Quantifying impacts of stony coral tissue loss disease on corals in Southeast Florida through surveys and 3D photogrammetry.

Since 2014, stony coral tissue loss disease (SCTLD) has contributed to substantial declines of reef-building corals in Florida. The emergence of this disease, which impacts over 20 scleractinian coral species, has generated a need for widespread reef monitoring and the implementation of novel survey and disease mitigation strategies. This study paired SCTLD prevalence assessments with colony-level monitoring to help improve understanding of disease dynamics on both individual coral colonies and at reef-wide scales. Benthic surveys were conducted throughout the northern Florida Reef Tract to monitor the presence/absence of disease, disease prevalence, and coral species affected by SCTLD. Observed SCTLD prevalence was lower in Jupiter and Palm Beach than in Lauderdale-by-the-Sea or St. Lucie Reef, but there were no significant changes in prevalence over time. To assess colony-level impacts of the disease, we optimized a low-cost, rapid 3D photogrammetry technique to fate-track infected Montastraea cavernosa coral colonies over four time points spanning nearly four months. Total colony area and healthy tissue area on fate-tracked colonies decreased significantly over time. However disease lesion area did not decrease over time and was not correlated with total colony area. Taken together these results suggest that targeted intervention efforts on larger colonies may maximize preservation of coral cover. Traditional coral surveys combined with 3D photogrammetry can provide greater insights into the spatiotemporal dynamics and impacts of coral diseases on individual colonies and coral communities than surveys or visual estimates of disease progression alone.

Introduction

Coral cover in the Tropical Western Atlantic (TWA) has declined over the last four decades [1,2], and coral diseases have been identified as one major driver of widespread coral decline throughout the region [3]. In the 1990s, white band disease dramatically reduced coral cover of Acropora cervicornis and Acropora palmata by 95% in the TWA [4]. At the same time, in the Florida Keys, white pox was responsible for up to a 70% reduction in A. palmata cover [5]. Additionally, increased disease prevalence following a coral bleaching event in 2005 caused a 60% decline in coral cover throughout the U.S. Virgin Islands [6].

Coral diseases, regardless of their host specificity, have contributed to the decline of many integral reef-building scleractinian species. While white pox, white band, and acute Montipora white syndrome are genus-specific, other diseases have a broad, even pan-oceanic host range [6-8]. For example, white plague type II affects 17 species of scleractinian coral across multiple genera [9]. Widespread decreases in coral cover from diseases result in an ecological shift from diverse, coral-dominated communities to more homogenous, algal-dominated communities [10,11]. Such shifts have the potential to create long-term changes in fish assemblages and fishery yields [12] as well as the loss of key ecosystem services such as fisheries habitat [13], coastal wave protection [14] and nutrient cycling [15] that can persist long after a coral disease event subsides.

Since 2014, the Florida Reef Tract (FRT) has experienced an ongoing outbreak of a newly-described coral disease responsible for widespread coral mortality throughout the TWA. Stony coral tissue loss disease (SCTLD) is characterized as a highly virulent disease that affects over 20 species of scleractinian corals in the TWA [16]. SCTLD first appeared in the summer of 2014 following the dredging of Government Cut in Miami-Dade County [17,18]. In subsequent years, reports of SCTLD infections have increased and spread from Miami-Dade County along the Florida Reef Tract (FRT) and into the wider TWA. To date, SCTLD has spread north to the northern terminus of the FRT in Martin County and south past the Marquesas Keys in Monroe County, with additional outbreaks observed in at least twelve territories throughout the TWA [19,20]. Spatial epidemiological modeling revealed that SCTLD is highly contagious with significant disease clusters as wide-spread as 140 km confirming the severity of this outbreak [21].

SCTLD manifests as lesions of necrotic tissue that spread across a colony, leaving behind denuded coral skeleton (Fig 1) [16]. SCTLD is histologically distinct from white plague, exhibiting a fast-acting, liquefactive necrosis after lesions develop deep within coral tissue and progress to the colony surface [22]. Across multiple host coral species, SCTLD-affected colonies demonstrate altered microbial communities relative to their apparently healthy counterparts [23-25]. While both microbial culture-based and sequencing studies are ongoing, no pathogen for SCTLD has been identified to date. Various intervention methods have been proposed and trialed to attempt to reduce immediate loss of coral cover due to colony mortality, including probiotic treatments, physical interventions, and various topical treatments [26]. The most popular has been a combination of physical intervention by trenching a firebreak around the infected tissue and covering the infected tissue with a topical antibiotic application [27]. Regardless of the method, the various proposed techniques require individual evaluation and continued monitoring to determine success.

Fig 1

SCTLD lesions on a colony of Montastraea cavernosa.

(a) Fate-tracked, SCTLD-infected Montastraea cavernosa with (b) rendered 3D model, (c) characteristic disease lesion, and (d) necrotic tissue.

Traditionally, coral disease monitoring has relied on a combination of benthic survey methods and individual colony monitoring, the latter often based on estimations of disease area, in situ linear measurements, or planar photography [28,29]. However, these techniques may incorporate a substantial observer bias or uncertainties associated with 2D estimations of 3D surfaces [30]. Underwater photogrammetry and 3D modeling of coral colonies is an emerging technique that offers the potential to enhance both the accuracy and speed of data collection. Structure-from-motion (SfM) photogrammetry derives 3D structure from a series of overlapping images, similar to established stereoscopic methods. Rather than requiring known a priori positions, SfM orientation is resolved based on common features extracted from overlapping images [31]. Conveniently, this can be done using a moving sensor, or from still images generated by an individual with a camera. Photogrammetry has been used in a number of coral reef applications including habitat characterizations [32-34], surface and volume measurements [35,36], and growth measurements [37,38]. SfM photogrammetry was first implemented on coral colonies using 10–20 overlapping photographs to generate 3D models [39]. As technology has progressed and computational power has become increasingly affordable, SfM photogrammetry techniques for the purpose of assessing various coral metrics have been refined and improved upon [35,38]. Changes in image capture and model reconstruction have improved measurement accuracy and precision, especially in regards to more established surface area methods such as the foil-wrapping method [30,35,38,40,41]. As such, SfM photogrammetry has emerged as a powerful and commonly-used tool for coral reef researchers.

In this study, we adapted a SfM photogrammetry technique [41] to track disease progression in fate-tracked Montastraea cavernosa colonies in Southeast Florida to gain insight into the colony-level dynamics of SCTLD. The study was designed to provide insight into colony- and community-level dynamics of this poorly understood disease through a combination of roving diver surveys and colony fate-tracking using SfM photogrammetry. The ultimate goal of this work is to increase widespread application of this and similar techniques to improve the design, implementation, and success of coral disease intervention, mitigation, and management strategies.

Methods

Disease prevalence surveys

Four locations across the northern Florida Reef Tract (NFRT) were selected for disease prevalence surveys: St. Lucie Reef (SLR), Jupiter (JUP), Palm Beach (PB), and Lauderdale-by-the-Sea (LBTS, Fig 2). The work conducted at SLR was done so under the St. Lucie Inlet State Park Permit 06261715, issued by St. Lucie Inlet State Park and FWC SAL-17-1960-SCRP, issued by Florida Fish and Wildlife Conservation Commission. The work conducted at JUP, PB, and LBTS was conducted under FWC SAL-17-1960-SCRP, issued by Florida Fish and Wildlife Conservation Commission. Following Hurricane Irma in September 2017, a rapid-response damage and disease survey effort was conducted throughout Southeast Florida [42]. Data from these initial surveys were used to select the sites for the present study based on occurrence of coral communities and prevalence of SCTLD (Fig 2).

Fig 2

Map of study locations throughout the Northern Florida Reef Tract Florida.

Red circles indicate roving diver survey sites and red triangles indicate sites where both roving diver surveys and coral fate-tracking occurred (Florida boundary and benthic hard bottom shapefile source: Florida Fish and Wildlife Conservation Commission-Fish and Wildlife Research Institute).

Roving diver disease surveys were conducted approximately monthly from November 2017 to June 2019. These surveys were designed to assess the greatest reef area possible, quantifying disease prevalence over an estimated range of 100–2000 m2 per survey dependent on underwater visibility. SCUBA divers swam for 20 min and recorded the species and disease status of every living coral colony ≥10 cm in diameter, and SCTLD abundance and prevalence were calculated from raw counts data. To assess multivariate variation in disease prevalence among sites and survey times, permutational analysis of variance (PERMANOVA; 9,999 permutations) was conducted in the R package vegan [43,44], with Bonferroni-corrected pairwise comparisons using the package pairwiseAdonis [45]. Non-parametric tests were implemented for all analyses as datasets were non-normal and normal distributions could not be achieved through transformation.

Fate-tracking of SCTLD-affected M. cavernosa

Benthic survey data indicated that disease incidence was too low at sites in Jupiter and Palm Beach, and that coral abundance was too low at sites in St. Lucie Reef, to conduct statistically robust fate-tracking studies in these locations. Consequently, three fate-tracking sites were established in Lauderdale-by-the-Sea (T328, BC1, and FTL4). These three sites are ~12 km from the nearby Hillsboro Inlet, less than 500 m from shore, and have been previously used for benthic and coral monitoring [46-48] within the NFRT (Fig 2). Montastraea cavernosa was selected for colony fate-tracking due to the abundance of infected colonies within the study sites. This coral species is considered intermediately susceptible to SCTLD, with onset of tissue loss occurring weeks to months later than what has been reported for highly susceptible species (e.g. Dendrogyra cylindrus, Meandrina meandrites, Colpophyllia natans). Lesions on infected M. cavernosa generally progress slower as compared to highly susceptible coral species, with total mortality occurring within months to years [16]. Montastraea cavernosa comprised 11.6% of the reported cases of SCTLD in the NFRT in late 2017 [42].

Twenty-four colonies of M. cavernosa affected with SCTLD were tagged with uniquely numbered cattle tags across the three sites on 24-August-2018 (T1). Sites were revisited on 11-Sep-2018 (T2), 8-Nov-2018 (T3), and 17-Dec-2018 (T4). Continuous video was taken for 3D model generation to quantify total colony surface area and disease lesion area for each time point.

Fate-tracked colonies were filmed using methods outlined in Young et al. [41], with the following modifications: Canon G16 cameras in Fantasea underwater housings were set on “Underwater mode,” 1080p and 60 frames per second (fps), and exposure was adjusted as needed based on ambient light conditions. One-meter, L-shaped PVC scale bars marked at 10 cm increments were placed at opposing right angles to frame the designated colony. A SCUBA diver swam approximately 1 m above the highest point of each coral colony and recorded continuous video while swimming repeated passes in a lawnmower pattern with the camera pointed directly downward. The number of adjacent passes varied depending on colony size, allowing for 60–70% overlap of filmed surface area. The camera was rotated 90° at the end of the first set of adjacent passes and another set of passes was completed perpendicular to the first set. The two complete sets of passes for a single colony required between 1–3 min of dive time, depending on colony size. Each sampling event produced an average of 14 GB of.mp4 video files, or approximately 425 MB of video file size per coral colony.

Video processing and 3D model generation protocols are described in full in our GitHub repository [49]. In summary, the free software package, FFmpeg (www.ffmpeg.org), was used to extract still frame images from videos of the fate-tracked colonies at a rate of 3 fps. Still images were then imported into Agisoft Metashape Standard Edition (Version 1.5.2, Agisoft LLC) software, which uses a proprietary algorithm that incorporates SfM and Brown’s lens distortion model to generate 3D models from 2D images [31,50]. Model generation in Metashape was conducted according to the manufacturer’s protocol in four general steps: 1) camera alignment, 2) dense point cloud generation, 3) mesh generation, and 4) texture overlay. Models were rendered on a 2018 Apple MacBook Pro with a 2.9 GHz processor, 16GB of RAM and a Radeon Pro Vega 16 4GB graphics card. A single model took approximately 40 min to render depending on the number of still images generated. Generated models were then exported as a.obj file and imported into the software Rhinoceros 3D (Robert McNeel & Associates) for analysis; the mean model file size was 64 MB.

Models were scaled using the PVC scale bars, then total colony surface area and disease lesion surface area were measured by hand-tracing polygons around coral tissue. SCTLD disease lesion area was defined as the stark white newly-dead coral tissue or skeleton with sloughing tissue that is characteristic of the disease margin. Both total colony area and disease lesion area were generated directly within Rhinoceros 3D, while healthy tissue area was calculated by subtracting disease lesion area from total colony area. Proportion of loss and rate of tissue loss per week were calculated for each pair of consecutive time points. To identify significant effects of time on healthy area, diseased area, total area, the proportion of tissue loss and the rate of tissue loss, Friedman’s rank sum tests were conducted using the package PMCMRplus [51]. Pairwise comparisons were made with Nemenyi tests in the package PMCMR [52]. Additionally, Spearman’s rank correlation analyses were used to correlate diseased area and total colony area and rate of tissue loss and total colony area.

3D model validation

Accuracy of model-generated surface area metrics was assessed by a validation experiment using a standardized square cupola prism with equal dimensions and surface features to represent a coral. The prism was constructed from ½” PVC using a top square of 25.4 x 25.4 cm, a bottom square of 47.6 x 47.6 cm, and height of 30.5 cm (S1 Fig). Angled sides were achieved using 45° PVC tees to better represent a mounding coral colony. Additionally, polygons of known surface areas (square: 40.32 cm2; rectangle: 12.90 cm2; circle: 45.60 cm2) were printed to scale on waterproof paper, and affixed to the top and sloped sides of the weighted prism.

The prism was filmed in a pool and at the three coral fate-tracking sites in Lauderdale-by-the-Sea (Fig 2) to quantify model error in pristine (pool) versus reef conditions (e.g. variable turbidity, currents, surge). Eight videos were recorded for the prism and boundary frames across the three reef sites (2–3 replicates per site), while six prism videos were captured in the pool. The prism was filmed at a depth of 2.8 m in the pool, 8.8 m at BC1, 5.2 m at T328, and 6.7 m at FTL4; filming distance remained the same for all videos. To assess accuracy of surface area measurements, 3D models were generated and analyzed using the same protocols described earlier, with polygons traced for each of the prism shapes based on the four corners for squares and rectangles, and the entire circumference for circles. Replicate surface area measurements of shapes were made for each model according to face of the prism (top and four sides), and number of shapes on each face (four for square and rectangle, one for circle), for a total n = 229 measurements for square, n = 232 for rectangle, and n = 59 for circle. Shape error was calculated as the absolute difference between measured shape surface areas and the corresponding template area. Shape error measurements were found to be highly right-skewed, therefore a square-root transformation was applied prior to statistical analyses. Multivariate homogeneity of variance was first tested using the betadisper function of the package vegan, and following identification of heterogeneous variance among sites due primarily to lower variance at site FTL4. A single-factor PERMANOVA was conducted across sites using the adonis function of the package vegan.

Additionally, a subset of four models from the colony fate-tracking dataset were extracted at three different frame rates (3–5 fps) across their respective time points to determine if frame rate affected model accuracy. A Kruskal-Wallis test was used to determine if model size differed across the different extraction frequencies.

Results

SCTLD prevalence varied significantly across location (Table 1), but not through time (Table 1, Fig 3). Pairwise comparisons indicated that disease prevalence was lower in JUP and PB than in SLR and LBTS (Table 1, Fig 3). In LBTS, disease prevalence remained relatively constant (10.9%) while in JUP and PB prevalence was consistently low, but peaked in March 2019 with SCTLD prevalence at 8.1% and 4.3%, respectively. Prevalence data from SLR was highly varied due to generally low coral cover observed during the survey period. For example, the highest reported disease prevalence of 43% at site SLR North in April 2019 was due to low coral cover; only 7 living corals were observed, 3 of which were diseased. The most abundant species throughout our surveys were M. cavernosa (63%, n = 3241), Porites astreoides (17%, n = 885), Siderastrea siderea (4%, n = 199), Stephanocoenia intersepta (2.3%, n = 118), and Agaricia agaricites (2.1%, n = 108). Colony abundance by species varied among the four locations (Kruskal-Wallis, H = 10.308, p = 0.016) with significantly lower abundance between SLR and all other locations (Dunn’s Test, all p < 0.05). Montastraea cavernosa was the most abundant species at all locations, except for SLR, where P. astreoides was the most abundant species (65%, n = 304). The species with more than 10 observations that had the highest disease prevalence were Pseudodiploria clivosa (36%, n = 47), A. agaricites (10%, n = 108), Dichocoenia stokesii (9%, n = 23), Orbicella faveolata (6%, n = 49) and M. cavernosa (5%, n = 3241).

Table 1

Results from univariate-permutational analysis of variance (PERMANOVA) comparing SCTLD prevalence from the roving diver surveys across location and time.

Test	Comparison	df	Psuedo-F	p-value*
PERMANOVA	Location	3	7.44	< 0.001
	Date	12	1.23	ns
	Location:Date	7	0.52	ns
Pairwise	LBTS–SLR		0.11	ns
	LBTS–JUP		26.76	< 0.001
	LBTS–PB		35.91	< 0.001
	SLR–JUP		7.12	0.0492
	SLR–PB		7.88	0.045
	JUP–PB		0.63	ns

*Non-significant p values listed as “ns”.

Pairwise comparisons from results of the univariate PERMANOVA across all four locations: St. Lucie Reef (SLR), Jupiter (JUP), Palm Beach (PB), and Lauderdale-by-the-Sea (LBTS).

Fig 3

Mean SCTLD prevalence across all four locations (St. Lucie Reef, Jupiter, Palm Beach and Lauderdale-by-the-Sea).

Points represent means; bars represent standard error.

Friedman’s rank sum tests indicated a significant decrease in total colony area (cm2) and healthy tissue area over time (Table 2, Fig 4). Disease lesion area varied significantly through time, however pairwise comparisons revealed that one time point comparison (T2–T4) was driving the variation (Table 2).

Table 2

Results of Friedman’s rank sum tests comparing total colony area, disease tissue area and healthy tissue area across time, with pairwise comparisons made using Nemenyi tests.

Test	Comparison	Test Statistic	p value*
Total Colony Area	Friedman’s	48.55	< 0.001
	T1 –T2	4.27	ns
	T1 –T3	7.59	< 0.001
	T1 –T4	9.01	< 0.001
	T2 –T3	3.32	ns
	T2 –T4	4.74	0.026
	T3 –T4	1.42	ns
Healthy Tissue Area	Friedman’s	41.29	< 0.001
	T1 –T2	4.27	0.026
	T1 –T3	7.59	< 0.001
	T1 –T4	9.01	< 0.001
	T2 –T3	3.32	ns
	T2 –T4	4.74	ns
	T3 –T4	1.42	ns
Disease Lesion Area	Friedman’s	14.02	0.003
	T1 –T2	4.27	ns
	T1 –T3	7.59	ns
	T1 –T4	9.01	ns
	T2 –T3	3.32	ns
	T2 –T4	4.74	0.012
	T3 –T4	1.42	ns

T1, 24-Aug-2018; T2, 11-Sep-2018; T3, 8-Nov-2018; T4, 17-Dec-2018.

*Non-significant p values listed as “ns”.

Fig 4

Comparison of tissue area through time.

Mean tissue areas through time (a) mean total colony area (cm2) (b) mean healthy tissue area (cm2), and (c) mean disease lesion area across all sites and through time for fate-tracked M. cavernosa colonies (n = 24). T1, 24-Aug-2018; T2, 11-Sep-2018; T3, 8-Nov-2018; T4, 17-Dec-2018.

A Friedman’s rank sum test revealed significant differences between rates of tissue loss through time (df2,71, Friedman χ2 = 6.25, p = 0.044). Average tissue loss (mean ± SEM) between T1–T2 was -98.2 ± 40.7 cm2 wk-1, -25.0 ± 7.7 cm2 wk-1 between T2–T3, and -45.1 ± 19.6 cm2 wk-1 between T3–T4 (Fig 5A). Rate of tissue loss was not significantly different among sites between any time point (S1 Table). On average, the proportion of tissue lost was 37.1 ± 7.2% over the course of this study, with three colonies experiencing complete mortality. There was no correlation between total colony area and disease lesion area (Spearman’s rank correlation rs = 0.15, p = 0.154, Fig 6), or between rates of tissue loss and total colony area (Spearman’s rank correlation rs = -0.046, p = 0.704, Fig 5B).

Fig 5

Comparison of rates of tissue loss.

(a) Mean change in tissue area on fate-tracked M. cavernosa colonies and (b) Spearman’s rank correlation between rate of tissue loss (cm2wk-1) and total colony area (cm2).

Fig 6

Correlations between total colony area and disease lesion area.

Spearman’s rank correlation between disease lesion area (cm2) and total colony area (cm2) for fate-tracked M. cavernosa colonies (n = 24).

