 1/10 of the original images. <a href="https://www.jove.com/files/ftp_upload/67700/67700fig04large.jpg" target="_blank">Please click here to view a larger version of this figure.</a>
<img alt="Coral reef analysis with 3D modeling, mean roughness index graph, percent cover, and urchin abundance." class="xfigimg" src="/files/ftp_upload/67700/67700fig5v2.jpg" data-altupdatedat="1788304762000" data-alt="Figure 5"> 
Figure 5: Ecological data extraction. Candidate data extracted from large-area imaging for each of the ecological workflows are shown in comparison to established in situ data collection methods. This includes (A,B) structural complexity, (C,D) community composition (error bars indicate standard error), (E,F) invertebrate density, and (G,H) size structure demographics. Abbreviation: LAI = large-area imaging. <a href="https://www.jove.com/files/ftp_upload/67700/67700fig5v2large.jpg" target="_blank">Please click here to view a larger version of this figure.</a>
<img alt="Coral reef aerial image, highlighting growth patterns, ecological survey, marine biodiversity analysis." class="xfigimg" src="/files/ftp_upload/67700/67700fig06.jpg" data-altupdatedat="1788304762000" data-alt="Figure 6"> 
Figure 6: Dense point cloud coregistration. An example visual comparison of two temporally coregistered dense point clouds. Areas of structural loss, likely from swell damage, are visually coded in red, as indicated by the magenta arrow. Areas of structural addition, typically attributed to growth of coral colonies, are visually coded in blue, as indicated by the yellow arrow. The coregistration workflow described here can still be used for reef areas as dynamic as seen here, where permanently installed Ground Control Points (GCPs) would be unreliable due to overgrowth or dislodgement. <a href="https://www.jove.com/files/ftp_upload/67700/67700fig06large.jpg" target="_blank">Please click here to view a larger version of this figure.</a>
<table border="1" fo:keep-together.within-page="1" fo:keep-with-next.within-page="always">
	<tbody>
		<tr>
			<td>Camera Feature</td>
			<td>Recommended Setting</td>
		</tr>
		<tr>
			<td>Focus</td>
			<td>Auto</td>
		</tr>
		<tr>
			<td rowspan="2">Shooting Mode</td>
			<td>P (Programmed Auto) for wide-angle lens</td>
		</tr>
		<tr>
			<td>A (Aperture-priority auto), set to an aperture of F8 for macro lens</td>
		</tr>
		<tr>
			<td>Release Mode</td>
			<td>S (Shutter-priority Auto)</td>
		</tr>
		<tr>
			<td>Autofocus Settings</td>
			<td>Autofocus central (AF-C), Single Central Point (S)</td>
		</tr>
		<tr>
			<td>Auto ISO Sensitivity Control</td>
			<td>ON</td>
		</tr>
		<tr>
			<td>Maximum ISO Sensitivity</td>
			<td>3200</td>
		</tr>
		<tr>
			<td>Minimum Shutter Speed</td>
			<td>1/320</td>
		</tr>
		<tr>
			<td>Image Quality</td>
			<td>RAW + JPEG</td>
		</tr>
		<tr>
			<td>Interval Timer</td>
			<td>1 s</td>
		</tr>
		<tr>
			<td>White Balance</td>
			<td>Custom</td>
		</tr>
	</tbody>
</table>
Table 1: Recommended camera settings. The following is a list of key camera settings used to optimize image quality. These settings prioritize the capture of in-focus images captured by a moving operator in underwater lighting conditions.
<table border="1" fo:keep-together.within-page="1" fo:keep-with-next.within-page="always">
	<tbody>
		<tr>
			<td colspan="2">Align Photos</td>
		</tr>
		<tr>
			<td>Accuracy</td>
			<td>High</td>
		</tr>
		<tr>
			<td>Generic Preselection</td>
			<td>No</td>
		</tr>
		<tr>
			<td>Key Point Limit</td>
			<td>5000</td>
		</tr>
		<tr>
			<td>Tie Point Limit</td>
			<td>0</td>
		</tr>
		<tr>
			<td>Guided Image Matching</td>
			<td>No</td>
		</tr>
		<tr>
			<td>Adaptive camera model fitting</td>
			<td>Yes</td>
		</tr>
		<tr>
			<td colspan="2">Build Dense Cloud</td>
		</tr>
		<tr>
			<td>Quality</td>
			<td>High</td>
		</tr>
		<tr>
			<td>Depth Filtering</td>
			<td>Mild</td>
		</tr>
		<tr>
			<td>Calculate Point Colors</td>
			<td>Yes</td>
		</tr>
		<tr>
			<td>Calculate Point Confidence</td>
