1 Host and environmental determinants of microbial community structure in the marine
2 phyllosphere
3
4 Margaret A. Vogel1, Olivia U. Mason2, and Thomas E. Miller1
5
6
7 Author Affiliations: 1Department of Biological Science, Florida State University, Tallahassee,
8 FL, 2Department of Earth, Ocean, and Atmospheric Science, Florida State University,
9 Tallahassee, FL
10
11 Corresponding Author: Margaret A. Vogel, Address 319 Stadium Drive, Tallahassee, FL
12 32301, Phone (850) 644-9823, Fax (850) 645-8447, Email [email protected]
13
14
15
16
17
18
19
20 Abstract
21 Although seagrasses are economically and ecologically critical foundation species, little
22 is known about their blade surface microbial communities and how those communities relate to
23 overall seagrass health. 16S rRNA gene sequencing (iTag) was used to examine the microbial
24 community composition and diversity on blade surfaces at five sites along a gradient of
25 freshwater input in the northern Gulf of Mexico. Additionally, seagrass surveys were performed
26 and environmental parameters were measured to characterize host characteristics and the in situ
27 conditions at each site. Results show that Thalassia testudinum (turtle grass) blades host unique
28 microbial communities that are distinct in composition and diversity from the water column. In
29 addition, compositional changes within these blade surface communities correlated with both
30 environmental conditions, including water depth, salinity, and temperature, and host
31 characteristics, including seagrass growth rates and blade nutrient composition. These
32 correlations may indicate that blade surface community composition changes with stressful
33 conditions either as a direct or indirect effect. Additionally, 15 microorganisms from five phyla
34 (Cyanobacteria, Proteobacteria, Bacteroidetes, Planctomycetes, and Chloroflexi) were present in
35 all blade surface samples, even after a large disturbance event (Hurricane Irma), and may
36 represent a core community for T. testudinum. Members of this core community may have
37 ecological importance for determining community structure or in performing key community
38 functions. Studies such as this are the first step to understanding what processes influence the
39 structure of marine phyllosphere communities in order to determine how these blade surface
40 communities relate to their host and to seagrass health as a whole.
41 Key Words
42 Seagrass; Bacteria; Core Communities; Phyllosphere; Microbial Ecology
43 Introduction
44 In recent years, there has been an increasing number of studies on the microbial
45 communities associated with plant hosts, especially of those communities associated with the
46 phyllosphere, or leaf surfaces (Lindow and Leveau 2002, Lindow and Brandl 2003, Vorholt
47 2012, Vandenkoornhuyse et al. 2015). We now know that the phyllosphere is a rich habitat that
48 can host up to 107 microbial cells per cm2 of leaf tissue and that these epiphytic microbial
49 communities can have a variety of relationships with their host plants ranging from beneficial to
50 pathogenic (Vorholt 2012). However, leaf associated microbial communities of aquatic plants,
51 including seagrasses, remain largely unexplored when compared to those of terrestrial plants.
52 Seagrasses are a polyphyletic group of angiosperms that colonized the marine
53 environment ~100 million years ago and currently have about 72 species distributed worldwide
54 (Hemminga and Duarte 2000, Short et al. 2011). Seagrasses can form dense monospecific or
55 mixed species meadows that serve as nurseries, feeding grounds, and habitats for a wide variety
56 of marine species from invertebrates to sea turtles and manatees. These foundation species also
57 support bacteria, algal epiphytes, and their grazers, creating highly productive ecosystems
58 (Zieman and Zieman 1989, Hemminga and Duarte 2000). In addition, seagrasses provide
59 valuable ecosystem services, such as stabilizing sediments and trapping and cycling nutrients
60 (Costanza et al. 1997, Duarte 2002, Barbier et al. 2011) and are important sites for blue carbon
61 sequestration (Fourqurean et al. 2012, Duarte et al. 2013). However, with rising anthropogenic
62 influence, eutrophication and degraded water quality are increasingly becoming threats to these
63 important habitats (Duarte 2002, Orth et al. 2006, Short et al. 2011) with seagrass coverage
64 declining at a rate of 110 km2 yr-1 worldwide since 1980 (Waycott et al. 2009).
65 Although there have been a few recent studies on the microbial communities that occur
66 on seagrass blade surfaces (Meija et al. 2016, Fahimipour et al. 2017, Crump et al. 2018, Ugarelli
67 et al. 2019), the interactions between these communities and the seagrass host remain poorly
68 understood. This is especially true for the tropical seagrass species Thalassia testudinum Banks
69 ex Kӧnig (turtle grass), for which there is only one previously published study that contains
70 information about its blade surface microbial communities (Ugarelli et al. 2019). Thalassia
71 testudinum is an important climax species and can be a dominant component of shallow waters
72 in the Caribbean, Western Atlantic, and Gulf of Mexico. In these areas, it can act as an
73 ecosystem engineer creating dense meadows which likely provide more ecosystem services than
74 other smaller seagrass species (Nordlund et al. 2016). Adding to their importance, tropical
75 seagrass meadows often occur adjacent to other critical habitats for biodiversity, such as coral
76 reefs and mangrove forests, and their presence has been correlated with a two-fold reduction of
77 disease levels in nearby corals (Lamb et al. 2017).
78 In the terrestrial phyllosphere, host-species has been found to be a significant driver of
79 variation in microbial community composition with more variation in leaf-associated
80 communities often occurring across plant-host species rather than within a host species even
81 across large spatial scales (Redford et al. 2010, Finkel et al. 2012, Laforest-Lapointe et al.
82 2016a). For instance, intra-specific variability of the microbial communities on Pinus ponderosa
83 was found to be less than inter-specific variability within and across continents (Redford et al.
84 2010). However, variation also occurs between these leaf-associated communities due to
85 environmental conditions, including precipitation/moisture, temperature, and salt content
86 (Jackson et al. 2006, Finkel et al. 2012, Vorholt et al. 2012, Laforest-Lapointe et al. 2016a,
87 2016b, 2017). In a study of seven tree species, the proportion of Alphaproteobacteria, a dominant
88 class in the natural plant microbiome, was found to decrease along a gradient of urban intensity
89 (Laforest-Lapointe et al. 2017). These compositional shifts can lead to changes in the
90 relationship between these leaf associated microbial communities and their host plant which can
91 ultimately affect host fitness and performance (Lindow and Leveau 2002, Vandenkoornhuyse et
92 al. 2015, Saleem et al. 2017). However, it is unknown how much variation exists within leaf-
93 associated microbial communities in the marine phyllosphere and what roles both biotic and
94 abiotic factors play in determining microbial community structure.
95 Characterizing the variation in blade surface microbial communities on seagrasses is the
96 first step to understanding the relative influence of host and environmental conditions on
97 community structure in the marine phyllosphere, which is essential to elucidating the potential
98 role of these microbial communities as a part of the seagrass holobiont. This study is the first to
99 use 16S rRNA amplicon sequencing (iTag: Illumina platform) to characterize the structure and
100 diversity of the microbial communities associated with T. testudinum blades and to examine
101 whether these microbial communities vary in composition with environmental and host
102 characteristics.
103
104 Methods
105 Sampling Location
106 This study took place in Apalachee Bay in the northern Gulf of Mexico along the Florida
107 Panhandle. The coastline of this area is mostly undeveloped with the St. Marks National Wildlife
108 Refuge occupying the adjacent land area. Five sites (ABT-1 – ABT-5) were established starting
109 near the mouth of the St. Marks River (30.07059°N, 84.16687°W) and extending south in a
110 linear fashinon approximately two miles into the bay (30.04194°N, 84.16634°W). These sites are
111 situated along a gradient of abiotic conditions caused by riverine input with the farthest site from
112 shore located on a shoal. Seagrasses in this region form dense meadows that have mixed species
113 composition. Thalassia testudinum is the dominant species at all study sites, however
114 Syringodium filiforme Kützing (manatee grass) is common along with small amounts of
115 Halodule wrightii Ascherson (shoal grass) and Halophila engelmannii Ascherson (star grass).
116 Microbial Sampling
117 To determine microbial community structure and diversity, samples were taken at all five sites
118 on three separate dates (22-Jul, 20-Aug, and 21-Sep-2016) capturing both spatial and temporal
119 variation. On each date, samples were taken from both the blade surface and water column with
120 all sites visited during a six-hour period. To sample water column communities, 1 liter of
121 seawater was collected from above the seagrass canopy, filtered using a sterile syringe with a 2.7
122 µM pre-filter, and microbial biomass was collected on a 0.22 μM Sterivex™ filter. To capture the
123 blade surface microbial communities, T. testudinum blades from five haphazardly chosen shoots
124 at least one meter apart were removed using sterile forceps at each site. The blade surface
125 microbial communities were then sampled using a sterile swab (PurFlock® Ultra, Puritan
126 Diagnostics, LLC). Microbial sampling was standardized by using the second oldest blade in the
127 shoot and only swabbing healthy tissue free of algal epiphytes. Microbial samples (swabs and
128 Sterivex™ filters) were immediately fixed in RNAlater® and placed on dry ice to be transported
129 to Florida State University where all samples were stored at -80°C until further processing.
130 A fourth set of samples for microbial analysis was collected the following year during
131 Sep-2017. During this sampling, blades were swabbed at all sites following the same methods as
132 the 2016 samplings; however, seagrass and environmental surveys were not repeated. The 2017
133 sampling was conducted two weeks after Hurricane Irma hit the Florida Gulf Coast. This was the
134 second hurricane to hit during the sampling the period with Hurricane Hermine making landfall
135 near the study sites on 2-Sep-2016. The effects of Hermine were restricted to increased turbidity
136 from storm water run-off as high water due to storm surge acted as a buffer to any physical
137 damage for the seagrass beds. In contrast, Hurricane Irma caused severe low tides leaving
138 seagrass beds exposed and causing large-scale die offs (MV personal observations).
139 Total Suspended Solid and Nutrient Analyses
140 During each microbial sampling, 0.22 μM filtered seawater was also collected at each site
141 and stored in acid-washed 30 ml Nalgene™ bottles for nutrient analysis. An additional 1 liter of
142 unfiltered seawater was collected into acid-washed 500 ml HDPE bottles for analysis of total
143 suspended solids (TSS). Environmental parameters, including water temperature, salinity,
144 conductivity, and dissolved oxygen, were also measured using a YSI Pro2030. In addition, T.
145 testudinum tissue samples were collected to determine tissue nutrient content.
146 Immediately following sample collection, 1 liter of seawater from each site was filtered
147 through a pre-weighed 0.7 μM filter using a vacuum pump. Filters were then dried at 60°C for 48
148 hours and re-weighed to determine total suspended solids content. Nutrient water samples were
149 kept frozen and sent to the Marine Chemistry Laboratory at the University of Washington,
150 School of Oceanography for analysis of PO4, NO3, NO2, and NH4 following the protocols of the
151 WOCE Hydrographic Program.
152 During each microbial sampling, additional healthy T. testudinum blades were
153 haphazardly chosen from a 20 m2 area at each site for nutrient analysis. Immediately following
154 the sample collection, T. testudinum blades were cleaned of epiphytes and debris and dried at
155 60°C for 72 hours. Dried samples were then ground to a fine powder using an acid-washed
156 mortar and pestle and stored in glass screw top vials. These samples were sent to the Stable
157 Isotope Ecology Laboratory of the Center for Applied Isotope Studies at the University of
158 Georgia, where they were analyzed for total carbon, nitrogen, and phosphorus content, as well as
159 stable carbon (δ13C) and nitrogen (δ15N) isotope ratios for indication of short-term and long-term
160 nutrient conditions at each site.
161 Seagrass Surveys
162 During the study period (Jul-Sep 2016), seagrass surveys were conducted to characterize
163 seagrass host condition. At each site, permanent transects were established running east to west
164 with 1 m2 quadrats 15 meters apart along the transect to determine seagrass abundance (percent
165 cover). Within each quadrat, ten blades were randomly chosen to be measured for
166 morphometrics (blade width and length) with only the second and third oldest blades used. Leaf
167 growth rates were measured as an indicator of seagrass growth and leaf turnover. Sexual
168 reproduction was not quantified as clonal reproduction is thought to be the dominant form of
169 growth form in the northern Gulf of Mexico (Phillips, 1960) and as only one inflorescence was
170 observed during the study period. Growth rates were measured using a modification of the leaf
171 marking method originally described by Zieman (1974) in which shoots were randomly chosen
172 at each site and marked two centimeters above the leaf sheath with a 1/16-inch hand punch.
173 Marked shoots were collected after three weeks and new material was measured by length and
174 dry weight. Water depth, temperature, and salinity were also measured at each site during these
175 surveys to further capture the variation in environmental conditions over the course of the study
176 period.
177 Microbial Community Analysis
178 DNA was extracted from the blade surface and water column microbial samples using a
179 phenol-chloroform extraction method (Gilles et al. 2015) and then purified using the QIAGEN
180 AllPrep™ DNA/RNA Mini Kit. 16S rRNA genes were amplified from DNA extracts in
181 duplicate in accordance with the protocol described by Caporaso et al. (2011, 2012) using a
182 modified annealing temperature of 60°C with the archaeal and bacterial primers 515F and 806R
183 (targets the V4 region of E. coli) modified by Apprill et al. (2015) and Parada et al. (2016).
184 During this stage, some samples from 2016 did not successfully amplify and were excluded from
185 sequencing. Amplicons were sequenced using an Illumina MiSeq at the University of Illinois
186 (2016) and at the Argonne National Laboratory (2017) in 250 x 250 b.p. mode. These sequences
187 will be available in NCBI’s SRA (accession XXX) and on the Mason server at
188 http://mason.eoas.fsu.edu. Raw sequences were demultiplexed using QIIME2 (Caporaso et al.
189 2010). Demultiplexed reads were quality filtered, including chimera removal, and joined using
190 DADA2 (Callahan et al. 2016). The resulting ASV (amplicon sequence variant) table was
191 filtered to remove any sequences resulting from mitochondrial or chloroplast DNA and
192 normalized using cumulative sum scaling (Paulson et al. 2013). Taxonomy was assigned using
193 the SILVA v. 132 (Yilmaz et al. 2014) database in QIIME2. Alpha diversity metrics were
194 obtained using QIIME2 after multiple rarefactions were performed on the data. Statistical
195 analyses were performed using R v 3.6.1 (R Core Team, 2018) and the vegan package (Oksanen
196 et al. 2018) to obtain measures of variation in community structure between sample types,
197 sampling dates, and sites. Microbial community dissimilarity was determined with Non-metric
198 Multidimensional Scaling (NMDS) ordination analysis with Bray-Curtis distance using sequence
199 counts. Ordination analyses were performed using the metaMDS command with the vegan
200 package (Oksanen et al. 2018) in R with 999 permutations and appropriate number of axes to
201 minimize stress. PERMANOVA (adonis) and environmental fitting (envfit) from the vegan
202 package in R were used in combination with the resulting ordinations. Shapiro-Wilkes test was
203 used to determine normality and bootstrapping was used in combination with statistical tests
204 when necessary to account for differences in sample size due to some samples failing to amplify.
205 All mean values are reported with plus or minus the standard error.
206
207 Results
208 Host-Plant and Site Characterization
209 The five study sites were characterized by differing environmental conditions due to the
210 gradient of freshwater input caused by the St Marks River. Salinity increased with distance from
211 shore ranging on average from 21 ppt at the first site (ABT-1) to 27 ppt at the fifth site (ABT-5;
212 Table 1). This salinity gradient might be expected to co-vary with water depth, however, due to
213 the bathymetry of the area, the fifth site has an average depth (1.2 m) similar to the first site (0.98
214 m; Table 1) with deeper sites in between. Additionally, water column nutrients (PO4, NO3, NO2,
215 and NH4) did not follow consistent trends with distance from shore (Table 1) and were not
216 significantly different by site or sampling date (Kruskall-Wallis, p<0.05). Dissolved oxygen
217 concentrations (DO; mg/l) and total suspended solids (TSS; mg/l) content followed opposite
218 patterns as DO concentrations increased then decreased and TSS decreased then increased with
219 distance from shore (Table 1). Thalassia testudinum growth morphology also differed between
220 the five sites. Water depth may be an important driver for the differences seen in blade
221 morphology as depth and blade length had a significant positive correlation (p<0.01, adjusted
222 R2=0.51). However, blade width does not follow this same trend with depth as blades at the first
223 site were found to be significantly smaller than at all other sites (Dunn’s Test with Bonferroni
224 correction, p<0.02). Additionally, seagrass growth rates were found to co-vary with blade width
225 and lengths. The fifth site from shore, ABT-5, which had greater blade lengths than the first site,
226 also had a higher growth rate by weight (2.05 ± 0.15 mg/day/shoot) than ABT-1 (1.57 ± 0.27
227 mg/day/shoot; Table 2). Growth rates were highest at sites ABT-4 and ABT-3 (5.38 ± 0.39 and
228 5.31 ± 0.57 mg/day/shoot, respectively), which also had the longest blades on average (41.56 ±
229 0.90 cm and 39.49 ± 0.93 cm, respectively) and greatest average water depth (Table 1). Blade
230 nutrient composition (%N, %C, %P, and δ13C content) was not found to be significantly different
231 due to site or sampling date; however mean δ15N was found to have significantly higher
232 concentrations at the farthest site (ABT-5) than the first site (ABT-1; Table 1).
233 Blade Surface Community Composition and Diversity
234 A total of 11,252 ASVs were found in T. testudinum blade surface samples (n=52) across
235 all sites during the 2016 samplings with samples dominated by members of the Proteobacteria,
236 Cyanobacteria, and Planctomycetes phyla. While at the species level there were no dominant
237 taxa, two bacterial classes, Gammaproteobacteria (21.56 ± 4.78%) and Alphaproteobacteria,
238 (20.26 ± 6.94%) each comprised approximately 20% of community abundance on average. The
239 majority of microorganisms (ASVs) comprised less than 1% of sample abundance, however
240 three ASVs belonging to the cyanobacterial family Cyanobiaceae were the exception, including
241 the cultured bacterium Synechococcus sp. CENA143. These ASVs comprised a combined 5% of
242 the community relative abundance on average. These three abundant cyanobacteria are also
243 closely related to known cyanobacteria found in mangrove systems (Rigonato et al., 2013; Silva
244 et al., 2014), including Synechococcus sp. CENA 172 and CENA 180 (100% similarity,
245 GenBank Accession KC695872.1 and KC695865.1). Other close relatives to these three ASVs
246 include cyanobacteria from the genera Synechococcus and Prochlorococcus as well as the
247 cyanobacterium Trichocoleus desertorum (87-89% similarity, GenBank Accession NR125697.1)
248 isolated from desert soils (Mühlsteinova et al., 2014).
249 On average, blade surface communities had a richness (chao1) of 792.12 ± 190.92 ASVs
250 with the maximum richness (1565.75 ASVs) occurring at the second site (ABT-2) from shore
251 and the lowest richness (436.03 ASVs) occurring at the fifth site (ABT-5) furthest from shore.
252 Across all sampling dates, richness was significantly lower at the fifth site (657.81 ± 118.99
253 ASVs) than at the second (873.76 ± 236.06 ASVs) and third sites (895.55 ± 165.01 ASVs;
254 Dunn’s Test with Bonferroni Correction, p<0.05). Diversity (Shannon-Weiner) showed a similar
255 pattern, with ABT-5 having significantly lower diversity (8.39 ± 0.29) than ABT-2 (8.80 ± 0.30)
256 and ABT-3 (8.86 ± 0.28; Dunn’s Test with Bonferroni Correction, p<0.05). Both richness and
257 diversity were significantly different by site location, however neither metric significantly
258 differed due to sampling date. Differences in community composition were visualized with
259 NMDS ordination analysis (Figure 1; stress=0.1039337, k=3) with each blade surface
260 community represented by a single point. Blade surface samples were found to differ
261 significantly in species composition due to site as well as having a significant interaction
262 between site and sampling date (PERMANOVA, p<0.01).
263 Water column samples (n=15) contained fewer ASVs with 1,083 total, however
264 approximately 80% of those ASVs were also present in blade surface communities. Individual
265 samples had significantly lower richness (215.92 ± 36.97 ASVs) and diversity (5.73 ± 0.40) than
266 the blade communities (Wilcoxon rank sum test, p<0.001). Community composition of water
267 column samples was found to be significantly different from that of the composition of blade
268 surface samples (PERMANOVA, p=0.001). Within water column samples, community
269 composition differed significantly by site (PERMANOVA, p<0.05; Figure 3), but not by
270 sampling date.