Absolute and relative shape error did not vary across the pool and three coral fate-tracking sites in Lauderdale-by-the-Sea (PERMANOVA: F3,279 = 0.733, p = 0.584). The absolute shape error across all sites was 1.7 ± 0.1 cm2 (pool: 1.4 ± 0.1 cm2; BC1: 2.0 ± 0.2 cm2; T328: 2.4 ± 0.3 cm2; FTL4: 1.8 ± 0.2 cm2), which corresponded to a relative shape error of 6.1 ± 0.3% (pool: 4.9 ± 0.3%; BC1: 7.0 ± 0.7%; T328: 8.7 ± 0.9%; FTL4: 6.9 ± 0.6%). Average absolute shape error across all sites was 1.70 ± 0.09 cm2, which corresponded to relative error of 6.13 ± 0.27%.

To assess potential effects of different frame rates used during model generation on colony-scale surface area measurements, a single-factor Kruskal-Wallis test indicated that frame rate had no significant impact on measured colony surface area (Kruskal-Wallis, H = 0.12, p = 0.94). This suggests that higher frame rates can be used from the initial video to improve poorly constructed models without affecting downstream spatial analyses. To keep processing time as low as possible, however, stills were extracted at 3 fps unless otherwise necessary.

Discussion

SCTLD is a unique coral disease due to its broad geographic extent, the number of coral species affected, and rapid rates of disease progression and spread [16,17,21,53]. SCTLD has been reported across ~400 km of the FRT, and now has been observed in at least 12 other territories throughout the TWA [20]. The geographic distribution of SCTLD continues to expand with new observations throughout the Caribbean [19,20]. SCTLD has been reported as far west as Belize and as far southeast as Martinique which are 1300 and 2500 km respectively from the northern terminus of the NFRT at St. Lucie Reef, FL [20]. Other coral diseases, such as white syndrome, have been observed over broad spatial extents (~1500 km) in the Great Barrier Reef [54]. Likewise, black band disease has been observed on coral reefs across ocean basins [55]. The overall mean disease prevalence for the NFRT observed in this study was 6%, which is lower than recorded in post Hurricane Irma surveys in 2017 [42] and previous SCTLD surveys in SE Florida [53]. Sites in Jupiter and Palm Beach had relatively low disease prevalence (1.7% and 1.0%) consistent with background disease levels (1–3%) within the TWA [56]. The highest values observed in this study were between 21–43% at SLR, but no site reached the highest reported disease prevalence values of 60% observed near Miami in 2014 [17]. The lower prevalence values reported in this study may be due in part to differences in species composition between the NFRT and southern regions of the FRT. The most abundant and susceptible species such as M. meandrites, D, stokesii, and Pseudodiploria strigosa [17,53] were comparatively sparse within our NFRT surveys. However, previous SCTLD impacts before our monitoring sites were established may also have affected the relative abundance of susceptible species [53,57].

We observed an increase in disease prevalence during the spring of 2018 which was unexpected, as prevalence for other described coral diseases such as white syndrome, white band, black band, and white pox often increases during the summer months as water temperatures increase [58-62]. SCTLD prevalence does not appear to have a positive correlation with temperature [53,58] as has been observed for other coral diseases [63-65], but potential environmental cofactors that may drive SCTLD prevalence need to be examined further.

At the colony level, disease progression was highly variable across fate-tracked M. cavernosa colonies near Lauderdale-by-the-Sea. As expected, total colony area and healthy tissue area decreased significantly over time, demonstrating the impact SCTLD can have on living coral cover over just a few months. Over the course of this study (115 days), colonies lost on average 37.1 ± 7.24% of their tissue surface area. Using a similar 3D photogrammetry method, Meiling et al. [58] saw proportional losses of 2.0 ± 0.11% day-1 of six SCTLD-infected M. cavernosa colonies in the US Virgin Islands. Additionally, Aeby et al. [25] noted subacute tissue loss in M. cavernosa was 34 ± 8.7% over 1 year using a semi-quantitative method to estimate tissue.

Disease lesion area did not correlate with total colony area, suggesting that larger colonies do not exhibit a higher proportion of diseased tissue compared to smaller colonies. As SCTLD progresses on coral colonies, loss of healthy tissue appears to be a more important indicator of disease virulence, rather than increase in disease lesion area. Importantly, rates of tissue loss did not correlate with total colony area. Therefore, larger corals may have a more favorable time horizon for potential intervention actions that could prevent total colony mortality. Conversely, smaller infected colonies may succumb more quickly to SCTLD. Considering the level of effort required for SCTLD interventions [66], intervention efforts focused on larger colonies with sufficient area of uninfected tissue remaining may result in more successful disease mitigation, and therefore reduction of coral cover loss due to colony mortality.

3D model generation as a fate-tracking method

This study tested a relatively inexpensive and rapid 3D model generation methodology described in Young et al. [41] as a means to track disease progression on a colony-level without substantial increases in expense, time, or computational requirements compared to observational and photographic techniques. Healthy and diseased surface area of 24 coral colonies were quantified and compared over time. Additionally, this 3D modeling technique was validated through the quantification of model error using a mock coral colony with standardized surface shapes. Shape error (i.e. variation between measured surface areas and template shapes) did not vary across different depths and turbidity conditions (S2 Fig). This The relative error of 6.13 ± 0.27% is likely an inflated estimate of model error, as the prism shapes used for accuracy assessments were printed on waterproof paper that could move with currents, rather than representing solid surfaces. There were some instances, however, where model generation was poor due to anomalies in the Metashape algorithm, inconsistent underwater filming, or most common, high turbidity, all of which may impact surface area measurements and therefore affect model error. Typically, these issues could be rectified by extracting still images from the video at an increased frame rate to ensure higher overlap among images, or by collecting replicate videos for each target. Notably, extracting at a higher fps improved poor models, but did not disproportionately affect surface area measurements from models that were constructed well using images extracted at 3 fps.

Due to Metashape’s proprietary algorithm, further adjustments within the model generation process to rectify poor models are limited. Stable, high-resolution image collection is therefore integral to successful downstream model generation. Alternatives to the lawn-mower video path used in this study may be more effective but also more time-intensive, particularly for tall (>1 m), highly rugose coral colonies. Discrete, overlapping photographs could be taken instead of continuous video [37], or video could be taken while swimming a circular pattern around the coral at varying angles [67]. Critically, while disease monitoring via linear extension measurements may be able to determine potential differences among colonies or treatments over time [16,28,68], linear extension does not accurately quantify tissue loss and may underestimate the progression of disease lesions on coral colonies. Quantitative 3D approaches such as the method presented here will improve our understanding of the ecology and impacts of SCTLD and other diseases on coral reef ecosystems, and may guide colony selection in future disease intervention strategies.

Conclusions

SCTLD affects a broad range of host taxa over broad geographic ranges relative to other described coral diseases. Based on the results from this study, larger colonies should be prioritized for SCTLD mitigation measures (i.e. disease intervention approaches) to maximize effort and potential treatment success when logistics and resources may be limited. Since SCTLD is a progressive and necrotic infection, the area of tissue loss, or proportion of tissue loss, may represent more impactful metrics for quantifying the severity of infection as opposed to disease lesion area or percent affected tissue. Additional studies incorporating longer timescales and multiple species are desirable to confirm these patterns of disease progression across all susceptible species. Rapid, cost-effective, and accurate methods using 3D models for quantifying coral surface area are a valuable approach for colony fate-tracking if high-resolution imagery can be obtained. It is recommended that managers and intervention specialists—particularly those focusing on SCTLD—adopt photogrammetric methods to enhance colony tracking methods and facilitate comparability across future studies and intervention trials.

Panel of prism template and deployment for underwater 3D photogrammetry.

A) Template used for five sides of prism, scaled to produce replicates of three standardized shapes: square (40.32 cm2), rectangle (12.90 cm2), and circle (45.60 cm2) simulating surface area measurements on a coral colony. B) Deployment of prism and scaling frames with diver recording continuous video in a lawnmower-pattern at an approximate distance of 1 m. C) Overhead view of prism and scaling frames, with 10 cm banded tape for scaling. D) Deployment of prism and scaling frames on a reef environment at Lauderdale-by-the-Sea, FL.

(TIF)

Click here for additional data file.

Panel of error measurements for each of the three prism shapes (square, rectangle, circle), calculated as the absolute difference between measured shape surface areas and template areas.

(TIF)

Click here for additional data file.

Kruskal-Wallis test comparing rate of tissue loss and site.

Non-significant p values are listed as “ns”.

(DOCX)

Click here for additional data file.

Dataset containing disease prevalence data from roving diver disease surveys.

(XLSX)

Click here for additional data file.

Dataset containing tissue area measurements from rendered 3D models of fate-tracked Montastraea cavernosa colonies.

(XLSX)

Click here for additional data file.

Dataset containing shape surface area and error measurements from prism deployment across three sites (pool, Lauderdale-by-the-Sea, St. Lucie Reef).

Error was calculated as the absolute difference between measured surface areas and actual area of the template shapes, then standardized as a percentage relative to template area.

(XLSX)

Click here for additional data file.

16 Oct 2020

PONE-D-20-25422

Quantifying impacts of stony coral tissue loss disease on corals in Southeast Florida through surveys and 3D photogrammetry

PLOS ONE

Dear Dr. Combs,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Reviewer 1 in particular notes that there is a significant literature on photogrammetry techniques on coral reefs, and further attention to this would improve both the introduction and discussion, rather than relying on citations to other taxa. They also require increased clarity in the methods regarding validation of the accuracy of the model, including what site-specific community parameters would impact the 3D model generation. Reviewer 2 also brings up a salient point here, which is that there was no direct comparison to 2D methods, so claims of increased accuracy are not demonstrated. They do recommend that a relatively rapid post-hoc statisitical comparison with 2D screen grabs could be considered, or more cautious language. I would suggest the former to increase the rigour of your study, but either would be sufficient. Reviewer 1 also questions the clarity of statistical analysis, and believes that further investigation into why shape area and error varies with location is either required, or explained in the text.

Both reviewer 1 and I are curious as to why the grouping of disease prevalence is at the county level rather than the location/site scale, particularly since you do not discuss different management, water quality or land use differences between counties (See Fig 3). By working at the location scale, this would allow comparison of more similar replicates, and would also be more accessible to a person not familiar with the geography of Florida. If you feel that county is an important part of the puzzle that makes it a more appropriate unit of replicate than location (more than just indicating a north - south gradient for example), it would be useful to explain this decision.

Both reviewers identify the recommendation of 'culling' as unjustified, or presented with too high a level of certainty, particularly since the effort, efficacy and description of what other interventions are available are not given (if there is only one colony in an area infected, but it is large, why not remove that if it leads to a better outcome than other interventions?). Without evidence presented here, or citations to small colonies dying more quickly or being disease vectors, this management recommendation goes far beyond the work.  Indeed, in Fig S2 It could be argued that there is some evidence that some larger colonies might lose tissue at a faster rate than some small colonies. In the discussion you also state that "disease lesion area did not correlate with total colony size suggesting that larger colonies do not exhibit a higher proportion of diseased tissue" - this could imply they show the same proportional loss or lower. I strongly suggest you present more data on proportional loss of colony tissue (not just absolute loss) if this is an argument you wish to make. Even so, suggesting culling small colonies as a management measure would still be extreme and would require a stronger argument with references to small colonies as vectors or disease reservoirs. The argument also begs the question - what is a small colony? Nor are what interventions that might be possible shy of culling the entire colony described.

Finally, there is a significant amount of relevant context and data in the SI. It would be good to see some of the tables and figures into the main text, particularly concerning the disease incidence, colony size and disease spread data.

Please submit your revised manuscript by Nov 30 2020 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

We look forward to receiving your revised manuscript.

Kind regards,

Fraser Andrew Januchowski-Hartley, Ph.D.

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2.  We note that Figure 2 in your submission contain map images which may be copyrighted. All PLOS content is published under the Creative Commons Attribution License (CC BY 4.0), which means that the manuscript, images, and Supporting Information files will be freely available online, and any third party is permitted to access, download, copy, distribute, and use these materials in any way, even commercially, with proper attribution. For these reasons, we cannot publish previously copyrighted maps or satellite images created using proprietary data, such as Google software (Google Maps, Street View, and Earth). For more information, see our copyright guidelines: http://journals.plos.org/plosone/s/licenses-and-copyright.

We require you to either (1) present written permission from the copyright holder to publish these figures specifically under the CC BY 4.0 license, or (2) remove the figures from your submission:

2.1.    You may seek permission from the original copyright holder of Figure 2 to publish the content specifically under the CC BY 4.0 license.

We recommend that you contact the original copyright holder with the Content Permission Form (http://journals.plos.org/plosone/s/file?id=7c09/content-permission-form.pdf) and the following text:

“I request permission for the open-access journal PLOS ONE to publish XXX under the Creative Commons Attribution License (CCAL) CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). Please be aware that this license allows unrestricted use and distribution, even commercially, by third parties. Please reply and provide explicit written permission to publish XXX under a CC BY license and complete the attached form.”

Please upload the completed Content Permission Form or other proof of granted permissions as an "Other" file with your submission.

In the figure caption of the copyrighted figure, please include the following text: “Reprinted from [ref] under a CC BY license, with permission from [name of publisher], original copyright [original copyright year].”

2.2.    If you are unable to obtain permission from the original copyright holder to publish these figures under the CC BY 4.0 license or if the copyright holder’s requirements are incompatible with the CC BY 4.0 license, please either i) remove the figure or ii) supply a replacement figure that complies with the CC BY 4.0 license. Please check copyright information on all replacement figures and update the figure caption with source information. If applicable, please specify in the figure caption text when a figure is similar but not identical to the original image and is therefore for illustrative purposes only.

The following resources for replacing copyrighted map figures may be helpful:

USGS National Map Viewer (public domain): http://viewer.nationalmap.gov/viewer/

The Gateway to Astronaut Photography of Earth (public domain): http://eol.jsc.nasa.gov/sseop/clickmap/

Maps at the CIA (public domain): https://www.cia.gov/library/publications/the-world-factbook/index.html and https://www.cia.gov/library/publications/cia-maps-publications/index.html

NASA Earth Observatory (public domain): http://earthobservatory.nasa.gov/

Landsat: http://landsat.visibleearth.nasa.gov/

USGS EROS (Earth Resources Observatory and Science (EROS) Center) (public domain): http://eros.usgs.gov/#

Natural Earth (public domain): http://www.naturalearthdata.com/

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Partly

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: No

Reviewer #2: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: Reviewer comments:

This study monitors SCTL disease prevalence at four locations in the Florida Reef Tract over three months and fate-tracks colony-level impacts using under water photogrammetry and 3D modelling of Montastrea cavernosa colonies at one location.

While the idea, study design and methodology is good, and I think it is great to see more applications of photogrammetry in coral reef research, I have concerns about the statistical evaluation of some results and especially about the model validation. As the fate-tracking is the main focus and the novelty of this work, a clear and detailed evaluation of the used methodology is important and this section should be improved before resubmission (see comments below).

In the introduction I would recommend that the authors give better credit to prior work on photogrammetry in coral reefs. Reading these papers will also help them to improve their model evaluation and compare the accuracy and precision of their methodology to similar applications.

Additional to the comments below I have attached a PDF with annotations, suggesting smaller changes where I think the text could be improved. Most of these suggestions should be easy to implement. If some time is invested on improving the model evaluation I believe this will be a nice contribution to the growing number of photogrammetry applications on coral reefs.

Introduction:

1) from L88: I think applications on cetacean and elasmobranch research are way out of the scope of this work. Better concentrate on referencing all the exciting research on coral reefs.

There is a lot of work that has been done on individual coral colonies to measure surface area, volume, carbon standing stock, coral growth and erosion rates ... I would recommend to read through them if the authors have not come across them before, also to get some inspiration for the section on accuracy and precision of 3D model building. I know all this literature can be overwhelming, but there is some good stuff in there and it will make your paper stronger. E.g.,

Bythell et al., 2001. Three-dimensional morphometric measurements of reef corals using underwater photogrammetry techniques. Coral Reefs

Cocito, Sgorbini, Peirano, & Valle, 2003. 3-D reconstruction of biological objects using underwater video technique and image processing. JEMBE

Courtney, Fisher, Raimondo, Oliver, & Davis, 2007. Estimating 3-dimensional colony surface area of field corals. JEMBE

Burns, Delparte, Gates, & Takabayashi, 2015. Utilizing underwater three-dimensional modeling to enhance ecological and biological studies of coral reefs.

Figueira et al., 2015. Accuracy and Precision of Habitat Structural Complexity Metrics Derived from Underwater Photogrammetry. Remote Sensing

Gutiérrez-Heredia, D'Helft, & Reynaud, 2015. Simple methods for interactive 3D modeling, measurements, and digital databases of coral skeletons. L&O methods

Lavy et al., 2015. A quick, easy and non-intrusive method for underwater volume and surface area evaluation of benthic organisms by 3D computer modelling. MEE

Ferreira et al. 2017. 3D photogrammetry quantifies growth and external erosion of individual coral colonies and skeletons. Sci Rep

Lange & Perry 2020. A quick, easy and non-invasive method to quantify coral growth rates using photogrammetry and 3D model comparisons. MEE

2) Also, in this section, I don’t think you need to cite aerial and scanner methods to prove your point, and the two studies you cite for coral growth did not use photogrammetry at all. Use Ferreira et al. 2017 and Lange & Perry 2020 if you want to keep this sentence. Alternatively you could explain photogrammetry and SfM first and then give examples of studies on reefs. Also see annotations in the PDF.

3) L. 104: optimized from what basis? You have not developed the photogrammetry technique, but a new application for it. Maybe say: “In this study we used photogrammetry techniques as described in [Young et al. ] in order to develop a new application, i.e to track SCTLD disease progression in M. cavernosa colonies.” or something similar?

Methods:

1) see annotations in PDF to improve clarity of the text.

2) L181: state version of Agisoft Metashape that you used as some versions seem to have issues and it will help for repeatability.

3) L. 215: I do not understand why site-specific “community parameters” (what is this anyways?) or even water quality should affect 3D model generation. This would mean that the method is not accurate and should not be used.

Related to this it does not make sense to compare the area of shapes among locations to evaluate the 3D models. So I am wondering why you did not just photograph the mock colony at the sites where you did the fate-tracking in order to calculate the accuracy of your methodology.

Concerning this there is two parameters you want to check:

A) accuracy, meaning how close to reality is your 3D model. Test this by comparing measured shape areas on your mock corals to known areas and calculate the mean error. (This is potentially how you got your 2.17 cm2?) If you notice that the error is bigger at a more turbid reef site, then you could conclude that the visibility affects the accuracy of model building/measurement. But direct comparison of shape areas among sites does not make sense.

B) precision, meaning how good is the reproducibility of your measurements. For this you should measure the surface of the same shape, or better the same coral colony, several times, including all the hand tracing around colonies etc. Then calculate the error, which will show you if you introduce considerable variability using your workflow.

You should calculate the coefficient of variation (error/average) in order to compare your errors to other studies doing surface area measurements. You cannot compare the error from small shapes (2 cm2) directly to the surface area of the colony.

L 240: I do not understand why these correlations were done and what they would tell us about the accuracy of the method.

Results:

1) I am wondering why did you choose to compare counties instead of locations as shown on the map? The latter would have the advantage that replications are more similar.

2) L 255-260: I don't think the figure supports these statements. I would say something along the line of “in Broward, disease prevalence stayed quite constant at around 10%, while in Palm Beach prevalence was usually very low, but peaked in March 2019 when ... of colonies were affected. Disease prevalence at Martin was very variable due to low numbers of live coral colonies...”

3) L 273: Has it been taken into account that colonies were measured repeatedly? Considering that the areas of your colonies are very different you might either have do more fancy statistics using repeated measures GLM/GAM (maybe using colony ID as fixed factor and site and time as explaining variables? sorry, not an expert myself) or you might have to calculate “loss of area” or “loss % of area” in order to make meaningful comparisons over time.

I know you calculated rate of change in healthy and diseased tissue, which I think might make more sense than comparing actual area. In general it is getting a bit confusing looking at so many parameters. I would suggest to think carefully which parameters are most useful in showing what you are interested in and rather use fewer but explain better what they mean.

E.g., I think the most interesting results and the best order in the section L277-309 would be (add numbers and statistics): “The rate of tissue loss did not differ among sites but was variable over time, with highest loss in the first observation interval. The diseased area however did stay constant over time, indicating that the lesion moves with the infected tissue.” Then add the correlations you think are useful. Not all are I think.

4) L287 and 288: S2 Fig should be S3 Fig?

5) L 312-L334: This whole section does not make sense to me. I do not understand why the areas of shapes should be different depending on location. This definitely does not increase trust in the method! It might in part be a relic of your different sample sizes and tests you do. Why do you run ANOVA then Kruskal Wallis and then 3-way PERMANOVA? I fail to understand what any of these significances mean and it is not explained in the results or discussion sections.

I would suggest to think carefully about what parameters actually tell you something about the accuracy/precision of the analysis (see comments above) and rewrite this section completely after the improved analysis.

Discussion:

This section could be improved by being very clear how the results compare to other studies and what can be said about the implications. See annotations in the PDF

1) L353: It is not clear what you are trying to say here. Also please make clear what you mean with L356-357. Next section can be shortened as suggested in PDF.

2) L363: This paragraph seems to be repeating discussions from the previous paragraph. Maybe you could combine those better without repeating yourself?

I think the order of these two paragraphs could be improved. E.g. “In the FRT, disease prevalence is typically higher (...%) than elsewhere in the TWA (~1-3%) [57]. Sites in Palm Beach in the present study showed relatively low background disease (6%), similar to prevalence across the NFRT after Hurricane Irma in Sept 2017 (6%???) (Walker 2018, 53). Highest values observed in this study were 20-45% at sites in Martin, but did not reach levels of up to 60% as reported in [17], likely due to low abundance of susceptible species after ongoing SCTLD impacts [53,58]”

3) L384: I thought Fig S2 showed that it does NOT correlate? Otherwise your next sentence does not make sense

4) I think the suggestion to cull small colonies is a bit drastic and not supported by your study. If there is other research supporting this approach please cite here.

5) L 384-393 could be condensed down to 1-2 sentences saying management should target larger colonies.

6) From L 403: This whole sections would have to be revised after analysis of accuracy/precision.

7) L406: It is not possible to compare the 2.17 cm2 shape error directly to the much larger colonies. Actually, this error seems very high considering that the shape on your mock coral is not very big?! Does that relate to about 10%? Calculate CV and compare to other studies. Also, if you have an error of about 10% you might want to check if the difference of total area/healthy area between time points would still be significant.

8) L409: you did not test the effect of colony size on model accuracy. Also, it does not make sense that water quality (do you mean turbidity/visibility?) affects the size of colonies. You are probably trying to say that lower visibility could result in lower model quality affecting measurements of surface area? Be precise in your wording.

Conclusions and general:

1) Different kinds of interventions should be mentioned in the introduction or discussion if it is the main discussion point in the conclusion.

2) L435: Why would interventions be unsuccessful in small colonies? They should work the same, just preserve less total area, right? So I understand prioritising big ones, but in sites were prevalence is low why not treat small ones too.

3) I think the discussion and conclusion sections will benefit from a second round of review after the improved model evaluation. Try to be very clear what the novelty of this work is (new application of photogrammetry method to accurately quantify tissue loss/disease progression over time) and what can be concluded from it (improve survey protocols, evaluate success of intervention methods... I don’t think culling of half the colonies is a good outcome here).

4) If you want people to take up this method, make the workflow easier to access. I know it is all in the GitHub repository, which is awesome, but I did not easily find the step-by-step guide and what you actually did to measure the areas. Maybe you could prepare a one-pager which is easy to print, stating the steps of image acquisition, model building and analysis to go into Suppl Methods. Or put a link in the methods which leads straight to the guide instead of the whole repository.

5) Also, it would be great if you could make the models (as .obj?) available in the repository.

Reviewer #2: The manuscript by Combs and colleagues presents the results of prevalence assessments and a time-series study of the progression of stony coral tissue loss disease (SCTLD) in the northern Florida Keys. The manuscript also presents the results of an assessment of the accuracy of a 3D photogrammetry modelling technique to track disease lesion progression, and describes the optimised method for use in future surveys. This manuscript was clear, concise, thorough, well organized, and well written. It provides sufficient detail for reproducibility studies and is timely given the on-going outbreak of this virulent and devastating coral disease. I hope that the method presented here can improve and speed-up survey techniques to improve our knowledge and ability to manage the disease.

I don’t have any significant concerns regarding any aspect of the manuscript. The most substantive comment I have is in reference to the conclusion that the model method presented here is “more accurate… than previously established methods”, when there wasn’t actually a quantitative comparison of this method with those previous methods. Thus, I think the conclusion is overreaching (summarised further below). This can either by addressed by a minor edit/softening of the language or a post-hoc statistical comparison of lesion size estimates made from the 2D video screen grabs. Otherwise, I only have minor comments and commend the authors on a well-done study.

General comments

- One of the major findings of this study is that the 3D model method presented is “more accurate data than previously established methods such as two-dimensional surface area estimation (ln 398-399).” I thought the assessment of the accuracy of the 3D method presented here was excellent and rigorous, but there was not a estimate of the disease lesions or the calibration templates/mock coral made from the same two-dimensional photographs to make a robust and unbiased statistical comparison of the two techniques. While I agree that this 3D model pipeline is achievable, it does still require ~40 min per colony of rendering, plus fairly expensive software and hardware for modelling and storage of large video files compared with the much simpler 2D method, so a quantitative comparison of the two methods would strengthen the argument and add justification for using the 3D model method presented here.

- The inclusion of more data from the disease surveys in the main text could be useful, and would balance the disease ecology with the methodological aspects of the study a bit better. Including which species were (and were not) affected at each site, and how disease prevalence changed over time by taxa within each site would be a nice addition to the main text. Line 365-7 mentions the “low abundance and species richness in Marin County sites compared with PB and Broward Counties”, but what is the density and species richness within each site?

- The manuscript is relatively sparse wrt figures and tables, with the vast majority of the information presented as supplementary. I would suggest including a few of these in the main text, particularly Fig S2 and possibly S4, again to balance out the ecology/methodology aspects of the manuscript.

Specific comments

- Line 193 – it is quite tricky to determine whether the stark white area of the lesion is still alive or newly dead, especially in an image/video. Would it be more appropriate to define the lesion area as the ‘stark white coral tissue or very newly dead white skeletal area’? If not, how was this distinction made? It is very difficult to tell in Figure 1, but the white lesion areas look like there is no live tissue to me.

- Line 282 – consider rounding the estimates of # of lesions per colony to the nearst whole number as that makes the most conceptual sense (or at least to the 10th which is probably a more appropriate significant digit)

- Line 360 – consider stating what the most susceptible species are.

- Line 384 – was the correlation between rates of tissue loss and total colony size positive or negative? Please specify.

- Line 388 – typo – suggest ‘greater preservation’

- Line 389 – is there justification for the statement that smaller colonies are more likely to succumb to SCTLD? It may take less time to succumb due to the rate of tissue loss, but is the likelihood of dying from the disease actually higher? Has much recovery been observed? I would suggest editing to “smaller infected colonies may succumb more quickly to SCTLD”

- Line 406 – perhaps you can expand on what you mean by ‘sufficient replication’? Does this mean replicate videos of each colony to render multiple models per colony?

- Line 413 – suggest ‘most commonly’

- Line 434 – suggest ‘regarding’ rather than ‘relative to’

- I wonder if it is worth commenting on how this model approach may work for other coral growth morphologies. Mcavernosa is a fairly simple mounding shape, but encrusting, foliose or branching corals may be much more difficult to accurately model?

- I have strong reservations about the mitigation technique of culling small colonies that is suggested in line 438 (and 389, see comment above). I understand that they may die more quickly than larger colonies and could act as sources of the disease. But the data presented here do not indicate that they are more susceptible (are there size-frequency distributions of healthy vs/ diseased corals in a population?), and the small colonies are often offspring of the few remaining colonies that have survived multiple selection events already and may be the best hope of the recovery of resilient communities, especially if there is any evidence that colonies can recover from this disease. Thus I much prefer the second option of focusing intervention resources on larger colonies and would urge caution when considering culling of smaller colonies.

- Fig 1. Typo in fig legend – add ‘a’ before ‘colony’

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

Submitted filename: PONE-D-20-25422_reviewer.pdf

Click here for additional data file.

18 Feb 2021

Editor:

Reviewer 1 in particular notes that there is a significant literature on photogrammetry techniques on coral reefs, and further attention to this would improve both the introduction and discussion, rather than relying on citations to other taxa. They also require increased clarity in the methods regarding validation of the accuracy of the model, including what site-specific community parameters would impact the 3D model generation. Reviewer 2 also brings up a salient point here, which is that there was no direct comparison to 2D methods, so claims of increased accuracy are not demonstrated. They do recommend that a relatively rapid post-hoc statistical comparison with 2D screen grabs could be considered, or more cautious language. I would suggest the former to increase the rigor of your study, but either would be sufficient. Reviewer 1 also questions the clarity of statistical analysis, and believes that further investigation into why shape area and error varies with location is either required, or explained in the text.

¬-We took only non-scaled reference photographs and thus cannot conduct a 2D comparison with our 3D data, as such we chose to use more cautious language throughout the manuscript.

Both reviewer 1 and I are curious as to why the grouping of disease prevalence is at the county level rather than the location/site scale, particularly since you do not discuss different management, water quality or land use differences between counties (See Fig 3). By working at the location scale, this would allow comparison of more similar replicates, and would also be more accessible to a person not familiar with the geography of Florida. If you feel that county is an important part of the puzzle that makes it a more appropriate unit of replicate than location (more than just indicating a north - south gradient for example), it would be useful to explain this decision.

-We agree with the editor and reviewers comments and reanalyzed the survey data to group at the Location level.

Both reviewers identify the recommendation of 'culling' as unjustified, or presented with too high a level of certainty, particularly since the effort, efficacy and description of what other interventions are available are not given (if there is only one colony in an area infected, but it is large, why not remove that if it leads to a better outcome than other interventions?). Without evidence presented here, or citations to small colonies dying more quickly or being disease vectors, this management recommendation goes far beyond the work. Indeed, in Fig S2 It could be argued that there is some evidence that some larger colonies might lose tissue at a faster rate than some small colonies. In the discussion you also state that "disease lesion area did not correlate with total colony size suggesting that larger colonies do not exhibit a higher proportion of diseased tissue" - this could imply they show the same proportional loss or lower. I strongly suggest you present more data on proportional loss of colony tissue (not just absolute loss) if this is an argument you wish to make. Even so, suggesting culling small colonies as a management measure would still be extreme and would require a stronger argument with references to small colonies as vectors or disease reservoirs. The argument also begs the question - what is a small colony? Nor are what interventions that might be possible shy of culling the entire colony described.

-We removed the suggestion of culling from the manuscript.

Finally, there is a significant amount of relevant context and data in the SI. It would be good to see some of the tables and figures into the main text, particularly concerning the disease incidence, colony size and disease spread data.

-We revised our results section to remove unnecessary metrics and focused on the more compelling outcomes from this study. We also included additional figures focusing on correlation between colony size and disease area as well as rate of tissue loss.

Reviewer 1:

Reviewer #1: Reviewer comments:

1) from L88: I think applications on cetacean and elasmobranch research are way out of the scope of this work. Better concentrate on referencing all the exciting research on coral reefs.

- We have removed the unnecessary references to cetacean and elasmobranch research and greatly increased our literature review concerning photogrammetry on corals and coral reefs. We appreciate the suggested journal articles.

2) Also, in this section, I don’t think you need to cite aerial and scanner methods to prove your point, and the two studies you cite for coral growth did not use photogrammetry at all. Use Ferreira et al. 2017 and Lange & Perry 2020 if you want to keep this sentence. Alternatively you could explain photogrammetry and SfM first and then give examples of studies on reefs. Also see annotations in the PDF.

- Reference to aerial and scanner methods have been excluded and the focus was shifted to in-water photogrammetry methods, we also corrected the two citations for Ferreira et al. 2017 and Lange & Perry 2020.

3) L. 104: optimized from what basis? You have not developed the photogrammetry technique, but a new application for it. Maybe say: “In this study we used photogrammetry techniques as described in [Young et al. ] in order to develop a new application, i.e to track SCTLD disease progression in M. cavernosa colonies.” or something similar?

-We revised the wording of this to better represent our work (L109).

Methods:

1) see annotations in PDF to improve clarity of the text.

-The PDF annotations were applied, they were very helpful and the authors thank the reviewer for their efforts to improve the clarity and writing throughout this manuscript.

2) L181: state version of Agisoft Metashape that you used as some versions seem to have issues and it will help for repeatability.

-This has been included (L181).

3) L. 215: I do not understand why site-specific “community parameters” (what is this anyways?) or even water quality should affect 3D model generation. This would mean that the method is not accurate and should not be used.

- Our intention was to compare surface area error from models constructed in varying environments (pristine pool versus nearshore coral reef with low visibility and high wave action). To make the model accuracy validation section more clear and applicable to the fate-tracking results, we have focused on comparisons of shape error between the pool models and models made at the fate-tracking sites in Lauderdale-by-the-Sea. We have rephrased this section to be more explicit starting at L214.

Related to this it does not make sense to compare the area of shapes among locations to evaluate the 3D models. So I am wondering why you did not just photograph the mock colony at the sites where you did the fate-tracking in order to calculate the accuracy of your methodology.

- Models with the mock coral colony were constructed at the same fate-tracking sites in the original version of the manuscript, but were not represented as separate sites in the original analysis. To improve the comparability of the model accuracy analyses to the fate-tracking portion, we have focused on the fate-tracking sites in Lauderdale-by-the-Sea only. Likewise, we have removed the analyses of overall shape areas and focused on the shape error associated with each model.

Concerning this there is two parameters you want to check:

A) accuracy, meaning how close to reality is your 3D model. Test this by comparing measured shape areas on your mock corals to known areas and calculate the mean error. (This is potentially how you got your 2.17 cm2?) If you notice that the error is bigger at a more turbid reef site, then you could conclude that the visibility affects the accuracy of model building/measurement. But direct comparison of shape areas among sites does not make sense.

B) precision, meaning how good is the reproducibility of your measurements. For this you should measure the surface of the same shape, or better the same coral colony, several times, including all the hand tracing around colonies etc. Then calculate the error, which will show you if you introduce considerable variability using your workflow.

You should calculate the coefficient of variation (error/average) in order to compare your errors to other studies doing surface area measurements. You cannot compare the error from small shapes (2 cm2) directly to the surface area of the colony.

- We have made substantial efforts in the data analysis and writing to make this section of the manuscript clearer and more applicable to the coral fate-tracking portion. First, we focused on comparison of model error across the pool and the three coral fate-tracking sites in Lauderdale-by-the-Sea. The statistical analyses were revised to a comparison of shape error (i.e. the difference between 3D modeled shape areas and the template) among sites (L214-222, L225-230, L308-3111). Replicate measurements of all the prism shapes were made within the original analyses, but this was not clear in the original version. The prism had five faces with the template shapes (top and four sides), each of which had multiple shapes (four squares and rectangles, one circle per face) (L222-225). Shape error is also represented in two ways, as 1) the absolute difference between measured shape areas and template area, and 2) percent error relative to template area (similar to previous studies, including those you suggested in the original review). The former is used for the statistical tests and for visual assessment of shape error in S4 Fig, and the latter is reported in the text (L311-314). These calculations result in mean errors of 1.70 sq. cm and 6.13%, respectively. The coefficient of variation was not calculated for these analyses, as this is typically used for comparisons of replicate 3D models or measurements. Each of the measurements in our prism dataset are statistically independent, and we feel provide for a robust assessment of model accuracy.

L 240: I do not understand why these correlations were done and what they would tell us about the accuracy of the method.

- This analysis has been removed, and the section has been rewritten according to the revised analysis of accuracy across sites.

Results:

1) I am wondering why did you choose to compare counties instead of locations as shown on the map? The latter would have the advantage that replications are more similar.

-The disease prevalence survey analyses were shifted from county level to location level as shown on the map.

2) L 255-260: I don't think the figure supports these statements. I would say something along the line of “in Broward, disease prevalence stayed quite constant at around 10%, while in Palm Beach prevalence was usually very low, but peaked in March 2019 when ... of colonies were affected. Disease prevalence at Martin was very variable due to low numbers of live coral colonies...”

-This section has been edited to improve clarity and better reflect our results and figures (L238-255).

3) L 273: Has it been taken into account that colonies were measured repeatedly? Considering that the areas of your colonies are very different you might either have do more fancy statistics using repeated measures GLM/GAM (maybe using colony ID as fixed factor and site and time as explaining variables? sorry, not an expert myself) or you might have to calculate “loss of area” or “loss % of area” in order to make meaningful comparisons over time.

I know you calculated rate of change in healthy and diseased tissue, which I think might make more sense than comparing actual area. In general it is getting a bit confusing looking at so many parameters. I would suggest to think carefully which parameters are most useful in showing what you are interested in and rather use fewer but explain better what they mean.

E.g., I think the most interesting results and the best order in the section L277-309 would be (add numbers and statistics): “The rate of tissue loss did not differ among sites but was variable over time, with highest loss in the first observation interval. The diseased area however did stay constant over time, indicating that the lesion moves with the infected tissue.” Then add the correlations you think are useful. Not all are I think.

-The Friedman’s test and subsequent post hoc analysis, the Nemenyi test, are appropriate since our data is not normally distributed; it is a non-parametric equivalent to a repeated measures analysis of variance. The section has been reduced in the number of metrics and tests run to increase overall clarity and focus on important, compelling results. (L289-297). Proportion of loss data was added, and the only correlation that was retained, as the authors thought this was the most compelling, was the correlation between disease lesion area and total colony size.

4) L287 and 288: S2 Fig should be S3 Fig?

- These were removed.

5) L 312-L334: This whole section does not make sense to me. I do not understand why the areas of shapes should be different depending on location. This definitely does not increase trust in the method! It might in part be a relic of your different sample sizes and tests you do. Why do you run ANOVA then Kruskal Wallis and then 3-way PERMANOVA? I fail to understand what any of these significances mean and it is not explained in the results or discussion sections.

- In the revised version of the manuscript, we have streamlined our assessments of model accuracy according to previous suggestions. We hope that the new sections present a better description of the analyses that were conducted. Also, following careful review of the statistical tests and 3D models, we identified a few outliers in the prism dataset that were the result of holes in the mesh leading to inaccurate surface area measurements, which appeared to be driving the significant differences presented in the original manuscript.

I would suggest to think carefully about what parameters actually tell you something about the accuracy/precision of the analysis (see comments above) and rewrite this section completely after the improved analysis.

- Thank you for your constructive comments, we hope that the revised sections address your concerns and suggestions.

Discussion:

This section could be improved by being very clear how the results compare to other studies and what can be said about the implications. See annotations in the PDF

1) L353: It is not clear what you are trying to say here. Also please make clear what you mean with L356-357. Next section can be shortened as suggested in PDF.

- This section, along with the section below, were combined and rewritten to improve clarity (L324-350).

2) L363: This paragraph seems to be repeating discussions from the previous paragraph. Maybe you could combine those better without repeating yourself?

I think the order of these two paragraphs could be improved. E.g. “In the FRT, disease prevalence is typically higher (...%) than elsewhere in the TWA (~1-3%) [57]. Sites in Palm Beach in the present study showed relatively low background disease (6%), similar to prevalence across the NFRT after Hurricane Irma in Sept 2017 (6%???) (Walker 2018, 53). Highest values observed in this study were 20-45% at sites in Martin, but did not reach levels of up to 60% as reported in [17], likely due to low abundance of susceptible species after ongoing SCTLD impacts [53,58]”

-The following section was changed using your suggestions to increase clarity (L332-338).

3) L384: I thought Fig S2 showed that it does NOT correlate? Otherwise your next sentence does not make sense

- This line was edited to correct the typo, the sentence should have read “rates of tissue loss did not correlate” as stated in the reviewer comment.

4) I think the suggestion to cull small colonies is a bit drastic and not supported by your study. If there is other research supporting this approach please cite here.

- The suggestion of culling has been removed from the manuscript

5) L384-393 could be condensed down to 1-2 sentences saying management should target larger colonies.

-This section was revised to omit culling and express the need to target larger colonies.

6) From L 403: This whole sections would have to be revised after analysis of accuracy/precision.

- Following the revised analysis, we have rewritten this section to emphasize the error rate in the prism measurements (L379-382), the potential causes (flexible shape templates and water movement/poor visibility; L382), and suggestions for improving model construction and underwater filming/photography (L384-391).

7) L406: It is not possible to compare the 2.17 cm2 shape error directly to the much larger colonies. Actually, this error seems very high considering that the shape on your mock coral is not very big?! Does that relate to about 10%? Calculate CV and compare to other studies. Also, if you have an error of about 10% you might want to check if the difference of total area/healthy area between time points would still be significant.

- Based on previous studies, we have opted to calculate error rate (%) for shape measurements relative to the corresponding template area. As a result, the global error rate is ~6%.

8) L409: you did not test the effect of colony size on model accuracy. Also, it does not make sense that water quality (do you mean turbidity/visibility?) affects the size of colonies. You are probably trying to say that lower visibility could result in lower model quality affecting measurements of surface area? Be precise in your wording.

- We have revised this section to be more specific regarding surface area measurements and potential effects of low visibility on model construction and error (L389).

Conclusions and general:

1) Different kinds of interventions should be mentioned in the introduction or discussion if it is the main discussion point in the conclusion.

-There has been a wide variety of intervention methods proposed, and refinement and evaluation is still ongoing, so we did not go into detail on the specific methods, but did include them in the introduction (L77-82).

2) L435: Why would interventions be unsuccessful in small colonies? They should work the same, just preserve less total area, right? So I understand prioritizing big ones, but in sites were prevalence is low why not treat small ones too.

-The reference to interventions being unsuccessful was removed and clarification was added to imply that larger colonies should be prioritized if present, but smaller colonies should not be ignored all together.

3) I think the discussion and conclusion sections will benefit from a second round of review after the improved model evaluation. Try to be very clear what the novelty of this work is (new application of photogrammetry method to accurately quantify tissue loss/disease progression over time) and what can be concluded from it (improve survey protocols, evaluate success of intervention methods... I don’t think culling of half the colonies is a good outcome here).

-The discussion and conclusions have been revised to focus more on the novelty of this work, comparable photogrammetry and SCTLD studies, and improving the management and intervention suggestions throughout.

4) If you want people to take up this method, make the workflow easier to access. I know it is all in the GitHub repository, which is awesome, but I did not easily find the step-by-step guide and what you actually did to measure the areas. Maybe you could prepare a one-pager which is easy to print, stating the steps of image acquisition, model building and analysis to go into Suppl Methods. Or put a link in the methods which leads straight to the guide instead of the whole repository.

- A reference to the protocol within our GitHub repository was added to the methods section. (L178)

5) Also, it would be great if you could make the models (as .obj?) available in the repository.

-Unfortunately, the models are too large to be housed on the GitHub repository, they will be available upon request.

Reviewer 2:

General comments

- One of the major findings of this study is that the 3D model method presented is “more accurate data than previously established methods such as two-dimensional surface area estimation (ln 398-399).” I thought the assessment of the accuracy of the 3D method presented here was excellent and rigorous, but there was not a estimate of the disease lesions or the calibration templates/mock coral made from the same two-dimensional photographs to make a robust and unbiased statistical comparison of the two techniques. While I agree that this 3D model pipeline is achievable, it does still require ~40 min per colony of rendering, plus fairly expensive software and hardware for modelling and storage of large video files compared with the much simpler 2D method, so a quantitative comparison of the two methods would strengthen the argument and add justification for using the 3D model method presented here.

-We took 2D reference photographs in order to double check possible abnormalities in the model (eg: if that is a Christmas tree worm or a small disease lesion) they are not scaled and therefore cannot be analyzed and used in a comparison of the two techniques. However, we have incorporated references to previous studies that have compared 3D photogrammetry to 2D and physical methods (tin foil method) that have shown 3D photogrammetry to be a more robust method.

- The inclusion of more data from the disease surveys in the main text could be useful, and would balance the disease ecology with the methodological aspects of the study a bit better. Including which species were (and were not) affected at each site, and how disease prevalence changed over time by taxa within each site would be a nice addition to the main text. Line 365-7 mentions the “low abundance and species richness in Marin County sites compared with PB and Broward Counties”, but what is the density and species richness within each site?

-We feel that adding broader ecological surveys of these reefs outside of SCTLD prevalence is outside the scope of this study. We included more species abundance data, mainly to highlight the lack of ‘highly susceptible’ species within our surveys is possibly due to SCTLD already and how different the coral communities are at SLR relative to the rest of our sites.

- The manuscript is relatively sparse wrt figures and tables, with the vast majority of the information presented as supplementary. I would suggest including a few of these in the main text, particularly Fig S2 and possibly S4, again to balance out the ecology/methodology aspects of the manuscript.

We included additional figures focusing on correlation between colony size and disease area as well as rate of tissue loss.

Several Specific comments

- Line 193 – it is quite tricky to determine whether the stark white area of the lesion is still alive or newly dead, especially in an image/video. Would it be more appropriate to define the lesion area as the ‘stark white coral tissue or very newly dead white skeletal area’? If not, how was this distinction made? It is very difficult to tell in Figure 1, but the white lesion areas look like there is no live tissue to me.

- We revised this to improve clarity (L193)

- Line 282 – consider rounding the estimates of # of lesions per colony to the nearst whole number as that makes the most conceptual sense (or at least to the 10th which is probably a more appropriate significant digit)

-These were removed from the analysis .

- Line 360 – consider stating what the most susceptible species are.

-This has been included (L340)

- Line 384 – was the correlation between rates of tissue loss and total colony size positive or negative? Please specify.

-This was a typo, it should have read “did not” correlate, it has been adjusted (L361)

- Line 388 – typo – suggest ‘greater preservation’

- This section has been rewritten

- Line 389 – is there justification for the statement that smaller colonies are more likely to succumb to SCTLD? It may take less time to succumb due to the rate of tissue loss, but is the likelihood of dying from the disease actually higher? Has much recovery been observed? I would suggest editing to “smaller infected colonies may succumb more quickly to SCTLD”

-This section was heavily revised as suggested and the statement regarding small colonies has been omitted.

- Line 406 – perhaps you can expand on what you mean by ‘sufficient replication’? Does this mean replicate videos of each colony to render multiple models per colony?

-This section has been heavily revised and this line has been removed.

- Line 413 – suggest ‘most commonly’

-This section has been heavily revised and this line has been removed.

- Line 434 – suggest ‘regarding’ rather than ‘relative to’

-This has been changed as suggested (L406)

- I wonder if it is worth commenting on how this model approach may work for other coral growth morphologies. Mcavernosa is a fairly simple mounding shape, but encrusting, foliose or branching corals may be much more difficult to accurately model?

- I have strong reservations about the mitigation technique of culling small colonies that is suggested in line 438 (and 389, see comment above). I understand that they may die more quickly than larger colonies and could act as sources of the disease. But the data presented here do not indicate that they are more susceptible (are there size-frequency distributions of healthy vs/ diseased corals in a population?), and the small colonies are often offspring of the few remaining colonies that have survived multiple selection events already and may be the best hope of the recovery of resilient communities, especially if there is any evidence that colonies can recover from this disease. Thus I much prefer the second option of focusing intervention resources on larger colonies and would urge caution when considering culling of smaller colonies.

-The suggestion of culling has been removed throughout the manuscript.

- Fig 1. Typo in fig legend – add ‘a’ before ‘colony’

-This has been changed as suggested.

Submitted filename: responseToReviewers.docx

Click here for additional data file.

31 Mar 2021

PONE-D-20-25422R1

Quantifying impacts of stony coral tissue loss disease on corals in Southeast Florida through surveys and 3D photogrammetry

PLOS ONE

Dear Dr. Combs,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit and is very close to appropriate standard for publication. Both reviewers are happy with how you have addressed their critiques and comments, but both do point out some minor errors and comments. In reviewer 1's case in particular, many of these are minor typographically errors or slight clarifications, but due to the process at PLoS One, you would not have the opportunity to correct them at a 'proofing' stage. Figure 5a also does not correspond to the Figure legend, and the panel appears to be a duplicate of Figure 6. Reviewer 2 would also like you to consider some additional disease ecology factors, and please see their attached pdf with some comments.

Please submit your revised manuscript by May 15 2021 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Fraser Andrew Januchowski-Hartley, Ph.D.

Academic Editor

PLOS ONE

Journal Requirements:

Please review your reference list to ensure that it is complete and correct. If you have cited papers that have been retracted, please include the rationale for doing so in the manuscript text, or remove these references and replace them with relevant current references. Any changes to the reference list should be mentioned in the rebuttal letter that accompanies your revised manuscript. If you need to cite a retracted article, indicate the article’s retracted status in the References list and also include a citation and full reference for the retraction notice.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: All comments have been addressed

Reviewer #2: (No Response)

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Partly

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: Reviewer 1 Round 2

The authors did a good job addressing my questions and suggestions in their revised version of the manuscript and improved the photogrammetry method section and discussion section significantly. Well done!

Please find below a couple of very minor suggestions I noted down while reading over the manuscript, otherwise I am happy to recommend publication as is.

Check Figures 5 and 6 and the captions and references in the text. Figure 5 includes the disease lesion area plot (but note different font size) but this is not represented in the figure caption and text references and Figure 6 depicts the same plot.

L173: hovered approximately 1 m above

L176: overlap of filmed surface area

L183: (link does not work)

L249: site SLR North in April 2019

L253: What do you mean with species abundance? Overall colony abundance? or number of species? Also, was this parameter higher or lower at SLR (says varying in L254)

L289: tissue area (get rid of s)

L313: This first sentence seems out of place. Maybe better integrates in method section as the reason to use PERMANOVA. Then start with “Absolute and relative shape error did not vary across the pool and ...”

L316: Absolute shape error

L318: corresponded to a relative shape error of

L322: on measured colony surface area

L332-334: consider to move this sentence behind L337 starting with “Other ... observed over similar spatial scales” (as you note similar distance to Belize/Martinique)

L342: no site (get rid of s)

L343: near Miami in ...(year)

L346: comparatively sparse (or absent)

L360: on a coral colony level over just ... over the course of this study

L361: why “in contrast”?

L374: sufficient area (get rid of s)

L379: progression on a colony-level

381: I would not consider this a complex morphology. Would rephrase to “Healthy and diseased surface area of 24 coral colonies were quantified and compared over time”

385: did not vary across different depth and turbidity conditions (S2 Fig.)

L386: Average absolute shape error

L387: to a relative error of ...

L389: rather than representing solid surfaces.

L404: may be able to determine

L406: the progression of disease

L411: what does non-discriminant mean? Do you mean “Compared to other coral diseases, impacts of SCTLD are not yet well described concerning host-specificity and spatio-temporal distribution”? Is that true? Lots of studies out on SCTLD

L412-416: would suggest to shorten and rephrase, e.g.: “Based on the results from this study, larger colonies should be prioritized for SCTLD mitigation measures.”

L417: remove alternative

L419-420:” ... longer timescales and multiple species are desirable to confirm observed patterns ... “ (3D mentioned in next sentence)

L422: not sure what redundancies and precautions you refer to. I suggest “... are a valuable approach for colony fate-tracking if high resolution imagery can be obtained.“

L425: more accurate than what?

Reviewer #2: The manuscript by Combs et al. reports the results of spatial and temporal field surveys of SCTLD in Florida and, using the survey results to inform site selection, undertook fate-tracking of individual Montastrea cavernosa colonies at the most impacted sites. The fate-tracking applied a structure-from-motion 3D photogrammetry technique to quantify rates of tissue loss and changes to lesion area and lesion number through time. The authors then used those data to improve our understanding of disease dynamics and make management recommendations.