			<td>Yes</td>
		</tr>
		<tr>
			<td colspan="2">Export Points</td>
		</tr>
		<tr>
			<td>File Type</td>
			<td>Stanford PLY</td>
		</tr>
		<tr>
			<td>Coordinates System</td>
			<td>Local Coordinates (m)</td>
		</tr>
		<tr>
			<td>Source Data</td>
			<td>Dense cloud</td>
		</tr>
		<tr>
			<td>Save point colors</td>
			<td>Yes</td>
		</tr>
		<tr>
			<td>Save point normal</td>
			<td>Yes</td>
		</tr>
		<tr>
			<td>Save point confidence</td>
			<td>Yes</td>
		</tr>
		<tr>
			<td>Save point classes</td>
			<td>No</td>
		</tr>
		<tr>
			<td>Convert colors to 8 bit RGB</td>
			<td>Yes</td>
		</tr>
		<tr>
			<td>Binary encoding</td>
			<td>Yes</td>
		</tr>
	</tbody>
</table>
Table 2: 3D Dense point cloud construction settings. A list of settings used in Agisoft Metashape to create and export a high-quality dense point cloud reconstruction.
Supplemental File 1: Roadmap. Example orthomosaic image marked with plot features and depths to aid in finding plot area for resurvey. <a href="https://www.jove.com/files/ftp_upload/67700/Supplemental%201%20Roadmap.pdf" target="_blank">Please click here to download this File.</a>
Supplemental File 2: Extract_meta.py. Script run in Agisoft Metashape to export camera pose and file directory information for use in Viscore to query the original images. <a href="https://www.jove.com/files/ftp_upload/67700/Supplemental%202%20extract_meta.py" target="_blank">Please click here to download this File.</a>
Supplemental File 3: Guide to Viscore. Software guide for Viscore, which includes workflows for model visualization, coregistration, and ecological analyses. <a href="https://www.jove.com/files/ftp_upload/67700/Supplemental_3_Guide_to_Viscore_Final.pdf" target="_blank">Please click here to download this File.</a>
Supplemental File 4: Rugosity_Functions.Rmd. Script used in R containing functions to process rugosity data extracted from Viscore. <a href="https://www.jove.com/files/ftp_upload/67700/Supplemental%204%20Rugosity_Functions.Rmd" target="_blank">Please click here to download this File.</a>
Supplemental File 5: Rugosity_Analysis.Rmd. Script used in R to calculate rugosity metrics. <a href="https://www.jove.com/files/ftp_upload/67700/Supplemental%205%20Rugosity_Analysis.Rmd" target="_blank">Please click here to download this File.</a>
Supplemental File 6: Quadrat_sampling.rpl.json. Script used in Viscore for quadrat density analysis workflow. <a href="https://www.jove.com/files/ftp_upload/67700/Supplemental%206%20Quadrat_sampling.repl.json" target="_blank">Please click here to download this File.</a>
Supplemental File 7: Density_taxo_lookup.json. Button lookup file for running quadrat sampling script to aggregate quadrat sampling data by taxonomic group. <a href="https://www.jove.com/files/ftp_upload/67700/Supplemental%207%20Density_taxo_lookup.json" target="_blank">Please click here to download this File.</a>
Supplemental File 8: Density_Analysis.R Script used in R to aggregate quadrat sampling data that calculates density by taxonomic group at a survey level. <a href="https://www.jove.com/files/ftp_upload/67700/Supplemental%208%20Density_Analysis.R" target="_blank">Please click here to download this File.</a>
Supplemental File 9: README.txt Example text file to be included with original imagery for data archiving that describes image capture methodology. <a href="https://www.jove.com/files/ftp_upload/67700/Supplemental%209%20README.txt" target="_blank">Please click here to download this File.</a>
Supplemental File 10: METADATA_KAH_2016-07_03.txt Example text file to be included with original imagery for data archiving that contains metadata fields for LAI survey. This includes fields for within survey scale and depth data for referencing as well as overall site metadata for geographic context. Abbreviation: LAI = large-area imaging. <a href="https://www.jove.com/files/ftp_upload/67700/Supplemental%2010%20METADATA_KAH_2016-07_03.txt" target="_blank">Please click here to download this File.</a>

This protocol covers a four-step large-area imaging survey methodology used to extract metrics of structural complexity, community composition, and population demographics for coral reef communities. The quality of imagery collected and integrated access to the source imagery are prioritized within each step of the protocol.

Digital imaging and processing technology have evolved to facilitate the expansion of large-area imaging surveys, which increase our capacity to study the ...

measuring-structure-composition-change-underwater-environments-with

Research

JoVE Journal

Environment

This protocol provides instructions on how to use a self-contained underwater velocimetry apparatus (SCUVA), which is designed for quantification of in situ animal-generated flows. In addition, this protocol addresses challenges posed by field conditions, and includes operator motion, predicting position of animals, and orientation of SCUVA.

Quantitatively Measuring In situ Flows using a Self-Contained Underwater Velocimetry Apparatus (SCUVA)

An integrated suite of imaging techniques has been applied to determine polyp morphology and tissue structure in the Caribbean corals Montastraeaannularis and M. faveolata. Fluorescence, serial block face, and two-photon confocal laser scanning microscopy have identified lobate structure, polyp walls, and estimated chromatophore and zooxanthellae densities and distributions.

Multimodal Optical Microscopy Methods Reveal Polyp Tissue Morphology and Structure in Caribbean Reef Building Corals

We describe a new method for counting fishes, and estimating relative abundance (MaxN) and fish density using rotating stereo-video camera systems. We also demonstrate how to use distance from camera (Z distance) to estimate species-specific detectability.

Development of New Methods for Quantifying Fish Density Using Underwater Stereo-video Tools

Image based surveying is an increasingly practical, non-invasive method to sample the marine environment. We present the protocol of a drop camera survey that estimates the abundance and distribution of the Atlantic sea scallop (Placopecten magellanicus). We discuss how this protocol can be generalized for application to other benthic macroinvertebrates.

Methods for Image-based Surveys of Benthic Macroinvertebrates and Their Habitat Exemplified by the Drop Camera Survey for the Atlantic Sea Scallop

We provide a detailed protocol for conducting underwater structure-from-motion photogrammetry surveys to generate 3D models and orthomosaics.

A Field Primer for Monitoring Benthic Ecosystems Using Structure-From-Motion Photogrammetry

Presented here is a protocol for collecting and processing underwater photogrammetry data, including a significantly simplified and fully automated image processing pipeline resulting in georeferenced and time-series aligned outputs ready for ecological data extraction, analysis, and application.

Reefshape: A System for the Efficient Collection and Automated Processing of Time-Series Underwater Photogrammetry Data for Benthic Habitat Monitoring

A method for the determination of fatty acid content and composition in microalgae based on mechanical cell disruption, solvent based lipid extraction, transesterification, and quantification and identification of fatty acids using gas chromatography is described. A tripentadecanoin internal standard is used to compensate for the possible losses during extraction and incomplete transesterification.

Analysis of Fatty Acid Content and Composition in Microalgae

More than half of proteins are small proteins (molecular mass &lt;200 kDa) that are challenging for both electron microscope imaging and three-dimensional reconstructions. Optimized negative staining is a robust and high-throughput protocol to obtain high contrast and relatively high resolution (~1 nm) images of small asymmetric proteins or complexes under different physiological conditions.

Optimized Negative Staining: a High-throughput Protocol for Examining Small and Asymmetric Protein Structure by Electron Microscopy

In planta measurement of nutrient and toxicant fluxes is essential to the study of plant nutrition and toxicity. Here, we cover radiotracer protocols for influx and efflux determination in intact plant roots, using potassium (K+) and ammonia/ammonium (NH3/NH4+) fluxes as examples. Advantages and limitations of such techniques are discussed.

Measuring Fluxes of Mineral Nutrients and Toxicants in Plants with Radioactive Tracers

Aquafaba is a viscous juice from canned chickpea that, when stirred vigorously, produces a relatively stable white froth or foam. The primary research goal is to identify the components of aquafaba that contribute viscosifying/thickening properties using nuclear magnetic resonance (NMR), ultrafiltration, electrophoresis, and peptide mass fingerprinting.

Composition and Properties of Aquafaba: Water Recovered from Commercially Canned Chickpeas

Learn a four-step large-area imaging protocol for coral reef ecology using underwater camera systems. This method enables detailed structural complexity, community composition, and demographic analyses of benthic marine ecosystems.

Measuring the Structure, Composition, and Change of Underwater Environments with Large-area Imaging

Explore More Videos