Progress 10/01/16 to 09/30/19
Outputs
Target Audience:Project coordination and sampling plans were developed by the Cornell Soil Health Team in coordination with scientific and technical experts encompassing Universities, Cooperative Extension, the USDA/NRCS, agricultural service providers, and farmers. Microbiome data will be shared with the Soil Health Team and its partners and analyzed in relation to results from soil health test results. Communication of results to program work teams within Cornell AES will facilitate information transmission to stakeholders on outcomes of the research. We expect that the proposed project will yield information on microbial indicators of soil health and that such information will be of great value for soil health testing. Improved soil health test data will provide farmers in New York and nationwide with better tools for managing soil productivity. Changes/Problems:
Nothing Reported
What opportunities for training and professional development has the project provided?A postdoctoral associate has been supervising this project. A PhD student has been involved in sample collection and analysis. In addition, a technician has been assisting in sample processing. These participants are learning modern methods of soil microbiome analysis. How have the results been disseminated to communities of interest?Results have been disseminated in peer reviewed publication and in presentations at national meetings. In addition, additional publications and communications are in preparation pending final data analysis. This project was designed to initiate research efforts that remain ongoing. What do you plan to do during the next reporting period to accomplish the goals?
Nothing Reported
Impacts
What was accomplished under these goals?
Soil health is a term used to describe the functional characteristics of soil that have agricultural and ecological value. These characteristics include the potential to store carbon, retain moisture, suppress disease, prevent erosion, and mitigate green-house gas and other off-site pollution, in addition to sustaining crop yields in the long-term. The purpose of our study was to test the predictive value of the soil microbiome for properties of soil health and identify differences in the ecological function of populations indicative of poor versus good health. Our study demonstrated that microbiome data is predictive of soil health and our ability to leverage our ecological understanding of soil microbes to deepen dimensions of soil health. Our results show that microbiome can resolve fine-scale differences in soil health, such as the over application of ammonia fertilizer, which have clear value for improving soil health management. The essential relationship between the soil microbiome and health was evident in the correlation between health rating and microbial activity (respiration), biomass (DNA yield) and community composition. These results emphasize the defining nature of SOM in assessments of soil health. We have shown that microbiome data can shed light on the quality of SOM, a capacity which will only improve as we further understand the roles of specific bacterial guilds. We also demonstrated that the soil microbiome is highly sensitive to capturing dimensions of soil disturbance, related to life-history strategies like growth rate predicated on rrn copy number. Finally, our study highlights the fallacy of assuming higher microbial diversity corresponds with better soil health management practices, as microbes are highly adaptive and changes in diversity due to soil disturbance may be best explained by the intermediate disturbance theorem. In summary, this work offers a range of meaningful insights into the use of soil microbiome data in measuring soil health. We aim to use this foundation to continue to improve soil health practices by refining our capacity to monitor and measure health metric. We also believe these approaches have great potential as our capacity to use machine learning as a predictive tool improves greatly as more data is collected. We collected soil health and microbiome data for 950 soil samples from agricultural sites across 40 of the contiguous states with the help of a network of trained professionals. We profiled the bacterial community using 16S rRNA gene amplicon sequencing as well as collecting information on the total soil DNA yield (a measure of biomass) and bacterial abundance (qPCR). We employed a variety of machine learning approaches to determine the predictive ability of the soil microbiome. We also curated a database of agricultural studies examining features of soil health, including disease suppressive soil (ex. PRJNA308986), crop-residue retention (PRJNA397131), soil aggregation (PRJNA447911), lignocellulose degrading populations, and tillage (PRJNA492265). We then tested whether bacterial groups identified in these studies were predictive of or associated with soil health. We identified trends in several major bacterial populations associated with properties of soil health and in ecological traits relevant to carbon cycling, such as the increase in ruderal bacteria in soils of poor health. A total of 949 soil samples were collected and processed. Many originated from the same geographic location, grouping into 161 unique sites (> 1 km radius apart) termed 'geo groups' with a median of two samples per group (nmax = 48). A total of 191 management groups where identified when samples within a geo group were differentiated by tillage practices. The aggregate health rating of a soil was significantly correlated with measures of biological activity and soil organic matter. Total DNA yield, a proxy for microbial biomass, was most strongly correlated with respiration (r = 0.63; p ~ 0) and aggregate stability (r = 0.62; p ~ 0), indicating the importance of microbial activity and biomass in soil health. Soil microbiomes grouped primarily by geographical proximity, yet variation within geo groups was apparent in soil health and microbial beta-diversity. After accounting for variation due to proximity, the degree of tillage, soil texture and amount of active carbon explained the greatest variation in soil microbiomes. We demonstrated that the soil microbiome has predictive value for soil health. A random forests classifier trained on 75% (ntrain =600 samples) and validated on 25% (nvalidate =199) exhibited quite high accuracy even after adjusting for chance prediction (i.e. the 'kappa' statistic). The microbiome data was most predictive of tillage status, able to predict 77% of the time whether a soil was no-till versus three other classes of tillage intensity. Soil health status was the next most predictable attribute of soil, a promising result given more sophisticated classification methods are available and being developed. The soil microbiome showed consistent effects of tillage, as expected given its predictive strength. The 'till-ome' was comprised of major populations of bacterial which were favored by tilling. Many of the member of the till-ome were also indicators of soil health. This was expected given that soil health was driven by various measures of soil organic matter (SOM), which were significantly lower in conventionally tilled and managed soils. Disentangling these co-related measures of soil health is possible based on a closer examination of the ecological associations of members of the microbiome. The value of leveraging publicly available data to interpret trends in microbiome data led us to build a database of 45 studies focused on agricultural soils and management practices, termed the ecoDB. Our work using the ecoDB is on-going and seeks to develop a set of taxa with known ecological-attributes that are predictive of soil health.
Publications
- Type:
Journal Articles
Status:
Published
Year Published:
2019
Citation:
Sirois, S. H. and Buckley, D. H. (2019) Factors governing extracellular DNA (eDNA) degradation dynamics in soil. Environmental Microbiology Reports. 11:173-184. doi.org/10.1111/1758-2229.12725
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2018
Citation:
Sara H. Sirois and Daniel H. Buckley (2018) Degradation dynamics and stabilization of extracellular DNA in soils vary in relation to moisture, temperature, habitat type, and management practice. Presented at the 17th International Symposium on Microbial Ecology.
|
Progress 10/01/17 to 09/30/18
Outputs
Target Audience:Project coordination and sampling plans have been developed by the Cornell Soil Health Team in coordination with scientific and technical experts encompassing Universities, Cooperative Extension, the USDA/NRCS, agricultural service providers, and farmers. Microbiome data will be shared with the Soil Health Team and its partners and analyzed in relation to results from soil health test results. Communication of results to program work teams within CUAES will facilitate information transmission to stakeholders on outcomes of the research. The national soil microbiome dataset that we develop will be of great value to agricultural science and environmental science research communities. This database will be unprecedented in scope and will provide valuable baseline information which will facilitate research to describe the contributions of specific microbes to critical soil processes and to investigate plant- microbiome interactions. We expect that the proposed project will yield information on microbial indicators of soil health and that such information will be of great value for soil health testing. Improved soil health test data will provide farmers in New York and nationwide with better tools for managing soil productivity. Changes/Problems:PI's FTE on the project was 4% What opportunities for training and professional development has the project provided?The postdoc who is working on this project has been learning machine learning methods for the analysis of soil microbiome data. How have the results been disseminated to communities of interest?
Nothing Reported
What do you plan to do during the next reporting period to accomplish the goals?We plan to sequence another 500 DNA samples for soil microbiome analyses. DNA will be extracted, amplified and prepared for sequencing. Sequences will be determined, quality control performed and the data analyzed. We will analyze the complete soil microbiome data set in relation to soil health data from the soil health lab to determine the degree to which microbiome composition or signatures are predictive of soil health. These results will be prepared for publication.
Impacts
What was accomplished under these goals?
We performed bacterial SSU rRNA gene amplicon sequencing of ~500 soil samples from the Cornell Soil Health Lab. Preliminary analyses were performed on these data to guide the next round of sampling efforts. The data were organized into a database along with soil health data to aid in analysis. We also planned and organized sampling efforts for the coming growing season. Preliminary results are promising. We can clearly identify region and soil characteristics from soil microbiome data. Furthermore, initial efforts at machine learning classification methods suggests that soil microbiome composition is predictive of soil health properties. Furthermore, it seems as though indicator organism analyses might be possible as certain taxa seem to be predictive of soil health.
Publications
|