Nova Southeastern University NSUWorks

HCNSO Student Theses and Dissertations HCNSO Student Work

7-19-2019 Microbial Analysis of Surfactant-Associated Bacteria in the Sea Surface Microlayer and Remote Sensing of Associated Slicks Georgia Parks Nova Southeastern University, [email protected]

Follow this and additional works at: https://nsuworks.nova.edu/occ_stuetd Part of the Marine Biology Commons, and the Oceanography and Atmospheric Sciences and Meteorology Commons Share Feedback About This Item

NSUWorks Citation Georgia Parks. 2019. Microbial Analysis of Surfactant-Associated Bacteria in the Sea Surface Microlayer and Remote Sensing of Associated Slicks. Master's thesis. Nova Southeastern University. Retrieved from NSUWorks, . (515) https://nsuworks.nova.edu/occ_stuetd/515.

This Thesis is brought to you by the HCNSO Student Work at NSUWorks. It has been accepted for inclusion in HCNSO Student Theses and Dissertations by an authorized administrator of NSUWorks. For more information, please contact [email protected]. Thesis of Georgia Parks

Submitted in Partial Fulfillment of the Requirements for the Degree of

Master of Science M.S. Marine Biology

Nova Southeastern University Halmos College of Natural Sciences and Oceanography

July 2019

Approved: Thesis Committee

Major Professor: Alexander Soloviev, Ph.D.

Committee Member: Joana Figueiredo, Ph.D.

Committee Member: Aurelien Tartar, Ph.D.

This thesis is available at NSUWorks: https://nsuworks.nova.edu/occ_stuetd/515

HALMOS COLLEGE OF NATURAL SCIENCES AND OCEANOGRAPHY

Microbial Analysis of Surfactant-Associated Bacteria in the Sea Surface Microlayer and Remote Sensing of Associated Slicks

By

Georgia Parks

Submitted to the Faculty of Halmos College of Natural Sciences and Oceanography in partial fulfillment of the requirements for the degree of Master of Science with a specialty in:

Marine Biology

Nova Southeastern University

Table of Contents List of Figures ...... 5

List of Tables ...... 7

Abstract ...... 9

Acknowledgements ...... 10

1. Introduction ...... 11

a. Sea Surface Microlayer ...... 11

b. Biosurfactants and Oil in the Sea Surface Microlayer ...... 12

c. Synthetic Aperture Radar ...... 13

d. Sea Surface Microlayer in Coral Reef Areas ...... 13

e. Sea Surface Microlayer and Diurnal Cycle ...... 14

2. Statement of Objectives ...... 15

3. Hypotheses ...... 15

4. Significance of Work ...... 15

5. Methods ...... 16

a. Sampling Method ...... 16

b. Experiment Overview ...... 17

c. DNA Analysis ...... 20 i. DNA Extraction ...... 20 ii. Next-Generation Sequencing ...... 20 iii. Statistical Analysis ...... 21

6. Results and Discussion ...... 22

a. Synthetic Aperture Radar Imagery Results ...... 23 i. SPLASH ...... 23 ii. Looe Key ...... 24

b. Synthetic Aperture Radar Imagery Analysis ...... 27 i. SPLASH ...... 27

3 ii. Looe Key ...... 27

c. Overall Bacteria Sequencing Results ...... 28 i. Illumina MiSeq DNA Sequencing Outputs ...... 28 ii. Beta Diversity ...... 31

d. Overall Bacteria Sequencing Analysis ...... 33 i. SPLASH ...... 33 ii. Looe Key ...... 33

e. Specific Bacteria Abundance Results ...... 34 i. SPLASH ...... 36 ii. Looe Key ...... 47

f. Specific Bacteria Abundance Analysis ...... 63 i. SPLASH ...... 64 ii. Looe Key ...... 67

8. Conclusions ...... 70

9. References ...... 72

10. Supplementary Material ...... 79

4 List of Figures Figure 1. Sea surface microlayer structure ...... 11 Figure 2. SAR signal scattering ...... 13 Figure 3. Filter collected from oil slick covered in brown film ...... 18 Figure 4. Three different types of oil seen during sample collection ...... 18 Figure 5. All locations of sample collection in the Gulf of Mexico during SPLASH ...... 19 Figure 6. All locations of sample collection during the Looe Key experiment ...... 19 Figure 7. Top-down view of muddy seagrass floor near Lois Key ...... 20 Figure 8. Sampling locations A-G from the SPLASH experiment in the Gulf of Mexico ...... 24 Figure 9. RADARSAT-2 SAR image on July 30th, 2018 at 11:14 UTC over Looe Key ...... 25 Figure 10. Polarimetric RADARSAT-2 image from July 30th, 2018 at 11:14 UTC focused on Lois Key area ...... 25 Figure 11. RADARSAT-2 SAR image on July 31st, 2018 at 23:34 UTC over Looe Key ...... 26 Figure 12. RADARSAT-2 SAR image on August 2nd, 2018 at 11:17 UTC over Looe Key .... 26 Figure 13. Alpha rarefaction curve for all Types of samples from SPLASH ...... 29 Figure 14. Alpha rarefaction curve for all Types of samples from Looe Key ...... 30 Figure 15. Bray Curtis PCoA plot showing Types of Samples at Sites during SPLASH ...... 31 Figure 16. Bray Curtis PCoA plot showing Types of Samples at Sites during Looe Key experiment ...... 32 Figure 17. Bray Curtis PCoA plot showing times of sampling during the Looe Key project .. 33 Figure 18. SPLASH results showing normalized ASVs per site for surfactant-associated bacterial genera ...... 36 Figure 19. Abundance of Alcanivorax from SPLASH ...... 37 Figure 20. Abundance of Alteromonas from SPLASH ...... 37 Figure 21. Abundance of Arthrobacter from SPLASH ...... 38 Figure 22. Abundance of Enterococcus from SPLASH ...... 38 Figure 23. Abundance of Halomonas from SPLASH ...... 39 Figure 24. Abundance of Tenacibaculum from SPLASH ...... 40 Figure 25. Abundance of Fusibacter from SPLASH ...... 41 Figure 26. Abundance of Hyphomonas from SPLASH ...... 41 Figure 27. Abundance of Methylophaga from SPLASH ...... 42

5 Figure 28. Abundance of Oleispira from SPLASH ...... 43 Figure 29. Abundance of Oleibacter from SPLASH ...... 43 Figure 30. Abundance of Rhodococcus from SPLASH ...... 44 Figure 31. Abundance of Synechococcus from SPLASH ...... 45 Figure 32. Abundance of Prochlorococcus from SPLASH ...... 45 Figure 33. Abundance of Vibrio from SPLASH ...... 46 Figure 34. Looe Key results showing normalized ASVs per site for surfactant-associated bacterial genera ...... 47 Figure 35. Abundance of Alcanivorax from Florida Keys experiment ...... 48 Figure 36. Effect of diurnal cycle on abundance of Alcanivorax from Looe Key ...... 48 Figure 37. Abundance of Alteromonas from Florida Keys experiment ...... 49 Figure 38. Effect of diurnal cycle on abundance of Alteromonas from Looe Key ...... 49 Figure 39. Abundance of Halomonas from Florida Keys experiment ...... 50 Figure 40. Effect of diurnal cycle on abundance of Halomonas from Looe Key ...... 50 Figure 41. Abundance of Tenacibaculum from Florida Keys experiment ...... 51 Figure 42. Effect of diurnal cycle on abundance of Tenacibaculum from Looe Key ...... 51 Figure 43. Abundance of Roseovarius from Florida Keys experiment ...... 52 Figure 44. Effect of diurnal cycle on abundance of Roseovarius from Looe Key ...... 52 Figure 45. Abundance of Winogradskyella from Florida Keys experiment ...... 53 Figure 46. Effect of diurnal cycle on abundance of Winogradskyella from Looe Key ...... 53 Figure 47. Abundance of Fusibacter from Florida Keys experiment ...... 54 Figure 48. Effect of diurnal cycle on abundance of Tenacibaculum from Looe Key ...... 54 Figure 49. Abundance of Hyphomonas from Florida Keys experiment ...... 55 Figure 50. Effect of diurnal cycle on abundance of Hyphomonas from Looe Key ...... 55 Figure 51. Abundance of Methylophaga from Florida Keys experiment ...... 56 Figure 52. Abundance of Oleibacter from Florida Keys experiment ...... 57 Figure 53. Effect of diurnal cycle on abundance of Oleibacter from Looe Key ...... 57 Figure 54. Abundance of Oleiphilus from Florida Keys experiment ...... 58 Figure 55. Effect of diurnal cycle on abundance of Oleiphilus from Looe Key ...... 58 Figure 56. Abundance of Erythrobacter from Florida Keys experiment ...... 59 Figure 57. Effect of diurnal cycle on abundance of Erythrobacter from Looe Key ...... 59

6 Figure 58. Abundance of Synechococcus from Florida Keys experiment ...... 60 Figure 59. Effect of diurnal cycle on abundance of Synechoccocus from Looe Key ...... 61 Figure 60. Abundance of Prochlorococcus from Florida Keys experiment ...... 61 Figure 61. Effect of diurnal cycle on abundance of Prochloroccocus from Looe Key ...... 62 Figure 62. Abundance of Vibrio from Florida Keys experiment ...... 62 Figure 63. Effect of diurnal cycle on abundance of Vibrio from Looe Key ...... 63

List of Tables Table 1. SPLASH project sample collection summary ...... 22 Table 2. Looe Key project sample collection summary ...... 23 Table 3. List of all surfactant- and oil-associated bacteria found in SPLASH and Looe Key samples ...... 35 Table S1. Alcanivorax confidence interval calculation from Florida Keys ...... 79 Table S2. Alteromonas confidence interval calculation from Florida Keys ...... 80 Table S3. Halomonas confidence interval calculation from Florida Keys ...... 81 Table S4. Tenacibaculum confidence interval calculation from Florida Keys ...... 82 Table S5. Roseovarius confidence interval calculation from Florida Keys ...... 83 Table S6 Winogradskyella confidence interval calculation from Florida Keys ...... 84 Table S7. Fusibacter confidence interval calculation from Florida Keys ...... 85 Table S8. Hyphomonas confidence interval calculation from Florida Keys ...... 86 Table S9. Methylophaga confidence interval calculation from Florida Keys ...... 87 Table S10. Vibrio confidence interval calculation from Florida Keys ...... 88 Table S11. Oleibacter confidence interval calculation from Florida Keys ...... 89 Table S12. Oleiphilus confidence interval calculation from Florida Keys ...... 90 Table S13. Erythrobacter confidence interval calculation from Florida Keys ...... 91 Table S14. Synechococcus confidence interval calculation from Florida Keys ...... 92 Table S15. Prochlorococcus confidence interval calculation from Florida Keys ...... 93 Table S16. Alcanivorax confidence interval calculation from SPLASH ...... 94 Table S17. Alteromonas confidence interval calculation from SPLASH ...... 95 Table S18. Arthrobacter confidence interval calculation from SPLASH ...... 96

7 Table S19. Enterococcus confidence interval calculation from SPLASH ...... 97 Table S20. Halomonas confidence interval calculation from SPLASH ...... 98 Table S21. Tenacibaculum confidence interval calculation from SPLASH ...... 99 Table S22. Fusibacter confidence interval calculation from SPLASH ...... 100 Table S23. Hyphomonas confidence interval calculation from SPLASH ...... 101 Table S24. Methylophaga confidence interval calculation from SPLASH ...... 102 Table S25. Oleispira confidence interval calculation from SPLASH ...... 103 Table S26. Oleibacter confidence interval calculation from SPLASH ...... 104 Table S27. Rhodococcus confidence interval calculation from SPLASH ...... 105 Table S28. Synechococcus confidence interval calculation from SPLASH ...... 106 Table S29. Prochlorococcus confidence interval calculation from SPLASH ...... 107 Table S30. Vibrio confidence interval calculation from SPLASH ...... 108 Table S31. Synthetic Aperture Radar imagery information during the SPLASH and Looe Key experiments...... 109

8 Abstract The sea-surface microlayer (SML) is the boundary layer at the air-sea interface where many biogeochemical processes occur. Many organisms (e.g., bacteria) produce surface active agents (surfactants) for life processes, which accumulate in the SML and dampen short gravity- capillary waves, resulting in sea surface slicks. Synthetic aperture radar (SAR) is capable of remotely sensing these features on the sea surface by measuring reflected backscatter from the ocean surface in microwaves. This study coordinates SAR overpasses with in situ SML and subsurface (SSW) microbial sample collection to guide subsequent analysis after 16s rRNA sequencing on the Illumina MiSeq. In April 2017, 138 SML and SSW samples were collected near a targeted oil-seep where the Taylor Platform was knocked down in the Gulf of Mexico, both in and out of visually-observed oil slicks. In July and August 2018, 220 SML and SSW samples were collected near the Looe Key coral reef and a coastal seagrass area. Analysis of microbial abundance and diversity between the two experiments shows that within oil slicks, surfactant- and oil-associated bacteria prefer to reside within the SSW rather than in the SML. In natural slicks in the coastal seagrass area, these bacteria are more abundant in the SML. Outside of these slicks, surfactant-associated bacteria are more abundant within the SML than the SSW. This suggests that the presence of oil reduces the habitability of the SML, whereas natural slicks created by foam and other surfactants creates a more habitable environment in the SML. With lower wind speed, abundance of these bacteria are greater, as increased wind speed results in a harsher environment. The diurnal cycle had an effect on the relative abundance of surfactant- associated bacteria in the SML and SSW. Our results demonstrate the usefulness of synthetic aperture radar to remotely sense sea surface slicks in coordination with in situ surfactant- associated bacteria data collection of the sea surface slicks.

Keywords: synthetic aperture radar, surfactant-associated bacteria, sea surface microlayer, slicks, Illumina MiSeq, 16s sequencing, Florida Keys

9 Acknowledgements Thank you to my major advisor, Dr. Alexander Soloviev, for his continued support during my time at NSU, and providing opportunities to conduct and present this research all over the world. Thank you to my committee member, Dr. Aurelien Tartar, for the use of his lab and guidance from the beginning with qPCR and overall microbial methodology. Thank you to my committee member, Dr. Joana Figueiredo, for her guidance and support, as well as instruction with statistical analysis. I am also grateful to Dr. Susanne Lehner, Egbert Schwarz (DLR), Hiu Shen and Will Perrie (BIO) for collection of SAR imagery and to Kathryn Howe, Dr. Cayla Dean, John Kluge, and Breanna Vanderplow for help with sample collection, data analysis, and day-to-day assistance. Thank you to Capt. Brian Ettinger for his help during preparation and field work in the Florida Keys. This thesis work was supported by the Gulf of Mexico Research Initiative (GoMRI), the Consortium for Advanced Research on Transport of Hydrocarbon in the Environment (CARTHE) and the Office of Naval Research award N00014-10-1-0938. Travel support was partially funded by the Pan Student Government Association Professional Development Grant and the Pan Ocean Remote Sensing Conference (PORSEC) 2018 travel grant. I am overwhelmingly grateful to my parents, siblings, extended family, and friends for their never-ending support and encouragement throughout my graduate studies.

10 1. Introduction a. Sea Surface Microlayer The sea surface microlayer (SML) represents the boundary layer that plays a significant role between the ocean and atmosphere, with a thickness on the order of 1 mm on the surface of the ocean (Cunliffe and Murrell, 2009). It is key in biogeochemical processes, cycling of particles, and air-sea gas and heat exchange on a global scale. Prior to recent findings, the SML was believed to be a more stable, classical model of organized “dry” (lipids, fatty acids) and “wet” (proteins, carbohydrates) surfactant layers (MacIntyre, 1974). The SML is now instead known to be a hydrated gelatinous layer formed from a complex structure of polysaccharides, proteins, and lipids (Wurl et al., 2008). The physical structure of the SML consists of a viscous sublayer (~1500 μm thick), thermal sublayer (~500 μm thick), and salinity diffusion sublayer (~50 μm thick) (Soloviev and Lukas, 2014) (Fig. 1).

Figure 1. Sea surface microlayer structure as a biochemical reactor and all inhabitants: (I) surfactant orientation and aggregation, (II) microbial structure and processes, and (III) photochemical transformation of dissolved organic matter. (Wurl et al., 2017)

The SML interacts with both the atmosphere and ocean. This boundary is an extreme environment, impacted by waves, wind, radiation, biological activity, and anthropogenic contaminants (e.g., oil spills). The SML is relatively physically stable under low wind speed

11 conditions because of surface tension at the air-water interface. The structure and behavior of the SML plays a role in global geochemistry as well as local events. b. Biosurfactants and Oil in the Sea Surface Microlayer Surface active agents (surfactants), capable of reducing surface tension, conjugate at the surface and dampen capillary waves and reduce air-sea gas exchange (Alpers et al., 1989). Surfactants breakdown hydrocarbons, which were observed during sampling in the Gulf of Mexico. Natural biosurfactants are the biological compounds that demonstrate high surface- active properties (Georgiou et al., 1992). The community of bacteria present within the sea surface microlayer has been termed the bacterioneuston (Franklin et al., 2005; Zaitsev, 2005). The SML of aquatic environments has aggregate-enriched biofilms that contain complex bacterioneuston, which are ecologically distinct from microbial communities residing below in the subsurface water (Cunliffe et al., 2011). Many species of bacteria can produce natural surfactants (i.e., biosurfactants) (Lang 2002). Biosurfactants are produced by bacteria for food capture, motility, protection, and aggregation (Cunliffe et al., 2013; Burch et al., 2010). Bacteria have also been found to produce biosurfactants that aid in oil-degradation within oil slicks in the natural marine environment. Hamilton et al. (2015), Kurata et al. (2016), and Howe et al. (2018) found that most surfactant- associated bacteria reside below the sea surface, producing surfactants that are then transported to the surface. Surfactants can be transported to the surface through simple diffusion, rising bubbles, convection and upwelling of underlying waters, and can be a sink for atmospheric deposition (Agogue et al., 2005). It has been suggested that surfactants biologically produced as transparent exopolymer particles, which are surface-active polysaccharides released by phytoplankton communities in the water column, form at the surface by coagulating and becoming enriched in relation to the underlying water and form slicks (Wurl et al., 2008). These surfactant films can potentially cause significant reduction in annual net flux estimates of air-sea gas exchange rates of greenhouse gases (Wurl et al., 2011). A slick’s damping capability depends on the concentration of the surfactants formed at the surface, along with environmental factors at the surface such as wind speed and wave breaking.

12 c. Synthetic Aperture Radar Synthetic aperture radar (SAR) is capable of taking images in high resolution (down to meters) in all-weather conditions. SAR is independent of sunlight, so can measure backscatter from the surface of the ocean during daytime and nighttime, and through clouds. Slicks can form at the surface of the ocean due to hydrodynamic and biological processes described above. These films can be visible in SAR, as the slicks reflect less microwave backscatter to the satellite than the surrounding rougher sea surface (Fig. 2) (Alpers and Espedal, 2004). Both surfactants and oil form thin films that become visible in SAR imagery as smooth dark shapes, spots or more complex stripes, due to the process of capillary wave-damping. By coordinating SAR imagery with in-situ measurements of the composition of these slicks, both from oil and biosurfactants, the information inferred from the images can be improved (Gade et al., 2013). SAR can also track surface manifestations of the processes and formations under the surface that modify surface gravity-capillary waves. A) B)

Figure 2. SAR signal scattering. A) Rough surface reflects some of the randomly scattered radar signal back to the receiving antenna, while (B) a smooth surface reflects the signal away from the receiving antenna. From Liew (2001).

In this study, European Space Agency’s (ESA) Sentinel 1 satellite, Canadian Space Agency’s (BIO) RADARSAT-2 satellite, and German Aerospace Center’s (DLR) TerraSAR-X satellite was used to visualize the presence of sea surface slicks influenced by surfactants. RADARSAT-2 is in C band with quad polarization, Sentinel 1 is in C band with dual polarization, and TerraSAR-X is in X band with dual polarization. d. Sea Surface Microlayer in Coral Reef Areas Coral reefs are important sensitive environments that are growing in attention in the public eye due to the changing ocean temperature and chemistry negatively affecting much of the

13 world’s reefs (Field et al., 2014; Hughes et al., 2018). Bacteria in the water column constitutes a food source for the corals (Gast et al., 1998). Several in situ studies have found a correlation between reef macroorganism cover and microbial community taxa in the water column (Dinsdale et al., 2008a; Kelly et al., 2014; Tout et al., 2014). Stony corals release organic matter as mucus for protection, feeding, and spawning (e.g. Harrison et al. 1984, Crossland 1987), in dissolved (Ferrier-Pages et al. 1998) or particulate form into their surroundings. This mucus either dissolves in the surrounding water or floats to the surface as less-soluble mucus (Naumann et al., 2009; Wild et al., 2010). The 50-80% of mucus that dissolves in the surrounding water provides a food source for planktonic bacteria (Ducklow and Mitchell, 1979). This mucus traps various particles, such as bacteria, becomes enriched in microorganisms and contributes to the composition of the SML (Wild et al., 2004b; Naumann et al., 2009). Looe Key Management Area remains one of the largest spur and groove reefs sites in the Florida Keys. Looe Key has been studied for many years, is a popular recreational dive site, and has been federally protected since 1981 (Donahue et al., 2008). The coral reef environment of Looe Key is known to be a biologically active community of many organisms, which can influence the surrounding microbial community and can be investigated through DNA analysis of surfactant associated bacteria collected from the SML. e. Sea Surface Microlayer and Diurnal Cycle Organisms migrate from the deep ocean to the surface and back daily, recognized as diel vertical migration (DVM) (Iwasa, 1982; Richards et al., 1996; Dean et al., 2016). Light (i.e., sunrise and sunset) is the main control cue that triggers DVM in the ocean, as it is the only factor that changes daily and coincides with the timing of migrations (Richards et al., 1996). Migration to the surface with ample food supply enables organisms to meet energy demands (Stich, 1981). Darkness allows these organisms to maneuver through surface waters to catch prey without being seen by their own predators (Enright, 1977). One study in the Baltic Sea on the effect of the diurnal cycle on the sea surface microlayer found that at night, the biodegradation of dissolved organic matter was stronger in the microlayer than in the underwater layer (Falkowska, 2001). The effect of the diurnal cycle on surfactant associated bacteria residing in and below the sea surface microlayer, specifically above a biologically active ecosystem of a coral reef, has not been studied before.

14

2. Statement of Objectives • Update sampling method to collect 220 samples in the sea surface microlayer above the Looe Key coral reef and a coastal seagrass area to improve statistical analysis. • Analyze the Florida Keys 2018 coral reef dataset and compare results with 138 samples collected in the sea surface microlayer in the Gulf of Mexico during the SPLASH (Submesoscale Processes and Lagrangian Analysis on the Shelf) 2017 Experiment. • Compare microlayer bacterial content in oil-slicks and outside of oil-slicks. • Study the diurnal effect on the abundance of surfactant associated bacteria. • Analyze bacterial composition of the sea surface microlayer and subsurface water. 3. Hypotheses • Surfactant- and oil-associated bacteria are in greater abundance in oil slicks than in natural slicks. • Surfactant- and oil-associated bacteria abundance depends on wind conditions. • Bacteria abundance differs between samples collected in the coral reef and in the seagrass area. • Relative abundance of the surfactant-associated bacteria between sea surface microlayer and subsurface water depends on time of day.

4. Significance of Work Similar to our skin as protection for our body, the sea surface microlayer is the first line of defense for the ocean. The bacterial composition of this layer and the underlying water affects air-sea processes such as momentum, heat, and gas exchange (Liss et al., 2005). Surfactants, when present, can dampen capillary waves and result in altering transfer rates (Alpers et al., 1989). Both oil and natural slicks formed by biosurfactants can be seen in SAR satellite imagery. Slicks formed during nighttime are not observable by passive light dependent sensors. By understanding the bacterial composition of the SML and SSW over a coral reef, distinction between oil slicks and natural slicks can be made in relation to SAR imaging analysis. The diurnal cycle plays an important role in regulating biological processes in the ocean such as diel

15 vertical migration, and this sampling technique can be used to specifically study its effect on the sea surface microlayer. Improvement of the polycarbonate membrane filter sampling technique and DNA extraction methods will lead to more precise measurements and lessen common contamination in bacterial research. Three advancements to SML research this project intends to achieve are to (1) improve the method of sea surface microlayer sample collection, (2) collect sea surface microlayer samples in an oil slick and over coral reef and seagrass ecosystems during synthetic aperture radar satellite overpasses, and (3) collect sea surface microlayer samples at nighttime and during daytime in the Florida Keys to detect diurnal effect. If these goals are attained, we will advance the method of the study of the SML, apply it to two contrasting ecosystems, and introduce diurnal effect.

5. Methods An innovative technique has been practiced and developed in previous years of this project to accurately sample the depth of the microlayer, obtain enough samples to be able to perform meaningful analysis, while reducing contamination during collection and DNA extraction. a. Sampling Method The method adapted for this study was used by Hamilton et al. (2015), Kurata et al. (2016), and Howe et al. (2018), with a 47 mm diameter polycarbonate membrane filter that collects approximately 35±5mm of surface material (Franklin et al., 2005). This filter was attached to a sterilized hook with a small loop of fishing line in a sterile bag prior to field work. In the field, the filter was attached to fishing pole by the fishing line and a paperclip with a piece of cork. It was cast away from the boat and wake and let rest on the surface for approximately 3 seconds, then brought back and collected with sterile forceps and placed directly into the MoBio DNA extraction tube. This is to reduce potential loss of material. The most successful aspects of this method in comparison to the glass-plate, mesh screen, and rotating-cylinder methods is the reduction of potential contamination by casting away from the boat and out of the wake, and increased efficiency due to quicker sampling with new filters for each sample. Compared to

16 previous experiments by Hamilton et al. (2015), Kurata et al. (2016), and Howe et al. (2018), a higher number of samples collected increased robustness of statistical analysis. An important addition was a small piece of cork above the clip that hooks the sampling device to the pole to aid in buoyancy and reduce submersion of the filter due to the weight of the hook. To collect SSW samples, a peristaltic pump reached an undisturbed area away from the boat using an extension pole to collect water from 0.2 m below the surface. The tube was sterilized prior to each sample collected, then water was run through the tubing for one minute before the water sample was collected into a sterile container. The filter was placed on the surface of this water in the container with sterile forceps for three seconds. It was then placed in a labeled 5ml MoBio DNA extraction tube. Samples were collected from both sunrise (approximately between 6 am and 8 am) and at night in darkness (approximately between 8 pm and 11 pm) during the Looe Key experiment. A total of 120 samples were collected from four different sites during the morning, and 90 samples were collected from three different sites during the nighttime (see Table 2 for exact times). Number of air controls was increased to improve analysis of field contamination. One filter was exposed to the air at each site during SPLASH, which was increased to ten filters exposed to the air at each sample site during Looe Key, for ten seconds to test for possible aerosol contaminants. Ten non-exposed filters never taken out of the lab were processed for laboratory and procedure contamination purposes in both experiments. Ten non-template controls (i.e., RNA free water) were sent with our samples to account for any contamination during sequencing. Samples were immediately put on dry ice in the field and transferred to a portable -80˚C freezer, then to the permanent -80°C freezer at Nova Southeastern University (NSU) to store prior to and after lab extraction. b. Experiment Overview During the SPLASH (Submesoscale Processes and Lagrangian Analysis on the Shelf) experiment with CARTHE (Consortium for Advanced Research on Transport of Hydrocarbon in the Environment) from April 19th to April 25th, 2017, a total of 138 samples were collected from seven sites in the Gulf of Mexico near a known oil-seep over the previously knocked-down Taylor Platform near the Mississippi river outflow (Figs. 3, 4 and 5). During the most recent experiment in the Florida Keys from July 28th to August 2nd, 2018, 220 samples were collected

17 near Looe Key and from a coastal seagrass ecosystem near Lois Key. Using a small boat, samples were collected from six different sites over the Looe Key coral reef; three sites were sampled from at night in darkness between 8 pm and 11 pm, two sites in the morning at sunrise between 6 am and 8 am (Fig. 6). Samples were collected from one site in a coastal seagrass area near Lois Key (Figs. 6 and 7).

Figure 3. Filter collected from oil slick covered in brown film.

A) B) C)

Figure 4. Three different types of oil seen during sample collection. A) Brown oil was present due to the interaction and accumulation against the boat; B) rainbow oil was commonly seen as a sheen on the surface alongside a distinct smell of oil; C) and a monomolecular film was seen although no visible oil was present.

18 Mississippi River Outflow

Figure 5. All locations of sample collection in the Gulf of Mexico during the SPLASH campaign.

Figure 6 shows all locations of sample collection during the Looe Key experiment, and Figure 7 shows the first six locations samples were collected from around the Looe Key coral reef. Figure 8 shows organisms seen during in situ investigation at Lois Key.

Figure 6. All locations of sample collection at Looe Key in the southeast corner of the image and Lois Key, a seagrass area, in the northwest corner of the image.

19

Figure 7. Top-down view of muddy seagrass floor near Lois Key, where samples were collected at site 7. This image shows a mixture of sea grasses and a sponge. c. DNA Analysis i. DNA Extraction The MoBio PowerWater DNA Isolation Kit was used to extract DNA from the polycarbonate membrane filters. Each filter was collected in the field and placed into the MoBio bead tube containing a unique bead mix, placed on dry ice, then transferred to a portable -80°C freezer and once back at the lab into a permanent -80°C freezer at the NSU Oceanographic Center. When ready for extraction using the MoBio DNEasy DNA Isolation Kit and associated protocol, a reformulated lysis buffer that isolates microorganisms from the filter was added and then the tube was rapidly vortexed to induce thorough lysis. A series of solutions were added that removed additional non-DNA organic and inorganic material and allowed for DNA binding. Between each step, the tube was centrifuged to enhance DNA purity and increase success of downstream applications. The last step allowed total genomic DNA to be captured on a silica spin column, which was washed and eluted from the spin column membrane for use in downstream applications (MoBio Laboratories Inc.). ii. Next-Generation Sequencing Next-generation sequencing has become widely used for investigating microbial community composition and has high efficiency and low costs per run compared to other

20 methods (Caporaso et al., 2011; Caporaso et al., 2012). Twenty µl of extracted DNA from each extracted sample were sent to Argonne National Laboratory (ANL) to be amplified and sequenced on the Illumina MiSeq platform on a 151 bp x 12 bp x 151 bp MiSeq run, targeting the 16S rRNA gene using customized primers 515F and 806R (forward 5’ TCGTCGGCAGCGTCAGATGTGTAAGAGACAGCCTACGGGNGGCWGCAG and reverse 5’ GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGGACTACHVGGGTATCTAATCC) (Caporaso et al., 2012). This followed the ANL amplification protocol of an initial denaturing step of 94 °C for 3 min, 35 cycles of 94°C for 45 s, 50°C for 60 s, and 72 C for 90 s, and an extension at 72°C for 10 min. High throughput sequencing of the 16S rRNA gene results in data that is a strong representation of the microbial population within each sample. iii. Statistical Analysis Raw reads were downloaded from ANL as FASTQ DNA sequence files and processed with Quantitative Insights into Microbial Ecology (QIIME 2 v. 2018.8.0) (Boylen et al., 2018). The three “qza” files were imported into QIIME2 using the “emp-import” command, then demultiplexed with the “demux” command, and quality filtered and trimmed using the DADA2 “dada2 denoise” command, creating a feature table. Sequences were quality filtered to remove chimeras, which are hybrid sequences that can be incorrectly interpreted as novel organisms, and scores under 25. A phylogenic tree was created using the “phylogeny fasttree” command, and the sequences were taxonomically classified through a learned Silva classifier (silva-132-99- 515- 806-nb-classifier.qza). From this point, the feature-table with taxonomy was exported and used in R processing (R Development Core Team 2008, RStudio Team 2015). Beta diversity was assessed using QIIME2 and R Studio (VEGAN and Phyloseq packages). A Bray-Curtis dissimilarity index was calculated which quantifies dissimilarities between sites (Faith et al., 1987). A PCoA run on the Bray-Curtis visualized the distance between the sites (Fig. 7). Counts of amplicon sequence variants (ASVs) were averaged between the number of samples collected at each site to normalize the data for comparison. Statistical significance was found using confidence intervals. Using the Wald confidence interval on a negative binomial distribution, 95% confidence intervals were calculated for each group of samples at each site and type in

Matlab (Arefi et al., 2016). Confidence interval (CIW) was calculated as:

�� = µ ± ∗ (µ + µ) ∗ �/ (1)

21 6. Results and Discussion Table 1. SPLASH project sample collection summary. Included is information about sample types, times and locations, environmental data, and satellite information. Date Site Time Coordinates Coordinates Wind Wave Condition* Satellite Comments 2017 (UTC) Start End (m/s) height (tube/sample Info (degrees, (degrees, (ft) number) (UTC) decimal decimal minutes) minutes) 4/16 - 12:30 - - 8.7 - - S @ Before 12:30 sampling 4/19 A 11:22- 29°0.847'N 29° 0.966'N 4-5 1-2 M: 202-203, TS @ No visible 12:45 89°45.209'W 89°47.708'W 206-213 12:08 slick W: 214-223 AC: 204 4/20 B 14:11- 28°56.066'N 28°56.113'N 6-8 3 M: 225-226, - No visible 15:31 88°58.357'W 88°59.229'W 228-229, slick 242-247 W: 230-232, 235-241 AC:234 4/20 C 15:54- 28°56.333'N 28°56.322'N 6 3 M: 249. 251- TS @ Slick 16:50 88°59.618'W 89°1.009'W 259 23:57; visible due W: 261-266, RS @ to presence 268, 270, 271 23:57; of oil AC: 267 4/21 D 15:20- 28°56.441'N 28°56.692'N 4 1-2 M: 270-279 - Slick 16:10 88°58.272'W 88°58.516'W W: 282-291, visible due 280, 281 to presence of oil 4/22 E 12:09- 28°56.332'N 28°56.387'N 2-3 1-2 M: 292-295, - Slick 13:10 88°58.060'W 88°58.045'W 299-302, visible due 304-310 to presence W: 312-314, of oil 316-322 AC: 315 4/22 F 15:44- 28°56.136'N 28°56.086'N 0-2 1 M: 323-327, - Intermittent 16:16 88°57.225'W 88°57.214'W 329-333 slick W: 334-343 AC: 344

4/25 G 12:23- 29°6.234'N 29°6.996'N 3-4 1 M: 348-360 TS @ No slick 13:15 89°54.175'W 89°53.074'W W: 361-367, 11:59 visible 369-370 AC: 372 *Condition: M- microlayer sample, W- water column sample AC-air control; TS- TerraSAR-X, RS- RADARSAT-2, S- Sentinel-1 (wind speed from NOAA NDBC - Buoy Station 42040)

22 Table 2. Looe Key project sample collection summary. Included is information about sample types, times and locations, environmental data, and satellite information. Date Site Time Coordinates Coordinates Wind Wave Condition* Satellite Comments 2018 (D- in Start End (m/s) height (tube/sample Info day; (UTC) (degrees, (degrees, (ft) number) (UTC) N- decimal decimal night) minutes) minutes) 7/29- 1-N 23:26- 24°32.805’N Same 3.5-5 1-2 M : 1-5 - No visible 30 00:04 81°24.089’W (moored) W : 6-10 slick AC: 11-15 7/30 2-D 11:31- 24°32.806N Same 4 2-3 M : 16-27 RS @ No visible 12:57 81°24.085W (moored) W : 28-37 11:14 slick AC: 38-47 7/30- 3-N 00:47- 24°33.026’N 24°33.091N 1-2 0.5-1 M : 48-57 - No visible 31 01:34 81°24.947’W 81°25.668W W : 58-67 slick AC: 68-77 7/31 4-D 22:46- 24°33.167’N 24°33.443N 4-5 2-4 M : 78-89 RS @ No visible 23:40 81°24.749’W 81°25.498W W : 90-99 23:34 slick AC:100-109 7/31- 5-N 01:41- 22°83.302’N 24°33.443N 6 2-4 M : 110-120 - No visible 8/1 02:27 81°24.145’W 81°25.498W W : 121-130 slick AC: 131- 140 8/2 6-D 10:58- 24°22.184’N 24°33.397N 7 3-5 M : 141-150 RS @ No visible 11:50 81°24.650’W 81°25.087W W : 151-160 11:27 slick AC: 161- 170 8/2 7 12:28- 24°36.996’N Same 7 1-2 M : 171-180 - Foam/slick 13:50 81°28.063’W (anchored) W : 181-190 visible on AC:191-200 surface *Condition: M-good microlayer sample, W-good water column sample AC-air control; TS- TerraSAR-X, RS-RADARSAT-2 a. Synthetic Aperture Radar Imagery Results i. SPLASH Three TerraSAR-X Stripmap images, one RADARSAT-2 image, and one Sentinel-1 image were taken over sites of sample collection during SPLASH. Only one image coincided with sampling, at site A on April 19th, 2017 at 11:08 UTC (Table 1, Fig. 9). Both the RADARSAT-2 and TerraSAR-X Stripmap images taken on April 20th, 2017 were collected at 22:57 UTC, approximately eight hours after sample collection at sites B and C, and fourteen hours before sample collection at site D shown in the inset SAR image (Table 1, Fig. 9). The distinct dark v-shaped feature indicates the surface expression of oil seeping from the knocked- down Taylor oil platform. One TerraSAR-X Stripmap image was collected on April 25th, 2017 at 10:59 UTC, 23 minutes before sample collection at site G. The Sentinel-1 image to the right

23 taken before sample collection on April 15th, 2017 12:30 UTC shows a variety of surface features throughout the Gulf of Mexico (Table 1, Fig. 9).

