Performing Department
Plant Pathology
Non Technical Summary
Modern plant breeding fundamentally relies on the observation, acquisition, and application of data to make informed decisions about the selection and release of varieties for farmers around the world. Although there have been improvements in both genetics and agronomics, the design and collection of data within modern field experiments has seen minimal advancement over the past decades. Individual plots are still primarily referenced by relative location in a row and column design and data is collected within the constraints and limitations of this layout. With state-of-the-art geo-position systems, we can now understand, collect, and analyze phenotypic data not only in a generic grid pattern, but in the dynamic geospatial context in which it actually exists. We propose a transformative conceptual shift in both the type and means of data collection for plant breeding field trials. To enable the collection, storage, analysis, and interpretation of phenotypic data on a geospatial basis, we will define open geospatial data structures, integrate support and analytical tools into Cassavabase, and create the breeder-friendly tools necessary to collect and utilize geospatial data in the field. Increasing the integrity, accessibility, and amount of data available to breeders will allow for more accurate and specific selection decisions with increased speed. Simplifying data collection and analysis workflows for breeders will enable the increased data integrity on larger population sizes required to produce adequate genetic gain to feed the world. Developing a framework for geospatial data will benefit all life sciences that rely on field studies.
Animal Health Component
0%
Research Effort Categories
Basic
20%
Applied
40%
Developmental
40%
Goals / Objectives
Goal: We aim to develop the tools and systems necessary to shift phenotypic data collection in plant breeding from an entry-basis to a geospatial-basis. To meet this goal, we will implement the following specific objectives for the project.Specific Objectives:Develop the apps and database structures needed for collection and use of geospatial data in plant breeding field trials.Create geospatial data models, database schema, and analysis tools.Implement 'geo-navigation' into the Field Book app by integrating algorithms that identify plants/plots based on user location.
Project Methods
1. Apps and algorithms for geospatial data collectionSurvey appThe Survey app will be the primary means by which breeders collect coordinate-based data. Collected plot coordinates will have dual purposes: 1) ground truth points for high throughput phenotyping research and 2) plot points for in-field plot navigation. We will finalize Survey development, evaluate its performance in breeding field environments, and release it to the breeding community. The current, unreleased version of Survey utilizes a deprecated method for interacting with external GNSS hardware and will be updated to ensure compatibility with future versions of Android. Survey will be fully-evaluated in regionally disparate breeding environments and distinct crops to ensure broad compatibility with different breeding program organizational styles as well as the diverse Android hardware ecosystem. As the BrAPI standards are updated to implement geospatial data (Objective 2), we will incorporate BrAPI data types into Survey along with the API calls necessary to directly transfer collected coordinate data to BrAPI compatible databases.Plot navigation algorithmsOur impact zone algorithm will be evaluated in active breeding trials to determine any modifications that must be made. A functional interface will be added that allows breeders to adjust the parameters of the algorithm to their specific breeding program. The final impact zone algorithm will be developed and delivered as an Android service which will allow simple integration into other field-based mobile applications.Smart Glasses interfacesTo facilitate the transition from on-device navigation to automatic plot navigation, we will implement our smart glasses interface into Field Book. This interface currently exists in a fork of the Field Book project and thus cannot be used or evaluated by breeders. Introducing this interface will promote the adoption and evaluation of smart glasses, providing us with the critical feedback needed to improve the interface.The current implementation of the smart glasses interface displays information from Field Book and can utilize the user's voice for navigation. We will integrate an additional voice interface that allows users to collect data directly from the glasses. This will require a functional and efficient workflow that is simultaneously protected from accidental data entry. Defining this workflow in such a way that it maintains the current efficiency of data collection on Field Book will be a priority.2. Data models, database schema, visualization, and analyticsGeospatial data modelsTo maximize the utility of geospatial data in breeding, data must be easy to exchange between different clients, whether those are servers or external hardware used for data collection. This will be accomplished by expanding the BrAPI standard with geospatial data types. Most objects in BrAPI don't currently include geospatial information and thus the proper implementation has not been discussed by the broader development community. Adding this metadata will facilitate the spatial identification of various experimental units including plots, plants, and fields. Standard JSON notation does exist for geospatial data, and we will work with the BrAPI coordinator to integrate these standards into BrAPI.Database integration and visualizationBreeding databases need the ability to store and query geospatial data. We will adapt the Cassavabase database system to accommodate these data types by modifying the relational backend. Every entity in the database (plant, plot, field, etc.) has a unique data object associated with a unique identifier. Relational relationships will allow us to add more data to these items including geospatial information. These modifications will make it possible to import data from our geospatial tools, query geospatial information, create dynamic map overlays that utilize the geospatial information to display phenotypic data, and visualize geospatial information based on user queries, such as color scales for certain phenotypic measurements, field conditions, and other patterns of interest. Data will be loaded either directly through BrAPI or via file uploads.Geospatial data analysis for breedingA novel functionality that we will enable by the new geospatial nature of the data within these databases is the analysis and correction of spatial field effects using geographic information rather than on the row-column grid design of the field experiment. As we have already demonstrated the power of these approaches, the geographic weighted regression for correction of spatial effects will be implemented as a user accessible tool for analysis within the database for initial primary analysis of raw data from the field.3. Geo-navigation in Field BookTo effectively increase the accuracy and integrity of data collected by plant breeders, we will develop a system that allows for in-field navigation when collecting phenotypic data in a breeding program.We will integrate the impact zone algorithm from objective 1 into Field Book which will allow for the identification and display of the specific plot a breeder is viewing based on their current location and directional bearing. We will add support for the data models from objective 2 to ensure that data transfer between Field Book and Cassavabase is seamless and incorporates the rich metadata that will be available.Field Book has evolved to be a fundamental component of breeding programs across the world. Since these changes will require major changes to the backend database, we will extensively evaluate the app for performance prior to release to ensure the broad network of breeders who rely on Field Book have no interruption to their phenotyping. Release of the additional features will be evaluated by collaborating breeders first through the Field Book beta versions that are already in place before a wider release to the public.