Effects of Neonicotinoid Seed Treatments on Phyllosphere and Soil Bacterial 3 Communities Over Time

The phyllosphere and soil are dynamic habitats for microbial communities. Non-pathogenic 23 microbiota, including leaf and soil beneficial bacteria, plays a crucial role in plant growth and 24 health, as well as in soil fertility and organic matter production. In sustainable agriculture, it is 25 important to understand the composition of these bacterial communities, their changes in response 26 to disturbances, and their resilience to agricultural practices. Widespread pesticide application may 27 have had non-target impacts on these beneficial microorganisms. Neonicotinoids are a family of 28 systemic insecticides being vastly used to control soil and foliar pests in recent decades. A few 29 studies have demonstrated the long-term and non-target effects of neonicotinoids on 30 agroecosystem microbiota, but the generality of these findings remains unclear. In this study, we 31 used 16S rRNA gene amplicon sequencing to characterize the effects of neonicotinoid seed 32 treatment on soil and phyllosphere bacterial community diversity, composition and temporal 33 dynamics in a three-year soybean/corn rotation in Quebec, Canada. cycle, in response to the pesticide application. Our results indicate that neonicotinoids have non-target effects on phyllosphere and soil 46 bacterial communities in a soybean-corn agroecosystem, especially potentially beneficial bacteria 47 that are vital for plant growth and improve soil fertility. Exploring the interactions among bacteria 48 and other organisms, as well as the bacterial functional responses to the pesticide treatment, may 49 enhance our understanding of these non-target effects and help us adapt agricultural practices to 50 control these impacts. 51


21
Background 22 The phyllosphere and soil are dynamic habitats for microbial communities. Non-pathogenic 23 microbiota, including leaf and soil beneficial bacteria, plays a crucial role in plant growth and 24 health, as well as in soil fertility and organic matter production. In sustainable agriculture, it is 25 important to understand the composition of these bacterial communities, their changes in response 26 to disturbances, and their resilience to agricultural practices. Widespread pesticide application may 27 have had non-target impacts on these beneficial microorganisms. Neonicotinoids are a family of 28 systemic insecticides being vastly used to control soil and foliar pests in recent decades. A few 29 studies have demonstrated the long-term and non-target effects of neonicotinoids on 30 agroecosystem microbiota, but the generality of these findings remains unclear. In this study, we 31 used 16S rRNA gene amplicon sequencing to characterize the effects of neonicotinoid seed 32 treatment on soil and phyllosphere bacterial community diversity, composition and temporal 33 dynamics in a three-year soybean/corn rotation in Quebec, Canada. 34 nicotine, they interrupt neural transmission in the nervous system by binding to the nicotinic 90 acetylcholine receptors (nAChRs). Because of the fundamental distinctions between the nAChRs 91 of invertebrates and vertebrates, neonicotinoids are selectively more toxic to invertebrates, like 92 insects [25,26]. In North America, neonicotinoids have mostly been used as seed treatments to 93 7 phyllosphere (Shannon index mean ± SE 7.0 ± 0.02 versus 4.2 ± 0.10, Wilcoxon adjusted P < 136 0.0001). The relative abundance of several bacterial families was strongly associated with soil 137 (such as Gemmatimonadaceae and Solibacteraceae), soybean phyllosphere (such as 138 Beijerinckiaceae and Rhizobiaceae) or corn phyllosphere (such as Sphingomonadaceae and 139 Hymenobacteraceae) (P < 0.001, envfit analysis of correlation between PCoA axes and variables, 140  Table 1). Time was a much greater 145 driver of community composition variation in the phyllosphere than in soil (15.7% versus 4.6%, 146 PERMANOVA P < 0.001, Table 1). Alpha diversity varied in time in the phyllosphere but not in 147 soil (Table 2). This effect in the phyllosphere was especially obvious between the first and the last 148 year of the rotation where diversity was highest in the last year (Shannon index mean ± SE 149 respectively, 4.0 ± 0.17 versus 4.8 ± 0.20, Wilcoxon adjusted P < 0.0001, Table 2) but we also 150 observed intra-annual variation in diversity (Table 2). 151

Effects of neonicotinoid seed treatment on bacterial communities 152
Neonicotinoid seed treatment showed complex effects on the composition of bacterial 153 a much stronger effect of the neonicotinoid seed treatment on the composition of the phyllosphere 160 communities in corn (5.3%) than in soybean (1.6%) (PERMANOVA P < 0.001 and P < 0.05 161 respectively, Table 3, Fig. 3). There was no significant difference in phyllosphere alpha diversity 162 between neonicotinoid treatments overall, but soil bacterial alpha diversity was significantly 163 higher in control versus neonicotinoid-treated samples (Shannon index mean ± SE 7.2 ± 0.02 164 versus 7.0 ± 0.03, Wilcoxon adjusted P < 0.001, Table 2). 165 The overall effect of neonicotinoid seed treatment on the temporal variation of bacterial 166 community composition and alpha diversity was weak. In the phyllosphere, although there was a 167 small significant effect of the interaction between neonicotinoid application and time (month and 168 year) on variation in community composition (1.4%, PERMANOVA P < 0.05, Table 3), the 169 impacts on inter-annual variation and specific interactions with individual host species were not 170 significant. The interaction of neonicotinoid seed treatment and time was slightly stronger in soil, 171 especially with month (2.4%, PERMANOVA P < 0.01, Table 3). Uncovering these effects by 172 studying each crop separately revealed that this month-to-month temporal variation in bacterial 173 community structure within a growing season was particularly important in corn (5.6%, 174 PERMANOVA P < 0.05, Table 3). Similarly, while the interaction between neonicotinoid seed 175 treatment with time had no significant effect on bacterial alpha diversity in the phyllosphere, soil 176 alpha diversity was significantly reduced in the neonicotinoid-treated samples in July and August 177 (interaction between neonicotinoid seed treatment and month: linear regression analysis of 9 phyllosphere and soil bacterial ASVs. Overall, we detected 34 bacterial ASVs in the phyllosphere 183 and 294 in soil that were significantly differentially abundant between the control and 184 neonicotinoid-treated samples. In the phyllosphere, 22 ASVs (mainly Bacteroidetes) were more 185 abundant, and 12 (mainly Proteobacteria) were less abundant in response to neonicotinoid seed 186 treatment (Table 4). The genera Hymenobacter (13 ASVs) and Pseudomonas (4 ASVs) were 187 particularly favored by neonicotinoid treatment, while the genera Arsenophonus (4 ASVs) and 188 Skermanella (3 ASVs) among others decreased in abundance in neonicotinoid-treated samples 189 (DESeq2, Fig. 4A overall community composition and diversity. In particular, we have shown a role for temporal 207 variation, alone and in interaction with habitat and host species, as an important driver of bacterial 208 community composition variation, especially in the phyllosphere. While succession of microbial 209 communities in the phyllosphere has been documented previously [7,9,56], here we have shown 210 that even in a rotation of annual crops, the patterns of bacterial succession within and among years 211 are an important driver of community structure. 212 We have shown that neonicotinoid seed treatments have a non-target impact on bacterial 213 community structure and diversity in a soybean/corn agroecosystem, in particular on the 214 taxonomic composition of soil bacterial communities over the growing season. Phyllosphere and 215 soil bacteria exhibit different patterns of community composition, alpha diversity and temporal 216 variation throughout the growing season and in response to neonicotinoid application. In the 217 phyllosphere, host plant species and time are stronger drivers of bacterial community variation 218 than neonicotinoid seed treatment; however, neonicotinoids interact with these parameters to 219 influence the phyllosphere bacterial community composition. Overall, soil bacteria exhibited 220 stronger changes in community composition and a significant decline in bacterial alpha diversity 221 in response to neonicotinoid treatment, while phyllosphere bacteria responses to neonicotinoids 222 were weaker. Our results complement previous lab-based studies of neonicotinoid effects on 223 bacterial communities [47, 57, 58], providing some of the first field-based evidence that 224 neonicotinoids impact bacterial diversity in agroecosystems. 225 Overall, soil bacterial communities were more affected by neonicotinoid pesticide 226 treatment than phyllosphere bacterial communities. Neonicotinoid effects on soil bacterial 227 community composition and diversity varied greatly in time, with the impacts of neonicotinoid 228 application on the soil bacterial community composition and alpha diversity most pronounced in 229 the middle of the growing season. We suggest that this could be explained by the fact that 230 neonicotinoids' active period is much shorter in plants [34,35] than in soils, where they potentially 231 persist for months or years [30,37]. Despite the reported accumulation potential of neonicotinoids 232 in soils over time [59], we did not observe any significant inter-annual difference in bacterial 233 diversity among years in interaction with the pesticide treatment, perhaps due to degradation or 234 leaching of the neonicotinoids [60, 61]. 235 We also observed that the more homogenous the bacterial community composition is, the 236 more it is altered by the neonicotinoid application (soil more than phyllosphere and corn 237 phyllosphere more than soybean phyllosphere). We need further studies to determine if the 238 homogeneity of the bacterial communities resulted in less resilience in response to perturbations 239 or if less variability within groups allowed us to notice more changes in the communities. 240 In addition to community-wide responses of bacteria to the neonicotinoid treatment, 241 numerous bacterial taxa increased or decreased in relative abundance in response to 242 neonicotinoids. Bacterial taxa that were favored by the pesticide treatment include several genera 243 that are known to be potentially involved in neonicotinoid degradation (e.g. with an experimental design that represents real farming conditions in a crop rotation. Despite the 264 fact that neonicotinoids target invertebrates, we observed non-target impacts of neonicotinoids on 265 bacterial communities of the phyllosphere and soil, especially the beneficial bacteria that are 266 crucial for plant growth and health and soil fertility and quality. Future studies to identify the 267 genomic and physiological features associated with tolerance of neonicotinoids will be required to 268 understand the mechanistic reasons for these associations. Investigating the biological interactions 269 among bacteria and other micro-and macro-organisms that are affected by pesticides will help us 270 to better understand the non-target effects of pesticides on microbial diversity and how to control 271 them with better agricultural practices.

Study Site 275
We cultivated a three-year rotation of soybean (2016 and 2018) and corn (2017)  glyphosate was applied twice during each growing season (before seeding and one month after it) 287 to control weeds. The corn field was also fertilized with 400 kg/ha NPK (15-15-15) before seeding 288 and 222 kg/ha N (27.5%) one month after seeding. Soil physicochemical properties (e.g. pH, etc.) 289 were constant across the experimental field and did not differ between the growing seasons. 290

Sample Collection 291
To study the phyllosphere bacteria (the bacteria collected from the leaf surface in our case), 292 each year we collected 48 samples (two samples per plot at three sampling times during the 293 growing seasons), for a total of 144 samples. The three annual sampling occasions happened in 294 July, August and September. We sampled 50-100 g of healthy mature middle leaves of 6-10 close 295 plants from the two middle rows of each plot. We then stored each sample in a sterile plastic bag 296 and transferred it to the laboratory in a cooler, surrounded by ice packs. We immediately collected 297 the bacterial cells from the leaves by washing them in a 0.85% saline solution and agitating the 298 solution using a stomacher at 250 rpm for 30 sec. We then transferred the solutions to 50-ml tubes, 299 centrifuged them at 4,000 g for 20 minutes and discarded the supernatants. We kept the remaining 300 pellets at -80 °C until use. 301 To study the soil bacteria, we sampled bulk soil (soil that does not adhere to plant roots) 302 from the upper 12-15 cm layer of soil with a corer (2 cm in diameter) from the soil around the 303 same plants that we sampled for the phyllosphere. For each soil sample, we collected soil from six 304 different spots, in a zigzag pattern and at 10 cm from the plants, and then mixed and pooled them 305 into one 400-500 g sample [8,67]. We transferred samples to the laboratory in a cooler and stored 306 at -80 °C until use. Each year, we collected 48 soil samples (two samples per plot at the same three 307 sampling times as phyllosphere), for a total of 144 samples. 308

DNA extraction 309
We extracted DNA from the samples of phyllosphere (pellets containing bacterial cells) 310 and soil (directly) using MoBio PowerSoil DNA isolation kit (QIAGEN). Considering the high 311 amount of material to be extracted from each soil sample, we extracted DNA twice, each time from 312 0.5 g of the same sample, and pooled the extractions together in order to better capture soil bacterial 313 community variation. The rest of the extraction was performed according to the manufacturer's 314 instructions. Then, we measured the concentration and quality of the extracted DNA using Qubit 315 (Thermo Fisher Scientific) and Nanodrop (Thermo Fisher Scientific) prior to storing them at -80 316

°C. 317
Bacterial DNA amplification 318 Following previously described protocols [6, 10, 68], we amplified the V5-V6 we selected cutoffs to rarefy samples based on inspection of rarefaction curves for phyllosphere 379 and soil samples, choosing rarefaction cutoffs that approached saturation in the ASV rarefaction 380 curve while keeping as many samples as possible. We first rarefied the soybean and corn 381 phyllosphere and soil samples to 5,000 reads per sample, which excluded 12 samples that 382 contained insufficient numbers of sequences and 699 ASVs. We then made a subset of non-treated 383 (control) samples (119 samples and 13,042 ASVs) to study the soybean and corn phyllosphere and 384 soil bacterial community diversity and composition. We also made a subset of phyllosphere 385 samples (110 samples and 6,695 ASVs) to study the variations in the phyllosphere bacterial 386 community diversity and composition in response to neonicotinoid seed treatment. Since soil 387 samples had more sequences per sample than phyllosphere samples, we rarefied the dataset again, 388 this time to 10,000 reads per sample, which excluded 22 samples that contained insufficient 389 numbers of sequences and 195 ASVs. Therefore, we subset soil samples to study the effects of neonicotinoid seed treatment on their bacterial diversity and composition (132 samples and 13,137 391 ASVs). Overall, quality control and filtering, decontamination, and rarefaction procedures at 5,000 392 and 10,000 cutoffs (Fig. 5) respectively eliminated 41% and 39% of the low-quality ASVs and 393 20% and 23% of the samples (including all the 15 negative controls). We then analyzed these 394 datasets using different R packages. 395

Characterization of phyllosphere and soil bacterial composition and diversity 397
To identify the bacterial composition and diversity of the soybean and corn phyllosphere 398 and soil, we analyzed the non-neonicotinoid treated (control) samples that were rarefied to 5,000 interactions suggested by our model. We adjusted the p-values using Holm's method.

Effects of neonicotinoid seed treatment on bacterial taxonomic composition 459
To determine the differentially abundant ASVs and taxa between control and 460 neonicotinoid-treated samples in each habitat, we performed a differential expression analysis of 461 sequence data (DESeq2 [81]) using the Wald significance test with a local fit type and compared 462 the results by estimating the log2 fold changes. We analyzed the non-rarefied and non-normalized

Consent for publication 486
Not applicable 487

Availability of data and materials 488
We have deposited the raw sequences at the NCBI Sequence Read Archive (SRA): PRJNA662376. 489 Our scripts to perform the analyses of the current study are available in the following GitHub 490 repository: https://github.com/memoll/acadie_16s. 491

Competing interests 492
The authors declare no conflict of interest.     Significance levels for each variable are given by: *** P < 0.001; ** P < 0.01; * P < 0.   given by: *** P < 0.001; ** P < 0.01; * P < 0.05; NS, P ≥ 0.05. 783  Table 4 Phyllosphere and soil bacterial phyla associated with control and neonicotinoid seed 785 treatment. Differential expression analysis of sequence data (DESeq2) identified the bacterial 786 phyla of the ASVs that are significantly differentially abundant (adjusted P < 0.05) between control 787 and neonicotinoid-treated samples of soybean and corn phyllosphere and soil in a three-year 788 rotation in L'Acadie.