Source: MICHIGAN STATE UNIV submitted to NRP
PRECISION AGRICULTURE TOOLS FOR OPTIMIZING ALFALFA PRODUCTION AND MARKETING
Sponsoring Institution
National Institute of Food and Agriculture
Project Status
COMPLETE
Funding Source
Reporting Frequency
Annual
Accession No.
1017002
Grant No.
2018-70005-28738
Cumulative Award Amt.
$299,977.00
Proposal No.
2018-03912
Multistate No.
(N/A)
Project Start Date
Sep 1, 2018
Project End Date
Aug 31, 2023
Grant Year
2018
Program Code
[AFRP]- Alfalfa and Forage Program
Recipient Organization
MICHIGAN STATE UNIV
(N/A)
EAST LANSING,MI 48824
Performing Department
PLANT SOIL MICROBIAL
Non Technical Summary
Alfalfa was the third most valuable field crop in the USA in 2017, but acreage has been decreasing over the last ten years, primarily losing ground to the greater feed energy production per acre of corn silage. This land use change has societal implications because alfalfa provides important soil-related ecosystem services when included in crop rotations, such as biological N fixation, soil carbon sequestration, reduced soil erosion, and improved soil structure, none of which are provided by corn. High-tech, big-data, and precision agricultural technologies are available to assist growers of other field crops, but methodologies suitable for alfalfa production have generally lagged behind. Alfalfa's multiple harvests per year require a constant compromise between yield and quality because the feeding value of a perennial forage declines as yield increases. Timing of each harvest is therefore critical. Cost-effective remote sensing technologies that can predict yield and quality potential of field of alfalfa in real time could be invaluable to informing harvest decisions and thus improving management and profitability of alfalfa. We propose to begin the process of developing pre-harvest remote sensing technologies for alfalfa by building a forage spectral library that can ultimately be used for drone- and satellite-based remote sensing systems, high performance computing, and modeling to permit the rapid and efficient application of precision agriculture solutions on a landscape scale in diverse managed alfalfa and forage systems. On the other end of the value chain from growers, we have stakeholders who feed alfalfa, to whom post-harvest forage quality is the most important factor. Near-infrared spectroscopy (NIRS) evolved as a laboratory-scale technology for assessing post-harvest forage quality, but instrumentation was too large and costly to be feasible for on-farm use. Today, NIRS technology is marketed directly to farmers in units as small as a cell phone. Farmer are bombarded with advertising for these new devices, but there is little objective data comparing accuracy or precision of these products versus laboratory-scale NIRS instruments. The primary uses for post-harvest analysis of alfalfa nutritive value are hay pricing and ration balancing. Being able to predict forage constituents on-farm can potentially save time and money over shipping samples to a testing lab, but a small accuracy error can potentially have large negative impacts on livestock performance. Therefore, there is a pressing need for objective comparative information to assist in use of these new devices. We will collect this data, develop educational programs to train students and farmers in technology use, and develop new calibrations designed for hand-held units. The project will conclude with a conference that brings together forage -based precision agriculture technology.This project will bring the precision agriculture revolution to alfalfa management by developing and adapting cutting edge spectroscopic technologies for use on-farm. The development of inexpensive, accurate, real-time producer-accessible forage mapping tools and forage analysis is a game-changing improvement over the current situation. By enhancing profitability, the project should increase alfalfa acreage along with its positive ecosystem services, thus addressing an important component of the broader question of how to sustainably feed the ever-growing world population in the face of increasing global demand for meat and dairy.
Animal Health Component
50%
Research Effort Categories
Basic
50%
Applied
50%
Developmental
(N/A)
Classification

Knowledge Area (KA)Subject of Investigation (SOI)Field of Science (FOS)Percent
2051640106050%
2051640201040%
2051640200010%
Goals / Objectives
Our project goal is to address the explosion of precision technology related to two emerging issues: 1) effective use of existing hand-held and small farm-scale benchtop units to estimate post-harvest forage nutritive value on farm, and 2) development of new landscape-scale remote sensing technologies for pre-harvest estimation of alfalfa yield and quality.Specifically, our objectives are:identify spectral signatures for pre-harvest alfalfa and its primary companion grasses,use spectral unmixing algorithms to determine pre-harvest yield, nutritive composition, and abundance of alfalfa and grass in a mixture,develop extension materials to assist in adoption of precision technologies in alfalfa production.