Figure 8. Sampling locations A-G from the SPLASH experiment in the Gulf of Mexico. All SAR images taken over the course of the experiment are superimposed on the map. The small rectangular images are TerraSAR-X, the single square image is RADARSAT-2, and the largest image on the right side of the map is Sentinel-1. ii. Looe Key Three RADARSAT-2 images were collected over the course of this project. The image collected on July 30th was taken at 11:15 am UTC, 16 minutes before sample collection began on site at site 1 (Table 2, Fig. 10). Two images collected on July 31st and August 2nd coincided perfectly with timing of sample collection (Table 2, Figs. 12 and 13). All images show no prominent slicks near the Looe Key coral reef, shown in the yellow circle. There are slick-like features surrounding Lois Key in each of the images. Figure 11 shows a close polarized image of Lois Key with slick-like features.

24

Figure 9. RADARSAT-2 synthetic aperture radar image taken in HH mode on July 30th, 2018 at 11:14 UTC over Looe Key. Sampling at Looe Key inside of the yellow circle occurred 16 minutes before overpass. Wind speed during sampling was 4 m/s.

Figure 10. Polarimetric RADARSAT-2 image from July 30th, 2018 at 11:14 UTC focused on Lois Key area. This shows slick-like features at the surface surrounding the islands. Upon in-situ observation, the area consists of seagrass and sponges, and the surface was covered in foam and slicks during sampling on August 2nd, 2018.

25

Figure 11. RADARSAT-2 synthetic aperture radar image taken in HH mode on July 31st, 2018 at 23:34 UTC over Looe Key. Sampling occurred at Looe Key during overpass inside yellow circle. Wind speed during sampling was 4-5 m/s.

Figure 12. RADARSAT-2 synthetic aperture radar image taken in HH mode on August 2nd, 2018 at 11:17 UTC over Looe Key. Sampling occurred at Looe Key during overpass inside yellow circle. Wind speed during sampling was 7 m/s.

26 b. Synthetic Aperture Radar Imagery Analysis i. SPLASH Three TerraSAR-X images, one Sentinel-1 image, and one RADARSAT-2 image were collected over locations where sampling occurred in the Gulf of Mexico (Fig. 9). One TerraSAR- X image taken on April 19th coincided perfectly with sampling at site A, where no slicks were visible. The dark zone at the bottom of this image is the signature of the Mississippi river outflow, emanating from the main channel bringing fresh water and runoff, potentially containing surfactants. One TerraSAR-X image taken on April 20th, 2017 was approximately eight hours after sampling occurred at site B and C, and fourteen hours before sampling occurred at site D, shown as the superimposed SAR image (Table 1, Fig. 9). The RADARSAT-2 image shown as the larger square underneath the TerraSAR-X image was taken at the same time and shows the same surface expression. This image shows the distinct dark v-shaped feature indicating the surface expression of oil seeping from the knocked-down oil platform and a distinct front, confirmed by in situ visual observations. Samples were collected in a relatively small oil source area on three consecutive days, under 6 m/s, 4 m/s, and 2-3 m/s wind speed (Table 1, sites C, D, and E). We did collect samples from just outside of the relatively narrow obvious oil source area where intermittent slicks were visually detected, under 0-2 m/s wind speed (Table 1, site F). The SAR image taken eight hours after sampling at site C and 14 hours before sampling at sites D and E was showing a much wider slick, supposedly because of lower wind (3.4 m/s). The monomolecular surfactant/oil films existing under low wind speed conditions may not be clearly seen visually but still affect surface gravity-capillary waves and SAR imagery, as a result. The oil’s influence on the samples collected in and near this oil slick has been considered when analyzing the bacterial abundance in the samples. The slick feature spreading throughout the Gulf of Mexico seen in the Sentinel-1 image to the right of the leaking Taylor Platform is associated with the ocean currents. It is visualized in SAR due to presence of slick films (Fig. 9). ii. Looe Key Four RADARSAT-2 images were captured during field work in the Florida Keys. Two RADARSAT-2 images were taken while sampling occurred at Looe Key on July 30th and August 2nd (Figs. 12, 13). One image was captured on July 29th shortly (16 min) before the

27 sampling began (Fig. 10). These images were acquired in order to detect surface slicks; unfortunately, no slicks were visually observed during sampling nor seen in the SAR images in the area of sampling at the Looe Key coral reef (Fig. 10, 12 and 13, indicated by the yellow circle). In two images from July 30th and 31st, slick-like features were seen surrounding a small island, Lois Key, in the northwest corner of both images. Because slicks were present in both images from two different days, it could be assumed that these features were persistent in this location during our experiment period. A polarized SAR image from July 30th shows distinct slicks surrounding the area, without any obvious source such as an oil rig, boats, or river outflow (Fig. 11). Upon in situ exploration, we discovered a shallow seagrass area with many sponges, small fish, and decaying biological material including a starfish (Fig. 8). During observation, slicks and foam were present on the ocean surface. We collected samples at this site (Site 7) and included them in our analysis. The use of SAR in the project strayed from our original purpose of detecting coral-associated slicks at Looe Key, but detected an area of slicks at Lois Key (which was subsequently sampled in situ). The SAR imagery captured during the satellite overpass on July 29th and 30th guided in situ measurements of the slick area near Lois Key on July 31st. c. Overall Bacteria Sequencing Results i. Illumina MiSeq DNA Sequencing Outputs A total of 138 samples from the SPLASH experiment were sequenced on the Illumina MiSeq and processed in QIIME 2 resulting in 8,358,529 raw reads with a mean of 54,630 reads per sample. Samples with fewer than 1,400 reads were dropped due to lack of adequate coverage. An alpha rarefaction curve for sample location shows full sampling depth for all locations sampled (Fig. 14). A total of 11,693 unique sequences were found after processing with DADA2 (Callahan et al., 2017). When collapsed at the genus level, 1054 specific genera were identified.

28

Figure 13. Alpha rarefaction curve for all Types of samples from SPLASH created in QIIME2. All Types reach a plateau in the graph showing complete sampling depth, meaning enough samples were collected to represent the complete microbiome at the site locations. AC, NE, and NTC do not increase as much as the SML and SSW due to the nature of the control sample ideally having little contamination.

A total of 220 samples from the Looe Key experiment were sequenced on the Illumina MiSeq and processed in QIIME 2 resulting in 6,495,578 raw reads with a mean of 29,660 reads per sample. Samples with fewer than 1,400 reads were dropped due to lack of adequate coverage. An alpha rarefaction curve for sample location shows full sampling depth for all locations sampled (Fig. 15). A total of 12,554 unique features were found after processing through DADA2 (Callahan et al., 2017). When collapsed at the genus level, 1237 specific genera were identified. Average number of reads per sample and observed OTUs shown in the alpha rarefaction curve in Figure 15 are relatively low due to the small amount of bacteria content collected in each sample using the filter and high number of controls that ideally show no contamination.

29

Figure 14. Alpha rarefaction curve for all Types of samples in Looe Key created in QIIME2. All Types reach a plateau in the graph showing complete sampling depth, meaning enough samples were collected to represent the complete microbiome at the site locations. AC, NE, and NTC do not increase as much as the SML and SSW due to the nature of the control sample ideally having little contamination.

Beta Diversity represents the diversity between groups of samples. In both projects, samples were grouped by site and type (e.g., SML at site 1 and SSW at site 1), and in the Looe Key project also by sampling time. Beta Diversity was measured using the Bray-Curtis metric, which quantifies compositional diversity between groups based on counts. This method takes into account the number of similar ASVs and number of non-similar ASVs and quantifies the difference between the samples. The resulting value of the Bray-Curtis dissimilarity index is between 0 and 1, with 0 representing two samples that are exactly the same and 1 representing two samples that do not share any common ASVs. A principle coordinate analysis (PCoA) was performed in R studio using the VEGAN and Phyloseq packages (McMurdie and Holmes, 2013; Oksanen et al., 2018). This ordination method, also known as classical multidimensional scaling, takes the Bray-Curtis dissimilarity values in the matrix and creates new axes: PC1 that explains the most amount of variation present in the dataset, and PC2 that explains the second most amount of variation, etc. This visualization is common in microbiology studies, as the large datasets resulting from sequencing are challenging to present clearly. The purpose of this plot is to take highly complex data in a high-dimensional space and represent it in a low-dimensional space without loss of too much information. The interpretation of a PCoA plot is that the samples

30 clustered closest together on these new axes are most similar, and the samples that are farther away are less similar. ii. Beta Diversity SPLASH Dispersion within all site/types were found to be significantly different from each other

(p<0.001, F13,12=8.1499) from the PERMANOVA (adonis) test in the VEGAN package in R (Callahan et al., 2017). A PCoA was run on the Bray-Curtis matrix to visualize the distance between sites and types of samples (Fig. 16).

By Type and Site Location

Figure 15. Bray Curtis PCoA plot showing sites/types during SPLASH: each site is shown with a different color, sea surface microlayer (SML) is represented by a circle, and subsurface water (SSW) by a triangle. Ellipsoids represent the standard deviation surrounding each group, with the circle filled with the corresponding group color indicating centroid of the group.

31 Looe Key Dispersion within all site/types were found to be significantly different from each other

(p<0.001, F13,12=6.4505) and dispersion within all times were found to be significantly different from each other (p=0.001, F2,1=5.1107) from the PERMANOVA (adonis) test in the VEGAN package in R (Callahan et al., 2017). A PCoA was run on the Bray-Curtis matrix to visualize the distance between sites and types of samples (Fig. 17), and time of sample collection (Fig. 18).

By Type and Site Location

Figure 16. Bray Curtis PCoA plot showing sites/types during Looe Key experiment: each site is shown with a different color, sea surface microlayer (SML) is represented by a circle, and subsurface water (SSW) by a triangle. Ellipsoids represent the standard deviation surrounding each group, with the circle filled with the corresponding group color indicating centroid of the group.

32

Figure 17. Bray Curtis PCoA plot showing times of sampling during the Looe Key project. Times are each shown with a different color, and types are shown with different symbols. Ellipsoids represent the standard deviation surrounding each group, with the circle filled with the corresponding group color indicating centroid of the group. d. Overall Bacteria Sequencing Analysis i. SPLASH A Bray Curtis Dissimilarity PCoA plot in Figure 16 shows SPLASH samples grouped by the combination of site and type. Sites A and G, both from west of the Mississippi river outflow, seem to be most similar. Sites B, D, E and F, all located near the oil seep, are most similar to each other. This indicates that the presence of oil significantly shifts the microbial composition of the surface waters. ii. Looe Key The Looe Key PCoA for site dissimilarity plot shows that site 7, in the seagrass slick, was most different in composition from the rest of the sampling sites over the coral reef (Fig. 17). The Bray Curtis Dissimilarity PCoA plot shown in Figure 18 reveals that the time of sampling

33 (morning/light or night/darkness) did not have a strong effect on the overall microbial community diversity over the coral reef, as all groups, including SML and SSW at each time, overlap. It can be concluded that time of sampling does not have as much of an effect on the microbial community diversity as site type and location. We cannot make conclusions on the effect of the diurnal cycle on the bacteria in the coastal seagrass area because we only collected samples in the morning. The PCoA Dissimilarity plot for Site (Fig. 17) shows greater distinct differences between samples grouped by site than in the PCoA Dissimilarity Plot for Time (Fig. 18). Note, these experiments were conducted in a shallow water area; while the diurnal cycle of the bacterial abundance may be prominent in the open ocean (e.g., due to diel vertical migration of zooplankton). e. Specific Bacteria Abundance Results A total of 12 surfactant- and oil-associated bacteria were selected for further analysis from the 1054 genera of bacteria identified in the SPLASH samples, and 12 surfactant- and oil- associated bacteria were selected from 1237 genera of bacteria identified from the Looe Key samples (Table 3, Figs. 19 and 36). Eight genera of bacteria were found in both SPLASH and Looe Key samples, and four genera were exclusively found in the SPLASH samples and four genera were exclusively found in the Looe Key samples (Table 3). Bacteria that were not shared between the two experiments were either found on control filters in one of the experiments or were just not detected in taxonomic identification after sequencing.

34 Table 3. List of all surfactant- and oil-associated bacteria found in SPLASH and Looe Key samples. Eight genera were found in samples from both experiment, and four were found exclusively Looe Key samples and four in the SPLASH experiment samples. SPLASH Both Looe Key Rhodococcus Alteromonas Roseovarius Oleispira Tenacibaculum Oleiphilus Enterococcus Methylophaga Erythrobacter Arthrobacter Hyphomonas Winogradskyella

Fusibacter

Oleibacter

Alcanivorax

Halomonas

The average abundance for each bacteria genus at each site and type (SML or SSW) was calculated in Matlab with a confidence interval based on the negative binomial distribution and Wald formula (Arefi et al., 2016). This confidence interval calculation was used to distinguish significance between the SML and SSW at each site. All raw values from calculations of confidence intervals for all genera of bacteria are included in the supplementary materials (Tables S1- S35). We can be 95% confident that the true mean of the abundance of the genera of interest lies within this interval. Where the intervals do not overlap between the SML and SSW at each site, there is a significant difference between them. The y-axes of these graphs show counts of the unique ASVs corresponding to the bacteria genus found in the samples. The counts of ASVs represent the abundance of each of the bacteria. These do not represent discrete numbers of specific bacteria found within the sample, rather the abundance of a particular sequence after amplification corresponding to a genus, relative to other samples. Counts of these specific surfactant-associated bacteria are found in much lower abundance than other, more common non-surfactant associated bacteria, so using discrete counts of ASVs rather than relative abundance is the best method in visualizing the abundance.

35 i. SPLASH

5000 Oil Slick 4500

Marinobacter 4000 Arthrobacter 3500 Enterococcus Alteromonas 3000 Tenacibaculum 2500 Methylophaga Hyphomonas 2000 Fusibacter 1500

Average Number of ASVs Oleispira Oleibacter 1000 Rhodococcus 500 Alcanivorax Halomonas 0 NE AC NTC SML SSW SML SSW SML SSW SML SSW SML SSW SML SSW SML SSW A B C D E F G

Figure 18. SPLASH results showing average ASVs per site for surfactant-associated bacterial genera, normalized by averaging counts by sample number at each site and type. Here SML is sea surface microlayer, SSW is subsurface water, NE is non-exposed, AC is air control, and ASV is amplicon sequence variant. The data is separated by site on the x-axis, with type indicated within each site category. The orange box outlines the four sites that were in an oil slick.

36 Oil Slick

* *

Figure 19. SPLASH abundance of surfactant-associated and oil-degrading bacteria Alcanivorax with average number of ASVs on the y-axis and sites on the x-axis, with colors corresponding to type of sample shown in the legend. Confidence intervals calculated using the negative binomial Wald interval (Eq. 1). Significance between SML and SSW at each site shown by asterisk.

Oil Slick

* * *

Figure 20. SPLASH abundance of surfactant-associated and oil-degrading bacteria Alteromonas with average number of ASVs on the y-axis and sites on the x-axis, with colors corresponding to type of sample shown in the legend. Confidence intervals calculated using the negative binomial Wald interval (Eq. 1). Significance between SML and SSW at each site shown by asterisk.

37 Oil Slick

* * *

Figure 21. SPLASH abundance of surfactant-associated and oil-degrading bacteria Arthrobacter with average number of ASVs on the y-axis and sites on the x-axis, with colors corresponding to type of sample shown in the legend. Confidence intervals calculated using the negative binomial Wald interval (Eq. 1). Significance between SML and SSW at each site shown by asterisk.

Oil Slick

* *

Figure 22. SPLASH abundance of Enterococcus with average number of ASVs on the y-axis and sites on the x-axis, with colors corresponding to type of sample shown in the legend. Confidence intervals calculated using the negative binomial Wald interval (Eq. 1). Significance between SML and SSW at each site shown by asterisk.

38 Oil Slick

* * *

Figure 23. SPLASH abundance of surfactant-associated and oil-degrading bacteria Halomonas with average number of ASVs on the y-axis and sites on the x-axis, with colors corresponding to type of sample shown in the legend. Confidence intervals calculated using the negative binomial Wald interval (Eq. 1). Significance between SML and SSW at each site shown by asterisk.

39 Oil Slick

* * *

Figure 24. SPLASH abundance of surfactant-associated and oil-degrading bacteria Tenacibaculum with average number of ASVs on the y-axis and sites on the x-axis, with colors corresponding to type of sample shown in the legend. Confidence intervals calculated using the negative binomial Wald interval (Eq. 1). Significance between SML and SSW at each site shown by asterisk.

40 Oil Slick

* *

Figure 25. SPLASH abundance of surfactant-associated and oil-degrading bacteria Fusibacter with average number of ASVs on the y-axis and sites on the x-axis, with colors corresponding to type of sample shown in the legend. Confidence intervals calculated using the negative binomial Wald interval (Eq. 1). Significance between SML and SSW at each site shown by asterisk.

Oil Slick

* * *

Figure 26. SPLASH abundance of surfactant-associated and oil-degrading bacteria Hyphomonas with average number of ASVs on the y-axis and sites on the x-axis, with colors corresponding to type of sample shown in the legend. Confidence intervals calculated using the negative binomial Wald interval (Eq. 1). Significance between SML and SSW at each site shown by asterisk.

41 Oil Slick

* *

Figure 27. SPLASH abundance of surfactant-associated and oil-degrading bacteria Methylophaga with average number of ASVs on the y-axis and sites on the x-axis, with colors corresponding to type of sample shown in the legend. Confidence intervals calculated using the negative binomial Wald interval (Eq. 1). Significance between SML and SSW at each site shown by asterisk.

42 Oil Slick

* *

Figure 28. SPLASH abundance of surfactant-associated and oil-degrading bacteria Oleispira with average number of ASVs on the y-axis and sites on the x-axis, with colors corresponding to type of sample shown in the legend. Confidence intervals calculated using the negative binomial Wald interval (Eq. 1). Significance between SML and SSW at each site shown by asterisk.

Oil Slick

* * * *

Figure 29. SPLASH abundance of surfactant-associated and oil-degrading bacteria Oleibacter with average number of ASVs on the y-axis and sites on the x-axis, with colors corresponding to type of sample shown in the legend. Confidence intervals calculated using the negative binomial Wald interval (Eq. 1). Significance between SML and SSW at each site shown by asterisk.

43

Oil Slick

* * * *

Figure 30. SPLASH abundance of surfactant-associated and oil-degrading bacteria Rhodococcus with average number of ASVs on the y-axis and sites on the x-axis, with colors corresponding to type of sample shown in the legend. Confidence intervals calculated using the negative binomial Wald interval (Eq. 1). Significance between SML and SSW at each site shown by asterisk.

44 Oil Slick

Figure 31. SPLASH abundance of Synechococcus represented with average number of ASVs on the y-axis and sites on the x-axis, with colors corresponding to type of sample shown in the legend. Confidence intervals calculated using the negative binomial Wald interval (Eq. 1). Significance between SML and SSW at each site shown by asterisk.

Oil Slick

Figure 32. SPLASH abundance of Prochlorococcus represented with average number of ASVs on the y-axis and sites on the x-axis, with colors corresponding to type of sample shown in the legend. Confidence intervals calculated using the negative binomial Wald interval (Eq. 1). Significance between SML and SSW at each site shown by asterisk.

45 Oil Slick

* * *

Figure 33. SPLASH abundance of Vibrio represented with average number of ASVs on the y- axis and sites on the x-axis, with colors corresponding to type of sample shown in the legend. Confidence intervals calculated using the negative binomial Wald interval (Eq. 1). Significance between SML and SSW at each site shown by asterisk.

46 ii. Looe Key

350

1-2 m/s Slick

300

7 m/s

250 Erythrobacter Alteromonas 7 m/s Tenacibaculum 200 Winogradskyella Methylophaga Roseovarius 3.5-5 m/s 150 4-6 m/s Hyphomonas Fusibacter Average Number of ASVs Oleibacter 100 Oleiphilus

6 m/s Alcanivorax Halomonas 4 m/s 50

0 NE AC AC AC AC AC AC AC NTC SML SSW SML SSW SML SSW SML SSW SML SSW SML SSW SML SSW 1 2 3 4 5 6 7

Figure 34. Counts of ASVs (amplicon sequence variants) of all surfactant associated bacteria at each of the sites and types of samples normalized by averaging sample count at each site from the Looe Key experiment. Here SML is sea surface microlayer, SSW is subsurface water, AC is air control, NE is non-exposed, and NTC is non-template control. The data is separated by site on the x-axis, with type indicated within each site category.

47 Slick

* * * *

Figure 35. Looe Key abundance of surfactant-associated and oil-degrading bacteria Alcanivorax with average number of ASVs on the y-axis and sites on the x-axis, with colors corresponding to type of sample shown in the legend. Confidence intervals calculated using the negative binomial Wald interval (Eq. 1). Significance between SML and SSW at each site shown by asterisk.

*

Figure 36. Looe Key abundance of Alcanivorax between day and night in average number of ASVs. Significant difference between SML and SSW in each category based on the 95% Wald confidence interval is shown by a star.

48 Slick

* * * * * *

Figure 37. Abundance of surfactant-associated bacteria Alteromonas represented with average number of ASVs on the y-axis and sites on the x-axis, with colors corresponding to type of sample shown in the legend. Confidence intervals calculated using the negative binomial Wald interval (Eq. 1). Significance between SML and SSW at each site shown by asterisk.

*

Figure 38. Abundance of Alteromonas between day and night in average number of ASVs. Significant difference between SML and SSW in each category based on the 95% Wald confidence interval is shown by a star.

49 Slick

* * * * * *

Figure 39. Abundance of surfactant-associated and oil-degrading bacteria Halomonas represented with average number of ASVs on the y-axis and sites on the x-axis, with colors corresponding to type of sample shown in the legend. Confidence intervals calculated using the negative binomial Wald interval (Eq. 1). Significance between SML and SSW at each site shown by asterisk.

*

Figure 40. Abundance of Halomonas between day and night in average number of ASVs. Significant difference between SML and SSW in each category based on the 95% Wald confidence interval is shown by a star.

50 Slick

* * * *

Figure 41. Abundance of Tenacibaculum genus represented with average number of ASVs on the y-axis and sites on the x-axis, with colors corresponding to type of sample shown in the legend. Confidence intervals calculated using the negative binomial Wald interval (Eq. 1). Significance between SML and SSW at each site shown by asterisk.

* *

Figure 42. Abundance of Tenacibaculum between day and night in average number of ASVs. Significant difference between SML and SSW in each category based on the 95% Wald confidence interval is shown by a star.

51 Slick

*

Figure 43. Abundance of Roseovarius genus represented with average number of ASVs on the y-axis and sites on the x-axis, with colors corresponding to type of sample shown in the legend. Confidence intervals calculated using the negative binomial Wald interval (Eq. 1). Significance between SML and SSW at each site shown by asterisk.

*

Figure 44. Abundance of Roseovarius between day and night in average number of ASVs. Significant difference between SML and SSW in each category based on the 95% Wald confidence interval is shown by a star.

52 Slick

* * * * *

Figure 45. Abundance of Winogradskyella genus represented with average number of ASVs on the y-axis and sites on the x-axis, with colors corresponding to type of sample shown in the legend. Confidence intervals calculated using the negative binomial Wald interval (Eq. 1). Significance between SML and SSW at each site shown by asterisk.

*

Figure 46. Abundance of Winogradskyella between day and night in average number of ASVs. Significant difference between SML and SSW in each category based on the 95% Wald confidence interval is shown by a star.

53 Slick

* * *

Figure 47. Abundance of Fusibacter genus represented with average number of ASVs on the y- axis and sites on the x-axis, with colors corresponding to type of sample shown in the legend. Confidence intervals calculated using the negative binomial Wald interval (Eq. 1). Significance between SML and SSW at each site shown by asterisk.

*

Figure 48. Abundance of Fusibacter between day and night in average number of ASVs. Significant difference between SML and SSW in each category based on the 95% Wald confidence interval is shown by a star.

54 Slick

* * * * * *

Figure 49. Abundance of Hyphomonas genus represented with average number of ASVs on the y-axis and sites on the x-axis, with colors corresponding to type of sample shown in the legend. Confidence intervals calculated using the negative binomial Wald interval (Eq. 1). Significance between SML and SSW at each site shown by asterisk.

* *

Figure 50. Abundance of Hyphomonas between day and night in average number of ASVs. Significant difference between SML and SSW in each category based on the 95% Wald confidence interval is shown by a star.

55 Slick

*

Figure 51. Abundance of Methylophaga genus represented with average number of ASVs on the y-axis and sites on the x-axis, with colors corresponding to type of sample shown in the legend. Confidence intervals calculated using the negative binomial Wald interval (Eq. 1). Significance between SML and SSW at each site shown by asterisk.

Methylophaga was not present in any samples collected from the coral reef area, and therefore cannot be included in day and night analysis.

56 Slick

* * * *

Figure 52. Abundance of Oleibacter represented with average number of ASVs on the y-axis and sites on the x-axis, with colors corresponding to type of sample shown in the legend. Confidence intervals calculated using the negative binomial Wald interval (Eq. 1). Significance between SML and SSW at each site shown by asterisk.

* *

Figure 53. Abundance of Oleibacter between day and night in average number of ASVs. Significant difference between SML and SSW in each category based on the 95% Wald confidence interval is shown by a star.

57 Slick

*

Figure 54. Abundance of Oleiphilus represented with average number of ASVs on the y-axis and sites on the x-axis, with colors corresponding to type of sample shown in the legend. Confidence intervals calculated using the negative binomial Wald interval (Eq. 1). Significance between SML and SSW at each site shown by asterisk.

*

Figure 55. Abundance of Oleiphilus between day and night in average number of ASVs. Significant difference between SML and SSW in each category based on the 95% Wald confidence interval is shown by a star.

58 Slick

* * *

Figure 56. Abundance of Erythrobacter genus represented with average number of ASVs on the y-axis and sites on the x-axis, with colors corresponding to type of sample shown in the legend. Confidence intervals calculated using the negative binomial Wald interval (Eq. 1). Significance between SML and SSW at each site shown by asterisk.

* *

Figure 57. Abundance of Erythrobacter between day and night in average number of ASVs. Significant difference between SML and SSW in each category based on the 95% Wald confidence interval is shown by a star.

59 Non-surfactant associated bacteria

Slick

*

Figure 58. Abundance of Synechococcus represented with average number of ASVs on the y- axis and sites on the x-axis, with colors corresponding to type of sample shown in the legend. Confidence intervals calculated using the negative binomial Wald interval (Eq. 1). Significance between SML and SSW at each site shown by asterisk.

60 *

Figure 59. Abundance of Synechococcus between night and day represented with average number of ASVs on the y-axis and times of sampling on the x-axis, with colors corresponding to type of sample in the legend. Confidence intervals calculated using the negative binomial Wald interval (Eq. 1). Significance between SML and SSW at each site shown by asterisk.

Slick

* *

Figure 60. Abundance of Prochlorococcus genus represented with average number of ASVs on the y-axis and sites on the x-axis, with colors corresponding to type of sample shown in the legend. Confidence intervals calculated using the negative binomial Wald interval (Eq. 1). Significance between SML and SSW at each site shown by asterisk.

61 * *

Figure 61. Abundance of Prochlorococcus between night and day represented with average number of ASVs on the y-axis and times of sampling on the x-axis, with colors corresponding to type of sample in the legend. Confidence intervals calculated using the negative binomial Wald interval (Eq. 1). Significance between SML and SSW at each site shown by asterisk.

Slick

* *

Figure 62. Abundance of Vibrio represented with average number of ASVs on the y-axis and sites on the x-axis, with colors corresponding to type of sample shown in the legend. Confidence intervals calculated using the negative binomial Wald interval (Eq. 1). Significance between SML and SSW at each site shown by asterisk.

62 *

Figure 63. Abundance of Vibrio between night and day represented with average number of ASVs on the y-axis and times of sampling on the x-axis, with colors corresponding to type of sample in the legend. Confidence intervals calculated using the negative binomial Wald interval (Eq. 1). Significance between SML and SSW at each site shown by asterisk. f. Specific Bacteria Abundance Analysis In total, 16 surfactant- and oil-associated bacteria were chosen for further analysis within the SPLASH and Looe Key samples (see Table 3, Figs. 19 and 35). These bacteria were found to have little to no contamination on the control filters and were important in relation to both surfactant production and oil degradation. Species within the genus Alcanivorax produce biosurfactants and participate in bioremediation of oil‐contaminated marine environments (Passeri et al., 1992; Yakimov et al., 1998; Kasai et al., 2002; Satpute et al., 2010). Alteromonas species produce surfactants that exhibit oil emulsifying activity, and also increase in abundance when coral mucus was added to the water (Bozzi et al., 1996; Vincent et al., 2004; Allers et al., 2008). Marine strains of Halomonas produce surfactants that possess high emulsifying activities and enhance biodegredation of oil-contaminated surface waters (Pepi et al., 2005; Gutiérrez et al., 2007a; Satpute et al., 2010). The genera Methylophaga, Winogradskyella, Hyphomonas, Roseovarius, Tenacibaculum, and Fusibacter have been correlated with the degradation of naphthalene and fluorene in Gulf of Mexico soils after the Deepwater Horizon Spill (Kappell et al., 2014).

63 Oleibacter can degrade petroleum aliphatic hydrocarbons (Teramoto et al., 2011; Santisi et al., 2013), Oleiphilus can degrade hydrocarbons, is active in bioremediation after crude spills, and indicates polluted environments (Golyshin et al., 2002; Ibacache-Quiroga, Ojeda, and Dinamarca, 2018), and Oleispira is present in oil-degrading microbial communities usually associated with crude-oil enrichments of Antarctic seawater (Yakimov et al. 2003). Erythrobacter was found in oil mousses collected in the Gulf of Mexico and may contribute to aromatic hydrocarbon degradation in oil‐contaminated beaches (Wang et al., 2010; Liu and Liu, 2013). Three non-surfactant-associated bacteria were analyzed in relation to environmental conditions associated with the sites, including effect of day and night and association with organisms in the water below. Photosynthetic prokaryotes Synechococcus and Prochlorococcus are responsible for a substantial fraction of the ocean’s major primary production and are the most abundant members of marine picophytoplankton. (Scanlan and West, 2002). Prochlorococcus is less ubiquitous than Synechococcus and usually absent from brackish and well-mixed waters (Partensky, Blanchot and Vaulot, 1999). The genus Vibrio is abundant within coral associated microbial communities and frequently associated with coral diseases, such as white syndrome, white band, and yellow band (Bourne and Munn, 2005; Koren and Rosenberg, 2006; Allers et al., 2008; Apprill, Hughnen, and Mincer, 2013; Tout et al., 2015). i. SPLASH Sites A and G were located farthest away from the oil seep and west of the Mississippi river outflow, and sites B, C, D, E, and F were all located near the oil seep (Fig. 8). Oil sheens seen on the surface and the distinct smell of oil were observed at sites C, D, E, and F while sampling. Wind speed was highest at site C in the oil slick (6 m/s, see Table 1). Wind speed was lowest at site F (0-2 m/s, see Table 1). Samples were collected from inside the oil slicks at each of these three sites. Sampling at site F occurred soon after sampling at site E, where slicks were intermittently seen during sampling but were just outside of the main oil slick. The abundance of seventeen specific surfactant- and oil-associated bacteria found in the SPLASH samples were analyzed.

64

Surfactant-Associated Bacteria When combined together, there is a clear correlation between greater abundance of these surfactant- and oil-associated bacteria with oil presence. Combined abundance was greatest in the SSW at sites D, E and F where slicks were visually observed and wind speeds were low (Table 1, Fig. 8). Wind speed was lowest at site F (0-2 m/s, Table 1), and abundance of surfactant- and oil-associated bacteria was greater in the SSW at site F than in any other site (Fig. 18), supporting the idea that rougher surface conditions results in lower abundance of bacteria. Within the oil slicks, abundance of surfactant- and oil-associated bacteria was greater in the SSW, whereas outside of slicks, most bacteria genera were more abundant in the SML. Several studies have found that surfactants are transported from the subsurface water to the surface through bubbles and physical processes (Agogue et al., 2005), which are supported by our findings (Naoko et al., 2015; Hamilton et al., 2016; Howe et al., 2018; Parks et al., 2019). Alteromonas and Halomonas were significantly enriched in the SSW, and Oleibacter and Oleispira were significantly enriched in the SML in the oil slicks and were in low to no abundance at any site outside of the slicks (Figs. 20, 23, 28, 29). Specifically, Alteromonas was in an order of magnitude greater abundance in samples collected in the oil slicks compared to sites with no oil. Jin et al. (2012) found that this genus includes species that are key agents of polycyclic aromatic hydrocarbon biodegradation in crude-oil contaminated coastal sediments. One potential reason for its significant increase in the presence of oil could be because other genera are not able to survive in oil, so this genus has more space and resources to grow. Another reason could be that oil provides a substantial amount of energy and encourages quick growth and multiplication. Focusing on these four genera of bacteria related to oil degradation could lead to further bioremediation research. In contrast, abundance of Alcanivorax was significantly greater in the SML than the SSW in the oil slicks, and greatest in both the SML and SSW at site G, possibly influenced by surfactant input from the runoff in the Mississippi River (Fig. 19). Tenacibaculum was in greatest abundance in the SML at site C and in the SSW at site E, both in the oil slicks (Fig. 24). Fusibacter was found only in the SSW at site C and in the SML at site D, both in the oil slicks (Fig. 25). Rhodococcus was found significantly enriched in the SML at site A, away from the oil seep and site F in the intermittent slicks and enriched in the SSW at site E within the oil slick

65 (Fig. 30). This genus shows no distinct pattern in correlation with presence of oil. Methylophaga and Enterococcus were only found in the SML, but without any clear association with presence of oil (Figs. 27, 22). Arthrobacter and Hyphomonas were found in both the SML and SSW at some sites and did not show any clear pattern between location (Figs. 21, 26). Outside of the oil slicks, abundance of surfactant-associated bacteria was greater in the SML. This follows results of previous work with a subset of surfactant-associated bacteria by Hamilton et al. (2015), Kurata et al. (2016), and Howe et al. (2018), which suggested that within no-oil slicks, surfactant-associated bacteria is typically in greater abundance within the SSW. There could be important physical processes that transport surfactants produced by these genera into the microlayer from below and/or laterally redistributed within the microlayer. The oil- degrading bacteria present can either break down the oil from within the slick or reside in the SSW, and further understanding of which genera are capable of this is necessary to understand the processes associated with natural oil-spill breakdown in the marine environment, specifically with Alteromonas species. Das and Chandran (2011) state that bioremediation is a promising technology for treating oil-contaminated sites, because it can be more cost-effective than other methods and will lead to complete mineralization.

Non-Surfactant Associated Bacteria Synechococcus was found at all sites, with no significant difference between SML and SSW at any site, and no clear pattern is discernable in relation to oil presence (Fig. 31). Prochlorococcus does follow the assumption that it is generally absent in brackish or well-mixed waters, as there was none found at sites A and G, west of the Mississippi River outflow influenced by this freshwater runoff (Fig. 32). Vibrio was found in low abundance at all sites, except for in the SML at both sites A and G, away from the oil seep and influenced by the Mississippi River runoff (Fig. 33). This could indicate that either the runoff could be a source of this bacteria or is associated with the presence of Vibrio and absence of Prochlorococcus. Kelly (1982) found that Vibrio sp. was more frequently isolated from the Gulf of Mexico coast waters during summer months and in low salinity water, which could indicate the influence of freshwater at these sites as well.

66 Overall, these findings indicate that although sites B, C, D, E and F were located closest to the mouth of the Mississippi River, sites A and G were more influenced by the runoff due to the absence of Prochlorococcus and presence of Vibrio. ii. Looe Key Sites 1 through 6 were located near the Looe Key coral reef, where there were no slicks seen by eye or in SAR imagery. Samples were collected from sites 1, 3 and 5 during the night in darkness, whereas samples were collected from sites 2 and 4 during morning in daylight. Site 7 was located in a seagrass area near Lois Key, were samples were collected from slicks and foam on the sea surface. Wind speed was lowest at site 3 (1-2 m/s, see Table 2). Surfactant-Associated Bacteria Overall, surfactant-associated bacteria abundance was generally greater in the SML than the SSW at all sites including the slick area at site 7. This follows the SPLASH results and previous studies regarding non-slick environments, as the abundance of these bacteria were in greater abundance in the SML than SSW. The main difference between these samples and samples from SPLASH was that in the oil slicks in the Gulf of Mexico, abundance was greater in the SSW than in the SML. Abundance of surfactant-associated bacteria was greatest in the SML at site 3 (Fig. 34), in the lowest wind speed (1-2 m/s, see Table 2) compared to the rest of the sites with greater wind speed conditions. Lower wind speed provides a more habitable environment for bacteria at the surface. Alcanivorax was enriched in the SSW near the coral reef and was greatest in abundance in both the SML and SSW in the seagrass area slicks (Fig. 35). We found that species in this genus tended to reside in the SML over the coral reef, opposite from the SPLASH results where they were in greater abundance in the SSW in the presence of oil. In both the Looe Key and SPLASH experiments, Alcanivorax was most abundant at sites 7 near the coastal area and site G west of the Mississippi river outflow, influenced by proximity to and/or runoff from land. Both Alteromonas and Halomonas were generally found in almost all samples over the coral reef (Figs. 38 and 40). Oil-degrading bacteria Methylophaga, Winogradskyella, Hyphomonas, Roseovarius, and Tenacibaculum were found almost exclusively in the SML at sites over the coral reef (Figs. 51, 46, 46, 49, 42). Without the presence of an oil slick at the surface, these oil-degrading bacteria

67 may be more capable of residing within the SML, in contrast to residing in the SSW from the SPLASH samples in the oil slick (Fig. 9). Fusibacter, Oleibacter, Oleiphilus, and Erythrobacter were in greatest abundance in the SML at site three, where wind speed was lowest (Figs. 47, 52, 54, 56). This again supports the idea that lower wind speed creates a more habitable surface environment for oil-associated bacteria. Overall, abundance of these oil-degrading and surfactant-associated bacteria at every site was greater in the SML than in the SSW (Figs. 34), indicating the importance of oil-degrading bacteria at the surface regardless of oil slick presence. Although no oil or slicks were seen by eye or in SAR at sites 1-6 over the coral reef, these oil- associated bacteria were still present. Most of these surfactant- and oil-associated bacteria are very low in counts of ASVs (<10), so it may be difficult to confidently infer meaningful conclusions other than that these bacteria are present in marginal frequencies in these samples. Further investigation with more sample counts, increasing statistical power, could be done in the future.

Non-Surfactant Associated Bacteria Synechococcus was greatest in abundance in the SSW at sites where sample collection occurred late at night compared to the morning (Figs. 58, 59). This suggests that the diurnal cycle affects the abundance of these important primary producers in the water column, either better supporting a habitat in the water column for them or reducing predation during nighttime. Prochlorococcus is absent from the seagrass slick area slick entirely, which suggests that this water is brackish or well-mixed compared to the open water near the coral reef where it was more abundant (Fig. 60). This agrees with the SPLASH results, that Prochloroccocus was absent from both sites that were located west of the Mississippi River outflow in the freshwater runoff (Fig. 3). Vibrio was relatively more abundant than many of the other surfactant-associated bacteria at all sites and was at most sites enriched in the SML (Fig. 62). This suggests that this genus is important in both coral reef and turbid seagrass ecosystems. Whether the stony coral tissue loss disease spreading through the Florida Keys is associated with this genus could be studied further by analyzing samples from the corals. Several genera of surfactant-associated bacteria were found in great abundance on the air controls, non-exposed controls, and non-template controls both in the Gulf of Mexico and Looe Key. These genera include Corynebacterium, Novosphingobium, Bacillus, Pseudomonas, and

68 Acinetobacter. These genera could have been introduced at any point during filter preparation, sample collection, or DNA extraction and sequencing. Due to their presence on the control filters in greater abundance than any of the other samples, reliable results could not be inferred, so they were not included in analysis.

Effect of Diurnal Cycle on Bacteria To determine the effect of the diurnal cycle on the bacteria genera, samples were grouped by timing (day and night), and compared between the SML and SSW. The relative abundance (SML to SSW) of both the 12 surfactant-associated bacteria and 3 non-surfactant associated bacteria does depend on time of day. The bacteria genera in greatest abundance, Halomonas, Alcanivorax, and Alteromonas were significantly less abundant in the SML than SSW during daytime (Figs. 36, 38, 40). In contrast, Erythrobacter, Oleiphilus, Oleibacter, and Roseovarius, which were found almost exclusively in the SML at all sites, were in greater abundance during nighttime than during daytime (Figs. 58, 55, 53, 44). These seven genera of bacteria seem to be more sensitive to light, as they either reside more in the SSW than the SML during the day or reside more in the SML at night than the SML during the day. Winogradskyella, Hyphomonas, Tenacibaculum were in greater abundance in the SML during daytime than during nighttime, which is different than expected from the previous groups of bacteria (Figs. 46, 50). There is no literature on the effect of the diurnal cycle on surfactant-associated bacteria, so this finding could lead to further research on surfactant production depending on time of day. This could sffect sea surface slick expression in SAR, as the satellites pass over an area of interest at sunrise and sunset. The diurnal cycle also had a significant effect on the relative abundance of the three non- surfactant associated bacteria. Vibrio was greater in the SML than the SSW during the nighttime. These primary producers, Prochlorococcus and Synechococcus, follow a different pattern than the surfactant-associated bacteria, being in greater abundance in the SSW than the SML during the nighttime. Hood et al. (2016) found the Synechococcus ceased growth and protein synthesis in dark, so this could be due to diel vertical migration bringing up these two bacteria attached to larger zooplankton, or another reason which could be an area of further research. An important result from this, is that the difference between daytime and nighttime abundance of surfactant-associated bacteria in the SML and SSW can have an effect on the

69 presence of surfactants and, as a result, on the visibility of ocean features in SAR satellite imagery depending on time of day.

8. Conclusions In this project, I analyzed 138 samples from the SPLASH experiment in the Gulf of Mexico and collected and analyzed 220 samples from the Looe Key experiment. In the Gulf of Mexico, samples were collected inside and outside of oil spills. In the Looe Key experiment, samples were collected above the Looe Key coral reef and a coastal seagrass area. Illumina MiSeq sequencing was used to analyze bacterial composition of all samples. For the Looe Key experiment, we also advanced the sampling method. Three TerraSAR-X synthetic aperture radar satellite images around sampling times, one RADARSAT-2 image, and one Sentinel-1 image were acquired in the Gulf of Mexico. We acquired two RADARSAT-2 synthetic aperture radar satellite images coinciding exactly with sampling times and one image before sampling over the Looe Key coral reef. We collected 54 samples in the oil spill and 84 samples outside of the oil spill in the Gulf of Mexico during the SPLASH experiment. In SPLASH we focused on comparing the bacterial content in the oil spill and outside of the oil spill. We collected 90 samples at night and 120 samples during day at Looe Key to compare microbial results between times of sampling, as well as from the sea surface microlayer and subsurface water to compare day and night measurements. In the Looe Key experiment, we focused on bacterial content associated with the coral reef and the diurnal cycle. Samples were also collected from natural slick in a coastal seagrass area. When combined together, the 12 surfactant-associated bacteria found in the SPLASH samples were in greater abundance in the SSW than the SML in the oil slicks. Outside of the oil slicks, abundance was greater in the SML than the SSW. The 12 surfactant-associated bacteria from Looe Key samples were found to be enriched in the SML when combined together. This suggests that the presence of oil creates a harsher environment, leading to a greater abundance of bacteria residing in the SSW instead of the SML. Natural slicks present in the seagrass area during Looe Key did not follow this pattern, as bacteria resided primarily in the SML. The genera of bacteria that are responsible for the largest abundance in SPLASH samples compared to Looe Key (Alteromonas, Oleibacter, and Halomonas) are strongly associated with the presence of oil, and they were in greatest abundance in the oil slicks than in any other site in

70 both the SPLASH and Looe Key samples. The lack of oil and slicks in the Looe Key samples may lead to the lower abundance of surfactant-associated bacteria in Looe Key. Relative abundance of the surfactant-associated bacteria between SML and SSW depends on time of day. The difference between daytime and nighttime abundance of surfactant- associated bacteria in the SML and SSW can have an effect on the presence of surfactants and, as a result, the visibility of ocean features in SAR satellite imagery. Ultimately, within non-slick and natural slick environments, surfactant-associated bacteria reside in greater abundance within the SML, whereas in oil slicks there is greater abundance of these bacteria within the SSW. As the SML is a harsh environment, impacted by wind, breaking waves, and UV rays, bacteria genera found in the SML samples are potentially more resistant and capable of surviving these elements. The specific bacteria, such as Oleibacter and Alteromonas, enriched in in samples from the oil-slicks during the SPLASH project, can be studied further to understand their abilities and functions in the presence of oil. These surfactant- associated bacteria influence fine scale interactions occurring within the SML and SSW that can be remotely detected by SAR, due to their ability to dampen surface gravity-capillary waves. This work supports the importance of using SAR in coordination with in situ work. This project successfully combined SAR to remotely sense the surface features of the ocean in productive areas with next-generation sequencing and allowed for comparison to results from oil slicks. Practical applications of this work include oil spill tracking and bioremediation, environmental monitoring with SAR satellites, air-sea interaction, and ocean circulation.

71 9. References Agogué, H., Casamayor, E. O., Bourrain, M., Obernosterer, I., Joux, F., Herndl, G. J., & Lebaron, P. (2005). A survey on bacteria inhabiting the sea surface microlayer of coastal ecosystems. FEMS Microbiology Ecology, 54(2), 269-280.

Allers, E., Niesner, C., Wild, C., & Pernthaler, J. (2008). Microbes Enriched in Seawater after Addition of Coral Mucus. Applied and Environmental Microbiology, 74(10), 3274. Retrieved from http://aem.asm.org/content/74/10/3274.abstract. doi:10.1128/AEM.01870-07

Alpers, W., & Hühnerfuss, H. (1989). The damping of ocean waves by surface films: A new look at an old problem. Journal of Geophysical Research: Oceans, 94(C5), 6251-6265.

Alpers, W., & Espedal, H. A. (2004). Oils and surfactants. Synthetic aperture radar marine user’s manual, 263-275.

Apprill, A., Hughen, K., & Mincer, T. (2013). Major similarities in the bacterial communities associated with lesioned and healthy F ungiidae corals. Environmental Microbiology, 15(7), 2063-2072.

Arefi, M., Borzadaran, G. M., & Vaghei, Y. (2016). Interval estimations for the mean of the generalized Poisson and generalized negative binomial distributions. Communications in Statistics- Simulation and Computation, 45(6), 1838-1864.

Bonthond, G., Merselis, D. G., Dougan, K. E., Graff, T., Todd, W., Fourqurean, J. W., & Rodriguez- Lanetty, M. (2018). Inter-domain microbial diversity within the coral holobiont Siderastrea siderea from two depth habitats. PeerJ, 6, e4323.

Bourne, D. G., & Munn, C. B. (2005). Diversity of bacteria associated with the coral Pocillopora damicornis from the Great Barrier Reef. Environmental Microbiology, 7(8), 1162-1174.

Bolyen, E., Rideout, J. R., Dillon, M. R., Bokulich, N. A., Abnet, C., Al-Ghalith, G. A., . . . Asnicar, F. (2018). QIIME 2: Reproducible, interactive, scalable, and extensible microbiome data science. PeerJ.

Bozzi, L., Milas, M., & Rinaudo, M. (1996). Characterization and solution properties of a new exopolysaccharide excreted by the bacterium Alteromonas sp. strain 1644. International journal of biological macromolecules, 18(1-2), 9-17.

Burch, A. Y., Shimada, B. K., Browne, P. J., & Lindow, S. E. (2010). Novel high-throughput detection method to assess bacterial surfactant production. Appl. Environ. Microbiol., 76(16), 5363-5372.

Callahan, B. J., McMurdie, P. J., & Holmes, S. P. (2017). Exact sequence variants should replace operational taxonomic units in marker-gene data analysis. The ISME journal, 11(12), 2639.

Caporaso, J. G., Lauber, C. L., Walters, W. A., Berg-Lyons, D., Huntley, J., Fierer, N., . . . Bauer, M. (2012). Ultra-high-throughput microbial community analysis on the Illumina HiSeq and MiSeq platforms. The ISME journal, 6(8), 1621.

72 Caporaso, J. G., Lauber, C. L., Walters, W. A., Berg-Lyons, D., Lozupone, C. A., Turnbaugh, P. J., . . . Knight, R. (2011). Global patterns of 16S rRNA diversity at a depth of millions of sequences per sample. Proceedings of the National Academy of Sciences, 108(Supplement 1), 4516-4522.

Ceh, J., Van Keulen, M., & Bourne, D. G. (2011). Coral-associated bacterial communities on Ningaloo Reef, Western Australia. FEMS Microbiology Ecology, 75(1), 134-144.

Crossland, C. (1987). In situ release of mucus and DOC-lipid from the corals Acropora variabilis and Stylophora pistillata in different light regimes. Coral Reefs, 6, 35–42

Cunliffe, M., & Murrell, J. C. (2009). The sea-surface microlayer is a gelatinous biofilm. The ISME journal, 3, 1001. Retrieved from https://doi.org/10.1038/ismej.2009.69. doi:10.1038/ismej.2009.69

Cunliffe, M., Engel, A., Frka, S., Gašparović, B., Guitart, C., Murrell, J. C., . . . Wurl, O. (2013). Sea surface microlayers: A unified physicochemical and biological perspective of the air–ocean interface. Progress in Oceanography, 109, 104-116.

Cunliffe, M., Upstill-Goddard, R. C., & Murrell, J. C. (2011). Microbiology of aquatic surface microlayers. FEMS microbiology reviews, 35(2), 233-246.

Das, N., & Chandran, P. (2011). Microbial Degradation of Petroleum Hydrocarbon Contaminants: An Overview. Biotechnology Research International, 2011. Retrieved from http://dx.doi.org/10.4061/2011/941810. doi:10.4061/2011/941810

Dean, C., Soloviev, A., Hirons, A., Frank, T., & Wood, J. (2016). Biomixing due to diel vertical migrations of zooplankton: Comparison of computational fluid dynamics model with observations. Ocean Modelling, 98, 51-64.

Dinsdale, E. A., Edwards, R. A., Hall, D., Angly, F., Breitbart, M., Brulc, J. M., . . . Li, L. (2008). Functional metagenomic profiling of nine biomes. Nature, 452(7187), 629.

Donahue, S., Acosta, A., Akins, L., Ault, J., Bohnsack, J., Boyer, J., . . . Delaney, J. (2008). The state of coral reef ecosystems of the Florida Keys. The State of Coral Reef Ecosystems of the United States and Pacific Freely Associated States: NOAA Technical Memorandum NOS NCCOS, 73, 161-187.

Doyle, S. M., Whitaker, E. A., De Pascuale, V., Wade, T. L., Knap, A. H., Santschi, P. H., . . . Sylvan, J. B. (2018). Rapid Formation of Microbe-Oil Aggregates and Changes in Community Composition in Coastal Surface Water Following Exposure to Oil and the Dispersant Corexit. Frontiers in Microbiology, 9(689). Retrieved from https://www.frontiersin.org/article/10.3389/fmicb.2018.00689. doi:10.3389/fmicb.2018.00689

Ducklow, H. W., & Mitchell, R. (1979). Bacterial populations and adaptations in the mucus layers on living corals1. Limnology and Oceanography, 24(4), 715-725. Retrieved from https://doi.org/10.4319/lo.1979.24.4.0715. doi:10.4319/lo.1979.24.4.0715

73 Engel, A. S., & Gupta, A. A. (2014). Regime shift in sandy beach microbial communities following Deepwater Horizon oil spill remediation efforts. PloS one, 9(7), e102934.

Enright, J. (1977). Diurnal vertical migration: Adaptive significance and timing. Part 1. Selective advantage: A metabolic model 1. Limnology and Oceanography, 22(5), 856-872.

Faith, D. P., Minchin, P. R., & Belbin, L. (1987). Compositional dissimilarity as a robust measure of ecological distance. Vegetatio, 69(1-3), 57-68.

Falkowska, L. (2001). 12-hour cycle of matter transformation in the sea surface microlayer in the offshore waters of the Gdańsk Basin (Baltic Sea) during spring. Oceanologia, 43(2).

Ferrier-Pages, C., Gattuso, J.P., Cauwet, G., Jaubert, J., Allemand D. (1998). Release of dissolved organic carbon and nitrogen by the zooxanthellate coral Galaxea fascicularis. Mar Ecol Prog Ser, 172, 265–274

Franklin, M. P., McDonald, I. R., Bourne, D. G., Owens, N. J., Upstill‐Goddard, R. C., & Murrell, J. C. (2005). Bacterial diversity in the bacterioneuston (sea surface microlayer): the bacterioneuston through the looking glass. Environmental Microbiology, 7(5), 723-736.

Gade, M., Byfield, V., Ermakov, S., Lavrova, O., & Mitnik, L. (2013). Slicks as indicators for marine processes. Oceanography, 26(2), 138-149.

Gajigan, A. P., Diaz, L. A., & Conaco, C. (2017). Resilience of the prokaryotic microbial community of Acropora digitifera to elevated temperature. MicrobiologyOpen, 6(4), e00478.

Gast, G. J., Wiegman, S., Wieringa, E., van Duyl, F. C., & Bak, R. P. (1998). Bacteria in coral reef water types: removal of cells, stimulation of growth and mineralization. Marine Ecology Progress Series, 167, 37-45.

Georgiou, G., Lin, S.-C., & Sharma, M. M. (1992). Surface–active compounds from microorganisms. Bio/technology, 10(1), 60.

Golyshin, P. N., Chernikova, T. N., Abraham, W.-R., Lünsdorf, H., Timmis, K. N., & Yakimov, M. M. (2002). Oleiphilaceae fam. nov., to include Oleiphilus messinensis gen. nov., sp. nov., a novel marine bacterium that obligately utilizes hydrocarbons. International Journal of Systematic and Evolutionary Microbiology, 52(3), 901-911.

Gutierrez, T., Mulloy, B., Black, K., & Green, D. (2007). Glycoprotein emulsifiers from two marine Halomonas species: chemical and physical characterization. Journal of applied microbiology, 103(5), 1716-1727.

Hamilton, B., Dean, C., Kurata, N., Vella, K., Soloviev, A., Tartar, A., . . . Lehner, S. (2015). Surfactant associated bacteria in the sea surface microlayer: Case studies in the Straits of Florida and the Gulf of Mexico. Canadian Journal of Remote Sensing, 41(2), 135-143.

Harrison, P.L., Babcock, R.C., Bull, G.D., Oliver, J.K., Wallace, C.C., Willis, B.L. (1984). Mass spawning in tropical reef corals. Science 223(4641), 1186–1189.

74 Hood, R. D., Higgins, S. A., Flamholz, A., Nichols, R. J., & Savage, D. F. (2016). The stringent response regulates adaptation to darkness in the cyanobacterium Synechococcus elongatus. Proceedings of the National Academy of Sciences, 113(33), E4867. Retrieved from http://www.pnas.org/content/113/33/E4867.abstract. doi:10.1073/pnas.1524915113

Howe, K., Dean, C. W., Kluge, J. A., Soloviev, A., Tartar, A., Shivji, M. S., ... & Perrie, W. (2018). Relative abundance of Bacillus spp., surfactant-associated bacterium present in a natural sea slick observed by satellite SAR imagery over the Gulf of Mexico. Elementa: Science of the Anthropocene, 6, 1.

Ibacache-Quiroga, C., Ojeda, J., & Dinamarca, M. A. (2018). 16S rRNA Amplicon Sequencing of Seawater Microbiota from Quintero Bay, Chile, Affected by Oil Spills, Shows the Presence of an Oil-Degrading Marine Bacterial Guild Structured by the Bacterial Genera Alcanivorax, Cobetia, Halomonas, and Oleiphilus. Microbiol Resour Announc, 7(21), e01366-01318.

Iwasa, Y. (1982). Vertical migration of zooplankton: a game between predator and prey. The American Naturalist, 120(2), 171-180.

Kappell, A. D., Wei, Y., Newton, R. J., Van Nostrand, J. D., Zhou, J., McLellan, S. L., & Hristova, K. R. (2014). The polycyclic aromatic hydrocarbon degradation potential of Gulf of Mexico native coastal microbial communities after the Deepwater Horizon oil spill. Frontiers in Microbiology, 5, 205.

Kasai, Y., Kishira, H., Sasaki, T., Syutsubo, K., Watanabe, K., & Harayama, S. (2002). Predominant growth of Alcanivorax strains in oil‐contaminated and nutrient‐supplemented sea water. Environmental Microbiology, 4(3), 141-147.

Kelly, M. T. (1982). Effect of temperature and salinity on Vibrio (Beneckea) vulnificus occurrence in a Gulf Coast environment. Applied and Environmental Microbiology, 44(4), 820. Retrieved from http://aem.asm.org/content/44/4/820.abstract.

Kelly, L. W., Williams, G. J., Barott, K. L., Carlson, C. A., Dinsdale, E. A., Edwards, R. A., . . . McDole, T. (2014). Local genomic adaptation of coral reef-associated microbiomes to gradients of natural variability and anthropogenic stressors. Proceedings of the National Academy of Sciences, 111(28), 10227-10232.

Koren, O., & Rosenberg, E. (2006). Bacteria associated with mucus and tissues of the coral Oculina patagonica in summer and winter. Appl. Environ. Microbiol., 72(8), 5254-5259.

Kurata, N., Vella, K., Hamilton, B., Shivji, M., Soloviev, A., Matt, S., . . . Perrie, W. (2016). Surfactant- associated bacteria in the near-surface layer of the ocean. Scientific reports, 6, 19123.

Lang, S. (2002). Biological amphiphiles (microbial biosurfactants). Current Opinion in Colloid & Interface Science, 7(1-2), 12-20.

Liew, S. C. (2001). Principles of remote sensing. Centre for Remote Imaging, Sensing and Processing, National University of Singapore, Singapore, 119260.

75 Liss, P. S., Liss, P. S., & Duce, R. A. (Eds.). (2005). The sea surface and global change. Cambridge University Press (2).

Liu, Z., & Liu, J. (2013). Evaluating bacterial community structures in oil collected from the sea surface and sediment in the northern Gulf of Mexico after the Deepwater Horizon oil spill. MicrobiologyOpen, 2(3), 492-504.

MacIntyre, F. (1974). The top millimeter of the ocean. Scientific American, 230(5), 62-77.

McMurdie, P. J., & Holmes, S. (2013). phyloseq: an R package for reproducible interactive analysis and graphics of microbiome census data. PloS one, 8(4), e61217.

Naumann, M. S., Richter, C., el-Zibdah, M., & Wild, C. (2009). Coral mucus as an efficient trap for picoplanktonic cyanobacteria: implications for pelagic–benthic coupling in the reef ecosystem. Marine Ecology Progress Series, 385, 65-76.

Oksanen, J., Blanchet, F., Friendly, M., Kindt, R., Legendre, P., McGlinn, D., . . . Solymos, P. (2018). vegan: community ecology package. R package version 2.4–4. 2017. In.

Partensky, F., Blanchot, J., & Vaulot, D. (1999). Differential distribution and ecology of Prochlorococcus and Synechococcus in oceanic waters: a review. Bulletin-Institut Oceanographique Monaco-Numero Special-, 457-476.

Passeri, A., Schmidt, M., Haffner, T., Wray, V., Lang, S., & Wagner, F. (1992). Marine biosurfactants. IV. Production, characterization and biosynthesis of an anionic glucose lipid from the marine bacterial strain MM1. Applied Microbiology and Biotechnology, 37(3), 281-286.

Pepi, M., Cesaro, A., Liut, G., & Baldi, F. (2005). An antarctic psychrotrophic bacterium Halomonas sp. ANT-3b, growing on n-hexadecane, produces a new emulsyfying glycolipid. FEMS Microbiology Ecology, 53(1), 157-166.

Richards, S. A., Possingham, H. P., & Noye, J. (1996). Diel vertical migration: modelling light-mediated mechanisms. Journal of Plankton Research, 18(12), 2199-2222.

Santisi, S., Catania, V., Catalfamo, M., Yakimov, M., Quatrini, P., Mancini, G., & Cappello, S. (2013). Hydrocarbonoclastic bacteria from a polluted harbour in Sicily (Italy): isolation, identification and characterization of their biotechnological potential. Paper presented at the First EMBO Conference on Aquatic Microbial Ecology SAME 13.

Satpute, S. K., Banat, I. M., Dhakephalkar, P. K., Banpurkar, A. G., & Chopade, B. A. (2010). Biosurfactants, bioemulsifiers and exopolysaccharides from marine microorganisms. Biotechnology advances, 28(4), 436-450.

Scanlan, D. J., & West, N. J. (2002). Molecular ecology of the marine cyanobacterial genera Prochlorococcus and Synechococcus. FEMS Microbiology Ecology, 40(1), 1-12.

Soloviev, A. V., Lukas, R., Donelan, M. A., Haus, B. K., & Ginis, I. (2014). The air-sea interface and surface stress under tropical cyclones. Scientific reports, 4, 5306.

76 Stich, H.-B., & Lampert, W. (1981). Predator evasion as an explanation of diurnal vertical migration by zooplankton. Nature, 293(5831), 396.

Teramoto, M., Ohuchi, M., Hatmanti, A., Darmayati, Y., Widyastuti, Y., Harayama, S., & Fukunaga, Y. (2011). Oleibacter marinus gen. nov., sp. nov., a bacterium that degrades petroleum aliphatic hydrocarbons in a tropical marine environment. International Journal of Systematic and Evolutionary Microbiology, 61(2), 375-380.

Tout, J., Jeffries, T. C., Webster, N. S., Stocker, R., Ralph, P. J., & Seymour, J. R. (2014). Variability in microbial community composition and function between different niches within a coral reef. Microbial ecology, 67(3), 540-552.

Tout, J., Siboni, N., Messer, L. F., Garren, M., Stocker, R., Webster, N. S., . . . Seymour, J. R. (2015). Increased seawater temperature increases the abundance and alters the structure of natural Vibrio populations associated with the coral Pocillopora damicornis. Frontiers in Microbiology, 6, 432.

Vincent, P., Pignet, P., Talmont, F., Bozzi, L., Fournet, B., Guezennec, J., . . . Prieur, D. (1994). Production and characterization of an exopolysaccharide excreted by a deep-sea hydrothermal vent bacterium isolated from the polychaete annelid Alvinella pompejana. Appl. Environ. Microbiol., 60(11), 4134-4141.

Wang, L., Wang, W., Lai, Q., & Shao, Z. (2010). Gene diversity of CYP153A and AlkB alkane hydroxylases in oil‐degrading bacteria isolated from the Atlantic Ocean. Environmental Microbiology, 12(5), 1230-1242.

Wild, C., Huettel, M., Klueter, A., Kremb, S. G., Rasheed, M. Y., & Jørgensen, B. B. (2004). Coral mucus functions as an energy carrier and particle trap in the reef ecosystem. Nature, 428(6978), 66.

Wild, C., Niggl, W., Naumann, M. S., & Haas, A. F. (2010). Organic matter release by Red Sea coral reef organisms—potential effects on microbial activity and in situ O2 availability. Marine Ecology Progress Series, 411, 61-71.

Wurl, O., Ekau, W., Landing, W. M., & Zappa, C. J. (2017). Sea surface microlayer in a changing ocean–A perspective. Elem Sci Anth, 5.

Wurl, O., & Holmes, M. (2008). The gelatinous nature of the sea-surface microlayer. Marine Chemistry, 110(1-2), 89-97.

Wurl, O., Wurl, E., Miller, L., Johnson, K., & Vagle, S. (2011). Formation and global distribution of sea-surface microlayers. Biogeosciences, 8(1), 121-135.

Yakimov, M. M., Denaro, R., Genovese, M., Cappello, S., D’Auria, G., Chernikova, T. N., . . . Giluliano, L. (2005). Natural microbial diversity in superficial sediments of Milazzo Harbor (Sicily) and community successions during microcosm enrichment with various hydrocarbons. Environmental Microbiology, 7(9), 1426-1441.

77 Yakimov, M. M., Golyshin, P. N., Lang, S., Moore, E. R., Abraham, W.-R., Lünsdorf, H., & Timmis, K. N. (1998). Alcanivorax borkumensis gen. nov., sp. nov., a new, hydrocarbon-degrading and surfactant-producing marine bacterium. International Journal of Systematic and Evolutionary Microbiology, 48(2), 339-348.

Zaitsev, Y., & Liss, P. (2005). Neuston of seas and oceans. The sea surface and global change. Cambridge University Press, Cambridge, 371-382.

Zhang, Z., Liu, L., Liu, C., & Cai, W. (2003). Studies on the sea surface microlayer: II. The layer of sudden change of physical and chemical properties. Journal of colloid and interface science, 264(1), 148-159.

78 10. Supplementary Material Looe Key Bacteria Confidence Interval Calculations Table S1. Alcanivorax abundance from Looe Key samples with negative binomial distribution and calculation of Wald confidence intervals. Alcanivorax SML T T/2 N MIN_CI MAX_CI

1 0 2.776 1.388 5 0 0 2 0 2.201 1.1005 12 0 0 3 20.3 2.262 1.131 10 12.86295 27.73705 4 15 2.201 1.1005 12 10.07841 19.92159 5 32.54545 2.228 1.114 11 21.44729 43.64361 6 57.1 2.262 1.131 10 36.49993 77.70007 7 162.3 2.262 1.131 10 104.0743 220.5257 SSW T T/2 N MIN_CI MAX_CI 1 16.6 2.776 1.388 5 5.990013 27.20999 2 0 2.262 1.131 10 0 0 3 69 2.262 1.131 10 44.14372 93.85628 4 120.7 2.262 1.131 10 77.35275 164.0472 5 0 2.262 1.131 10 0 0 6 51.1 2.262 1.131 10 32.64594 69.55406 7 173.7 2.262 1.131 10 111.397 236.003 AC T T/2 N MIN_CI MAX_CI 1 0 2.776 1.388 5 0 0 2 0 2.262 1.131 10 0 0 3 0 2.262 1.131 10 0 0 4 0 2.262 1.131 10 0 0 5 0 12.706 6.353 2 0 0 6 0 2.262 1.131 10 0 0 7 0 2.262 1.131 10 0 0 NE 0 2.306 1.153 9 0 0 NTC 0 2.262 1.131 10 0 0

79 Table S2. Alteromonas abundance from Looe Key samples with negative binomial distribution and calculation of Wald confidence intervals. Alteromonas SML T T/2 N MIN_CI MAX_CI

1 22.6 2.776 1.388 5 8.264439 36.93556 2 0 2.201 1.1005 12 0 0 3 27.2 2.262 1.131 10 17.29461 37.10539 4 2 2.201 1.1005 12 1.221829 2.778171 5 7.181818 2.228 1.114 11 4.607092 9.756544 6 0 2.262 1.131 10 0 0 7 42 2.262 1.131 10 26.80077 57.19923 SSW T T/2 N MIN_CI MAX_CI 1 128.4 2.776 1.388 5 48.38819 208.4118 2 0 2.262 1.131 10 0 0 3 0 2.262 1.131 10 0 0 4 0 2.262 1.131 10 0 0 5 0 2.262 1.131 10 0 0 6 78.1 2.262 1.131 10 49.989 106.211 7 6.5 2.262 1.131 10 4.00282 8.99718 AC T T/2 N MIN_CI MAX_CI 1 0 2.776 1.388 5 0 0 2 0 2.262 1.131 10 0 0 3 0 2.262 1.131 10 0 0 4 0 2.262 1.131 10 0 0 5 0 12.706 6.353 2 0 0 6 0 2.262 1.131 10 0 0 7 0 2.262 1.131 10 0 0 NE 0 2.306 1.153 9 0 0 NTC 0 2.262 1.131 10 0 0

80 Table S3. Halomonas abundance from Looe Key samples with negative binomial distribution and calculation of Wald confidence intervals. Halomonas SML T T/2 N MIN_CI MAX_CI 1 38.8 2.776 1.388 5 14.40719 63.19281 2 0 2.201 1.1005 12 0 0 3 0 2.262 1.131 10 0 0 4 1.583333 2.201 1.1005 12 0.940829 2.225837 5 12.18182 2.228 1.114 11 7.925516 16.43812 6 0 2.262 1.131 10 0 0 7 3.9 2.262 1.131 10 2.336518 5.463482 SSW T T/2 N MIN_CI MAX_CI 1 0 2.776 1.388 5 0 0 2 14.6 2.262 1.131 10 9.202392 19.99761 3 33.7 2.262 1.131 10 21.46955 45.93045 4 0 2.262 1.131 10 0 0 5 0.3 2.262 1.131 10 0.076645 0.523355 6 48.90909 2.262 1.131 10 31.23866 66.57952 7 3.8 2.262 1.131 10 2.272522 5.327478 AC T T/2 N MIN_CI MAX_CI 1 0 2.776 1.388 5 0 0 2 0 2.262 1.131 10 0 0 3 0 2.262 1.131 10 0 0 4 0 2.262 1.131 10 0 0 5 0 12.706 6.353 2 0 0 6 0 2.262 1.131 10 0 0 7 0 2.262 1.131 10 0 0 NE 0 2.306 1.153 9 0 0 NTC 0 2.262 1.131 10 0 0

81 Table S4. Tenacibaculum abundance from Looe Key samples with negative binomial distribution and calculation of Wald confidence intervals. Tenacibaculum SML T T/2 N MIN_CI MAX_CI 1 0 2.776 1.388 5 0 0 2 49.75 2.201 1.1005 12 33.78702 65.71298 3 4.6 2.262 1.131 10 2.784754 6.415246 4 0 2.201 1.1005 12 0 0 5 0 2.228 1.114 11 0 0 6 17.1 2.262 1.131 10 10.80784 23.39216 7 4.3 2.262 1.131 10 2.592602 6.007398 SSW T T/2 N MIN_CI MAX_CI 1 0 2.776 1.388 5 0 0 2 0 2.262 1.131 10 0 0 3 0 2.262 1.131 10 0 0 4 0 2.262 1.131 10 0 0 5 0 2.262 1.131 10 0 0 6 0 2.262 1.131 10 0 0 7 0 2.262 1.131 10 0 0 AC T T/2 N MIN_CI MAX_CI 1 0 2.776 1.388 5 0 0 2 0 2.262 1.131 10 0 0 3 0 2.262 1.131 10 0 0 4 0 2.262 1.131 10 0 0 5 0 12.706 6.353 2 0 0 6 0 2.262 1.131 10 0 0 7 0 2.262 1.131 10 0 0 NE 0 2.306 1.153 9 0 0 NTC 0 2.262 1.131 10 0 0

82 Table S5. Roseovarius abundance from Looe Key samples with negative binomial distribution and calculation of Wald confidence intervals. Roseovarius SML T T/2 N MIN_CI MAX_CI 1 0 2.776 1.388 5 0 0 2 0 2.201 1.1005 12 0 0 3 1.9 2.262 1.131 10 1.060466 2.739534 4 0 2.201 1.1005 12 0 0 5 0 2.228 1.114 11 0 0 6 0 2.262 1.131 10 0 0 7 0 2.262 1.131 10 0 0 SSW T T/2 N MIN_CI MAX_CI 1 0 2.776 1.388 5 0 0 2 0 2.262 1.131 10 0 0 3 0 2.262 1.131 10 0 0 4 0 2.262 1.131 10 0 0 5 0 2.262 1.131 10 0 0 6 0 2.262 1.131 10 0 0 7 0 2.262 1.131 10 0 0 AC T T/2 N MIN_CI MAX_CI 1 0 2.776 1.388 5 0 0 2 0 2.262 1.131 10 0 0 3 0 2.262 1.131 10 0 0 4 0 2.262 1.131 10 0 0 5 0 12.706 6.353 2 0 0 6 0 2.262 1.131 10 0 0 7 0 2.262 1.131 10 0 0 NE 0 2.306 1.153 9 0 0 NTC 0 2.262 1.131 10 0 0

83 Table S6. Winogradskyella abundance from Looe Key samples with negative binomial distribution and calculation of Wald confidence intervals. Winogradskyella SML T T/2 N MIN_CI MAX_CI 1 0 2.776 1.388 5 0 0 2 0 2.201 1.1005 12 0 0 3 17.1 2.262 1.131 10 10.80784 23.39216 4 26.16667 2.201 1.1005 12 17.6965 34.63683 5 2.272727 2.228 1.114 11 1.356681 3.188774 6 22.2 2.262 1.131 10 14.08323 30.31677 7 2.5 2.262 1.131 10 1.442046 3.557954 SSW T T/2 N MIN_CI MAX_CI 1 0 2.776 1.388 5 0 0 2 2.9 2.262 1.131 10 1.697199 4.102801 3 19.4 2.262 1.131 10 12.28494 26.51506 4 0 2.262 1.131 10 0 0 5 0 2.262 1.131 10 0 0 6 0 2.262 1.131 10 0 0 7 0 2.262 1.131 10 0 0 AC T T/2 N MIN_CI MAX_CI 1 0 2.776 1.388 5 0 0 2 0 2.262 1.131 10 0 0 3 0 2.262 1.131 10 0 0 4 0 2.262 1.131 10 0 0 5 0 12.706 6.353 2 0 0 6 0 2.262 1.131 10 0 0 7 0 2.262 1.131 10 0 0 NE 0 2.306 1.153 9 0 0 NTC 0 2.262 1.131 10 0 0

84 Table S7. Fusibacter abundance from Looe Key samples with negative binomial distribution and calculation of Wald confidence intervals. Fusibacter SML T T/2 N MIN_CI MAX_CI

1 0 2.776 1.388 5 0 0 2 0 2.201 1.1005 12 0 0 3 0 2.262 1.131 10 0 0 4 0 2.201 1.1005 12 0 0 5 0 2.228 1.114 11 0 0 6 1.7 2.262 1.131 10 0.933753 2.466247 7 41.2 2.262 1.131 10 26.28692 56.11308 SSW T T/2 N MIN_CI MAX_CI 1 0 2.776 1.388 5 0 0 2 0 2.262 1.131 10 0 0 3 0 2.262 1.131 10 0 0 4 17.1 2.262 1.131 10 10.80784 23.39216 5 0 2.262 1.131 10 0 0 6 0 2.262 1.131 10 0 0 7 0 2.262 1.131 10 0 0 AC T T/2 N MIN_CI MAX_CI 1 0 2.776 1.388 5 0 0 2 0 2.262 1.131 10 0 0 3 0 2.262 1.131 10 0 0 4 0 2.262 1.131 10 0 0 5 0 12.706 6.353 2 0 0 6 0 2.262 1.131 10 0 0 7 0 2.262 1.131 10 0 0 NE 0 2.306 1.153 9 0 0 NTC 0 2.262 1.131 10 0 0

85 Table S8. Hyphomonas abundance from Looe Key samples with negative binomial distribution and calculation of Wald confidence intervals. Hyphomonas SML T T/2 N MIN_CI MAX_CI

1 83.2 2.776 1.388 5 31.24562 135.1544 2 0 2.201 1.1005 12 0 0 3 40.4 2.262 1.131 10 25.77306 55.02694 4 38.83333 2.201 1.1005 12 26.33865 51.32801 5 2.727273 2.228 1.114 11 1.656373 3.798172 6 87.2 2.262 1.131 10 55.83429 118.5657 7 7 2.262 1.131 10 4.323566 9.676434 SSW T T/2 N MIN_CI MAX_CI 1 0 2.776 1.388 5 0 0 2 0 2.262 1.131 10 0 0 3 0 2.262 1.131 10 0 0 4 0 2.262 1.131 10 0 0 5 0 2.262 1.131 10 0 0 6 0 2.262 1.131 10 0 0 7 0 2.262 1.131 10 0 0 AC T T/2 N MIN_CI MAX_CI 1 0 2.776 1.388 5 0 0 2 0 2.262 1.131 10 0 0 3 0 2.262 1.131 10 0 0 4 0 2.262 1.131 10 0 0 5 0 12.706 6.353 2 0 0 6 0 2.262 1.131 10 0 0 7 0 2.262 1.131 10 0 0 NE 0 2.306 1.153 9 0 0 NTC 0 2.262 1.131 10 0 0

86 Table S9. Methylophaga abundance from Looe Key samples with negative binomial distribution and calculation of Wald confidence intervals. Methylophaga SML T T/2 N MIN_CI MAX_CI

1 0 2.776 1.388 5 0 0 2 0 2.201 1.1005 12 0 0 3 0 2.262 1.131 10 0 0 4 0 2.201 1.1005 12 0 0 5 0 2.228 1.114 11 0 0 6 0 2.262 1.131 10 0 0 7 7.7 2.262 1.131 10 4.772698 10.6273 SSW T T/2 N MIN_CI MAX_CI 1 0 2.776 1.388 5 0 0 2 0 2.262 1.131 10 0 0 3 0 2.262 1.131 10 0 0 4 0 2.262 1.131 10 0 0 5 0 2.262 1.131 10 0 0 6 0 2.262 1.131 10 0 0 7 0 2.262 1.131 10 0 0 AC T T/2 N MIN_CI MAX_CI 1 0 2.776 1.388 5 0 0 2 0 2.262 1.131 10 0 0 3 0 2.262 1.131 10 0 0 4 0 2.262 1.131 10 0 0 5 0 12.706 6.353 2 0 0 6 0 2.262 1.131 10 0 0 7 0 2.262 1.131 10 0 0 NE 0 2.306 1.153 9 0 0 NTC 0 2.262 1.131 10 0 0

87 Table S10. Vibrio abundance from Looe Key samples with negative binomial distribution and calculation of Wald confidence intervals. Vibrio SML T T/2 N MIN_CI MAX_CI 1 145.2 2.776 1.388 5 54.75981 235.6402 2 120.4167 2.201 1.1005 12 82.00334 158.83 3 330.3 2.262 1.131 10 211.9883 448.6117 4 281.4167 2.201 1.1005 12 191.8556 370.9778 5 184.4545 2.228 1.114 11 122.3316 246.5775 6 311.4 2.262 1.131 10 199.848 422.952 7 405.2 2.262 1.131 10 260.1 550.3 SSW T T/2 N MIN_CI MAX_CI 1 133.4 2.776 1.388 5 50.2845 216.5155 2 199.5 2.262 1.131 10 127.9695 271.0305 3 11.9 2.262 1.131 10 7.468702 16.3313 4 219.3 2.262 1.131 10 140.6879 297.9121 5 115.4 2.262 1.131 10 73.94833 156.8517 6 178.2 2.262 1.131 10 114.2876 242.1124 7 148.2 2.262 1.131 10 95.01721 201.3828 AC T T/2 N MIN_CI MAX_CI 1 27.6 2.776 1.388 5 10.16018 45.03982 2 0 2.262 1.131 10 0 0 3 0 2.262 1.131 10 0 0 4 6.6 2.262 1.131 10 4.066964 9.133036 5 0 12.706 6.353 2 0 0 6 8.8 2.262 1.131 10 5.478632 12.12137 7 0 2.262 1.131 10 0 0 NE 0 2.306 1.153 9 0 0 NTC 0 2.262 1.131 10 0 0

88 Table S11. Oleibacter abundance from Looe Key samples with negative binomial distribution and calculation of Wald confidence intervals. Oleibacter SML T T/2 N MIN_CI MAX_CI

1 0 2.776 1.388 5 0 0 2 0 2.201 1.1005 12 0 0 3 118.7 2.262 1.131 10 76.06807 161.3319 4 1.75 2.201 1.1005 12 1.053077 2.446923 5 12.72727 2.228 1.114 11 8.287624 17.16692 6 2.7 2.262 1.131 10 1.569566 3.830434 7 0 2.262 1.131 10 0 0 SSW T T/2 N MIN_CI MAX_CI 1 0 2.776 1.388 5 0 0 2 0 2.262 1.131 10 0 0 3 7.1 2.262 1.131 10 4.387721 9.812279 4 0 2.262 1.131 10 0 0 5 0 2.262 1.131 10 0 0 6 0 2.262 1.131 10 0 0 7 0 2.262 1.131 10 0 0 AC T T/2 N MIN_CI MAX_CI 1 0 2.776 1.388 5 0 0 2 0 2.262 1.131 10 0 0 3 0 2.262 1.131 10 0 0 4 0 2.262 1.131 10 0 0 5 0 12.706 6.353 2 0 0 6 0 2.262 1.131 10 0 0 7 0 2.262 1.131 10 0 0 NE 0 2.306 1.153 9 0 0 NTC 0 2.262 1.131 10 0 0

89 Table S12. Oleiphilus abundance from Looe Key samples with negative binomial distribution and calculation of Wald confidence intervals. Oleiphilus SML T T/2 N MIN_CI MAX_CI 1 0 2.776 1.388 5 0 0 2 0 2.201 1.1005 12 0 0 3 37.5 2.262 1.131 10 23.91034 51.08966 4 0 2.201 1.1005 12 0 0 5 0 2.228 1.114 11 0 0 6 0 2.262 1.131 10 0 0 7 0 2.262 1.131 10 0 0 SSW T T/2 N MIN_CI MAX_CI 1 0 2.776 1.388 5 0 0 2 0 2.262 1.131 10 0 0 3 0 2.262 1.131 10 0 0 4 0 2.262 1.131 10 0 0 5 0 2.262 1.131 10 0 0 6 0 2.262 1.131 10 0 0 7 0 2.262 1.131 10 0 0 AC T T/2 N MIN_CI MAX_CI 1 0 2.776 1.388 5 0 0 2 0 2.262 1.131 10 0 0 3 0 2.262 1.131 10 0 0 4 0 2.262 1.131 10 0 0 5 0 12.706 6.353 2 0 0 6 0 2.262 1.131 10 0 0 7 0 2.262 1.131 10 0 0 NE 0 2.306 1.153 9 0 0 NTC 0 2.262 1.131 10 0 0

90 Table S13. Erythrobacter abundance from Looe Key samples with negative binomial distribution and calculation of Wald confidence intervals. Erythrobacter SML T T/2 N MIN_CI MAX_CI

1 0 2.776 1.388 5 0 0 2 0 2.201 1.1005 12 0 0 3 48.2 2.262 1.131 10 30.78319 65.61681 4 5.666667 2.201 1.1005 12 3.714047 7.619286 5 0 2.228 1.114 11 0 0 6 18.6 2.262 1.131 10 11.77116 25.42884 7 0 2.262 1.131 10 0 0 SSW T T/2 N MIN_CI MAX_CI 1 0 2.776 1.388 5 0 0 2 0 2.262 1.131 10 0 0 3 0 2.262 1.131 10 0 0 4 0 2.262 1.131 10 0 0 5 0 2.262 1.131 10 0 0 6 0 2.262 1.131 10 0 0 7 0 2.262 1.131 10 0 0 AC T T/2 N MIN_CI MAX_CI 1 0 2.776 1.388 5 0 0 2 0 2.262 1.131 10 0 0 3 0 2.262 1.131 10 0 0 4 0 2.262 1.131 10 0 0 5 0 12.706 6.353 2 0 0 6 0 2.262 1.131 10 0 0 7 0 2.262 1.131 10 0 0 NE 0 2.306 1.153 9 0 0 NTC 0 2.262 1.131 10 0 0

91 Table S14. Synechococcus abundance from Looe Key samples with negative binomial distribution and calculation of Wald confidence intervals. Synechococcus SML T T/2 N MIN_CI MAX_CI

1 939.8 2.776 1.388 5 356.1253 1523.475 2 1022.167 2.201 1.1005 12 697.2788 1347.055 3 3014.3 2.262 1.131 10 1936.046 4092.554 4 2777.833 2.201 1.1005 12 1895.193 3660.474 5 3063.455 2.228 1.114 11 2034.322 4092.587 6 3010.7 2.262 1.131 10 1933.733 4087.667 7 2404.1 2.262 1.131 10 1544.086 3264.114 SSW T T/2 N MIN_CI MAX_CI 1 3096.8 2.776 1.388 5 1174.205 5019.395 2 1842.6 2.262 1.131 10 1183.409 2501.791 3 4113.3 2.262 1.131 10 2641.985 5584.615 4 2774.4 2.262 1.131 10 1781.947 3766.853 5 4633.4 2.262 1.131 10 2976.069 6290.731 6 3598.7 2.262 1.131 10 2311.433 4885.967 7 925.5 2.262 1.131 10 594.3128 1256.687 AC T T/2 N MIN_CI MAX_CI 1 0 2.776 1.388 5 0 0 2 36.3 2.262 1.131 10 23.13956 49.46044 3 15.7 2.262 1.131 10 9.908772 21.49123 4 28 2.262 1.131 10 17.80844 38.19156 5 0 12.706 6.353 2 0 0 6 57.2 2.262 1.131 10 36.56416 77.83584 7 0 2.262 1.131 10 0 0 NE 0 2.306 1.153 9 0 0 NTC 0 2.262 1.131 10 0 0

92 Table S15. Prochlorococcus abundance from Looe Key samples with negative binomial distribution and calculation of Wald confidence intervals. Prochlorococcus SML T T/2 N MIN_CI MAX_CI

1 81.6 2.776 1.388 5 30.63881 132.5612 2 112.6667 2.201 1.1005 12 76.71544 148.6179 3 227.7 2.262 1.131 10 146.0836 309.3164 4 126.4167 2.201 1.1005 12 86.09721 166.7361 5 577.7273 2.228 1.114 11 383.5103 771.9443 6 267.8 2.262 1.131 10 171.8417 363.7583 7 0 2.262 1.131 10 0 0 SSW T T/2 N MIN_CI MAX_CI 1 286.2 2.776 1.388 5 108.2363 464.1637 2 301.1 2.262 1.131 10 193.2318 408.9682 3 341.6 2.262 1.131 10 219.2468 463.9532 4 197 2.262 1.131 10 126.3636 267.6364 5 885.1 2.262 1.131 10 568.362 1201.838 6 576.1 2.262 1.131 10 369.877 782.323 7 0 2.262 1.131 10 0 0 AC T T/2 N MIN_CI MAX_CI 1 0 2.776 1.388 5 0 0 2 0 2.262 1.131 10 0 0 3 0 2.262 1.131 10 0 0 4 0 2.262 1.131 10 0 0 5 0 12.706 6.353 2 0 0 6 0 2.262 1.131 10 0 0 7 0 2.262 1.131 10 0 0 NE 0 2.306 1.153 9 0 0 NTC 0 2.262 1.131 10 0 0

93 SPLASH Bacteria Confidence Interval Calculations

Table S16. Alcanivorax abundance from SPLASH samples with negative binomial distribution and calculation of Wald confidence intervals. Alcanivorax SML T T/2 N MIN_CI MAX_CI A 0 2.262 1.131 10 0 0 B 49.4444444 2.306 1.153 9 30.2500916 68.6387973 C 0 2.262 1.131 10 0 0 D 70.8181818 2.228 1.114 11 46.8641602 94.7722035 E 122 2.16 1.08 14 86.6416322 157.358368 F 0 2.262 1.131 10 0 0 G 206.692308 2.179 1.0895 13 144.084583 269.300032 SSW T T/2 N MIN_CI MAX_CI A 0.55555556 2.306 1.153 9 0.19827074 0.91284037 B 16 2.262 1.131 10 10.1014257 21.8985743 C 0 2.447 1.2235 7 0 0 D 0 2.262 1.131 10 0 0 E 6.7 2.262 1.131 10 4.13111082 9.26888918 F 0 2.262 1.131 10 0 0 G 172.111111 2.306 1.153 9 105.771186 238.451036 AC 0 2.571 1.2855 6 0 0 NE 0 2.571 1.2855 6 0 0 NTC 0 4.303 2.1515 3 0 0

94 Table S17. Alteromonas abundance from SPLASH samples with negative binomial distribution and calculation of Wald confidence intervals. Alteromonas SML T T/2 N MIN_CI MAX_CI A 0.4 2.262 1.131 10 0.13235655 0.66764345 B 252.444444 2.306 1.153 9 155.229653 349.659236 C 36 2.262 1.131 10 22.9468684 49.0531316 D 175.909091 2.228 1.114 11 116.656402 235.16178 E 156.785714 2.16 1.08 14 111.386658 202.18477 F 458.3 2.262 1.131 10 294.208624 622.391376 G 18.8461538 2.179 1.0895 13 13.0022226 24.6900851 SSW T T/2 N MIN_CI MAX_CI A 59.8888889 2.306 1.153 9 36.6802215 83.0975563 B 252.7 2.262 1.131 10 162.142284 343.257716 C 43.2857143 2.447 1.2235 7 23.0387893 63.5326393 D 2630.9 2.262 1.131 10 1689.77033 3572.02967 E 1043 2.262 1.131 10 669.788508 1416.21149 F 4148.9 2.262 1.131 10 2664.85215 5632.94785 G 197.777778 2.306 1.153 9 121.573261 273.982295 AC 166.5 2.571 1.2855 6 78.8582622 254.141738 NE 147.833333 2.571 1.2855 6 69.9879709 225.678696 NTC 0 4.303 2.1515 3 0 0

95 Table S18. Arthrobacter abundance from SPLASH samples with negative binomial distribution and calculation of Wald confidence intervals. Arthrobacter SML T T/2 N MIN_CI MAX_CI A 10.9 2.262 1.131 10 6.82667245 14.9733276 B 0 2.306 1.153 9 0 0 C 0 2.262 1.131 10 0 0 D 10.5454545 2.228 1.114 11 6.83927019 14.2516389 E 0 2.16 1.08 14 0 0 F 0 2.262 1.131 10 0 0 G 0 2.179 1.0895 13 0 0 SSW T T/2 N MIN_CI MAX_CI A 0 2.306 1.153 9 0 0 B 0 2.262 1.131 10 0 0 C 0 2.447 1.2235 7 0 0 D 0 2.262 1.131 10 0 0 E 0 2.262 1.131 10 0 0 F 4.1 2.262 1.131 10 2.46454115 5.73545885 G 0 2.306 1.153 9 0 0 AC 0 2.571 1.2855 6 0 0 NE 0 2.571 1.2855 6 0 0 NTC 0 4.303 2.1515 3 0 0

96 Table S19. Enterococcus abundance from SPLASH samples with negative binomial distribution and calculation of Wald confidence intervals. Enterococcus SML T T/2 N MIN_CI MAX_CI A 0 2.262 1.131 10 0 0 B 0 2.306 1.153 9 0 0 C 5.3 2.262 1.131 10 3.23333153 7.36666847 D 0 2.228 1.114 11 0 0 E 8.57142857 2.16 1.08 14 5.95701238 11.1858448 F 0 2.262 1.131 10 0 0 G 0 2.179 1.0895 13 0 0 SSW T T/2 N MIN_CI MAX_CI A 0 2.306 1.153 9 0 0 B 0 2.262 1.131 10 0 0 C 0 2.447 1.2235 7 0 0 D 0 2.262 1.131 10 0 0 E 0 2.262 1.131 10 0 0 F 0 2.262 1.131 10 0 0 G 0 2.306 1.153 9 0 0 AC 0 2.571 1.2855 6 0 0 NE 0 2.571 1.2855 6 0 0 NTC 0 4.303 2.1515 3 0 0

97 Table S20. Halomonas abundance from SPLASH samples with negative binomial distribution and calculation of Wald confidence intervals. Halmonas SML T T/2 N MIN_CI MAX_CI A 18.4 2.262 1.131 10 11.6427127 25.1572873 B 89.3333333 2.306 1.153 9 54.8079237 123.858743 C 0 2.262 1.131 10 0 0 D 31.6363636 2.228 1.114 11 20.8435915 42.4291358 E 30.2857143 2.16 1.08 14 21.4008319 39.1705967 F 95.5 2.262 1.131 10 61.1657198 129.83428 G 1 2.179 1.0895 13 0.57266294 1.42733706 SSW T T/2 N MIN_CI MAX_CI A 0 2.306 1.153 9 0 0 B 31.1 2.262 1.131 10 19.799561 42.400439 C 17 2.447 1.2235 7 8.91061199 25.089388 D 400.2 2.262 1.131 10 256.888313 543.511687 E 233.5 2.262 1.131 10 149.809248 317.190752 F 560.3 2.262 1.131 10 359.727939 760.872061 G 21.2222222 2.306 1.153 9 12.8758601 29.5685844 AC 8 2.571 1.2855 6 3.54689737 12.4531026 NE 0 2.571 1.2855 6 0 0 NTC 0 4.303 2.1515 3 0 0

98 Table 21. Tenacibaculum abundance from SPLASH samples with negative binomial distribution and calculation of Wald confidence intervals. Tenacibaculum SML T T/2 N MIN_CI MAX_CI A 0 2.262 1.131 10 0 0 B 0 2.306 1.153 9 0 0 C 94.6 2.262 1.131 10 60.5876124 128.612388 D 4.45454545 2.228 1.114 11 2.79889052 6.11020039 E 0.5 2.16 1.08 14 0.25002857 0.74997143 F 0 2.262 1.131 10 0 0 G 0 2.179 1.0895 13 0 0 SSW T T/2 N MIN_CI MAX_CI A 0 2.306 1.153 9 0 0 B 2.3 2.262 1.131 10 1.31466595 3.28533405 C 4.85714286 2.447 1.2235 7 2.39060207 7.32368364 D 6.8 2.262 1.131 10 4.1952601 9.4047399 E 145.8 2.262 1.131 10 93.4755834 198.124417 F 0 2.262 1.131 10 0 0 G 0 2.306 1.153 9 0 0 AC 0 2.571 1.2855 6 0 0 NE 0 2.571 1.2855 6 0 0 NTC 0 4.303 2.1515 3 0 0

99 Table S22. Fusibacter abundance from SPLASH samples with negative binomial distribution and calculation of Wald confidence intervals. Fusibacter SML T T/2 N MIN_CI MAX_CI A 0 2.262 1.131 10 0 0 B 0 2.306 1.153 9 0 0 C 0 2.262 1.131 10 0 0 D 5.81818182 2.228 1.114 11 3.70266354 7.9337001 E 0 2.16 1.08 14 0 0 F 0 2.262 1.131 10 0 0 G 0 2.179 1.0895 13 0 0 SSW T T/2 N MIN_CI MAX_CI A 0 2.306 1.153 9 0 0 B 0 2.262 1.131 10 0 0 C 5.42857143 2.447 1.2235 7 2.69673326 8.1604096 D 0 2.262 1.131 10 0 0 E 0 2.262 1.131 10 0 0 F 0 2.262 1.131 10 0 0 G 0 2.306 1.153 9 0 0 AC 0 2.571 1.2855 6 0 0 NE 0 2.571 1.2855 6 0 0 NTC 0 4.303 2.1515 3 0 0

100 Table S23. Hyphomonas abundance from SPLASH samples with negative binomial distribution and calculation of Wald confidence intervals. Hyphomonas SML T T/2 N MIN_CI MAX_CI A 0 2.262 1.131 10 0 0 B 0 2.306 1.153 9 0 0 C 12.6 2.262 1.131 10 7.91815176 17.2818482 D 0 2.228 1.114 11 0 0 E 0 2.16 1.08 14 0 0 F 0 2.262 1.131 10 0 0 G 1.30769231 2.179 1.0895 13 0.78276726 1.83261735 SSW T T/2 N MIN_CI MAX_CI A 4.55555556 2.306 1.153 9 2.62206314 6.48904797 B 0 2.262 1.131 10 0 0 C 0 2.447 1.2235 7 0 0 D 0 2.262 1.131 10 0 0 E 0 2.262 1.131 10 0 0 F 0 2.262 1.131 10 0 0 G 0 2.306 1.153 9 0 0 AC 0 2.571 1.2855 6 0 0 NE 0 2.571 1.2855 6 0 0 NTC 0 4.303 2.1515 3 0 0

101 Table S24. Methylophaga abundance from SPLASH samples with negative binomial distribution and calculation of Wald confidence intervals. Methylophaga SML T T/2 N MIN_CI MAX_CI A 0 2.262 1.131 10 0 0 B 0 2.306 1.153 9 0 0 C 0 2.262 1.131 10 0 0 D 0 2.228 1.114 11 0 0 E 7.35714286 2.16 1.08 14 5.09383701 9.6204487 F 0 2.262 1.131 10 0 0 G 1.84615385 2.179 1.0895 13 1.15349611 2.53881158 SSW T T/2 N MIN_CI MAX_CI A 0 2.306 1.153 9 0 0 B 0 2.262 1.131 10 0 0 C 0 2.447 1.2235 7 0 0 D 0 2.262 1.131 10 0 0 E 0 2.262 1.131 10 0 0 F 0 2.262 1.131 10 0 0 G 0 2.306 1.153 9 0 0 AC 0 2.571 1.2855 6 0 0 NE 0 2.571 1.2855 6 0 0 NTC 0 4.303 2.1515 3 0 0

102 Table S25. Oleispira abundance from SPLASH samples with negative binomial distribution and calculation of Wald confidence intervals. Oleispira SML T T/2 N MIN_CI MAX_CI A 0 2.262 1.131 10 0 0 B 0 2.306 1.153 9 0 0 C 0 2.262 1.131 10 0 0 D 190.090909 2.228 1.114 11 126.074761 254.107057 E 229.357143 2.16 1.08 14 163.010842 295.703444 F 0 2.262 1.131 10 0 0 G 0 2.179 1.0895 13 0 0 SSW T T/2 N MIN_CI MAX_CI A 0 2.306 1.153 9 0 0 B 0 2.262 1.131 10 0 0 C 0 2.447 1.2235 7 0 0 D 0 2.262 1.131 10 0 0 E 0 2.262 1.131 10 0 0 F 0 2.262 1.131 10 0 0 G 0 2.306 1.153 9 0 0 AC 0 2.571 1.2855 6 0 0 NE 0 2.571 1.2855 6 0 0 NTC 0 4.303 2.1515 3 0 0

103 Table S26. Oleibacter abundance from SPLASH samples with negative binomial distribution and calculation of Wald confidence intervals. Oleibacter SML T T/2 N MIN_CI MAX_CI A 0 2.262 1.131 10 0 0 B 0 2.306 1.153 9 0 0 C 17.4 2.262 1.131 10 11.0004986 23.7995014 D 447 2.228 1.114 11 296.692166 597.307834 E 789.571429 2.16 1.08 14 561.523565 1017.61929 F 0 2.262 1.131 10 0 0 G 0 2.179 1.0895 13 0 0 SSW T T/2 N MIN_CI MAX_CI A 10.6666667 2.306 1.153 9 6.37924884 14.9540845 B 0 2.262 1.131 10 0 0 C 0 2.447 1.2235 7 0 0 D 0 2.262 1.131 10 0 0 E 12.5 2.262 1.131 10 7.85394341 17.1460566 F 0 2.262 1.131 10 0 0 G 0 2.306 1.153 9 0 0 AC 0 2.571 1.2855 6 0 0 NE 0 2.571 1.2855 6 0 0 NTC 0 4.303 2.1515 3 0 0

104 Table S27. Rhodococcus abundance from SPLASH samples with negative binomial distribution and calculation of Wald confidence intervals. Rhodococcus SML T T/2 N MIN_CI MAX_CI A 34.5 2.262 1.131 10 21.9834013 47.0165987 B 19.4444444 2.306 1.153 9 11.7815388 27.1073501 C 3.7 2.262 1.131 10 2.2085373 5.1914627 D 2.54545455 2.228 1.114 11 1.53641664 3.55449245 E 2.71428571 2.16 1.08 14 1.79780111 3.63077031 F 7.1 2.262 1.131 10 4.38772145 9.81227855 G 0 2.179 1.0895 13 0 0 SSW T T/2 N MIN_CI MAX_CI A 0.77777778 2.306 1.153 9 0.32584426 1.2297113 B 3.7 2.262 1.131 10 2.2085373 5.1914627 C 0 2.447 1.2235 7 0 0 D 3.7 2.262 1.131 10 2.2085373 5.1914627 E 34 2.262 1.131 10 21.6622466 46.3377534 F 0 2.262 1.131 10 0 0 G 0 2.306 1.153 9 0 0 AC 0 2.571 1.2855 6 0 0 NE 0 2.571 1.2855 6 0 0 NTC 0 4.303 2.1515 3 0 0

105 Table S28. Synechococcus abundance from SPLASH samples with negative binomial distribution and calculation of Wald confidence intervals. Synechococcus SML T T/2 N MIN_CI MAX_CI A 400.5 2.262 1.131 10 257.081017 543.918983 B 229.333333 2.306 1.153 9 141.000931 317.665735 C 1070.2 2.262 1.131 10 687.260329 1453.13967 D 1168.72727 2.228 1.114 11 776.002999 1561.45155 E 710.928571 2.16 1.08 14 505.580356 916.276787 F 400.5 2.262 1.131 10 257.081017 543.918983 G 1891.23077 2.179 1.0895 13 1319.60096 2462.86058 SSW T T/2 N MIN_CI MAX_CI A 551.222222 2.306 1.153 9 339.177069 763.267376 B 455.9 2.262 1.131 10 292.666993 619.133007 C 2399.71429 2.447 1.2235 7 1289.76034 3509.66823 D 729 2.262 1.131 10 468.091758 989.908242 E 305.3 2.262 1.131 10 195.929674 414.670326 F 252 2.262 1.131 10 161.692642 342.307358 G 1457.44444 2.306 1.153 9 897.107829 2017.78106 AC 103.333333 2.571 1.2855 6 48.8419019 157.824765 NE 0 2.571 1.2855 6 0 0 NTC 0 4.303 2.1515 3 0 0

106 Table S29. Prochlorococcus abundance from SPLASH samples with negative binomial distribution and calculation of Wald confidence intervals. Prochlorococcus SML T T/2 N MIN_CI MAX_CI A 1.4 2.262 1.131 10 0.74441012 2.05558988 B 1876 2.306 1.153 9 1154.79853 2597.20147 C 976 2.262 1.131 10 626.751302 1325.2487 D 1505.54545 2.228 1.114 11 999.689456 2011.40145 E 2131.78571 2.16 1.08 14 1516.31822 2747.25321 F 1267.7 2.262 1.131 10 814.123735 1721.27626 G 3.30769231 2.179 1.0895 13 2.16707371 4.4483109

SSW T T/2 N MIN_CI MAX_CI A 0 2.306 1.153 9 0 0 B 3031.3 2.262 1.131 10 1946.96582 4115.63418 C 204 2.447 1.2235 7 109.431398 298.568602 D 3773.4 2.262 1.131 10 2423.65108 5123.14892 E 2677.2 2.262 1.131 10 1719.51096 3634.88904 F 2179.8 2.262 1.131 10 1400.00787 2959.59213 G 0 2.306 1.153 9 0 0 AC 145.666667 2.571 1.2855 6 68.9583844 222.374949 NE 3.5 2.571 1.2855 6 1.41725196 5.58274804 NTC 0 4.303 2.1515 3 0 0

107 Table S30. Vibrio abundance from SPLASH samples with negative binomial distribution and calculation of Wald confidence intervals. Vibrio SML T T/2 N MIN_CI MAX_CI A 2809.4 2.262 1.131 10 1804.42916 3814.37084 B 106.111111 2.306 1.153 9 65.137358 147.084864 C 100.6 2.262 1.131 10 64.4416629 136.758337 D 317.363636 2.228 1.114 11 210.598574 424.128699 E 32.3571429 2.16 1.08 14 22.8742849 41.8400008 F 120.4 2.262 1.131 10 77.1600491 163.639951 G 971.538462 2.179 1.0895 13 677.814789 1265.26213 SSW T T/2 N MIN_CI MAX_CI A 202.111111 2.306 1.153 9 124.241145 279.981078 B 129.5 2.262 1.131 10 83.0053755 175.994625 C 59.7142857 2.447 1.2235 7 31.8697796 87.5587919 D 91.9 2.262 1.131 10 58.8532909 124.946709 E 5.3 2.262 1.131 10 3.23333153 7.36666847 F 19.1 2.262 1.131 10 12.0922707 26.1077293 G 140.666667 2.306 1.153 9 86.4119514 194.921382 AC 177.333333 2.571 1.2855 6 84.0062038 270.660463 NE 0 2.571 1.2855 6 0 0 NTC 0 4.303 2.1515 3 0 0

108 SPLASH and Looe Key Synthetic Aperture Radar Information

Table S31. Synthetic Aperture Radar imagery information during the SPLASH and Looe Key experiments. Satellite Date Time Incidence Coordinate Polarization Cruise (UTC) Angle Point Coordinates Sentinel-1 4/16/17 1:30 30.669°- 27.95°N90.77°W VH/VV 46.127° 28.37°N88.19°W 29.46°N91.10°W 29.87°N88.49°W TerraSAR-X 4/19/17 12:08 19.888°- 29.31°N89.74°W HH/VV 21.82° 29.28°N89.56°W 28.81°N89.85°W 28.78°N89.67°W TerraSAR-X 4/20/17 23:57 42.649°- 29.17°N89.09°W HH/VV 43.844° 29.19°N88.93°W 28.66°N88.99°W

SPLASH 28.69°N88.83°W RADARSAT- 4/20/17 23:57 32.696°- 29.20°N89.32°W HH/VV 2 35.666° 29.20°N88.77°W HV/VH 28.89°N89.32°W 28.89°N89.77W TerraSAR-X 4/25/17 12:00 40.902°- 29.41°N90.06°W HH/VV 42.152° 29.41°N89.79°W 28.87°N90.06°W 28.87°N89.79°W RADARSAT- 07/30/18 11:15 44.61 24.38°N81.36°W HH 2 24.61°N81.32°W 24.65°N81.55°W 24.42°N81.59°W RADARSAT- 07/31/18 23:34 41.23° 24.44°N81.30°W HH 2 24.68°N81.35°W 24.64°N81.58°W

Looe Key Looe 24.40°N81.53°W RADARSAT- 08/02/18 11:27 25.91° 24.40°N81.25°W HH 2 24.63°N81.20°W 24.68°N81.43°W 24.44°N81.48°W

109