Mapping seagrass with BioBase and saving Florida Manatees

Guest Blog by Robert M. Baker, CPG, PG (a) and Penelope R. Baker (b)

(a) Professional geologist at RMBAKER LLC and Navico BioBase Ambassador

(b) Stanford University student, Ecology and Evolutionary Biology, Wildlife Photographer

BioBase is a cloud software that directly supports the preservation of our aquatic environments. Words like preservation and conservation directly imply things like careful planning, measuring and monitoring, treatment and rehabilitation – actionable strategies for the good of animals, plants and natural resources where BioBase can play an important role. BioBase offers an opportunity to observe natural systems, like seagrasses, not easily seen otherwise and does so effortlessly and affordably.

Despite occupying only 0.1 percent of the ocean floor, seagrass meadows store up to 18 percent of the ocean’s carbon, making seagrass preservation a key strategy in mitigating climate change (e.g., “Blue Carbon” initiatives).

However, according to rough estimates from the UN Environment Program, seagrasses are declining globally by approximately 7 percent per year. Very few technologies and efforts are currently underway to map the status and trends of seagrass survival more confidently.

Seagrasses that occupy some 2.5 million acres of shallow estuaries in Florida are mapped by aerial photography typically every few years. The Florida Fish and Wildlife Conservation Commission (FWC) Seagrass Integrated Mapping and Monitoring (SIMM) program collects and centralizes the distribution of seagrass condition reports statewide, although most of the actual mapping and reporting activities are derived through the efforts and budgets of the state’s five water management districts. The last SIMM reports for Sarasota Bay and the northern Indian River Lagoon (IRL) were published in 2016 and 2018, respectively.

The West Indian manatee (Trichechus manatus) is an herbivore that depends on seagrass for its diet. Manatee mortalities have risen from 520 in 2016 to 1101 in 2021, up to nearly 10 percent of the total statewide population of these charismatic megafauna (FWC 2021 #2; Figure 1). Many manatee deaths were attributed to common causes such as watercraft collisions, entanglement in flood control structures, cold stress or natural causes, etc. However, the 2021 spike, particularly in eastern counties along the IRL, was classified by FWC as an “unusual mortality event (UME)” likely linked to starvation. The IRL, along with other coastal estuaries, has experienced significant losses of seagrasses in recent years due to water pollution, algal blooms, extreme weather and heat.

Figure 1. A basemap of Florida counties colored according to the total number of recorded 2021 manatee deaths as per the FWC website. Also shown are BioBase test areas visited by the author.

Brevard County recorded the most 2021 manatee deaths in Florida at 359 (FWC 2021 #1). In the absence of abundant forage short-term feeding efforts, led by FWC and partners, continue to provide wintertime access to food. This temporary feeding response was recently deployed in Brevard County as an emergency measure, with manatees eating at the response station for the first time on January 20, 2022 (FWC 2022 #1). As many as 140 manatees have been rescued statewide in 2021 with some sent to SeaWorld® in Orlando, FL for rehabilitation. State scientists have historically resisted widespread replanting of seagrasses until estuary water quality trends could be reversed with notable improvement on a grand scale, but the UME has prompted the FWC to reevaluate such a long term strategy regarding habitat restoration (FWC 2022 #1). Interestingly, there is currently a robust debate among scientists, activists, lobbyists and legislators regarding proposals to establish seagrass mitigation banking in public waterways despite the hindrances for seagrass survival presented by pervasively poor water quality conditions (Pittman, 2022).  BioBase would be a powerful tool to monitor the effectiveness of any future seagrass mitigation banking programs.

In 2018 the SIMM report for the northern IRL indicated a 56 percent loss of seagrass acreage from 2009 to 2017 (Morris et al., 2018). The most recent GIS data layers for seagrasses in the IRL were from St. Johns River Water Management District (SJRWMD) 2019 aerial surveys. Data from scheduled 2021 surveys was not publicly available as of this writing. Visual observations and reconnaissance sonar data (Figure 2) collected by the author in 2021 while fishing in Mosquito Lagoon (northernmost estuary of IRL system) suggested the continuation of widespread die-off’s of seagrass subsequent to the 2019 aerial flights and beyond the losses reported in the 2018 SIMM report.

Figure 2. (a) Shown is a BioBase reconnaissance survey situated in the northern mangrove archipelago of Mosquito Lagoon at the northern end of the IRL. The SJRWMD 2019 aerial survey indicated the pervasive presence of “continuous seagrass” meadows, with some isolated areas without grasses that were targeted as part of the test. (b) In the spring of 2021 sonar data from within the proposed testing area did not demonstrate any patterns associated with seagrasses.

BioBase EcoSound Seagrass Mapping Proof-of-Concept

BioBase mapping techniques were used in a small test area within Sarasota Bay, located in Sarasota County just to the south of Tampa Bay on the Gulf coast of Florida, as a possible proof-of-concept exploration for the BioBase technology. Sarasota Bay was chosen primarily because seagrasses were known to be present and Southwest Florida Water Management District (SWFWMD) datasets provided GIS reference polygons from as recently as 2020.

Figure 3 shows the transect data collected across an area of known seagrass with variable water depths. Tide-adjusted water depths ranged from about 4 to 9 feet during the survey. Tidal adjustments were automatically performed by BioBase during the initial processing of each sonar log file after import. A tidal station in Bradenton, FL (Manatee County) just to the north of Sarasota Bay was used to make corrections. The survey work began on a low tide where water depths were below the MLLW 0 foot datum and ended about 1.14 feet above MLLW. The total tidal range across the entire time span of the survey work was 1.38 feet.

Figure 3. Three sonar logs were collected in Sarasota Bay along 5.2 miles of transects in a few hours of water time. Survey speed was about 5 to 6 mph. The sequence of data collection was blue track to red track to yellow track, with a midday time gap between the blue and red. Note the dark colors associated with seagrass meadows atop the central sand bar.

Figure 4 shows a screen capture image from a Lowrance HDS-16 Live used to collect the Sarasota Bay data. While navigating a search-and-rescue reference grid, sonar logs at 200Khz and 800Khz were continuously monitored. An important part of the sonar monitoring was to ensure that the bottom tracking was reasonable and accurate. The bottom tracking line is shown in the figure mapped on the 200Khz sonar panel. Other important values to monitor during a survey are speed, depth and voltage supply. To maintain time efficiency a speed of 6 to 7 mph is necessary, but for vegetation mapping speed must never exceed 10 mph. When surveying waters that can become shallow in a hurry it’s important to keep track of depth. Maintaining a proper voltage supply to the entire sonar system is also a critical part of performing sonar surveys for extended hours.

Figure 4. Shown is a screen-capture of the Lowrance HDS-16 Live monitor during the 2021 test survey of Sarasota Bay (note the sonar feed came through a networked HDS-9 Live). Notice that the survey track (yellow) and the east-to-west search-and-rescue grid lines (black) are spaced about 125-130 feet apart. The vessel was located at the end of the yellow track shown as an arrow surrounded by a circular compass reference dial. The sonar data is shown as an 800Khz downscan image in the top right panel, and a 200Khz primary image with bottom tracking on the bottom right panel.

Standard BioBase outputs from the Sarasota Bay test survey are shown in Figure 5. The three images shown (bathymetry, hardness and vegetation biovolume, respectively) represent typical image outputs from within the online BioBase app. Each image represents a grid model created using a sophisticated kriging algorithm, with point data along each survey transect used as the input data into the model.

The primary bathymetric feature in Figure 5(a) was an approximately north-to-south trending shoaling sandbar, mostly parallel to the eastern shoreline, with water depths atop the bar at approximately 6 to 7 feet. This feature was visible in color aerial photographs (FDOT 2020 images with a resolution of 0.5 ft per pixel) as a dark region of reflectivity surrounded by lighter colored sands (see Figure 3). This darker coloring as seen in Figure 3 generally corresponded to the presence of seagrasses on the shoaling feature.

Figure 5. (a) The Sarasota Bay test area resultant grid model of bathymetry in 1-foot contour intervals. (b) Hardness as an index value from 0 (light tan) to 0.5 (brownish red) showed softer bottom sediments in shallower water areas. (c) Vegetation was mapped primarily in areas of shallower water atop a shoaling sand bar, but pockets of “patchy” grasses were noted in slightly deeper waters.

Similar north-to-south trending features are noted in the hardness and vegetation maps (Figures 5[b] and 5[c], respectively), but with slightly different shapes suggesting that bottom composition and grass growth were not entirely controlled by the physiography of the bottom. The presence of vegetation atop the sandbar appears to correspond with a softer bottom, while patchy pockets of seagrass in deeper waters on the flanks of the sandbar were often found associated with harder bottoms. This relationship may certainly have been caused by a variety of viable ecological factors (such as vegetation species and associated substrate preference), but was most likely a sonar artifact created by an understandably lesser-than-sharp bottom contact where vegetation was both thicker and taller atop the sandbar.

GIS layers obtained from the SWFWMD 2020 aerial photo mapping of seagrasses are overlain across the Sarasota Bay study area as an independent reference (Figure 6). The three notable areas mapped by the SWFWMD survey indicated continuous seagrass meadows, areas of patchy and discontinuous seagrass growth, and a third area of different spectral color response that was presumably ground-truthed by the surveyor to correspond to something submerged but unrelated to seagrass. On several occasions during our Sarasota Bay test survey a dark colored bottom was observed even when no grasses were visible in the sonar, and so the SWFWMD “submerged but not seagrass” pattern may correspond to this observation.

Figure 6. GIS layer data from the SWFWMD 2020 aerial photo survey of seagrasses indicated three notable regions within the Sarasota Bay test area. These patterns consisted of “continuous seagrass”, “patchy seagrass” and some other coloring interpreted to be “submerged” but “other than seagrass.”

The BioBase point data for vegetation was exported and gridded by third party software (Surfer®) using a trend-driven kriging approach. The flexibility to export and process BioBase datasets for customized applications is one the great features of the BioBase app. The biovolume percentages were displayed as a colored contour model layer with the SWFWMD 2020 aerial survey GIS layers as a basemap. Vegetation contours less than 4 percent were rendered transparent in order to highlight the extents of taller and more notable grasses detected by the sonar survey.  This graphical processing manipulation was arbitrarily chosen for this publication, and it should be noted that actual use of BioBase for assessment and monitoring of aquatic resources requires a rigorous approach with standardization of field equipment and processing techniques (Radomski and Holbrook, 2015).

There is remarkable consistency and correlation between the BioBase derived vegetation extents and the extent of “continuous seagrass” mapped by the SWFWMD 2020 survey (Figure 6). The aerial photo surveys had the advantages of being large-scale continuous spatial datasets processed by a toolbox of spectral analytical techniques. The BioBase analyses had the advantage of being derived from proven acoustic in situ geophysical observations, albeit along transects spaced some 125-130 feet apart. Both methods appear very effective where the water was shallowest, where the vegetation formed a contiguous body, and where the coloring of the aerial imagery showed strong contrasts.

In the case of “patchy seagrass”, the BioBase derived maps appear to reveal more reliable details. In all cases where sonar data was collected within a 2020 SWFWMD patchy seagrass polygon, there was little to no seagrass to be observed by sonar in the fall of 2021. The sonar logs did show occasional localized thin veneers, or residuum, of possible vegetative matter occupying less than 4 percent of the water column. Additional patchy seagrass was mapped by the BioBase sonar logs in areas where the 2020 aerial photo analysis indicated vegetation-free “water.” It may be possible that the inconsistencies observed in the mapping of patchy seagrass were entirely a result of actual habitat changes within the test area from 2020 to late 2021, but the likely answer is that in situ measurements using sonar presented a superior result when aerial spectral analyses were less certain.

Figure 7. The SWFWMD 2020 GIS layers overlain by the BioBase generated seagrass regions were remarkably consistent atop the central sandbar and the eastern flats where there was continuous seagrass. The BioBase sonar results generally did find and map patchy seagrasses at various locations, but generally not where they were mapped by the 2020 SWFWMD survey.
Figure 8. Locations of sonar image examples (Figure 8) mapped with both the 2020 SWFWMD GIS layers and the 2021 BioBase vegetation layer.

The locations of excerpted sonar images are shown in Figure 8, and the sonar images themselves in Figure 9. Unlike the images in Figure 4 that were captured while the survey was in progress, the Figure 9 images are parts of a BioBase processed sonar file that can be viewed and edited while using the online app. For each sonar image example, the mapped colored dots (small crosses) formed the BioBase derived points along each transect that were then used to create the grids. For this survey these points were generally spaced by the app to be about 6.5 feet apart, but that spacing will tend to vary depending on the range of bathymetric relief. When no vegetation was detected within the water column, the BioBase app mapped both a bottom point and a vegetation point at the same location. When vegetation was detected, the vegetation point was mapped at the same location as a bottom point but higher up within the water column. The biovolume percentage calculated by BioBase was derived from the depth differences of these two points. Observation of each sonar image suggested a remarkable level of precision by the BioBase app at picking both the bottom and the top of vegetation, even with most vegetation occupying less than about 30 percent of the water column.

In Figure 9(a) the bottom picks were very slightly elevated up into the vegetative canopy. This had little impact on the biovolume percentage, but may have biased the hardness model along the central sandbar ever so slightly toward a softer bottom than warranted by the sediments alone. Bottom hardness can be an expression of silt and organic matter build-up as a thin veneer, but the accumulation of fine sediments will typically occur in lower energy environments sometimes aligned with relatively deeper water. When the vegetation is thick the bottom tracking algorithm may wander upward and generate a pick amplitude slightly lower than that of the actual bottom. This lower amplitude of the bottom pick can influence the indexing of bottom hardness. Edits to the vertical position of the bottom pick can be performed within the BioBase app if an adjustment is desired or warranted.

In Figure 9(c) the downscan image at 800Khz was displayed rather than the standard 200Khz data. The higher resolution of the downscan channel may prove more beneficial when interpreting small biovolume percentages, or when working in shallower waters. The downscan channel can only be used to interpret the vegetation picks, with all bottom tracking interpretations always derived from the 200Khz sonar logs. Future versions of the BioBase app may allow a rescan of the bottom pick, and perhaps this step will be available for the downscan signal.

Figure 9(d) and (e) show patchy seagrasses not mapped by the 2020 SWFWMD aerial surveys. Figure 9(e) in particular shows areas of no vegetation west of a mapped patch. The no vegetation region was mapped as patchy seagrass in 2020.

The observations presented as part of our Sarasota Bay test survey confirm that BioBase is a robust seagrass assessment tool. Other remote sensing assessment techniques in deep or turbid environments cannot assess seagrass abundance to the level of detail that BioBase can. Timely manatee conservation measures cannot be implemented without a firm understanding of the status and trends of seagrass growth (and taking actions to reverse declines). Social mapping techniques that take advantage of C-MAP Genesis and BioBase databases may be able to map seagrass boundaries effectively, perhaps as an early warning of die-offs leading to future UME’s, in a fashion similar to the effective mapping of bryozoan reef systems in Australia (Flynn and Dutta, 2021). BioBase truly offers a powerful but low-cost, easy-to-use tool in the marine explorer’s toolbox.

Figure 9. (a), (b) and (c) These sonar images are captured from the BioBase app output, and indicate seagrasses where they were continuous atop the shoaling sand bar. (d) This BioBase app captured image shows patchy seagrasses northeast of the sand bar in an area of deeper water where the 2020 aerial photo survey indicated only “open water.” (e) This BioBase app captured sonar image indicates thin patchy seagrasses that coincided with a 2020 aerial photo “open water” area, while 2020 patchy seagrasses were not seen on the western portion of the same transect.

Works cited

FWC 2021 #1 – Preliminary 2021 Manatee Mortality Table by County.

FWC 2021 #2 – 2021 Preliminary Manatee Mortality Table with 5-Year Summary

FWC 2033 #1 – 2022 Manatee Mortality Event Along the East Coast: 2020-2022 Weekly Updates and FAQ

Flynn, A. and Dutka, T. (2021). Social mapping of Australian bays and conservation of Fish Aggregating Bryozoans. guest blog post, October 14, 2021.

Greene, A., Rahman, A. F., Kline, R., & Rahman, M. S. (2018). Side scan sonar: A cost-efficient alternative method for measuring seagrass cover in shallow environments. Estuarine, Coastal and Shelf Science, 207(November 2017), 250–258.

Morris, L, Hall. L, Chamberlain, R., and Jacoby, C. (2018). Summary Report for the Northern Indian River Lagoon, in Seagrass Integrated Mapping and Monitoring Program Mapping and Monitoring Report No. 3, Yarbro, L. and Carlson Jr., P., eds.  FWC Technical Report 17, Version 3.

Pérez Espinosa, I., Gallegos Martínez, M. E., Ressl, R. A., Valderrama Landeros, L. H., & Cárdenas, G. H. (2019). Spatial distribution of seagrasses and submerged aquatic vegetation of los Petenes, Campeche. Terra Digitalis, 3(2), 1–11.

Pittman, C., (2022). Bills would help consultants make a killing off of killing off Florida’s seagrass. Florida Phoenix online commentary on January 20, 2022.

Radomski, P., and Holbrook, B. (2015). A comparison of two hydroacoustic methods for estimating submerged macrophyte distribution and abundance: A cautionary note. Journal of Aquatic Plant Management, v. 53, pages 151-159.

Rahnemoonfar, M., Rahman, A. F., Kline, R. J., and Greene, A. (2018). Peer-Reviewed Technical Communication Automatic Seagrass Disturbance Pattern Identification on Sonar Images. IEEE Journal of Oceanic Engineering, 1(March).

Author: biobasemaps

BioBase is a cloud platform for the automated mapping of aquatic habitats (lakes, rivers, ponds, coasts). Standard algorithms process sonar datafiles (EcoSound) and high resolution satellite imagery (EcoSat). Depth and vegetation maps and data reports are rapidly created and stored in a private cloud account for analysis, and sharing. This blog highlights a range of internal and external research, frequently asked questions, feature descriptions and highlights, tips and tricks, and photo galleries.

Leave a Reply

Translate »
%d bloggers like this: