This project investigated the effects of long-term organic and conventional management on bacterial community structures in agricultural soils using 16S metagenomic sequencing. Due to the cost of sequencing at the time the experiment was designed, we were not able to use sufficient replication to overcome the enormous variability arising from the heterogeneity of the soil environment, so we did not detect differences in structure or diversity due to management, crop, or sampling site. Bacterial community structure did differ by sampling site, suggesting that soil type plays a greater role in structuring bacterial communities than management system or crop. However, we generated 22 detailed phylogenetic profiles of agricultural soil bacterial communities. We also used the information obtained in this experiment about expected variability in 16S profiles in the design of a follow-up experiment (currently in progress) that investigates the effects of cover crops and organic fertilizers on soil microbial community structure and function.
The impact of agricultural practices, particularly organic farming, on soil biological properties has long been a subject of interest, but only in the last few years have researchers begun to use molecular techniques to investigate the impact of agricultural practices on soil microbial communities. These techniques enable investigation of entire communities, including those microbes that cannot be grown in culture, which constitute the large majority of soil populations. Recent developments in high-throughput next-generations sequencing have allowed whole-community, or metagenomic, DNA sequencing. This project uses metagenomic 16S sequencing to investigate the structure and diversity of soil bacterial communities under long-term organic and conventional management. To our knowledge, this is the first study to date using 16S sequencing in organic field crop systems.
The stated objective of this project were 1) to construct a set of phylogenetic profiles of soil communities under a broad range of management practices, which will allow us to determine which bacterial populations are favored by which practices, and 2) to use these profiles to make predictions about the metabolic capabilities of these communities, which can be used to guide further research into the impacts of organic and conventional practices on soil biological functions.
This project did achieve the objective of generating multiple snapshot-style profiles of soil bacterial communities across a wide range of soil types and agronomic practices. Because we were not able to detect differences in microbial community structure associated with management system or crop, we were not able to use our phylogenetic profiles to predict differences in soil microbial function.
Since this project was initiated, the cost of metagenomic sequencing has dropped dramatically, allowing us to design a follow-up experiment that makes use of replication and subsampling to address the diversity and heterogeneity of soil communities. The follow-up experiment, which was initiated in 2011, investigates the relationship between soil microbial community structure and metabolic function in a much more explicit and statistically rigorous manner than was possible within the constraints of the original experiment.
Soil samples were collected from five sites (Table 1) from fields that had been under organic management for at least 15 years, as well as fields that had been conventionally managed during the same period. At some sites, samples were collected from multiple rotations or phases of rotation (Table 2).
All samples were collected between September 25th and October 20, 2010. For each treatment, ten cores were taken from a single field or experimental repetition. Cores were taken at least 2 m. from the edge of the field or plot, and were distributed throughout the whole plot where possible, or a portion of the plot where necessary to ensure that all cores were taken from the same soil type. At Lamberton, variability of soil types within the VICMS plot area did not allow complete matching of soil types across treatments.
Cores were taken using a 3/4” soil probe. Debris was removed from the soil surface and the probe was inserted to a depth of 10 cm, or shallower where soil was very hard. The core was removed intact when possible, and placed in a Ziploc bag. Where it was not possible to remove the core from the probe intact, the top 2 cm were cut from the core using a knife and placed in the bag, and the rest of the core was discarded. The core or loose soil was wrapped tightly in the bag, sealed with masking tape, and immediately placed on ice. Gloves, probe, and knife were wiped clean and sterilized with 70% isopropyl alcohol spray between treatments. Samples were transported to the laboratory in a cooler and frozen immediately upon arrival.
Before processing, the top 2 cm were cut from cores that had been stored intact. The remainder of the core was discarded. For each treatment, soil from all cores was blended into a single sample. Samples were sieved to homogenize.
DNA extraction and sequencing
DNA was extracted from samples using the MoBio PowerSoil kit (MoBio Laboratories, Carlsbad, CA), following kit instructions and using nuclease-free H20 for the final elution step. Samples were PCR amplified in triplicate using Illumina V6 ID tag primers 1-12, excluding ID2, to create amplicons of approximately 104-bp covering the V6 hypervariable region of bacterial 16S rDNA. Multiplexed amplicon libraries were sequenced on Illumina HiSeq by the National Center for Genome Resources (Santa Fe, NM).
Data processing was carried out using mothur (Schloss et al., 2012). Sequences were paired-end aligned and screened for quality. Sequences were removed if they had a quality score < 35 over a window of 50 nt, had a mismatch to a barcode sequence, had >1 mismatch to a primer sequence, had homopolymers > 8 nt, or had an ambiguous base (N). Singleton sequences and sequences identified as chimeras by UCHIME (Edgar et al., 2011) were removed. Sequence read number was normalized by random subsampling to 139,725 reads per sample. Sequences were aligned to the SILVA database (Quast et al., 2013), and sequences corresponding to chloroplast lineages were removed. Sequences were clustered into OTUs using the furthest-neighbor algorithm at 97% similarity. OTUs were classified using the Ribosomal Database Project ver. 9 database (Cole et al. 2014).
Data analysis was performed in mothur using Bray-Curtis distance matrices. A subset of fourteen samples was identified that represented seven pairs, each pair matched by location, current crop, and rotation (to the greatest extent possible), and differing by management system. A further subset of eleven samples was identified, representing five sample pairs matched for location, rotation (where possible), and management system, but differing in current crop (corn or soy); one pair also incorporated an extra sample from a similar rotation. Because experimental replication was not available due to the prohibitive cost of sequencing at the time that this study was designed, sample pairs within the designated subsets were treated as replicates for the purpose of this analysis. Differences among management systems, crops, and locations were assessed using analysis of molecular variance (AMOVA) and analysis of similarity (ANOSIM).
We obtained phylogenetic profiles of 22 sampled soils representing organic, conventional, and prairie management at five locations with diverse soil types. The depth of sequencing in this study was extremely high, making these profiles an important resource for understanding the structure of agricultural soil communities. This additional depth is not superfluous, as less abundant taxa may be highly active and responsible for key functional features of the community (Zarraonaindia et al. 2013), suggesting the need to sequence at high depth and incorporate less abundant taxa into analyses.
In this project, the expense of sequencing at the time that the experiment was designed prohibited traditional replication, which constrained our ability to make claims about the statistical significance of observed effects. We did not detect statistically dignificant differences in diversity or structure between microbial communities associated with organic and conventional systems. We were also unable to detect significant differences with crop, another factors which has been shown by previous research to affect microbial community structure. Therefore, we cannot conclude that there are no differences in bacterial community structure between organic and conventional soils, but rather that this experiment did not include sufficient replication to detect the “signal” of treatment differences through the “noise” of variability attributable to the extremely heterogeneous nature of the soil environment and its microbial habitats. Community structure did, however, differ among sampling sites. This indicates that the effects of location, including soil type and weather conditions, played a greater role in structuring soil bacterial communities than management system or preceding crop. This is consistent with the findings of previous studies conducted with non-sequencing-based community profiling methods (Garbeva et al., 2004; Widmer et al., 2006).
The difficulty encountered in this experiment has guided us in the design of the next phase of this inquiry, which is currently underway. In this phase, we are investigating the impacts of cover crops and organic fertilizers on soil microbial community structure and function, incorporating sequencing data as well as data on soil respiration, nitrogen mineralization, and enzyme activity.
Educational & Outreach Activities
We have presented results from this project at the MOSES organic farming conference, and discussed the project at the Southwest Research and Outreach Center’s annual organic field day. A publication is also being prepared for submission to peer-reviewed journals in the field of sustainable agriculture. The project has drawn considerable interest from growers, and helped to lay groundwork for future research. As data from our follow-up experiment is analyzed, peer-reviewed publications will be developed as well as articles for grower publications.
Short-term outcomes proposed for this project were: 1) construction of a set of phylogenetic profiles of soil microbial communities under a broad range of management practices and 2) use of these profiles to make predictions about the metabolic capabilities of these communities, which can be used to guide further research into the impacts of organic and conventional practices on soil biological functions. Intermediate-term proposed outcomes were: 1) increased understanding of the effects of organic management on soil biotic communities and 2) empowerment of growers to foresee the impacts of particular inputs and practices on soil functions relevant to crop success, leading to better long-term maintenance of soil fertility and farm productivity.
We were not able to detect differences in bacterial community structure associated with management systems; therefore, the findings of this project may not directly affect the information available to inform farmer decision-making. However, our findings reinforced the increasingly crucial understanding of the soil environment as a diverse and heterogeneous habitat with spatial variation at a variety of scales. Methodological insight was also gained, as the magnitude of the variance observed in this experiment will inform future experimental design to ensure that sampling and replication are sufficient to capture treatment differences. Our findings also underscore the importance of soil type as a factor structuring soil microbial communities.
This project also represents an unusually deep sequencing of the sampled soils, providing a profile of the communities in the our sampled fields that includes rare taxa that may be missed by shallower sequencing. To date, few studies in agricultural soil have approached this depth of sequencing, making these profiles an unusually detailed resource for describing the communities affecting cropping systems.
An economic analysis was beyond the scope of this project. However, many functions performed by soil microbial communities are of economic importance to organic growers. We feel that there is strong potential for microbial community analysis to contribute to understanding of nutrient cycling, weed and disease suppression, and other microbially mediated soil functions, ultimately enabling more efficient use of puchased inputs.
At this stage, our results do not indicate recommendations for farmer adoption. However, interest in this project among growers at field day and conference presentations has been high, and we are engaged in conversations with farmers and other researchers about how newly available and affordable metagenomic sequencing may be useful to growers. We are interested both in how these techniques can contribute to scientific understanding of the effects of particular practices, which can then be translated into recommendations to growers; and in the potential of metagenomic profiling of field soils to contribute, alongside routine soil tests, to farmer decision-making regarding crop selection and nutrient application.
Areas needing additional study
Although we were unable to distinguish between bacterial communities from organic and conventional soils, this work was highly informative for future metagenomic investigations of soil communities. Our subsequent experiments, which investigate the effects of cover crops and fertilizers on soil community structure and function in organic systems, have drawn on our findings in this project in determining appropriate levels of replication for detecting community differences. If possible, we would also like to revisit these sites at a later date to sample with replication, and repeat our analysis. We also believe that our observation of apparently systematic differences between the subsets of bacterial taxa that were more abundant in conventional samples and those that were more abundant in organic samples warrants further investigation.