Project Methods
Objective 1. Identify spectral signatures for pre-harvest alfalfa, companion grasses, and mixtures:Task 1.1. Collecting spectral and supporting data from alfalfa and grass monocultures: We will utilize a high spectral-resolution field spectroradiometer with a spectral range of 350-2500 nm and with bandwidth resolution of 3 nm at 700 nm and 8 nm at 1400/2100 nm. Spectra will be collected from existing alfalfa and grass variety test plots and alfalfa management trials at three locations representing the range of growing conditions in Michigan (East Lansing, Lake City, and Chatham). Sampled plots will be selected from existing test plots in Michigan to provide a cross-section of commercially important traits and characteristics, including stand ages, fall dormancy ratings, pest and disease resistance ratings, glyphosate resistance, and reduced lignin traits. Supporting data, including weather conditions, forage height, chlorophyll content, leaf area index, photosynthetically active radiation, soil moisture, vegetation coverage, forage yield, moisture content, nutritive composition, and phenological stage will be obtained from each plot. Nutritive constituents will be predicted using near infrared reflectance spectroscopy (NIRS, Foss 6500, Eden Prairie, MN) and a universal mixed forage equation (NIRS Forage and Feed Testing Consortium, Hillsboro, WI) and validated using laboratory chemistry for 20 randomly selected samples per year. This prediction-validation method reduces cost of nutritive value analysis and can produce acceptable prediction equations in remote sensing applications.Task 1.2. Performing statistical analysis to predict alfalfa yield and nutritive composition, and to discriminate forage species: The pre-harvest predictions of greatest interest to alfalfa growers are yield and nutritive value because these help determine when to harvest. We will use partial least squares regression (PLSR) with a stratified random sampling of 70% training and 30% test data to test if we can accurately predict alfalfa yield and nutritive constituents from the pre-harvest plant spectra. This process is similar to generation of global equations for NIRS constituent predictions in testing labs, but the uncontrolled nature of the field environment presents a greater challenge.Task 1.3. Designing the forage spectral library: We propose using NoSQL (Not Only SQL) databases. Using NoSQL databases is gaining momentum with companies such as Microsoft, Oracle, and Amazon. Cyber-Platforms based on NoSQL databases, such as MongoDB, provide a flexible and dynamic schema that can be changed on the fly. In this platform, records with different sets of properties can be stored, which is very useful for spectral databases containing landscape- and field-scale data with different sets of attributes for each spectra.Objective 2. Use spectral unmixing algorithms to determine pre-harvest yield, nutritive composition, and abundance of alfalfa and grass in a mixture:Task 2.1. Establishing and maintaining alfalfa mixtures for spectra collection: Plots will be established in spring 2019 at Hickory Corners and East Lansing, MI, to provide two different growing environments for evaluation of spectral prediction of pre-harvest yield and nutritive composition in alfalfa and alfalfa-grass mixtures. This is small plot trial with a randomized complete block design and four replications per location. Forage treatments include three alfalfa varieties (KF4020, KF401B, KF425HD) that vary in morphological and biochemical traits and three grass species ('Echelon' orchardgrass, 'Liherold' meadow fescue, and 'Kora' tall fescue) planted alone and in all possible binary combinations of grass and alfalfa for a total of 15 treatments per replication. Binary mixtures will be planted at a seeding ratio of 75% alfalfa and 25% grass based on the seeding rates for each monoculture.Task 2.2. Collecting spectra and supporting data of alfalfa mixtures: Spectra, forage nutritive composition, and supporting data will be collected from the alfalfa-grass mixtures at the first and second cutting in the establishment (2019) and first production year (2020), using the same methodologies described in Objective 1. In addition, we will manually determine botanical composition of each species in the mixtures. Task 2.3. Developing/validating hyperspectral unmixing algorithms in order to determine yield, nutritive constituents, and the abundance of forage species in the mixture: We will follow the same procedure as Effort 1.2 to develop equations to predict pre-harvest forage nutritive value and yield. Due to possible challenges imposed by alfalfa mixtures, various linear and nonlinear unmixing algorithms will be tested. Because we will also have continuous variables (fractional cover of each species predicted) we will calculate correlations between observed and predicted percent cover for species, functional group, and community.Objective 3. Develop extension materials to assist in adoption of precision agriculture technologies for alfalfa. We will create extension tools to assist growers and producers with assessment of alfalfa composition using existing on-farm technologies and in understanding how to use farm and landscape scale remote sensing technologies. Task 3.1. Use NIR optical sensing technologies to increase knowledge and skills for current producers and future stakeholders (undergraduate ag students) in forage crop quality testing. Mobile optical sensor instruments based on NIRS will be integrated into hay production workshops and field days in Wisconsin and Michigan. Use of devices in workshops will support knowledge transfer by demonstrating how to predicting forage quality constituents (CP, fiber, and other quality parameters) in real time for a range of forages differing in variety, preservation type, maturity, moisture, and weathering. This will help producers better understand the key to alfalfa crop optimization and forage profitability: that their profit depends on energy and protein content as much as it does on dry matter yield. Task 3.2. Development of a global NIRS calibration for "as-fed" whole moist alfalfa hay, haylage, and green chop. Calibration equations for as-fed (undried) forage samples are not readily available for NIRS measurements. Undergraduate students will collect data and develop these equations for alfalfa and alfalfa/grass mixed hay, haylage and greenchop as part of their training. The alfalfa varieties used will include conventional types as well as genetically modified and conventionally-bred reduced lignin alfalfa. As-fed samples will be scanned for NIR spectra using the Unity 2600XT NIR SpectraStar unit operating in the wavelength range 680-2600 nm/1nm increment. Sample constituents will be CP, in vitro true digestibility at 48 h, neutral detergent fiber, and acid detergent fiber. