Nectar Microbial Diversity and Changes Associated with Environmental Exposure.

25 Background: Plants are critical to global environmental health and food production strategies; 26 most plants utilise flowers as part of their reproduction cycle. Flowers attract pollinators using 27 a range of complex strategies and floral nectar is an essential component of this attraction 28 profile. Nectar is a nutrient rich liquid, containing a range of sugars, organic acids, amino acids, 29 lipids and vitamins, found to be a suitable habitat for a wide range of fungi, but so far, limited 30 bacterial diversity has been detected. Several antimicrobial properties and adverse 31 environmental conditions, such as high osmotic pressure present in the nectar were thought to 32 reduce bacterial numbers. 33 Results: This study reports the next generation sequencing analysis of the bacterial and fungal 34 diversity in flower nectar. This was achieved in four floral species native to the United 35 Kingdom 37 All flower species examined had a diverse bacterial and fungal populations present with a core 38 microbiome detected, dominated by Proteobacteria and Firmicutes phyla, while Basidomycota 39 were the most persistent fungal phyla in all of the floral nectar types sampled. However, many 40 unique bacterial and fungal species were detected at lower abundances. Furthermore, in N. 41 pseudonarcissus and D. purpurea floral nectar , the microbial diversity detected in the nectar 42 between flowers exposed to the environment versus non-environment exposed flowers, was 43 different. Conclusions: These results suggest that floral nectars in different plant species do contain a 45 distinct microbiome and the individual flower microbial community diversity may be affected 46 by floral nectar composition, insect visitation and other environmental factors.

Step 2 x 30, 72°C 10:00, 4°C ∞) in order to amplify the 16S rRNA gene from 155 species isolated from nectar. The amplified PCR products were sequenced using Sanger 156 sequencing (Source Bioscience, Nottingham, UK). The resulting sequences were trimmed, 157 edited, checked for quality and identified using BLAST [46]. 158

Analysis of Microbial Community Diversity by Illumina Sequencing. 159
To profile the microbial community, primers 799F (AACMGGATTAGATACCCKG, [ [49,50]) were used to 163 amplify the fungal ITS1 region. Sequencing was performed on the Illumina MiSeq V3 platform 164 (LGC Genomics GmbH, Berlin, Germany). Barcode sequences adapters and primer dimer 165 products were removed from the resulting sequence fragments using Illumina bcl2fastq 1.8.4 Sequences of 16S rRNA genes were pre-processing by picking operational taxonomic units 170 (OTUs) using Mothur 1.35. 1 [51]. Ambiguous 16S rRNA gene sequences were removed, short 171 alignments filtered out and the remaining sequences aligned using the 16S rRNA gene Mothur-172 Silva SEED r119 reference. Chimeras were eliminated using the UCHIME algorithm [ Chao1 metrics [57]. 189 types collected (Figure 1). Lamium album (30.5%), environment exposed N. pseudonarcissus 194 (26%), H. non-scripta (22.5%) and environment exposed D. purpurea (27.7%), all had similar 195 values ( Figure 1). However, the non-environment exposed nectar in N. pseudonarcissus and 196 D. purpurea had considerably lower values, 2.5% and 4% consecutively ( Figure 1). When 197 Kruskal-Wallis analysis was applied to the four different environment exposed floral species 198 (L. album, N. pseudonarcissus, H. non-scripta, D. purpurea) significant differences were found 199 (P= 0.021) between % sucrose content. Further, significant differences were observed between 200 both environmental conditions (non-environment exposed and environment exposed) of 201  Table 1). The nectar from L. album had the highest observed detectable species richness and N. pseudonarcissus (Non-environment exposed) 1 1 42 98 H. non-scripta (Environment exposed)  Acetobacteraceae (87%), compared to environment exposed flowers which were dominated 233 by Enterobacteriaceae (42%), Acetobacteraceae (24%), Clostridiales family Xl; genus 234 Clostridium (18%) and Enterobacteriaceae; genus Erwinia (14%). Environmentally exposed 235 N. pseudonarcissus had higher detectable species richness than the nectar collected in 236 laboratory conditions (Figure 2A). The environment exposed nectar included 237 Enterobacteraceae; genus Erwina (82%) and Pseudomonadaceae (18%), however with nectar 238 collected in the laboratory 99% of sequences were similar to Enterobacteriaceae. H. non-239 scripta had the lowest detectable species richness ( Figure 2A), with 98% of detectable bacterial 240 diversity from sequences most similar to Enterobacteriaceae (98%); genus Erwinia. 241 Comparing environment exposed and non-environment exposed nectar in N. psuedonarcissus 242 and D. purpurea, a core microbiome was detected ( Figure 3) with seventeen orders of bacteria 243 consistently present in both species of flower, in environment exposed and non-exposed achieved for H. non-scripta and non-environment exposed N. pseudonarcissus but further 254 sequencing would reveal more diversity in the other samples. The Chao1 estimator was applied 255 to OTU distributions (Table 1) to estimate alpha diversity. 256 Lamium album displayed the highest detectable fungal species diversity and the environment 257 exposed N. pseudonarcissus had the lowest (Figure 2 SI). The greatest difference between Sobs 258 and SChao1 was detected in the environment exposed N. pseudonarcissus, indicating that more 259 unique OTUs could be detected with further sequencing analysis. Fungal sequences with high 260 sequence similarity to Basidiomycota dominated all four plant species. Within this phyla 261 Tremellomycetes were observed to be the dominant class, except environment exposed D. 262 purpurea which was dominated by sequences similar to Agaricales. 263 Analysis of fungal genera indicated that environment exposed D. purpurea had less detectable 264 diversity than non-environment exposed flowers (Figure 2 SI). The only genera present in both 265 environments, exposed and non-environment exposed, were sequences similar to 266 Cystofilobasidium, but these sequences were more numerically dominant in non-environment 267 exposed flowers (Figure 2 SI). Other than Cystofilobasidium non-environment exposed D.  Figure 4A). When comparing environment exposed 271 and non-environment exposed N. pseudonarcissus, the environment exposed flowers had 272 higher detectable species richness ( Figure 4A), being dominated by sequences similar to 273 Cryptococcus (49%), Geastrum (16%) and Hyphoderma (11%), whilst non-environment 274 exposed flowers were solely dominated by sequences similar to Guehomyces (100%), and was 275 the only genera to be found in both environment exposed and non-environment exposed  This study provides a profile of the nectar microbiome in four common plants, native to the 294 UK. Profiling of the 16S rRNA gene and ITS gene using NGS has allowed the detection of 295 complex bacterial and fungal communities in each floral nectar type. These populations 296 differed in composition across species and environmental conditions. Sequencing and 297 subsequent Chao1 indices provided evidence for higher bacterial community richness 298 compared to fungi, in all the flower species and conditions observed ( Table 1). 299 In examining bacteria detected, except for L. album which appears to have a very unique 300 composition, Proteobacteria were detected in the highest relative numbers in all floral species 301 ( Figure 2) (93% of the diversity in environment exposed D. purpurea, 72% in non-environment 302 exposed, 100% in environment exposed N. pseudonarcissus, 99% in non-environment exposed 303 N. pseudonarcissus and 100% in environment exposed H. non-scripta). However, in L. album, 304 sequences similar to Firmicutes were the dominant phylum (65%), with the highest detectable 305 species richness with dominant sequences being most similar to Clostridium (49%), Proteus 306 (10%) and Hydrotalea (6%). Hyacinthoides non scripta had the floral nectar with the lowest 307 detectable species diversity with the majority of OTUs being similar to Erwinia (33%). The 308 consistency of detection in nectar types indicates that members of the Proteobacteria are 309 common members of these microbiomes. Nectar communities have previously been found to 310 be dominated by Proteobacteria [58], also comprising the main phyla in three pollinator 311 exposed floral species (Borago officinalis, Centaurea cyanus and Symphytum officinale) but 312 not being detected in non-pollinator exposed 'bagged' flowers of the same species [59]. 313 There was a detected effect of environmental exposure, on flower nectar composition. 314 (18%). Both Erwinia and Pseudomonas have been documented in nectar environments [60, 61] 320 with both being the only two genera currently reported to be directly associated with pollinating 321 species acting as bacterial vectors [62]. Similarly, the environment exposed D. purpurea was 322 found to have a higher detectable OTU richness in comparison to non-environment exposed D. 323 purpurea (Figure 2). Environment exposed D. purpurea had sequences detected similar to 324 Erwinia (14%), Clostridium sensu stricto (7%) and Pseudomonas (3%), and these genera were 325 also observed in other floral species (H. non-scripta, environment exposed N. pseudonarcissus 326 and L. album). Environmental exposure does appear to alter the bacterial community detected, 327 however, overall, the same phyla (Proteobacteria and Firmicutes) remain dominant. This 328 dominance observed across differing floral species [37,14] could suggest the restrictive 329 environmental conditions found in nectar, e.g. high osmotic pressure [30] and presence of 330 hydrogen peroxide, [37] lead to the selection of specifically adapted nectar colonisers. 331 Although there was a clear variation in bacterial communities across plant species, 332 environmental conditions appear to affect community composition, as demonstrated by 333 differences between environment exposed and non-environment exposed samples of the same 334 plant species. However, from Figure 1, it can be observed that nectar collected under 335 environmental conditions had much higher % sucrose detected than the samples collected in 336 laboratory conditions. It could be possible that flowers of these species may take longer to 337 produce nectar with a higher sucrose content than the time that was allowed before collection. 338 Thus, the observed changes in microbial community composition between environment exposed and non-environment exposed species could be due to these reductions in sucrose 340 content of nectar and further analysis needs to be performed in order to investigate this. 341

Detected fungal diversity included sequences similar to Tremellomycetes (of the phylum 342
Basidiomycota) which were found to be dominant in all species, except environment exposed 343 D. purpurea which was dominated by Agaricales (Basidiomycota). Sequences similar to 344 Tremellomycetes accounted for the majority of diversity in non-environment exposed and 345 environment exposed N. pseudonarcissus (100% and 52% respectively), as well as 100% 346 detected in H. non-scripta, 83% in L. album, and 42% in non-environment exposed D. richness and H. non-scripta the lowest. Environment exposed N. pseudonarcissus was found 354 to have a higher detectable fungal species richness in comparison with non-environment 355 exposed N. pseudonarcissus. The nectar of the environment exposed flower was dominated by 356 sequences similar to Cryptococcus (50%), Geartrum (16%) and Hyphoderma (11%), whilst 357 non-environment exposed flowers were almost completely dominated by Geotomyces (99.9%). 358 This was the sole genera present in both variations of N. pseudonarcissus, but present in much 359 higher numbers in non-environment exposed N. pseudonarcissus. These findings, along with 360 the bacterial community analysis, further support the idea that species richness identified in the 361 nectar changed not across the plant species studied, but also by the environmental conditions 362 to which the flowers are exposed. in several previous studies [37,14]. However, this is the first study to examine bacterial and 383 fungal communities in floral nectar exposed to environmental factors and compared to nectar 384 from flowers that opened in controlled conditions in the laboratory. From the considerable 385 differences detected in the fungal and bacterial diversities, environmental conditions (most 386 likely pollinator visitation) may play a considerable role in shaping the microbial community 387 present [66,67]. Understanding the microbial diversity present in nectar is essential to 388 assessing their effects on the nutritional value of the nectar, how they strengthen or weaken the 389 plant-pollinator ecological mutualism, and therefore their effects on the efficacy of pollination 390 [8,68]. Further work should seek to fully assess the diversity and species richness of microbial 391 communities within the floral nectar of other plant species as well as the role of bacteria within 392 the yeast-plant-pollinator system and the effects of the environment of these plants upon  Total % mass sucrose for each floral species measured using the Brix scale via refractometry for each floral species. Hatched bars indicate floral nectar not exposed to environmental conditions. Error bars show standard deviation of means (n=3).

Figure 2: Relative abundance and OTU distribution of bacterial families
A) Relative abundance of bacterial families (>1%) detected by sequencing the 16S rRNA gene from different floral nectar in environment exposed and non-exposed conditions. Sequences were assigned to OTUs with over 97% sequence identity. Sequences not similar to any family and sequences <1% at family were assigned to unclassified/Other.

Figure 4: Relative abundance and OTU distribution of fungal genera
A) Relative abundance of fungal genera (>1%) detected by sequencing the ITS1 region from different floral nectar in environment exposed and non-exposed conditions. Sequences were assigned to OTUs with over 97% sequence identity. Sequences not similar to any family and sequences <1% at family were assigned to Unclassified/Other. B) OTU network showing distribution of all OTUs identified to class detected via sequencing the ITS2 rDNA region from samples of floral nectar. Node size indicates the dominance of reads assigned to an OTU and node colour indicates consensus taxonomy.

Figure 3: Bacterial orders present in nectar of N. pseudonarcissus and D. purpurea
A comparison of the bacterial orders detected in the nectar of daffodils and foxgloves.
OTUs were found consistently within N. pseudonarcissus (NP) and D. purpurea (DP) nectar in both non-environment exposed nectar (Non-EE) and environmentally exposed (EE) nectar. In the detection of a core microbiome seventeen bacterial classes were detected in all samples