The revised manuscript has addressed all the major concerns I raised in my previous review, including removing mention of culling, removing statements of direct comparisons with 2D methods that were not justified by the data, reanalysing the data at the location level, and including some of the supplementary data in the main text. Overall, I think the manuscript has been clarified and streamlined will be a useful addition to the literature on this important topic. I only have a few outstanding comments that warrant a minor review, most notably about the intervention recommendation, and have added some edits and comments in an annotated PDF, which I hope the authors will find useful.

• References: Some key references on disease ecology and SCTLD are missing, including Muller et al. 2020 Spatial Epidemiology of the Stony-Coral-Tissue-Loss Disease in Florida. Front Mar Sci 7:163; Meiling et al. (2020) and stony coral tissue loss disease (SCTLD) lesion progression slows in association with thermal stress. Frontiers in Marine Science, 7, 1128. More references are added in the annotated PDF file, and I urge the authors to review some of the key disease literature for the Caribbean in the discussion of ecological drivers including temperature, local pressures, etc.

• Intervention recommendation: Line 36, Line 374, Line 415: I think it is worth adding a caveat/qualifier here about what the end goal of the intervention is. If the goal is to minimise immediate loss of coral cover, then yes, perhaps it is best to prioritise the largest colonies. But, that doesn’t take into account other on-going processes. Firstly, one could argue that the smaller colonies represent younger and more locally adapted/stress tolerant genotypes, and applying intervention efforts across all colonies in a small high-value (heat tolerant)/well connected (larval source)/high-flow (i.e. pathogen source) area might be better than selecting only larger individuals in a broader area using the same amount of time/resources. Secondly, the disease appears to follow contagious transmission dynamics (Muller et al 2020), so if infected small colonies are left untreated in an area around larger colonies that are treated, it may remain a pathogen source to re-infect the larger colonies; I understand this argument led to your original suggestion of culling but that also doesn’t factor in other processes. I would urge you to add a qualifier as such and soften the language because the statement “should be prioritised” is quite strong when these other factors aren’t considered.

• Methods: Lines 142-147: The sentences “Statistical tests were run in the R statistical environment” and “non-parametric tests were implemented for all analyses unless otherwise noted”… seemed out of place and premature, because it isn’t clear what tests were run and why? I suggest moving line 146 up to start the section with “to assess variation in disease prevalence among sites and survey times”, and then you can explain the methodological details.

• Figures: It appears that Figures 5a and 6 are the same? Also, Figure 5b could be faceted by site to support the argument that tissue loss didn’t differ among sites.

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: Ines Lange

Reviewer #2: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

Submitted filename: PONE-D-20-25422_R1_V2.pdf

Click here for additional data file.

4 May 2021

Response to reviewers:

Reviewer #1: Reviewer 1 Round 2

The authors did a good job addressing my questions and suggestions in their revised version of the manuscript and improved the photogrammetry method section and discussion section significantly. Well done!

Please find below a couple of very minor suggestions I noted down while reading over the manuscript, otherwise I am happy to recommend publication as is.

- Thank you very much for taking the time to review this manuscript again, and for providing additional edits to improve readability. We have addressed your comments in the line-by-line responses below.

Check Figures 5 and 6 and the captions and references in the text. Figure 5 includes the disease lesion area plot (but note different font size) but this is not represented in the figure caption and text references and Figure 6 depicts the same plot.

- Thank you for catching this, this was the result of a small error in the R code used to generate figures. The correct figures have now been included.

L173: hovered approximately 1 m above

- Changed as suggested

L176: overlap of filmed surface area

- Changed as suggested

L183: (link does not work)

- Link has been fixed

L249: site SLR North in April 2019

- Changed as suggested

L253: What do you mean with species abundance? Overall colony abundance? or number of species? Also, was this parameter higher or lower at SLR (says varying in L254)

- Reworded to signify as colony abundance and lower abundance at SLR relative to all other sites.

L289: tissue area (get rid of s)

- Changed as suggested

L313: This first sentence seems out of place. Maybe better integrates in method section as the reason to use PERMANOVA. Then start with “Absolute and relative shape error did not vary across the pool and ...”

- Changed as suggested

L316: Absolute shape error

- Changed as suggested

L318: corresponded to a relative shape error of

- Changed as suggested

L322: on measured colony surface area

- Changed as suggested

L332-334: consider to move this sentence behind L337 starting with “Other ... observed over similar spatial scales” (as you note similar distance to Belize/Martinique)

- Changed as suggested

L342: no site (get rid of s)

- Changed as suggested

L343: near Miami in ...(year)

- Changed as suggested

L346: comparatively sparse (or absent)

- Changed as suggested

L360: on a coral colony level over just ... over the course of this study

- Changed as suggested

L361: why “in contrast”?

- This has been removed to avoid confusion.

L374: sufficient area (get rid of s)

- Changed as suggested

L379: progression on a colony-level

- Changed as suggested

381: I would not consider this a complex morphology. Would rephrase to “Healthy and diseased surface area of 24 coral colonies were quantified and compared over time”

- Changed as suggested

385: did not vary across different depth and turbidity conditions (S2 Fig.)

- Changed as suggested

L386: Average absolute shape error

- Changed as suggested

L387: to a relative error of ...

- Changed as suggested

L389: rather than representing solid surfaces.

- Changed as suggested

L404: may be able to determine

- Changed as suggested

L406: the progression of disease

- Changed as suggested

L411: what does non-discriminant mean? Do you mean “Compared to other coral diseases, impacts of SCTLD are not yet well described concerning host-specificity and spatio-temporal distribution”? Is that true? Lots of studies out on SCTLD

- This statement has been rewritten to emphasize that SCTLD has broad taxonomic and temporal impacts relative to other coral diseases.

L412-416: would suggest to shorten and rephrase, e.g.: “Based on the results from this study, larger colonies should be prioritized for SCTLD mitigation measures.”

- Changed as suggested

L417: remove alternative

- Changed as suggested

L419-420:” ... longer timescales and multiple species are desirable to confirm observed patterns ... “ (3D mentioned in next sentence)

- Changed as suggested

L422: not sure what redundancies and precautions you refer to. I suggest “... are a valuable approach for colony fate-tracking if high resolution imagery can be obtained.“

- Changed as suggested

L425: more accurate than what?

- Added language to clarify

Reviewer #2: The manuscript by Combs et al. reports the results of spatial and temporal field surveys of SCTLD in Florida and, using the survey results to inform site selection, undertook fate-tracking of individual Montastrea cavernosa colonies at the most impacted sites. The fate-tracking applied a structure-from-motion 3D photogrammetry technique to quantify rates of tissue loss and changes to lesion area and lesion number through time. The authors then used those data to improve our understanding of disease dynamics and make management recommendations.

The revised manuscript has addressed all the major concerns I raised in my previous review, including removing mention of culling, removing statements of direct comparisons with 2D methods that were not justified by the data, reanalysing the data at the location level, and including some of the supplementary data in the main text. Overall, I think the manuscript has been clarified and streamlined will be a useful addition to the literature on this important topic. I only have a few outstanding comments that warrant a minor review, most notably about the intervention recommendation, and have added some edits and comments in an annotated PDF, which I hope the authors will find useful.

- Thank you very much for taking the time to review this manuscript again, and for providing additional edits to improve readability. We have addressed your comments in the line-by-line responses below.

• References: Some key references on disease ecology and SCTLD are missing, including Muller et al. 2020 Spatial Epidemiology of the Stony-Coral-Tissue-Loss Disease in Florida. Front Mar Sci 7:163; Meiling et al. (2020) and stony coral tissue loss disease (SCTLD) lesion progression slows in association with thermal stress. Frontiers in Marine Science, 7, 1128. More references are added in the annotated PDF file, and I urge the authors to review some of the key disease literature for the Caribbean in the discussion of ecological drivers including temperature, local pressures, etc.

- Thank you for providing additional references to strengthen our conclusions. The Meiling et al. (2020) study was referenced in the first revision, but a glitch with our reference manager software caused a replacement of the Muller et al. (2020) study. We have added it back in, and have included your additional suggestions.

• Intervention recommendation: Line 36, Line 374, Line 415: I think it is worth adding a caveat/qualifier here about what the end goal of the intervention is. If the goal is to minimise immediate loss of coral cover, then yes, perhaps it is best to prioritise the largest colonies. But, that doesn’t take into account other on-going processes. Firstly, one could argue that the smaller colonies represent younger and more locally adapted/stress tolerant genotypes, and applying intervention efforts across all colonies in a small high-value (heat tolerant)/well connected (larval source)/high-flow (i.e. pathogen source) area might be better than selecting only larger individuals in a broader area using the same amount of time/resources. Secondly, the disease appears to follow contagious transmission dynamics (Muller et al 2020), so if infected small colonies are left untreated in an area around larger colonies that are treated, it may remain a pathogen source to re-infect the larger colonies; I understand this argument led to your original suggestion of culling but that also doesn’t factor in other processes. I would urge you to add a qualifier as such and soften the language because the statement “should be prioritised” is quite strong when these other factors aren’t considered.

- We have modified these statements to modify the language and to include caveats where relevant. Interventions have been proposed as a way to slow down immediate loss of coral cover by aiding in colony survival, but the groups leading these efforts recognize that intervention is not a sustainable approach over the long-term response to the disease outbreak. We have indicated this in the introduction and discussion at L81 and L375, respectively. Additionally, we have added a caveat in the conclusions at L413 stating that intervention efforts on larger colonies should be prioritized when resources are limited.

• Methods: Lines 142-147: The sentences “Statistical tests were run in the R statistical environment” and “non-parametric tests were implemented for all analyses unless otherwise noted”… seemed out of place and premature, because it isn’t clear what tests were run and why? I suggest moving line 146 up to start the section with “to assess variation in disease prevalence among sites and survey times”, and then you can explain the methodological details.

- Changed as suggested

• Figures: It appears that Figures 5a and 6 are the same? Also, Figure 5b could be faceted by site to support the argument that tissue loss didn’t differ among sites.

- Thank you for catching this, this was the result of a small error in the R code used to generate figures. The correct figures have now been included, and Figure 5b has been faceted by site.

Submitted filename: responseToReviewersV2.docx

Click here for additional data file.

19 May 2021

Quantifying impacts of stony coral tissue loss disease on corals in Southeast Florida through surveys and 3D photogrammetry

PONE-D-20-25422R2

Dear Dr. Combs,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Fraser Andrew Januchowski-Hartley, Ph.D.

Academic Editor

PLOS ONE

18 Jun 2021

PONE-D-20-25422R2

Quantifying impacts of stony coral tissue loss disease on corals in Southeast Florida through surveys and 3D photogrammetry

Dear Dr. Combs:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Fraser Andrew Januchowski-Hartley

Academic Editor

PLOS ONE