Advances in Phenomics: Single Location vs. Distributed Analysis of Genotype and Phenotype, Lecture notes of Biology

Download Advances in Phenomics: Single Location vs. Distributed Analysis of Genotype and Phenotype and more Lecture notes Biology in PDF only on Docsity! Phenomics: Genotype to Phenotype A report of the Phenomics workshop sponsored by the USDA and NSF 2011 NIFA-NSF Phenomics Workshop Report Page 2 Table of Contents Executive Summary ..............................................................................................................................................................3 Recommendations.................................................................................................................................................................4 Figure 1 ........................................................................................................................................................................................8 Figure 2 ........................................................................................................................................................................................9 Figure 3 ..................................................................................................................................................................................... 10 1. Introduction ...................................................................................................................................................................... 11 A. Phenomics: From Genotype to Phenotype ...................................................................................................... 11 B. How is phenomics different from phenotype? .............................................................................................. 12 2. What is needed to advance phenomics? ........................................................................................................... 13 A. The roles of ‘reference’ and ‘model’ systems .................................................................................................. 14 B. High throughput data types and workflows ................................................................................................. 15 C. Maximizing the value of phenomics .................................................................................................................. 17 D. Making data and software available ................................................................................................................ 18 E. Project scale: the effect of size.............................................................................................................................. 21 F. Success depends on a trained workforce ......................................................................................................... 22 3. Computation and modeling ...................................................................................................................................... 23 A. Data capture ................................................................................................................................................................ 23 B. Data integration ......................................................................................................................................................... 24 C. Data analysis and visualization........................................................................................................................... 26 D. Predictive modeling .................................................................................................................................................. 27 4. Summary ............................................................................................................................................................................. 28 References cited .................................................................................................................................................................. 29 Appendix .................................................................................................................................................................................. 36 NIFA-NSF Phenomics Workshop Report Page 5 fact many of the most ubiquitous and important technologies and techniques of biotechnology came from fundamental science, rather than targeted research. Maintaining the full research and development pipeline will allow the US research community to continue creating a strong base of fundamental knowledge that will serve as a foundation for work in economically important organisms and applications-oriented research. This approach is used by world-class research- driven industrial organizations and should be supported by funding strategies set by NSF, USDA and other government and private sources. C. Funding agencies should facilitate mechanisms encouraging communication and collaboration between groups working on fundamental research and those trying to solve 'real world' problems. These could range from relatively inexpensive Research Coordination Network type grant awards used at NSF (http://www.nsf.gov/pubs/2011/nsf11531/nsf11531.htm) through funding of research from early discovery to translation, or conversely by focusing a breadth of research activities on well-defined practical problems. D. Funding priorities should be set based on science and technology goals, including both discovery-oriented and application-oriented research. Despite significant acceleration of progress on a range of economically and societally relevant issues by increasing the emphasis on organisms directly involved in these issues, recent trends have gone too far toward excluding 'basic' research on reference organisms from programs at NSF and USDA. In the longer term, this works against an end-to-end model of innovation and application. Consistent strong support for US research on reference organisms would maintain or increase the competiveness of US research and development. 2. Balance in scale of projects Phenomics projects are by nature relatively large scale and infrastructure intensive and when done well generate large amounts of high quality data at relatively low cost. These projects usually require expertise from domain experts that do not traditionally collaborate, and are often in different institutions or units within an institution (such as Biology, Computer Science and Engineering). These and other factors can create barriers for collaboration and increase the amount of time, effort and cost required to achieve the desired goals. While large collaborative projects are often viewed favorably by funding agencies and university administrators, they can result in lack of recognition for individual participants and reduced training potential. NIFA-NSF Phenomics Workshop Report Page 6 Phenomics research would benefit from funding of a portfolio of projects that range from creation of specific enabling technologies or proof-of-concept studies in single laboratories or small groups, through large-scale phenotyping projects and development of data and germplasm resources. Having a vision for how the smaller projects should impact phenomics will be important prior to making requests for proposals so that the best science that fits into this larger vision can be funded. Innovations should be sought in experimental design and process, cyberinfrastructure and data analysis methods that can be applied to a broad range of organisms and growth conditions (ranging from controlled environment to field) and that increase data acquisition throughput, quality and utility. 3. Data Management Biologists face the challenge of developing efficient and robust computational and bioinformatic methods to reduce large and diverse phenomics datasets into representations that can be interpreted in a biological context. Datasets of long-term value require data standards and metadata descriptions of the experiments in a format that enables computational approaches to data analysis. Approaches should be sought for high-throughput data collection methods that promote high quality results and long-term utility of the data. Use of a laboratory information management system (LIMS) in large projects is essential to ensure collection of high quality data. Different project management models should be considered for large phenomics projects, depending on the desired outcomes. These could range from highly integrated projects at a single site to collaborative consortia of laboratories with world-class expertise in complementary areas of biology, phenotyping or data analysis (See Figure 2). As with all large projects, strong project management with milestones and quality metrics are essential. Funding is needed to meet data storage/archive needs and ensure the availability and utility of phenomic data for computational approaches, with a sustained long-term funding stream preferable to a high funding rate over a short duration. 4. Considerations for workforce training NIFA-NSF Phenomics Workshop Report Page 7 Large-scale phenomics projects should have integrated training plans that align the interests of the project with the needs of students and postdocs. Large-scale phenomics projects generally involve repetitive procedures, which are typically best done by IT, laboratory and field technical support staff, and in some cases by undergraduate students. Science training opportunities exist for graduate students and postdoctorals in planning the project to create data useful for asking important questions, creating enabling technologies and mining the phenotypic data. Initiatives in phenomics should include educational activities to enable biology trainees to think quantitatively and collaborate with experts in physical, computational and engineering sciences. Training also should be provided in management of, and participation in, large interdisciplinary and collaborative domestic and international projects. Training in computer science and experimental design, including statistics, is essential for biologists involved in phenomics research. A basic understanding of the logic and methods of programming, knowledge of command-line tools (e.g. UNIX shell), and a familiarity with the development of computational pipelines and workflows will be essential for scientists to acquire, analyze, and critically interpret genomic and phenomic data. Funding is needed for undergraduates, predoctorals and postdoctorals to be trained in computational thinking that will advance our ability to obtain, analyze and utilize large scale phenomics data. The recently announced National Plant Genome Initiative Postdoctoral Research Fellowship program (http://www.nsf.gov/pubs/2011/nsf11499/nsf11499.pdf) and USDA‟s National Institute of Food and Agriculture Fellowships (http://www.csrees.usda.gov/fo/afrinifafellowshipsgrantprogram.cfm) are examples of such programs, and there is need for others. NIFA-NSF Phenomics Workshop Report Page 10 Figure 3. Phenotypes are complex. This simplified schematic diagram attempts to illustrate the multidimensionality of phenomics experiments and data. Genetic diversity can come in many forms such as induced mutants, natural variants, results of genetic crosses and populations of organisms. The environmental conditions used will impact the traits being measure and the variety of phenotypes ('attributes') measurable is nearly limitless. Not shown in this diagram are the dimensions of time (for example developmental progression of the organism) and differences in phenotypes in various cell, tissue or organ-types. NIFA-NSF Phenomics Workshop Report Page 11 1. Introduction A. Phenomics: From Genotype to Phenotype One of the central principals of biology is the concept that a set of genetic instructions, or genotype, interacts with the environment to produce the characteristics, or phenotype, of an organism. Understanding how particular genotypes result in specific phenotypic properties is a core goal of modern biology, and enables development of organisms with commercially useful characteristics. However, prediction of phenotype from genotype is generally a difficult problem due to the large number of genes and gene products that contribute to most phenotypes in concert with complex and changeable environmental influences. The last 20 years have created a revolution in our understanding of genotype: while genomes typically are quite large, with millions or billions of nucleotides, the relative chemical simplicity of DNA lends itself to large-scale analysis. We can now determine genotypes down to the level of individual nucleotides in whole genomes, and entire genomes are now rapidly sequenced at steadily declining costs and ever increasing speed. Next generation resequencing methods provide opportunities to get the complete genotype and epigenotype not only of a single representative of a genus or species, but of many representatives of a phylogenetic group or population. High-density single nucleotide polymorphism genotyping, first pioneered in the human HapMap project, has become tractable for any organism and now is routinely applied to plants and microbes for the characterization of natural genetic variation and to support trait- driven efforts to clone and understand specific genes. Thus genome science is moving beyond the era of reference and model organisms to study in depth any microbe, animal or plant that has characteristics of interest to science and society. The study of phenotypes is quite different. Unlike a genotype, the phenotype of an organism can be described at many levels, from specific molecules to dynamic metabolic networks to complex cellular developmental and physiological systems, all the way to the aggregate or social behaviors of complex populations. Interactions with symbionts, pathogens or competing organisms create additional levels of phenotypic complexity. Moreover, phenotypes are dynamic NIFA-NSF Phenomics Workshop Report Page 12 and the timescales in which they change vary tremendously. Consider for instance the rapid responses of a bacterium to nutrient changes (Segall et al., 1986) or the dynamic changes in photosynthesis of a leaf as a single cloud passes over the sun (Murchie and Niyogi, 2011), compared to the slow morphological changes in long lived plants or even the lifelong changes in the outward appearance of a human being. Phenotypes rarely have a single discrete description, and most phenotypic characters are better described as continuous functions as opposed to the discrete „A,C,G,T‟ character codes of the genotype. Indeed a complete catalog of phenotypes (the phenome) can have essentially infinite complexity (See Figure 3). Now that digital DNA data are available in abundance, we face an acute need to quantify individual phenotypes in a way that can be explicitly matched to individual genotypes. If this challenge can be mastered, we face the promise of gaining a deeper insight into the components of complex traits such as yield or stress resistance in economically important plants and animals or population dynamics for microbes that play key roles in global nutrient cycling. This can extend to a systems level description of the underlying processes, ultimately enabling predictions of emergent phenotypes such as fitness and survival in studies of ecology and evolution or yield and stress tolerance or other traits of economic value. B. How is phenome different from phenotype? Phenomics, the study of the phenome, is a rapidly emerging area of science, which seeks to characterize phenotypes in a rigorous and formal way, and link these traits to the associated genes and gene variants (alleles). Examples of phenotypic parameters include gross morphological measures such as cell size, tree height or wheat yield, dynamic measures such as rate of cell division of a unicellular organism, metabolism or nutrient uptake, and molecular measures such as mass spectrometry fingerprints and transcript profiles. Formally, phenomics is the science of large-scale phenotypic data collection and analysis, whereas the phenome is the actual catalog of measurements. While it shares characteristics with classical mutant screening or quantitative trait analysis, it is distinguished from these traditional approaches in scale and scope (Winzeler et al., 1999; Lango Allen et al., 2010; Speliotes et al., 2010; Heffner et al., 2011; Lu et al., 2011; Nichols et al., 2011). First, phenomic studies NIFA-NSF Phenomics Workshop Report Page 15 traits such as seed yield are likely to be controlled by such a complex array of factors that they are best studied in the target organism in their usual production environments. The use of non-reference organisms for the study of specific biological processes is regaining popularity in part as a result of the revolution in sequencing and phenotyping technologies. Once an organism, organ or cell type is identified as being useful for study of a specific problem, transcriptome and/or genome sequence provides a starting place for characterizing the gene space and relative abundances of mRNAs. These sequences allow discovery of candidate genes (Cocuron et al., 2007), enable large scale proteomics analysis (Schilmiller et al., 2010) and a first pass construction of metabolic or regulatory networks (Keseler et al., 2009). Re-sequencing of genetic variants provides enormous numbers of molecular markers that can be used to find genes contributing to target phenotypes by traditional genetic mapping or more modern approaches such as genome-wide association studies (Rounsley and Last, 2010). Phenomics approaches such as digital descriptions of growth and development, physiological parameters, and protein and metabolite abundance may also increase the accessibility of non-reference model organisms. B. High throughput data types and workflows Some components currently considered key elements of a phenotype description such as gene expression profiles can be acquired by relatively accessible, high throughput technologies. In fact, large-scale transcript profiling studies in reference organisms have already proven of great value (EcoCyc for E. coli: http://ecocyc.org/; AtGenExpress for Arabidopsis: http://www.weigelworld.org/resources/microarray/AtGenExpress/ (Kilian et al., 2007). In contrast, technologies for characterizing downstream phenomics attributes are becoming increasingly powerful but are currently far from routine. i. Proteome and Metabolome Techniques for identification and quantification of large numbers of proteins and metabolites (Last et al., 2007) are increasingly sophisticated. However, many challenges exist for making these techniques accessible to the broad scientific community. These include expense of equipment, large range of concentrations and diverse chemical properties of these molecules. For example, analysis of stable, abundant and soluble metabolites and proteins is far easier than rare, NIFA-NSF Phenomics Workshop Report Page 16 labile and insoluble molecules. Although most protein and metabolite analysis methods require tissue extraction, well established methods exist for non-invasive measurements such as near infrared transmittance for seed metabolites (Velasco et al., 1999) and more experimental methods such as MALDI-ToF (matrix assisted laser desorption/ionization time of flight) mass spectrometry for spatial resolution of metabolites (Shroff et al., 2008). ii. Physiological attributes Physiological measurements of processes such as photosynthesis, nutrient uptake and transport can be reproducible and sophisticated (for example, see (Baxter et al., 2008), but achieving the necessary throughput is challenging. Spatial variation of physiological parameters can increasingly be approached through imaging technologies. This opens a new window of analysis, since in many cases the heterogeneity of the response in space and time is a key feature of the phenotype, which contains significant information about the underlying biological principles (Jansen et al., 2009; Walter et al., 2009). iii. Plant growth and development Growth and development of multicellular organisms can be measured by quantitative parameters like biomass or by spatially resolving technologies based on cameras and image analysis if experimental systems and computer algorithms capable of measuring the feature or process of interest can be developed. High throughput phenotyping platforms based on image analysis are available for laboratory and greenhouse settings (http://www.lemnatec.com/; http://www.plantaccelerator.org.au/) but their use is far from widespread. Quantification of below-ground structure and behavior is a major hurdle, though culture in or on gelled media make image-based techniques practicable (Fang et al., 2009; Brooks et al., 2010; Clark et al., 2011). Tomographic systems can provide insight in the dynamics of structure and function of root systems (Jahnke et al., 2009) but currently are not able to handle high throughput due to technical limitations and the large amount of data generated. iv. Phenomics in situ: measurements in the field Ecologists, breeders and systematists have been practicing phenotyping in the field for decades. High throughput phenotyping for certain target and correlated traits is routine, NIFA-NSF Phenomics Workshop Report Page 17 particularly in plant breeding where thousands of unique genotypes are evaluated seasonally. Remote sensing is increasingly powerful; for example canopy spectral reflectance is employed in plant breeding programs for measuring nitrogen- or water-use efficiency (Gutierrez et al., 2010). However, there are still many important traits that are difficult or costly to evaluate and phenomics technologies could bring new approaches that would enhance the identification of superior genotypes and effectively train prediction models. C. Maximizing the value of phenomics resources Because of their large scale, phenomics projects are resource intensive and generate large amounts of data. These large projects can be highly cost effective if the resulting data are of high quality and lasting utility to a large number of investigators. Several major factors contribute to the long term success of phenomics projects: the source of genetic diversity employed and whether it is preserved for future use; the quality of the growth conditions; the phenotypic assays performed; collection, storage and interpretation of data. Genetic diversity Germplasm collections are the starting point of many phenotypic investigations and are increasingly important as new phenotyping technologies emerge. Phenotyping tools are most useful for the task of understanding genetic function when they can be applied to the study of well-curated germplasm or genetic stocks appropriate to the problem being addressed. These collections can include lines generated by transgenesis or transposon mutagenesis, chemical or radiation induced mutagenesis, accessions of variants derived from natural populations and lines produced by crossing including breeding material, introgression lines and recombinant inbred lines (Eshed and Zamir, 1995; Alonso et al., 2003; Yu et al., 2008; Buckler et al., 2009). Production and maintenance of large sets of germplasm is time and labor intensive and careful thought should be given to the balance of cost and utility of a population. For example, sets of germplasm that can be used for large numbers of studies or to measure broad sets of phenotypes may be of higher value than those custom designed for specific projects or narrow phenotypic NIFA-NSF Phenomics Workshop Report Page 20 to enable extraction and computational analysis of datasets spanning multiple experiments. This model may be appropriate for specialized datasets with lower potential for widespread reuse and small, highly cohesive communities focused around specific research questions. (3) Individual project or community databases. Generic Model Organism Database tools could be used by individual communities or projects. Interoperability between different databases that capture the results, metadata, and provenance is essential and could be facilitated by the establishment of common controlled vocabularies for phenotypic measurements. This model has a variety of challenges including how to motivate researchers to contribute data, mechanisms to ensure that interoperability is maintained and approaches to fund long-term data curation and storage for many such resources. An increased reliance on computational approaches has resulted in development of a large number of software packages for a range of biological problems. However, documentation and widespread adaptability of the software is a major obstacle to re-use of code outside the developer‟s group. As with the reuse of data, software reuse is not always the optimal solution, but depends on the quality of available software, the effort required to adapt existing software vs. the effort of writing new iPlant Collaborative The iPlant Cyberinfrastructure Collaborative (http://www.iplantcollaborative.org/) offers both a compelling platform for sharing of source code for biological applications and an environment for building reproducible scientific workflows from published software. The phenomics community could be well served to encourage adoption of that platform by developers of bioinformatics applications and domain experts who can build best- practice workflows comprised of these software components. Encouraging participation of phenomics researchers and tool developers in iPlant-sponsored 'hack-a- thons' for collaborative code development and 'bring your own data' training workshops could ensure that this infrastructure contains tools appropriate to topics in phenomics. In addition, the iPlant platform encourages integration of visualization resources into its „software ecosystem‟ as well as development of new information visualization applications using the Stanford Protovis toolkit, R, and the Javascript InfoViz Toolkit. NIFA-NSF Phenomics Workshop Report Page 21 software, and the potential for continued refinement of pre-existing vs. newly-developed software. Nevertheless, availability of software and analysis pipelines for use and reuse will have a positive impact on phenomics research, allowing groups without in-house software expertise to carry out analyses, facilitating direct comparison of different datasets and allowing verification of published experimental results. Software tools and analysis pipelines could be made accessible for reuse by other groups within a community resource such as iPlant and documented as persistent publishable objects, referred to via a DOI or other identifier. This will serve dual purposes: First, experimental reproducibility can be enhanced because bioinformatics methods can be described not just in narrative terms, but in the context of a replayable series of events in an analytical infrastructure. Second, authors of tools and pipelines could receive publication credit as their DOIs are referenced in the literature. This should create incentives for such development activity. E. Project scale: the effects of size Until recently, most biological research involved single laboratories or small numbers of investigators who collaborated because of shared interests. The post-genome era dramatically changed this model. Now, research projects can vary in size from a small individual investigator group directed toward a defined focused project to large consortia of investigators working together toward a broad set of goals. Large group projects allow collaboration of domain experts in plant biology, microbiology, sequencing, high throughput omics, informatics and mathematical modeling. Bringing in expertise from disciplines such as statistics, engineering and computational sciences allows design of more efficient processes for obtaining high quality data and novel approaches to analyzing the large datasets. Each scale of project in the continuum has advantages and disadvantages to participants and to the broader science community. For example, traditional single investigator science allows a tremendous amount of freedom to the investigator, strong training in problem solving and hypothesis testing to early career participants, and can generate deep understanding of specific areas of science. However, the small size may limit both the breadth of approaches and opportunities for multi-disciplinary training and creation of biological resources and data of longer-term value. In contrast, consortia can tackle larger questions through applying diverse NIFA-NSF Phenomics Workshop Report Page 22 expertise and create multi-disciplinary training environments. However, successful management of these projects requires skills different from „smaller‟ science: competent and trusted leadership, formal project management, and resources devoted to data management. Communication between laboratories can divert time and effort from data gathering, analysis and dissemination. Large projects with set goals and repetitive operations can stifle creativity and training. In addition, the academic system tends to emphasize individual achievement, especially for early and mid-career scientists, where first author publications and grant funding record are of paramount importance. A proper balance of funding between the two has clear advantages. The large consortia can facilitate the larger scale experiments that would be beyond the means of the individual investigator. Funding of smaller projects enables the data to be mined extensively for hypothesis generation and testing. F. Success depends on a trained workforce 21 st Century Biology, including phenomics, critically depends upon a workforce trained differently from the traditional US model focused on graduate student and postdoctoral training in deep, and sometimes narrow, hypothesis-driven research. In addition to strong expertise in biology, the characterization of phenotypes is increasingly dependent on tools and activities at the interface of biological, computational and physical sciences. Any initiative in phenomics should include educational activities to enable biology students to think quantitatively and collaborate with scientists in domain areas such as chemistry, computer science and engineering. Most current plant biology curricula are sorely lacking in explicit training in computational methods. A basic understanding of the logic and methods of programming, knowledge of command-line tools (e.g. Unix shell), and a familiarity with the development of computational pipelines and workflows will be essential for scientists to acquire, analyze, and critically interpret genomic and phenomic data. Such training should include data management and curation, fundamentals of information visualization, an understanding of basic data types, and best practices in terms of methodology documentation. Making such training a requisite or strongly recommended part of coursework (similar to funding agency requirements for ethics NIFA-NSF Phenomics Workshop Report Page 25 necessitating the community-wide adoption of unique identifiers for such objects. The use of ontologies to describe phenomics data and metadata will be essential to make datasets searchable and reusable and facilitate data integration and computational analyses. Several efforts are already underway in this area (see Ontologies sidebar); thus the challenge will be to encourage or enforce community adoption and input into these ongoing efforts. Plant phenotype ontologies. Acronym Full Name Scope Link GO Gene Ontology Biological process, molecular function , subcellular localization www.geneontology.org PO Plant Ontology Plant anatomical parts and developmental stages www.plantontology.org TO Trait Ontology Cereal plant traits www.gramene.org/db/ontology/ search?id=TO:0000387 CO Crop Ontology Crop plants (anatomy, developmental stage, trait) www.generationcp.org/ontology PATO Phenotypic Attribute and Trait Ontology Phenotypic qualities obofoundry.org/wiki/index.php/ PATO: Main_Page CHeBI Chemical Entities of Biological Interest Chemical compounds www.ebi.ac.uk/chebi/ EnvO Environmental Ontology Habitat environmentontology.org EO Environment Plant growth conditions incl. temperature, growth media, light regime, etc. www.ebi.ac.uk/ontology- lookup/browse.do?ontName=E O UO Unit Ontology Units (describing length, volume, density, irradiance, temperature, etc) http://www.ebi.ac.uk/ontology- lookup/browse.do?ontName=U O NIFA-NSF Phenomics Workshop Report Page 26 C. Data Analysis and Visualization i. Visualization There is active development of visualization software within the greater scientific community for “omic” phenotypic data. For example, Cytoscape modules are widely used for visualization networks that incorporate co-expression, protein-protein interaction, and biochemical pathway data. The Generic Model Organism Database (GMOD) group (http://gmod.org/) also develops tools focused on genome and pathway data analysis and visualization. Gaggle (http://gaggle.systemsbiology.net/docs/) provides a way to share datasets across different analysis and visualization tools. However, the large number of underlying components that must be visualized in multicellular organisms or communities and the large scale of some phenomics datasets presents a computational challenge. Development of scalable, interactive methods is needed to view and interpret genome-scale phenotype data. Data presented via such tools should be seamlessly linked to other datasets and resources to allow more efficient data exploration and mining. ii. Image processing Image-based phenotyping offers a way to capture and extract not just morphological and gross developmental phenotype data, but also to interrogate physiological status through non- destructive close-range or remote-sensing technologies. A major roadblock is the lack of extensible algorithms for performing quantitation, feature extraction, and summarization. Assembly of the necessary data storage, transmission, and computational pipelines required is also difficult due to logistical impediments and computational requirements for advanced image analysis. These issues could be addressed by encouraging collaboration between image processing experts in the computer science and engineering domains and plant biologists through image classification contests, use-case marketplaces, and other networking opportunities. Adoption of extensible, high-throughput image analysis platforms such as the BISQUE system from The Center for Bioimage Informatics,UC Santa Barbara (http://www.bioimage.ucsb.edu/) may help address the logistical and scalability issues. NIFA-NSF Phenomics Workshop Report Page 27 D. Predictive Modeling Accurate prediction of an organism‟s phenotype from its genotype and environment is a stringent test of our understanding of a biological system. The ability to make such predictions has both fundamental scientific applications and practical benefits, providing a way to generate and test hypotheses about biological mechanisms as well as facilitating plant breeding and microbe engineering. i. Genomic selection Advancements in high-throughput genotyping are rapidly decreasing the cost of whole-genome genotyping while phenotyping costs are stable or increasing. This is driving the use of marker- assisted selection for major genes in plant and animal breeding. Commonly employed marker- assisted selection strategies, however, are not well suited for complex traits controlled by many loci of small effect (Meuwissen et al., 2001; McMullen et al., 2009). Genomic selection is an emerging technology complementary to marker-assisted selection, which uses phenotypes and thousands of genetic markers covering the entire genome to develop complex prediction models that are used to calculate genomic estimated breeding values for complex traits. These models can then be used to predict phenotypes based only on genotype in related populations. Because selections for multiple traits are based solely on whole-genome genotypes, multiple cycles of selection can be made without phenotyping, resulting in increased annual genetic gain (Heffner et al., 2011). Due to the complexities of such modeling, however, more research is needed to assess model accuracies for populations differing in linkage disequilibrium, distribution of QTL, size, marker density, and especially subpopulation structure. Further, these methods emphasize the importance of accurate, high throughput phenotyping for complex traits that drive the gains in efficiency. Because plants and animals differ in several aspects affecting GS strategies, the statistical approaches may be similar but the outcomes and applications will differ substantially. ii. Explicit mechanistic modeling While a predictive model lacking explicit biological mechanisms connecting genotype and environment with phenotype can serve a practical purpose in designing new crops, microbes and NIFA-NSF Phenomics Workshop Report Page 30 Upadyayula, N., Ware, D., Yates, H., Yu, J., Zhang, Z., Kresovich, S., and McMullen, M.D. (2009). The genetic architecture of maize flowering time. Science 325, 714-718. Clark, R.T., MacCurdy, R.B., Jung, J.K., Shaff, J.E., McCouch, S.R., Aneshansley, D.J., and Kochian, L.V. (2011). Three-dimensional root phenotyping with a novel imaging and software platform. Plant Physiol 156, 455-465. Cocuron, J.C., Lerouxel, O., Drakakaki, G., Alonso, A.P., Liepman, A.H., Keegstra, K., Raikhel, N., and Wilkerson, C.G. (2007). A gene from the cellulose synthase-like C family encodes a beta-1,4 glucan synthase. Proc Natl Acad Sci U S A 104, 8550-8555. Eshed, Y., and Zamir, D. (1995). An introgression line population of Lycopersicon pennellii in the cultivated tomato enables the identification and fine mapping of yield-associated QTL. Genetics 141, 1147-1162. Fang, S., Yan, X., and Liao, H. (2009). 3D reconstruction and dynamic modeling of root architecture in situ and its application to crop phosphorus research. Plant J 60, 1096-1108. Gerke, J., Lorenz, K., and Cohen, B. (2009). Genetic interactions between transcription factors cause natural variation in yeast. Science 323, 498-501. Gutierrez, M., Reynolds, M.P., and Klatt, A.R. (2010). Association of water spectral indices with plant and soil water relations in contrasting wheat genotypes. J. Exp. Bot. 61, 3291-3303. Heffner, E.L., Jannink, J.-L., and Sorrells, M.E. (2011). Genomic Selection Accuracy using Multifamily Prediction Models in a Wheat Breeding Program. Plant Genome 4, 65-75. Irie, S., Doi, S., Yorifuji, T., Takagi, M., and Yano, K. (1987). Nucleotide sequencing and characterization of the genes encoding benzene oxidation enzymes of Pseudomonas putida. J Bacteriol 169, 5174-5179. Jahnke, S., Menzel, M.I., van Dusschoten, D., Roeb, G.W., Buhler, J., Minwuyelet, S., Blumler, P., Temperton, V.M., Hombach, T., Streun, M., Beer, S., Khodaverdi, M., Ziemons, K., Coenen, H.H., and Schurr, U. (2009). Combined MRI-PET dissects dynamic changes in plant structures and functions. Plant J 59, 634-644. Jansen, M., Gilmer, F., Biskup, B., Nagel, K.A., Rascher, U., Fischbach, A., Briem, S., Dreissen, G., Tittmann, S., Braun, S., De Jaeger, I., Metzlaff, M., Schurr, U., Scharr, H., and Walter, A. (2009). Simultaneous phenotyping of leaf growth and chlorophyll fluorescence via GROWSCREEN FLUORO allows detection of stress tolerance in Arabidopsis thaliana and other rosette plants. Funct. Plant Biol. 36, 902-914. Keseler, I.M., Bonavides-Martinez, C., Collado-Vides, J., Gama-Castro, S., Gunsalus, R.P., Johnson, D.A., Krummenacker, M., Nolan, L.M., Paley, S., Paulsen, I.T., Peralta-Gil, M., Santos-Zavaleta, A., Shearer, A.G., and Karp, P.D. (2009). EcoCyc: a comprehensive view of Escherichia coli biology. Nucleic Acids Res 37, D464-470. Kilian, J., Whitehead, D., Horak, J., Wanke, D., Weinl, S., Batistic, O., D'Angelo, C., Bornberg-Bauer, E., Kudla, J., and Harter, K. (2007). The AtGenExpress NIFA-NSF Phenomics Workshop Report Page 31 global stress expression data set: protocols, evaluation and model data analysis of UV-B light, drought and cold stress responses. Plant J 50, 347-363. Lango Allen, H., Estrada, K., Lettre, G., Berndt, S.I., Weedon, M.N., Rivadeneira, F., Willer, C.J., Jackson, A.U., Vedantam, S., Raychaudhuri, S., Ferreira, T., Wood, A.R., Weyant, R.J., Segre, A.V., Speliotes, E.K., Wheeler, E., Soranzo, N., Park, J.H., Yang, J., Gudbjartsson, D., Heard-Costa, N.L., Randall, J.C., Qi, L., Vernon Smith, A., Magi, R., Pastinen, T., Liang, L., Heid, I.M., Luan, J., Thorleifsson, G., Winkler, T.W., Goddard, M.E., Sin Lo, K., Palmer, C., Workalemahu, T., Aulchenko, Y.S., Johansson, A., Zillikens, M.C., Feitosa, M.F., Esko, T., Johnson, T., Ketkar, S., Kraft, P., Mangino, M., Prokopenko, I., Absher, D., Albrecht, E., Ernst, F., Glazer, N.L., Hayward, C., Hottenga, J.J., Jacobs, K.B., Knowles, J.W., Kutalik, Z., Monda, K.L., Polasek, O., Preuss, M., Rayner, N.W., Robertson, N.R., Steinthorsdottir, V., Tyrer, J.P., Voight, B.F., Wiklund, F., Xu, J., Zhao, J.H., Nyholt, D.R., Pellikka, N., Perola, M., Perry, J.R., Surakka, I., Tammesoo, M.L., Altmaier, E.L., Amin, N., Aspelund, T., Bhangale, T., Boucher, G., Chasman, D.I., Chen, C., Coin, L., Cooper, M.N., Dixon, A.L., Gibson, Q., Grundberg, E., Hao, K., Juhani Junttila, M., Kaplan, L.M., Kettunen, J., Konig, I.R., Kwan, T., Lawrence, R.W., Levinson, D.F., Lorentzon, M., McKnight, B., Morris, A.P., Muller, M., Suh Ngwa, J., Purcell, S., Rafelt, S., Salem, R.M., Salvi, E., Sanna, S., Shi, J., Sovio, U., Thompson, J.R., Turchin, M.C., Vandenput, L., Verlaan, D.J., Vitart, V., White, C.C., Ziegler, A., Almgren, P., Balmforth, A.J., Campbell, H., Citterio, L., De Grandi, A., Dominiczak, A., Duan, J., Elliott, P., Elosua, R., Eriksson, J.G., Freimer, N.B., Geus, E.J., Glorioso, N., Haiqing, S., Hartikainen, A.L., Havulinna, A.S., Hicks, A.A., Hui, J., Igl, W., Illig, T., Jula, A., Kajantie, E., Kilpelainen, T.O., Koiranen, M., Kolcic, I., Koskinen, S., Kovacs, P., Laitinen, J., Liu, J., Lokki, M.L., Marusic, A., Maschio, A., Meitinger, T., Mulas, A., Pare, G., Parker, A.N., Peden, J.F., Petersmann, A., Pichler, I., Pietilainen, K.H., Pouta, A., Ridderstrale, M., Rotter, J.I., Sambrook, J.G., Sanders, A.R., Schmidt, C.O., Sinisalo, J., Smit, J.H., Stringham, H.M., Bragi Walters, G., Widen, E., Wild, S.H., Willemsen, G., Zagato, L., Zgaga, L., Zitting, P., Alavere, H., Farrall, M., McArdle, W.L., Nelis, M., Peters, M.J., Ripatti, S., van Meurs, J.B., Aben, K.K., Ardlie, K.G., Beckmann, J.S., Beilby, J.P., Bergman, R.N., Bergmann, S., Collins, F.S., Cusi, D., den Heijer, M., Eiriksdottir, G., Gejman, P.V., Hall, A.S., Hamsten, A., Huikuri, H.V., Iribarren, C., Kahonen, M., Kaprio, J., Kathiresan, S., Kiemeney, L., Kocher, T., Launer, L.J., Lehtimaki, T., Melander, O., Mosley, T.H., Jr., Musk, A.W., Nieminen, M.S., O'Donnell, C.J., Ohlsson, C., Oostra, B., Palmer, L.J., Raitakari, O., Ridker, P.M., Rioux, J.D., Rissanen, A., Rivolta, C., Schunkert, H., Shuldiner, A.R., Siscovick, D.S., Stumvoll, M., Tonjes, A., Tuomilehto, J., van Ommen, G.J., Viikari, J., Heath, A.C., Martin, N.G., Montgomery, G.W., Province, M.A., Kayser, M., Arnold, A.M., Atwood, L.D., Boerwinkle, E., Chanock, S.J., Deloukas, P., Gieger, C., Gronberg, H., Hall, P., Hattersley, A.T., Hengstenberg, C., Hoffman, W., Lathrop, G.M., Salomaa, V., Schreiber, S., Uda, M., Waterworth, D., Wright, A.F., Assimes, T.L., Barroso, I., Hofman, A., Mohlke, K.L., Boomsma, D.I., Caulfield, M.J., NIFA-NSF Phenomics Workshop Report Page 32 Cupples, L.A., Erdmann, J., Fox, C.S., Gudnason, V., Gyllensten, U., Harris, T.B., Hayes, R.B., Jarvelin, M.R., Mooser, V., Munroe, P.B., Ouwehand, W.H., Penninx, B.W., Pramstaller, P.P., Quertermous, T., Rudan, I., Samani, N.J., Spector, T.D., Volzke, H., Watkins, H., Wilson, J.F., Groop, L.C., Haritunians, T., Hu, F.B., Kaplan, R.C., Metspalu, A., North, K.E., Schlessinger, D., Wareham, N.J., Hunter, D.J., O'Connell, J.R., Strachan, D.P., Wichmann, H.E., Borecki, I.B., van Duijn, C.M., Schadt, E.E., Thorsteinsdottir, U., Peltonen, L., Uitterlinden, A.G., Visscher, P.M., Chatterjee, N., Loos, R.J., Boehnke, M., McCarthy, M.I., Ingelsson, E., Lindgren, C.M., Abecasis, G.R., Stefansson, K., Frayling, T.M., and Hirschhorn, J.N. (2010). Hundreds of variants clustered in genomic loci and biological pathways affect human height. Nature 467, 832-838. Last, R.L., Jones, A.D., and Shachar-Hill, Y. (2007). Towards the plant metabolome and beyond. Nat Rev Mol Cell Biol 8, 167-174. Lu, Y., Savage, L.J., Larson, M.D., Wilkerson, C.G., and Last, R.L. (2011). Chloroplast 2010: a database for large-scale phenotypic screening of Arabidopsis mutants. Plant Physiol 155, 1589-1600. Lu, Y., Savage, L.J., Ajjawi, I., Imre, K.M., Yoder, D.W., Benning, C., Dellapenna, D., Ohlrogge, J.B., Osteryoung, K.W., Weber, A.P., Wilkerson, C.G., and Last, R.L. (2008). New connections across pathways and cellular processes: industrialized mutant screening reveals novel associations between diverse phenotypes in Arabidopsis. Plant Physiol 146, 1482-1500. Massonnet, C., Vile, D., Fabre, J., Hannah, M.A., Caldana, C., Lisec, J., Beemster, G.T., Meyer, R.C., Messerli, G., Gronlund, J.T., Perkovic, J., Wigmore, E., May, S., Bevan, M.W., Meyer, C., Rubio-Diaz, S., Weigel, D., Micol, J.L., Buchanan-Wollaston, V., Fiorani, F., Walsh, S., Rinn, B., Gruissem, W., Hilson, P., Hennig, L., Willmitzer, L., and Granier, C. (2010). Probing the reproducibility of leaf growth and molecular phenotypes: a comparison of three Arabidopsis accessions cultivated in ten laboratories. Plant Physiol 152, 2142- 2157. McMullen, M.D., Kresovich, S., Villeda, H.S., Bradbury, P., Li, H., Sun, Q., Flint- Garcia, S., Thornsberry, J., Acharya, C., Bottoms, C., Brown, P., Browne, C., Eller, M., Guill, K., Harjes, C., Kroon, D., Lepak, N., Mitchell, S.E., Peterson, B., Pressoir, G., Romero, S., Oropeza Rosas, M., Salvo, S., Yates, H., Hanson, M., Jones, E., Smith, S., Glaubitz, J.C., Goodman, M., Ware, D., Holland, J.B., and Buckler, E.S. (2009). Genetic properties of the maize nested association mapping population. Science 325, 737-740. Meuwissen, T.H.E., Hayes, B.J., and Goddard, M.E. (2001). Prediction of Total Genetic Value Using Genome-Wide Dense Marker Maps. Genetics 157, 1819- 1829. Murchie, E.H., and Niyogi, K.K. (2011). Manipulation of photoprotection to improve plant photosynthesis. Plant Physiol 155, 86-92. Nichols, R.J., Sen, S., Choo, Y.J., Beltrao, P., Zietek, M., Chaba, R., Lee, S., Kazmierczak, K.M., Lee, K.J., Wong, A., Shales, M., Lovett, S., Winkler, M.E., Krogan, N.J., Typas, A., and Gross, C.A. (2011). Phenotypic landscape of a bacterial cell. Cell 144, 143-156. NIFA-NSF Phenomics Workshop Report Page 35 Barroso, I., Boehnke, M., Stefansson, K., North, K.E., McCarthy, M.I., Hirschhorn, J.N., Ingelsson, E., and Loos, R.J. (2010). Association analyses of 249,796 individuals reveal 18 new loci associated with body mass index. Nat Genet 42, 937-948. Velasco, L., Pérez-Vich, B., and Fernández-Martínez, J.M. (1999). Estimation of seed weight, oil content and fatty acid composition in intact single seeds of rapeseed (Brassica napus L.) by near-infrared reflectance spectroscopy. Euphytica 106, 79-85. Walter, A., Silk, W.K., and Schurr, U. (2009). Environmental effects on spatial and temporal patterns of leaf and root growth. Annu Rev Plant Biol 60, 279-304. Winzeler, E.A., Shoemaker, D.D., Astromoff, A., Liang, H., Anderson, K., Andre, B., Bangham, R., Benito, R., Boeke, J.D., Bussey, H., Chu, A.M., Connelly, C., Davis, K., Dietrich, F., Dow, S.W., El Bakkoury, M., Foury, F., Friend, S.H., Gentalen, E., Giaever, G., Hegemann, J.H., Jones, T., Laub, M., Liao, H., Liebundguth, N., Lockhart, D.J., Lucau-Danila, A., Lussier, M., M'Rabet, N., Menard, P., Mittmann, M., Pai, C., Rebischung, C., Revuelta, J.L., Riles, L., Roberts, C.J., Ross-MacDonald, P., Scherens, B., Snyder, M., Sookhai- Mahadeo, S., Storms, R.K., Veronneau, S., Voet, M., Volckaert, G., Ward, T.R., Wysocki, R., Yen, G.S., Yu, K., Zimmermann, K., Philippsen, P., Johnston, M., and Davis, R.W. (1999). Functional characterization of the S. cerevisiae genome by gene deletion and parallel analysis. Science 285, 901-906. Yu, J., Holland, J.B., McMullen, M.D., and Buckler, E.S. (2008). Genetic design and statistical power of nested association mapping in maize. Genetics 178, 539-551. NIFA-NSF Phenomics Workshop Report Page 36 Appendix: Opportunities and Challenges for Microbial and Plant Phenomics For both microbes and plants, critical decisions regarding experimental design and measurements usually can follow from a well defined starting point, which often derives from a practical agricultural, environmental or energy-related issue. A number of illustrative examples of goal- driven topics for phenomic research were mentioned in the main body of the White Paper and are provided here in more detail, as follows: The Microbiome The plant microbiome Microbial quorum sensing Plant fungal interactions in mycorrhizae Understanding microbial communities related to bioenergy Understanding microbial communities for environmental remediation Linking plant health to the rhizosphere microbiome The Plant Phenome General context Increasing yield Spatial and temporal understanding - disease resistance Root system developmental plasticity I. The Microbiome The Plant Microbiome. It has long been appreciated that plants and microbes form intimate associations. Studies of Agrobacterium tumefaciens and the crown gall tumors they cause spawned the entire field of plant genetic engineering (Valentine 2003). Symbiotic associations between rhizobia and legumes involve an intricate set of plant-bacterial communications resulting in root nodules that allow bacteria to receive carbon from the plant in exchange for fixed nitrogen (Downie 2010). Insights into how microbial pathogens interact with plants at the molecular level to induce innate immune responses or to cause disease are guiding the understanding of plant-pathogen co-evolution in natural and agricultural systems. Against this backdrop researchers understand that plants maintain complex communities of epiphytic, endophytic and rhizosphere microbes; plants have microbiomes. However, these communities are largely unstudied, and their influence on plant growth, yield and general health is essentially unknown. Filling this large knowledge gap will be a challenge. How do we define if a microbe is truly plant-associated? How do we identify microbial and plant phenotypes that are critical for maintaining stable interactions? How do we deal with extreme variation in plant-associated communities? Before these framework issues can be tackled we need to address the even more fundamental question of who is there. With recent technological advances in DNA sequencing and proteomics there is an unprecedented opportunity to inventory and functionally characterize plant microbiomes. Microbial quorum sensing. Cell-cell communication between microbes and their plant hosts. Many bacteria can perceive and respond to one another by a process called quorum NIFA-NSF Phenomics Workshop Report Page 37 sensing (Waters and Bassler 2005). This communication system influences colonization of plant and animal hosts by both symbionts and pathogens. Over 100 species of bacteria use small diffusible signaling molecules called N-acyl homoserinelactones (acyl-HSLs) to control gene expression by quorum sensing. Quorum sensing signal activity has been described for many plant-associated bacteria, including members of the Pseudomonas, Sinorhizobium, Mesorhizobium, and Bradyrhizobium genera (Dulla and Lindow 2009; Mathesius et al. 2003). It controls a variety of processes including motility, exopolysaccharide synthesis, plasmid transfer, and efficiencies of root nodulation and nitrogen-fixation. Acyl-HSLs can elicit responses in plants as well. In the legume Medicago truncatula, over 150 root proteins were differentially synthesized upon addition of acyl-HSLs. The identified proteins had predicted roles in host defense response, flavonoid metabolism, and hormone response, among others. The profiles of seed exudates also changed in response to acyl-HSL addition. In recent studies, the model plant Arabidopsis thaliana responded to acyl-HSL addition in a variety of ways; changes were seen in: root architecture, root hair development, and gene expression (Ortiz-Castro et al. 2008; von Rad et al. 2008). These results suggest that acyl-HSLs serve not only as intraspecies bacterial signals, but also as interkingdom signals to hosts. Plant fungal interactions in mycorrhizae - benefits to plants. Microbial interactions have the potential to greatly influence plant growth and phenotype. The earliest fossil records of fungi are in intimate association with plants (Taylor et al. 1995;Yuan et al. 2005) and phylogenetic reconstruction suggest that plant/fungal co-evolution has occurred over hundreds of millions of years (Taylor and Berbee 2006). Distinct plant-fungal associations, known as mycorrhizae, have evolved in roots to promote plant growth by mobilization of micronutrients in the soil and provide adaptive phenotypes such as drought tolerance (Gianinazzi et al. 2010). More recently, many other fungi that grow within plants (endophytic fungi) have been discovered that have the potential to influence plant phenotype (Saunders et al. 2010). For example, an endophytic fungus from a highly thermotolerant plant is necessary for plant thermotolerance; separation of the plant from the endophyte renders both organisms sensitive to high temperature (Redman et al. 2002). Another common endophyte of maize is able to ward of infection by pathogenic fungi (Lee et al. 2009). Comprehensive examination of plant endophytic fungi has only just begun, yet has great potential for the discovery of microbes which positively or negatively impact plant productivity and fitness. Understanding microbial communities related to bioenergy. Prior to NextGen sequencing methodologies, the challenge of characterizing microbial communities associated with plants was daunting. However, currently opportunities exist for determining how plants may affect microbial community structure of soils or plant surfaces and in turn how these microbes feed back to impact plant growth and development. Opportunities also exist for determining a comprehensive microbiome of degraded plant biomass in nature that will inform efforts aimed at use of these substrates for biofuels. The future of industrial microbiology lies in harnessing microbial communities to perform complex bioprocesses patterned on natural microbial communities (Sabra et al. 2010). Understanding microbial communities for environmental remediation. The phenotypic characterization of a plant microbiome is a daunting task. Is it possible to learn from others who have applied genomic and proteomic techniques to simpler natural systems? Jill Banfield at the NIFA-NSF Phenomics Workshop Report Page 40 Spatial and temporal understanding - disease resistance. A particularly appealing attribute of phenomic approaches is that automation allows for design of screens that better encompass responses over time. For example, rather than relying on „yes/no‟ phenotypes (typical of qualitative traits), novel approaches for improving plant disease resistance could take into account a continuum from disease to resistance (typical of multigenic or quantitative traits), and therefore, would rely on assessing large plant populations for the amount of disease development over time in response to multiple pathogens or types of pathogens. A specific example would the slow stem rusting genes for Puccinia graminis race Ug99 resistance. Both race-specific “gene- for-gene” and polygenic resistance seem to exist in wild wheat and barley germplasm. The task is to locate the genomic positions of resistance determinants and introgress those regions into elite backgrounds without dragging agronomically undesirable neighboring genes (Hiebert et al. 2011). A well coordinated effort combining large scale field tests of breeding materials with automated observation of carefully chosen small subsets of these materials may address critical aspects of plant architecture, reproductive development and disease progression. Automated conditions should simulate different environments (solar irradiation, humidity, soil types, etc) to maximize the relevance of large datasets to the selection of traits for specific geographic areas and climatic conditions. Variations may address problems such as flood, salinity or drought tolerance. Simultaneous measurement of physiological and morphological status over time (photosynthesis/respiration rates, growth rates, composition of root exudates, etc.) may provide holistic insights into fundamental mechanisms of plant response or adaptation. Root system developmental plasticity. An example of the complexity of morphological phenotyping is provided by root system responses to soil drying. Root system developmental responses (plasticity) in response to drought are complex (O‟Toole and Bland 1987); different types of roots respond differently, and responses will be determined not only by water status but other interacting variables. As an example, lateral (secondary) root proliferation (number and length) can be stimulated in response to mild water deficit (Read and Bartlett 1972; Jupp and Newman 1987), but inhibited as the soil dries further. Thus, the phenotype of lateral root proliferation can be evaluated only under specific mild water deficit conditions. In addition, the response will be spatially and temporally variable as the soil profile dries. Root hair proliferation can also be stimulated in response to water deficits (Vasellati et al. 2001), but many phenotyping systems may not have the precision to evaluate this response. The complexity of phenotyping root system developmental responses to soil drying is further compounded by other interacting variables such as soil type (Sponchiado et al. 1989), rate of drying, interaction with other stress conditions (soil strength, temperature, etc), interaction with nutritional status, and interaction with rhizosphere microorganisms. For example, phosphorous (P) deficiency also causes lateral root and root hair proliferation (Zhu and Lynch 2004; Zhu et al. 2010). Since soil drying causes decreased P mobility in the soil, soil drying and soil P status may have interacting effects that will also be altered by interacting mycorrhiza, which also influence P uptake (e.g., Zhu et al. 2005). The complexity of the question is further compounded by consideration of varying microbial populations, disease pressures, climate, soil types, etc in different regions. References Cited (Appendix) Denef VJ, Mueller RS, Banfield JF (2010) AMD biofilms: using model communities to study microbial evolution and ecological complexity in nature. ISME Journal 4: 599-610. NIFA-NSF Phenomics Workshop Report Page 41 Downie JA (2010) The roles of extracellular proteins, polysaccharides and signals in the interactions of rhizobia with legume roots. FEMS Microbiol Rev 34: 150-170. Dulla GF, Lindow SE (2009) Acyl-homoserine lactone-mediated cross talk among epiphytic bacteria modulates behavior of Pseudomonas syringae on leaves. ISME Journal 3: 825-834. Gianinazzi S, Gollotte A, Binet M-N, van Tuinen D, Redecker D, Wipf D (2010) Agroecology: the key role of arbuscular mycorrhizas in ecosystem services. Mycorrhiza 20: 519-530. Haas D, Defago G (2005). Biological control of soil-borne pathogens by fluorescent pseudomonads. Nature Reviews Microbiology 3: 307-319. Hiebert CW, Fetch TG, Zegeye T, Thomas JB, Somers DJ, Humphreys DG, McCallum BD, Cloutier S, Singh D, Knott DR (2011) Genetics and mapping of seedling resistance to Ug99 stem rust in Canadian wheat cultivars „Peace‟ and „AC Cadillac‟. Theor Appl Genetics 122: 143-149. Jupp AP, Newman EI (1987) Morphological and anatomical effects of severe drought on the roots of Lolium perenne L. Annals of Botany 105: 393-402. Lee K, Pan JJ, May G (2009) Endophytic Fusarium verticilliodes reduces disease severity caused by Ustilago maydis on maize. FEMS Microbiol Lett 299: 31-37. Mathesius U, Mulders S, Gao M, Teplitski M, Caetano-Anolles G, Rolfe BG, Bauer WD (2003) Extensive and specific responses of a eukaryote to bacterial quorum-sensing signals. Proc Natl Acad Sci USA 100: 1444-1449. Mendes R, Kruijt M, de Bruijn I, Dekkers E, van der Voort M, Schneider JH, Piceno YM, DeSantis TZ, Andersen GL, Bakker PA, Raaijmakers JM (2011) Deciphering the rhizosphere microbiome for disease-suppressive bacteria. Science 332: 1097-1100. O’Toole JC, Bland WL (1987) Genotypic variation in crop plant root systems. Advances in Agronomy 41: 91-145. Ortiz-Castro R, Martinez-Trujillo M, Lopez-Bucio J (2008) N-acyl-L-homoserine lactones: a class of bacterial quorum-sensing signals alter post-embryonic root development in Arabidopsis thaliana. Plant Cell Environ 31: 1497-1509. Read DJ, Bartlett EM (1972) The physiology of drought resistance in the soy-bean plant (Glycine max). I. The relationship between drought resistance and growth. Journal of Applied Ecology 9: 487-499. Redman R, Sheehan KB, Stout RG, Rodriguez TJ, Henson JM (2002) Thermotolerance generated by plant/fungal symbiosis. Science 298: 1581. Sabra W, Dietz D, Tjahjasari D, Zeng A-P (2010) Biosystems analysis and engineering of microbial consortia for industrial biotechnology. Eng Life Sci 10: 407-421. Saunders M, Glenn AE, Kohn LM (2010) Exploring the evolutionary ecology of fungal endophytes in agricultural systems: using functional traits to reveal mechanisms in community processes. Evolutionary Applications 3: 525–537. Sponchiado BN, White JW, Castillo JA, Jones PG (1989) Root growth of four common bean cultivars in relation to drought tolerance in environments with contrasting soil types. Experimental Agriculture 25: 249-257. Taylor JW, Berbee ML (2006) Dating divergences in the fungal tree of life: review and new analyses. Mycologia 98: 838-849. Taylor TN, Remy W, Hass H, Kerp H (1995) Fossil arbuscular mycorrhizae from the Early Devonian. Mycologia 87: 560-573. NIFA-NSF Phenomics Workshop Report Page 42 Valentine L (2003) Agrobacterium tumefaciens and the plant: the David and Goliath of modern genetics. Plant Physiol 133: 948-955. Vasellati V, Oesterheld M, Medan D, Loreti J (2001) Effects of flooding and drought on the anatomy of Paspalum dilatum. Annals of Botany 88: 355-360. von Rad U, Klein I, Dobrev PI, Kottova J, Zazimalova E, Fekete A, Hartmann A, Schmitt- Kopplin P, Durner J (2008) Response of Arabidopsis thaliana to N-hexanoyl-DL- homoserine-lactone, a bacterial quorum sensing molecule produced in the rhizosphere. Planta 229: 73-85. Waters CM, Bassler BL (2005) Quorum sensing: cell-to-cell communication in bacteria. Annu Rev Cell Dev Biol 21: 319-346. Yuan X, Xiao S, Taylor TN (2005) Lichen-like symbiosis 600 million years ago. Science 308: 1017-1020. Zhu J, Kaeppler SM, Lynch JP (2005) Topsoil foraging and phosphorus acquisition efficiency in maize (Zea mays). Functional Plant Biology 32: 749-762. Zhu J, Lynch JP (2004) The contribution of lateral rooting to phosphorus acquisition efficiency in maize (Zea mays) seedlings. Functional Plant Biology 31: 949-958. Zhu J, Zhang C, Lynch JP (2010) The utility of phenotypic plasticity of root hair length for phosphorus acquisition. Functional Plant Biology 37: 313-322.