271 Correlations with Environmental and Host Characteristics
272 Compositional differences between the blade surface microbial communities were found
273 to have significant correlations with environmental factors at each site, including water
274 temperature, average depth, salinity, total suspended solids concentration, and phosphate
275 concentration (Figure 1; Table 2; envfit with Bonferroni correction, p<0.05) as well as host
276 characteristics, including average blade length, average blade width, average percent cover, T.
277 testudinum growth rates, and stable isotope composition (δ15N and δ13C content) (Table 2; envfit
278 with Bonferroni correction, p<0.05). In addition, dissolved oxygen content in the water column
279 had a marginally significant correlation with blade surface composition (Table 2; envfit with
280 Bonferroni correction, p=0.054).
281 In addition, other community metrics of the blade surface samples, including diversity
282 and richness, were found to be correlated with both environmental and host parameters. Shannon
283 diversity of the blade surface communities was significantly correlated with total suspended
284 solids concentration, phosphate concentration, and blade δ15N content (Spearman’s Rank Order
285 Correlation with Bonferroni correction, p≤0.05). Richness (chao1) of the blade surface
286 communities was also significantly correlated with water column total suspended solids and
287 blade δ15N content (Spearman’s Rank Order Correlation with Bonferroni correction, p<0.05).
288 However, some of these correlations with environmental and host characteristics are confounded
289 as parameters, such as water depth and blade lengths, co-vary.
290 The composition of the water column microbial communities was significantly correlated
291 with several environmental parameters, including water temperature, salinity, dissolved oxygen,
292 and phosphate concentration (envfit with Bonferroni correction, p<0.05; Figure 2). Salinity and
293 water temperature explained the most variance with the highest R2 values of 0.93 and 0.75,
294 respectively. However, in contrast to blade surface communities, diversity and richness of the
295 water column communities did not significantly correlate with any of the environmental
296 conditions.
297 Core Community
298 Although blade surface communities contained many species in low abundances, 21
299 ASVs were present in 100% of the blade surface samples and combined comprised 8-21% of
300 community abundance. These 21 ASVs represent the following five different bacterial phyla in
301 order of decreasing average abundance- Cyanobacteria, Proteobacteria, Planctomycetes,
302 Chloroflexi, and Bacteroidetes. The three most abundant ASVs from the blade surface samples,
303 which were previously discussed, were also amongst these 21 core community members. The
304 combined abundance of these 21 ASVs was also found to significantly correlate with water
305 temperature at the time of sampling, average water depth, and seagrass growth rates (Spearman’s
306 Rank Order Correlation with Bonferroni correction, p<0.05). Additionally, combined
307 abundances were significantly different between sites (ANOVA, p<0.001) with the furthest two
308 sites (ABT-4 and ABT-5) having a significantly higher abundance of the core community than
309 the first three sites (ABT-1, ABT-2, and ABT-3; Tukey’s HSD, p<0.03). Additionally, members
310 of the core community were largely absent from the water column community. The only
311 exceptions were two ASVs that occurred in no more than two water samples and in extremely
312 low abundances (<0.006% on average).
313 September 2017 Sampling
314 The composition of the blade surface samples from September 2017 was found to be
315 significantly different from the prior year’s blade surface samples (PERMANOVA, p=0.001;
316 Figure 3). Of the total 12,493 ASVs found in the 2017 blade surface samples (n=25), 4,120
317 ASVs (33%) were shared with the previous year. Additionally, these samples contained 15 of the
318 21 ASVs (71%) that comprised the 2016 core community. These 15 core ASVs were present in
319 all 77 samples from both 2016 and 2017 and include members of the Cyanobacteria,
320 Proteobacteria, Bacteroidetes, Planctomycetes, and Chloroflexi phyla. In contrast to the 2016
321 blade surface samples, the four most dominant ASVs from 2017 represent uncultured bacteria
322 belonging to the family Rhodobacteraceae in the Proteobacteria phyla. These ASVs each made
323 up at least 1% of community abundance on average and together comprised 5.4% of total
324 abundance.
325 Both chao1 richness and Shannon diversity were found to be significantly higher in the
326 2017 blade surface samples than in the 2016 blade surface samples (Kruskall-Wallis, p<0.01).
327 Blade surface samples from 2017 had a mean Shannon diversity of 9.62 ± 0.10 with no
328 significant differences in diversity found between sites (Kruskall-Wallis, p>0.05). However,
329 chao1 richness was significantly different between sites (Kruskall-Wallis, p<0.05) with the
330 highest richness observed at ABT-1 (2690.61 ± 102.38) and lowest observed at ABT-3 (1909.45
331 ± 104.64).
332
333 Discussion
334 Blade Surface Microbial Communities
335 The composition of the microbial communities associated with T. testudinum blade
336 surfaces was found to be highly diverse and to vary significantly among sites and sampling dates.
337 These data show that significant variation does exist in these communities even on relatively
338 local spatial (~3 km) and temporal (~3 months) scale. The differences in community composition
339 were correlated with both environmental conditions and host characteristics. The significant
340 environmental conditions included water temperature, depth, and salinity, all of which are known
341 to affect T. testudinum growth and, outside of the optimum range, act as stressors for the host
342 plant (McMillan 1978, Zieman and Zieman 1989, Tomasko and Dawes 1990, Irlandi et al, 2002).
343 Additionally, host characteristics such as blade length and width, growth rates, and percent cover
344 can be used as measures for seagrass health indicating that these communities may change with
345 more or less favorable conditions for the host. However, as this study is limited to correlations, it
346 is not known whether it is the environment or host that is the cause for these changes in
347 composition. For instance, low salinity has been shown to result in thinner T. testudinum blade
348 widths (Irlandi et al. 2002) and, since microbial community composition correlated significantly
349 with both factors, it is not known if these differences are a result of the environment or the host’s
350 condition. Manipulative studies are needed to investigate the separate influences of host plant
351 and environmental conditions on these microbial communities.
352 Additionally, compositional changes within the communities also resulted in differences
353 in richness and diversity between the sites with the fifth site having lower diversity and richness
354 than the second and third sites. This suggests that diversity and richness of the blade microbial
355 communities also change with less favorable conditions for the host. Although the fifth site has
356 more optimal salinities for T. testudinum growth, it is the shallowest site and tends to be clearer
357 than the sites closer to shore, which can result in high light stress (Schubert et al. 2015). These
358 stressful conditions could also be affecting the interactions between the host plant and surface
359 microbial communities, resulting in lower microbial diversity. Alternatively, the effect could be
360 in the opposite direction with the third and second site being more stressful for the host due to
361 lower salinities and decreased light availability. In that case, increased microbial diversity could
362 reflect greater functional diversity of the microbial community which may be beneficial under
363 less optimal conditions. Understanding changes in composition and diversity in these surface
364 microbial communities in a stress context will help elucidate the role of these microbial
365 communities in relation to seagrass health.
366 Although the blade surface communities vary across space and time, the composition of
367 these communities was significantly different from that of the water column in all cases. The
368 blade surface communities also had higher species diversity and richness than the water column
369 communities. This suggests that not only are there multiple source populations other than the
370 water column (likely the sediment community), but that the host plant may be a driver in
371 structuring microbial community structure on its blade surfaces. Whether this influence is limited
372 to offering a substrate for facilitating an attached lifestyle or if feedbacks between the host and
373 microbial community exist is yet to be determined. Ugarelli et al. (2019) also found that the
374 microbial communities on T. testudinum blades were distinct from the water column microbial
375 communities, however the number of ASVs found in the T. testudinum phyllosphere (3,347)
376 were much lower. The higher number of ASVs observed in this study (over 11,000) could be due
377 to differences in sampling methods as well the larger sample size. In the temperate seagrass
378 Zostera marina, Crump et al. (2018) also found that the blade associated microbial communities
379 were distinct from that of the water column but differ from Fahimipour et al. (2017), which did
380 not find significant differences between the two. In addition, Crump et al. (2018) found that on
381 average 13% of eelgrass phyllosphere communities were dominated by a single OTU, which
382 differs from this study that used ASVs as the taxonomic unit and found the highest relative
383 abundance to be 2% of the community. However, it is not surprising that temperate seagrasses
384 would have different leaf microbiomes than tropical seagrasses. Additionally, comparisons
385 across seagrass microbiome studies are not always appropriate due to differences in sampling
386 methods, sequencing platforms, and data pipelines used.
387 A Possible ‘Core’ Community
388 Fifteen ASVs were consistently found in all blade surface samples from both 2016 and
389 2017. These 15 microorganisms may comprise a ‘core’ community that is unique to and occurs
390 on all Thalassia testudinum blades. While no one ASV dominated blade surface samples, the
391 three most abundant ASVs from each of the 2016 and 2017 samplings were also present in 100%
392 of the blade surface samples and include in this core community. Their ubiquity among blade
393 surface samples and dominance of community abundance may indicate that these ASVs are
394 ecologically significant in determining community structure and identity or that they may
395 perform key community functions. Members of this core community may undergo processes that
396 are beneficial or deleterious to the host and could affect the maintenance of a healthy microbiota
397 on these leaf surfaces. Members of this core community may also be important key stone species
398 involved in biofilm production or in determining community structure (Shade and Handelsman
399 2012, Herren and McMahon 2018). Additional sampling needs to be conducted to confirm the
400 presence of this core community on greater spatial and temporal scales. However, the presence
401 of these 15 ASVs even after the substantial seagrass die-off that occurred with Hurricane Irma
402 suggests that the micro-organisms comprising the core community may have particular
403 importance within these blade surface communities. These microbes may be early colonizers of
404 seagrass blades after a disturbance or during new growth and persist through priority effects.
405 Alternatively, these microbes may result from species-level selection as the microbiome
406 develops. However, it may also be that the taxonomic identity of core community members is
407 less important than their functional identity as microbial species from various taxonomic groups
408 can have similar metabolisms and perform similar functions. Functional analysis of these
409 communities will not only help to understand the relationships they have with their seagrass host,
410 but also to examine if assembly of core community members is dictated by functional identity
411 rather than taxonomic identity (Burke et al. 2011). Functional analyses of these communities
412 using a combination of metagenomics and metatranscriptomics is needed in order to assess the
413 relative importance of these two factors in determining T. testudinum blade surface communities
414 as well as to identify functions encoded in the core community members.
415 Although the microbial communities on T. testudinum blades shared many ASVs
416 between 2016 and 2017, community composition between years did show marked differences,
417 including higher richness and diversity in 2017. These differences could be due to the large
418 disturbance event (i.e. Hurricane Irma) that occurred prior to the sampling and the 2017 blade
419 surface community may represent an early colonizing community. Interestingly, the four ASVs
420 that had the highest relative abundance (1-1.5%) in 2017 were all Alphaproteobateria belonging
421 to the family Rhodobacteraceae. Additionally, three of these four ASVs were present in 100% of
422 samples from both 2016 and 2017 (n=77) and the fourth was only missing from one sample.
423 These three Rhodobacteraceae ASVs are closely related to other known Rhodobacteraceae
424 members isolated from saline environments around the world, including the western Pacific,
425 North Sea, and Atlantic Ocean. One closely related cultured relative, Oceanica granulosus
426 (100% similarity, GenBank Accession AY242897.1), is known to accumulate poly-β-
427 hydroxybutyrate (PHB) which is produced in response to physiological stress due to nutrient
428 limitation (Cho and Giovannoni 2004). This may indicate that the blade surfaces are nutrient
429 limited environments similar to leaf surfaces in the terrestrial phyllosphere (Vorholt 2012).
430 Members of the Rhodobacteraceae family are also often found in the marine environment and
431 have been found to be dominant community members early on in biofilm formation within
432 marine environments (Jones et al. 2007, Dang et al. 2008, Elifantz et al. 2013). While always
433 present in the core community, these three Rhodobacteraceae may increase in community
434 relative abundance during times of biofilm formation, including post-disturbance or during early
435 successional stages, with Cyanobacteria becoming more abundant in later successional stages.
436 This study shows that T. testudinum blade surfaces host rich microbial communities that
437 are distinct from that of the water column. Further, these communities exhibit changes in
438 composition that correlate with characteristics of both their environment and host which is
439 similar to what is known about terrestrial phyllosphere communities. This correlative study
440 provides the foundation for experiments to more directly investigate the relationship between T.
441 testudinum health and the blade surface microbiome. These types of studies will also indicate
442 whether or not the microbial communities found on submerged, marine plants and their
443 relationships to those host plants are similar to the host-microbe dynamics in other systems, such
444 as the terrestrial phyllosphere or other marine vegetation, including algae. Additionally,
445 understanding this relationship between blade microbiomes and the host plant may be imperative
446 to understanding seagrass health as a whole and the importance of microbiomes in maintaining
447 healthy coastal ecosystems. These future studies may have important implications for seagrass
448 conservation and management as the loss of seagrass will inevitably lead to greater ecological
449 and economic losses (Waycott et al. 2009).
450
451 Acknowledgements
452 This study was conducted under Florida Fish and Wildlife Commission permit number SAL-16-
453 1799C-SR. This study was made possible through funding from the PADI Foundation (Grant
454 #21843), the American Museum of Natural History Lerner Gray Grants, and the Florida Sea
455 Grant Scholars Program.
456
457 References
458 Apprill A, McNally S, Parsons R, Weber L (2015) Minor revision to V4 region SSU rRNA 806R
459 gene primer greatly increases detection of SAR11 bacterioplankton. Aquat Micro Ecol
460 75: 129-137.
461 Barbier EB, Hacker SD, Kennedy C, Koch EW, Stier AC, Silliman BR (2011) The value of
462 estuarine and coastal ecosystem services. Ecological Monographs 81(2): 169-193.
463 Burke C, Steinberg P, Rusch D, Kjelleberg S, Thomas T (2011) Bacterial community assembly
464 based on functional genes rather than species. Proc Natl Acad Sci 108(34): 14288-14293.
465 Callahan BJ, McMurdie PJ, Rosen M.J, Han AW, Johnson AJA, Holmes SP (2016) DADA2:
466 High-resolution sample inference from Illumina amplicon data. Nat Methods 13(7): 581-
467 583.
468 Caporaso JG, Kuczynski J, Stombaugh J, Bittinger K, Bushman FD, Costello EK, Fierer N, Pẽna
469 AG, Goodrich JK, Gordon JI, Huttley GA, Kelley ST, Knights D, Koenig JE, Ley RE,
470 Lozupone CA, McDonald D, Muegge BD, Pirrung M, Reeder J, Sevinsky JR, Turnbaugh
471 PJ, Walters WA, Widmann, J, Yatsunenko T, Zaneveld J, Knight R (2010) QIIME allows
472 analysis of high-throughput community sequencing data. Nat Methods 7: 335-336.
473 Caporaso JG, Lauber CL, Walters WA, Berg-Lyons D, Lozupone CA, Turnbaugh PJ, Fierer N,
474 Knight R (2011) Global patterns of 16S rRNA diversity at a depth of millions of
475 sequences per sample. Proc Natl Acad Sci USA 108: 4516-4522.
476 Caporaso JG, Lauber CL, Walters WA, Berg-Lyons D, Huntley J, Fierer N, Owens SM, Betley J,
477 Fraser L, Bauer M, Gormley N, Gilbert JA, Smith G, Knight R (2012) Ultra-high-
478 throughput microbial community analysis on the Illumina HiSeq and MiSeq platforms.
479 ISME J 6: 1621-1624.
480 Cho JC and Giovannoni SJ (2004) Oceanicola granulosus gen. nov., sp. nov. and Oceanicola
481 batsensis sp. nov., poly-β-hydroxybutyrate-producing marine bacteria in the order
482 ‘Rhodobacterales’. Int J Syst Evol Microbiol 54: 1129-1136.
483 Costanza R, d’Arge R, de Groot R, Farber S, Grasso M, Hannon B, Limburg K, Naeem S,
484 O’Neill RV, Paurelo J, Raskin RG, Sutton P, van den Belt M (1997) The value of the
485 world’s ecosystem services and natural capital. Nature 387: 253-260.
486 Crump BC, Wojahn, JM, Tomas F, Mueller RS (2018) Metatranscriptomics and Amplicon
487 Sequencing Reveal Mutualisms in Seagrass Microbiomes. Front Microbiol 9: article 388.
488 Dang H, Li T, Chen M, Huang G (2008) Cross-ocean distribution of Rhodobacterales Bacteria
489 as primary surface colonizers in temperate coastal marine waters. Appl Environ
490 Microbiol 74(1): 52-60.
491 Duarte CM (2002) The future of seagrass meadows. Environ Conserv 29(2): 192-206.
492 Duarte CM, Losada IJ, Hendriks IE, Mazarrasa I, Marba N (2013) The role of coastal plant
493 communities for climate change mitigation and adaptation. Nat Clim Change 3: 961-968.
494 Elifantz H, Horn G, Ayon M, Cohen Y, Minz D (2013) Rhodobacteraceae are the key members
495 of the microbial community of the initial biofilm formed in Eastern Mediterranean
496 coastal seawater. FEMS Microbiol Ecol 85: 348-357.
497 Fahimipour AK, Kardish MR, Lang JM, Green JL, Eisen JA, Stachowicz JJ (2017) Global-Scale
498 Structure of the Eelgrass Microbiome. Appl Environ Microbiol 83(12): e03391-16.
499 Finkel OM, Burch AY, Elad T, Huse SM, Lindow SE, Post AF, Belkin S (2012) Distance-decay
500 relationships partially determine diversity patterns of phyllosphere bacteria on Tamrix
501 trees across the Sonoran Desert. Appl Environ Microbiol 78(17): 6187-6193.
502 Fourqurean JW, Duarte CM, Kennedy H, Marba N, Holmer M, Mateo MA, Apostolaki ET,
503 Kendrick GA, Krause-Jensen D, McGlathery KJ, Serrano O (2012) Seagrass ecosystems
504 as a globally significant carbon stock. Nat Geosci 5: 505-509.
505 Gilles LE, Thrash CJ, deRada S, Rabalais NN, Mason OU (2015) Archaeal enrichment in the
506 hypoxic zone in the northern Gulf of Mexico. Environ Microbiol 17(10): 3847-3856.
507 Hemminga MA, Duarte CM (2000) Seagrass Ecology. Cambridge University Press. Cambridge,
508 United Kingdom, 298 pp.
509 Herren CM, McMahon KD (2018) Keystone taxa predict compositional change in microbial
510 communities. Environ Microbiol 20(6): 2207-2217.
511 Irlandi E, Orlando B, Macia S, Biber P, Jones T, Kaufman L, Lirman D, Patterson ET (2002)
512 The influence of freshwater runoff on biomass, morphometrics, and production of
513 Thalassia testudinum. Aquat Bot 72: 67-78.
514 Jackson EF, Echlin HL, Jackson CR (2006) Changes in the phyllosphere community of the
515 resurrection fern, Polypodium polypodioides, associated with rainfall and wetting. FEMS
516 58: 236-246.
517 Jones PR, Cottrell MT, Kirchman DL, Dexter SC (2007) Bacterial community structure of
518 biofilms on artificial surfaces in an estuary. Microb Ecol 53: 153-162.
519 Laforest-Lapointe I, Messier C, Kembel SW (2016a) Host species identity, site, and time drive
520 temperate tree phyllosphere bacterial community structure. Microbiome 4(27): 1-10.
521 Laforest-Lapointe I, Messier C, Kembel SW (2016b) Tree phyllosphere bacterial communities:
522 exploring the magnitude of intra-and inter-individual variation among host species. PeerJ
523 4:e2367. DOI 10.7717/peerj.2367.
524 Laforest-Lapointe I, Messier C, Kembel SW (2017) Tree leaf bacterial community structure and
525 diversity differ along a gradient of urban intensity. mSystems 2(6): e00087-17.
526 Lamb JB, van de Water JAJM., Bourne DG, Altier C, Hein MY, Fiorenza EA, Abu N, Jompa J,
527 Harvell CD (2017) Seagrass ecosystems reduce exposure to bacterial pathogens of
528 humans, fishes, and invertebrates. Science 355: 731-733.
529 Lindow SE, Brandl. MT (2003) Microbiology of the Phyllosphere. Appl Environ Microbiol
530 69(4): 1875-1883.
531 Lindow SE, Leveau JHJ (2002) Phyllosphere microbiology. Curr Opin Microbiol 13: 238-243.
532 McMillan C (1978) Morphogeographic varaiation under controlled conditions in five seagrasses,
533 Thalassia testudinum, Halodule wrigthii, Syringodium filiforme, Halophia engelmanni,
534 and Zostera marina. Aquat Bot 4: 169-189.
535 Mejia AY, Rotinia A, Lacasella F, Bookman R, Thaller MC, Shem-Tov R, Winters G, Migliore
536 L (2016) Assessing the ecological status of seagrasses using morphology, biochemical
537 descriptors and microbial community analyses. A study in Halophila stipulacea (Forsk.)
538 Aschers meadows in the northern Red Sea. Ecol Indic 60: 1150-1163.
539 Mühlsteinova R, Johansen JR, Pietrasiak N, Martin MP, Osorio-Santos K, Warren SD (2014)
540 Polyphasic characterization of Trichocoleus desertorum sp. nov. (Pseudanabaenales,
541 Cyanobacteria) from desert soils and phylogenetic placement of the genus Trichocoleus.
542 Phytotaxa 163(5): 241-261.
543 Nordlund LM, Koch EW, Barbier EB, Creed JC (2016) Seagrass Ecosystem Services and their
544 Variability across Genera and Geographical Regions. PLoS ONE 11(10): e0163091.
545 Oksanen J, Blanchet FG, Friendly M, Kindt R, Legendre P, McGlinn D, Minchin PR, O’Hara
546 RB, Simpson Gl, Solymos P, Stevens MHH, Szoecs E, Wagner H (2018) vegan:
547 Community Ecology Package. R package version 2.5-2. https:/CRAN.R-
548 project.org/package=vegan.
549 Orth RJ, Carruthers TJB, Dennison WC, Duarte CM, Fourqurean JW, Heck KL, Hughes AR,
550 Kendrick GA, Kenworthy WJ, Olyarnik S, Short FT, Waycott M, Williams SL (2006) A
551 global crisis for seagrass ecosystems. BioSci 56(12): 987-996.
552 Parada AE, Needham DM, Fuhrman JA (2016) Every base matters: assessing small subunit
553 rRNA primers for marine microbiomes with mock communities, time series and global
554 field samples. Environ Microbiol 18(5): 1403-1414.
555 Paulson JN, Stine OC, Bravo HC, Pop M (2013) Differential abundance analysis for microbial
556 marker-gene surveys. Nat Methods 10: 1200-1202.
557 Phillips RC (1960) Observations on the Ecology and Distribution of the Florida Seagrasses.
558 Professional Papers Series: Number Two. Florida State Board of Conservation Marine
559 Laboratory. St. Petersburg, FL, 76 pp.
560 R Core Team (2018) R: A language and environment for statistical computing. R Foundation for
561 Statistical Computing, Vienna, Austria.
562 Redford AJ, Bowers RM, Knight R, Linhart Y, Fierer N (2010) The ecology of the phyllosphere:
563 geographic and phylogenetic variability in the distribution of bacteria on tree leaves.
564 Environ Microbiol 12(11): 2885-2893.
565 Rigonato J, Kent AD, Alvarenga DO, Andreote FD, Beirigo RM, Vidal-Torrado P, Fiore MF
566 (2013) Drivers of cyanobacterial diversity and community composition in mangrove soils
567 in south-east Brazi. Environ Microbiol 15(4): 1103-1114.
568 Saleem M, Meckes N, Pervaiz ZH, Traw MB (2017) Microbial interactions in the phyllosphere
569 increase plant performance under herbivore biotic stress. Front Microbiol 8(41). DOI:
570 10.3389/fmicb.2017.00041.
571 Schemske DW, Mittelbach GG, Cornell HV, Sobel JM, Roy K (2009) Is there a latitudinal
572 gradient in the importance of biotic interactions? Annu Rev Ecol Evol Syst 40: 245-269.
573 Schubert N, Colombo-Pallota MF, Enriquez S (2015) Leaf and canopy scale characterization of
574 the photoprotective response to high-light stress of the seagrass Thalassia testudinum.
575 Limnol and Oceanogr 60: 286-302.
576 Shade A, Handelsman J (2012) Beyond the Venn diagram: the hunt for a core microbiome.
577 Environ Microbiol 14(1): 4-12.
578 Short FT, Polidoro B, Livingstone SR, Carpenter KE, Bandeira S, Bujang JS, Calumpong HP,
579 Carruthers TJB, Coles RG, Dennison WC, Erftemeijer PLA, Fortes MD, Freeman AS,
580 Jagtap TG, Kamal AHM, Kendrick GA, Kenworthy WJ, La Nafie YA, Nasution IM,
581 Orth RJ, Prathep A, Saciangco JC, van Tussenbroek B, Vergara SG, Waycott M, Zieman
582 JC (2011) Extinction risk assessment of the world’s seagrass species. Biol Conserv 144:
583 1961-1971.
584 Silva CSP, Genuário DB, Vaz MGMV, Fiore FM (2014) Phylogeny of culturable cyanobacteria
585 from Brazilian mangroves. Syst Appl Microbiol 37(2): 100-112.
586 Tomasko DA, Dawes CJ (1990) Influences of season and water depth on the clonal biology of
587 the seagrass Thalassia testudinum. Mar Biol 105: 345-351.
588 Ugarelli K, Laas P, Stingl U (2019) The microbial communities of leaves and roots associated
589 with turtle grass (Thalassia testudinum) and manatee grass (Syringodium filiforme) are
590 distinct from seawater and sediment communities but are similar between species and
591 sampling sites. Microorganisms 7(4): doi:10.3390/microorganisms7010004.
592 Vandenkoornhuyse P, Quaiser A, Duhamel M, Le Van A, Dufresne A (2015) The importance of
593 the microbiome of the plant holobiont. Tansley Review. New Phytol 206: 1196-1206.
594 Vorholt JA (2012) Microbial life in the phyllosphere. Nat Rev Micrbiol 10: 828-840.
595 Waycott M, Duarte CM, Carruthers TJB, Orth RJ, Dennison WC, Olyarnik, S, Calladine A,
596 Fourqurean JW, Heck Jr KL, Hughes AR, Kendrick GA, Kenworthy WJ, Short FT,
597 Williams SL (2009) Accelerating loss of seagrass across the globe threatens coastal
598 ecosystems. Proc Natl Acad Sci USA 106: 12377-12381.
599 Yilmaz P, Parfrey LW, Yarza P, Gerken J, Pruesse E, Quast C, Schweer T, Peplies J, Ludwig W,
600 Glӧckner FO (2014) The SILVA and “All-species Living Tree Project (LTP)” taxonomic
601 frameworks. Nucleic Acids Res 42: D643-D648.
602 Zieman JC (1974) Methods for the Study of Growth and Production of Turtle Grass, Thalassia
603 testudinum Kӧnig. Aquaculture 4: 139-143.
604 Zieman JC, Zieman RT (1989) The ecology of the seagrass meadows of the west coast of
605 Florida: a community profile. Biological Report 85(7.25). U.S. Fish and Wildlife Service.
606 Washington, D.C., 155 pp.
607
608
609
610
611
612
613 Table 1. Site Characterization. Mean water column and host parameters (± S.E.) measured at
614 each site during seagrass surveys and microbial samplings during 2016.
Water Column
Parameters
Depth Temp Salinity (‰) DO (mg/l) TSS PO4 NO3 NO2 NH4
(m) . (°C) (mg/l) (µg/l) (µg/l) (µg/l) (µg/l)
ABT- 0.98 ± 30.7 ± 21.0 ± 1.8 4.33 ± 1.07 15.88 ± 1.9 ± 4.1 ± 1.7 ± 56.8 ±
1 0.08 0.8 4.98 0.5 2.6 0.6 28.0
ABT- 1.6 ± 30.6 ± 22.7 ± 1.8 4.98 ± 0.57 9.77 ± 2.0 ± 0.1 ± 1.7 ± 34.7 ±
2 0.20 0.4 1.60 0.1 0.1 0.8 6.2
ABT- 1.85 ± 30.8 ± 24.9 ± 2.2 6.01 ± 0.04 9.62 ± 1.9 ± 1.3 ± 2.2 ± 121.9
3 0.15 0.1 1.65 0.5 0.6 1.0 ± 50.0
ABT- 2.18 ± 30.8 ± 26.0 ± 2.0 5.60 ± 0.42 11.53 ± 1.1 ± 0.8 ± 0.6 ± 8.8 ±
4 0.13 0.2 3.07 0.1 0.5 0.1 0.7
ABT- 1.2 ± 0.0 30.8 ± 27.3 ± 2.0 5.55 ± 0.43 19.76 ± 1.3 ± 0.1 ± 0.9 ± 13.0 ±
5 0.4 2.47 0.3 0.1 0.6 7.1
Seagrass Host
Parameters
Length Width GR GR Cover TP TN TC δ15N δ13C
(cm) (cm) (mg/shoot/day (cm/shoot/day (%) (%) (%) (%) (‰) (‰)
) )
ABT- 29.9 ± 0.4 ± 1.57 ± 0.27 0.9 ± 0.2 46 ± 12 0.146 2.24 ± 33.93 1.26 ± -16.12
1 0.8 0.0 ± 0.01 0.08 ± 0.28 0.32 ± 0.73
ABT- 31.3 ± 0.7 ± 4.31 ± 0.26 1.3 ± 0.1 66 ± 5 0.133 2.11 ± 33.66 2.22 ± -12.79
2 0.8 0.1 ± 0.00 0.01 ± 0.14 0.13 ± 0.75
ABT- 43.6 ± 0.7 ± 5.31 ± 0.57 1.8 ± 0.1 73 ± 10 0.136 2.14 ± 33.52 1.63 ± -12.76
3 1.2 0.0 ± 0.01 0.05 ± 0.27 0.36 ± 0.77
ABT- 40.0 ± 0.7 ± 5.38 ± 0.39 1.6 ± 0.1 55 ± 11 0.130 2.02 ± 33.52 3.00 ± -11.93
4 1.3 0.0 ± 0.00 0.11 ± 0.28 0.12 ± 0.82
ABT- 21.5 ± 0.7 ± 2.05 ± 0.15 0.7 ± 0.0 86 ± 12 0.138 2.11 ± 36.55 3.02 ± -12.50
5 0.4 0.0 ± 0.01 0.14 ± 2.96 0.04 ± 0.79
615
616 Table 2. Environmental fitting results. Environmental fitting (envfit, vegan package) on the
617 NMDS ordination was used to determine significant environmental and host characteristics
618 (p<0.05, indicated by asterisk). Bonferroni correction was applied to p-values to account for
619 multiple comparisons.
NMDS1 NMDS2 R2 p-value corrected p-value
Host Characteristics
Avg. Length (cm) -0.37028 -0.92892 0.3974 0.001 0.018 *
Avg. Width (cm) 0.24096 -0.97054 0.4877 0.001 0.018 *
Growth Rate (g/shoot/day) -0.21823 -0.9759 0.4596 0.001 0.018 *
Avg. Cover (%) 0.96926 0.24605 0.2523 0.002 0.036 *
Total N (%) -0.99989 -0.01481 0.0636 0.200 1.000
Total C (%) 0.78755 0.61625 0.1329 0.025 0.450
Total P (%) -0.89918 0.43758 0.0789 0.136 1.000
δ15N (‰) 0.99711 0.07593 0.3810 0.001 0.018 *
δ13C (‰) 0.79302 -0.60919 0.5883 0.001 0.018 *
Abiotic Characteristics
Water Temp. (°C) 0.81146 -0.58440 0.5076 0.001 0.018 *
Avg. Depth (m) 0.00464 -0.99999 0.4144 0.001 0.018 *
Dissolved Oxygen (mg/l) 0.74554 -0.66646 0.2075 0.003 0.054
Salinity (‰) 0.96511 0.26185 0.2539 0.001 0.018 *
Total Suspended Solids (mg/l) 0.56713 0.82363 0.3463 0.001 0.018 *
Phosphate (µg/l) -0.90468 0.42609 0.2643 0.002 0.036 *
Nitrate (µg/l) -0.44827 -0.89390 0.1733 0.012 0.216
Nitrite (µg/l) -0.71556 -0.69855 0.1008 0.059 1.000
Ammonium (µg/l) -0.92074 -0.39018 0.0965 0.079 1.000
620
621 Figure legends
622 Fig. 1 Non-metric Multidimensional Scaling (NMDS) ordination of 16S rRNA iTag sequence
623 data (stress=0.10, k=3) from 2016. Ordination shows T. testudinum blade surface communities as
624 single points with vectors showing significant environmental (water temperature, average depth,
625 salinity, total suspended solids content, and phosphate concentration) and host (average blade
626 length and width, average growth rates, average Thalassia testudinum cover, blade δ15N and δ13C
627 content) fitted factors (envfit, corrected p<0.05). Symbol color represents sites (ABT-1: yellow,
628 ABT-2: red, ABT-3: purple, ABT-4: blue, ABT-5: green) and symbol shape represents sampling
629 event (Jul: square, Aug: circle, Sep: triangle)
630
631 Fig. 2 Non-metric Multidimensional Scaling (NMDS) ordination of 16S rRNA iTag sequence
632 data (stress=0.06, k=2) from 2016. Ordination shows water column communities as single points
633 with symbol color represents sites (ABT-1: yellow, ABT-2: red, ABT-3: purple, ABT-4: blue,
634 ABT-5: green) and symbol shape represents sampling event (Jul: square, Aug: circle, Sep:
635 triangle). Vectors show environmental fitted factors (salinity, temperature, dissolved oxygen
636 content, and phosphate concentration) that were found to be significant (envfit, corrected p<0.05)
637
638 Fig. 3 Non-metric Multidimensional Scaling (NMDS) ordination of 16S rRNA iTag sequence
639 data (stress=0.10, k=2) for all blade surface samples from both 2016 and 2017. Ordination shows
640 blade surface communities as single points with symbol colors representing site (ABT-1: yellow,
641 ABT-2: red, ABT-3: purple, ABT-4: blue, ABT-5: green) and symbol shapes representing
642 sampling event (Jul-2016: square, Aug-2016: circle, Sep-2016: triangle, Sep-2017: cross)
643 Figure 1
644
645
646
647
648
649
650 Figure 2
651
652
653
654
655
656
657 Figure 3
658
659