Data for: Differential responses of soil bacterial communities to a prescribed fire in a paired restored and remnant prairie system

Published: 04-03-2021| Version 1 | DOI: 10.17632/4w32gzgvsv.1
Contributors:
Kathryn Docherty,
Zachary Whitacre

Description

Restoration of prairie and oak savanna systems on former agricultural land is an important effort for improving soil health and ecosystem services. Restored and remnant systems in the same region are often managed using the same strategies, including implementation of prescribed fire. Yet, soil microbial communities and functions in these two types of systems are typically very different due to past land-use history and current plant community composition. In this study, we investigated the responses of soil bacterial communities, enzyme activities and putative functional pathways to the effects of prescribed fire in a paired remnant and restored prairie system located in southwest Michigan, USA. We examined the immediate effects of fire one day after the fire as well as the longer-term effects in a time series extending to 11-months after the fire. Our results indicate that the soil bacterial communities in the remnant were immediately responsive in composition but not in function. Additionally, remnant community composition in burned plots returned to the composition in the control within one month, indicating resiliency in this community. In contrast, soil bacterial communities in the restoration shifted both compositionally and functionally one day after the fire, and continued to differ after 11-months. Past land-use effects on bacterial community composition and site-level heterogeneity, coupled with present-day differences in plant communities and litter quantity, mediated the different responses in the two systems. Our results suggest that land management plans aimed at increasing soil functional resiliency may require different management strategies for restored and remnant prairies.

Files

Steps to reproduce

We conducted this study an agricultural field that had been restored to prairie and an adjacent remnant. Both the fields were burned on 09/25/2013. Burn breaks were established to yield burned and control plots in each prairie. We chose the locations of five 1 x 1 m plots in both the burned and control areas. We collected triplicate 10-cm soil cores from each plot pre-fire, 1-day (09/2013), 1-month (10/2013), 7-months (04/2014) and 11-months (08/2014) after the fire in all plots. We measured soil temperature at the time of collection. (Univariate Table) Cores were used for all soil abiotic and microbial analyses. We collected triplicate 20-cm cores to measure belowground biomass (BGB) from each plot. (Univariate Table) In 08/2014 we measured plant composition in four transects using a point-intercept method (Bonham 1989) (Plant Community Table) We removed a 50-g subset of soil to freeze for DNA extraction. We sieved soil through a 2 mm sieve and measured pH, soil water content (SWC), soil organic matter (SOM) and conducted 0.5 M NaHCO3 extraction for total phosphorus (TP) analysis and 2 M KCl extractions for NH4 and NO3 analysis. We used colorimetric assays to measure nutrient concentrations. We assessed the extracellular enzyme activities of cellobiohydrolase, β-glucosidase, N-acetylglucosaminidase, and phosphatase using fluorescent-linked substrates (Gutknecht et al., 2010). We measured fluorescence using a microplate fluorometer (Cary Eclipse). (Univariate Table) We extracted genomic DNA with a PowerSoil DNA Isolation Kit (MoBio, Carlsbad, CA, USA). Amplicon preparation of the V4 region of Bacterial 16S rRNA was performed using the 515/806 primer pair (Kozich et al. 2013). Mi-Seq (Illumina, San Diego, CA, USA) sequencing was conducted at the MSU Genomics Core Facility in two Mi-Seq runs. Run 1: pre-fire, 09/2013, and 10/2013; Run 2: 04/2014 and 08/2014. Fastq files are publicly available at NCBI-SRA under PRJNA706578 and PRJNA706648. We processed the sequences using the QIIME2 version 2019.7 (Bolyen et al., 2019); DADA2 was used to merge paired reads, filter by sequence quality, denoise, create an Amplicon Sequence Variant (ASV) table, and remove chimeras (Callahan et al., 2017). Forward and reverse reads were truncated at the 15th base pair from the 5´ end to remove low quality regions of the sequences. From the 3´ end, the sequences were truncated at the 249th and 231st bp for the forward and reverse reads, respectively. Singleton sequences were removed. ASVs were taxonomically assigned using the Naïve Bayes Classifier using the SILVA (version 132) 99% OTU database (Quast et al., 2013). (ASV Table) We used PICRUSt to predict metagenomics pathways (KEGG 1-3) using Galaxy (Langille et al. 2013). We created an OTU table using the Greengenes version 13.5 closed-reference database (DeSantis et al. 2006). We normalized data to account for multiple copies of 16S rRNA in bacteria. (OTU Table, PiCRUST Table)