Pollinators mediate floral microbial diversity and network under agrochemical disturbance
Data files
Mar 19, 2021 version files 1.55 MB
-
ASVs_Bacteria_20210311.csv
-
ASVs_Fungi_20210311.csv
-
FloralMicrobiome_Bacteria_16S_CommunityMatrix_20210311.csv
-
FloralMicrobiome_Bacteria_16S_metadata_20210311.xlsx
-
FloralMicrobiome_Fungi_ITS_CommunityMatrix_20210311.csv
-
FloralMicrobiome_Fungi_ITS_metadata_20210311.xlsx
-
Plants_Pollinators_FloralTraits_20210311.xlsx
Abstract
Methods
Floral microbiome collection and sequencing
Two flowers were collected from each plant sample each week two days post agrochemical application. Following Wei and Ashman (2018), flowers (without pedicel) were collected into a sterile 15 mL centrifuge tube using ethanol-rinsed forceps and were transferred to a -20 ºC freezer within two hours. For microbial DNA extraction, we pooled flowers standardized by size (N = 1–4) per plant sample per 2-wk time period and obtained 246 total samples. Epiphytic microbes were collected by sonicating flowers in 3 mL phosphate-buffered saline at 40 kHz for 10 min and vortexing for 5 min, and were pelleted by centrifuging at 13,300 rpm for 5 min. Microbial DNA was extracted using Quick-DNA Fecal/Soil Microbe Kits (Zymo Research, Irvine, CA). Two negative controls without flower samples were included in the process of microbe isolation and DNA extraction. Samples and negative controls were sent to Argonne National Laboratory for bacterial (16S rRNA V5–V6 region, 799f–1115r primer pair) (Redford, Bowers, Knight, Linhart, & Fierer, 2010) and fungal (ITS1f–ITS2) (Smith & Peay, 2014) library preparation. Because the negative controls failed in PCRs in library preparation, the 246 samples were sequenced on two lanes of Illumina MiSeq paired-end 250 bp.
Microbial sequence processing
Demultiplexed paired-end (PE) reads were used for detecting bacterial and fungal amplicon sequence variants (ASVs) using package DADA2 v1.12.1 (Callahan et al., 2016) in R v3.6.0 (R Core Team, 2019) and QIIME 2 v2019.4 (Bolyen et al., 2019). For bacterial ASV analysis in DADA2, PE reads were trimmed and filtered [truncLen = c(245, 245), trimLeft = c(10, 0), maxN = 0, truncQ = 2] after initial quality inspection. Then end-specific variants were identified after taking into account sequence errors, prior to joining the PE reads (minOverlap = 20, maxMismatch = 4) for ASV detection. Bacterial ASVs were further filtered against chimeras and assigned with taxonomic identification based on the SILVA reference database (132 release) implemented in DADA2. For fungal ASV analysis in DADA2, the PE reads were first screened to remove potential primer contaminations. Due to the low-quality end (~50 bp) of the fungal reads, we then truncated reads at 200 bp during quality filtering [truncLen = c(200, 200), maxN = 0, truncQ = 2] to ensure the accuracy in ASV detection. By doing so, our method provided a conservative estimate of fungal ASVs, due to the potential loss of information based on ITS length variation (Callahan et al., 2016). After chimera removal, fungal taxonomic assignment was conducted based on the UNITE reference database (v8.0 dynamic release) using QIIME 2.
Bacterial and fungal ASV tables were further filtered separately before conversion into microbial community matrices using package phyloseq (McMurdie & Holmes, 2013). First, we removed non-focal ASVs (Archaea, chloroplasts and mitochondria). Second, we filtered out low-depth samples (<100 reads; N = 13 and 8 samples for bacterial and fungal data set, respectively). Third, we normalized per-sample reads to the same number (i.e. the median reads, 18788 and 37516, bacterial and fungal data set, respectively) following Wei and Ashman (2018). Lastly, we removed low-frequency ASVs (<0.001% of total observations). The final community matrices consisted of 1237 and 1165 ASVs for bacteria (N = 223) and fungi (N = 240), respectively.
Usage notes
This dataset contains seven data files:
(1) Plants_Pollinators_FloralTraits_20210311.xlsx
This file contains the experimental data of each plant sample, including strawberry genotype (1, Mara Des Bois; 2, Albion; 3, Portola; 4, San Andreas), agrochemical treatment (C, water control; B, bactericide; F, fungicide; BF, bactericide and fungicide), time period (the first vs. last 2 wk, Time 1 vs. Time 2; coded as wk12 vs. wk34 in the data), experimental block (1–40), visitation (i.e. number of visiting pollinators per ~36 min per plant sample during a 2-wk time period) of individual pollinator functional groups (bees, wasps, flies, butterflies, beetles, true bugs and other taxa), and floral traits (i.e. flower size, pollen production per flower, and flower abundance).
(2) FloralMicrobiome_Bacteria_16S_metadata_20210311.xlsx
This file contains the experimental data of the 223 samples that were included in the final analyses of floral bacterial communities.
(3) FloralMicrobiome_Bacteria_16S_CommunityMatrix_20210311.csv
This file contains the bacterial community matrix of 1237 amplicon sequence variants (ASVs) across the 223 plant samples.
(4) FloralMicrobiome_Fungi_ITS_metadata_20210311.xlsx
This file contains the experimental data of the 240 samples that were included in the final analyses of floral fungal communities.
(5) FloralMicrobiome_Fungi_ITS_CommunityMatrix_20210311.csv
This file contains the fungal community matrix of 1165 amplicon sequence variants (ASVs) across the 240 plant samples.
(6) ASVs_Bacteria_20210311.csv
This file contains the taxonomical identification of the 1237 bacterial ASVs.
(7) ASVs_Fungi_20210311.csv
This file contains the taxonomical identification of the 1165 fungal ASVs.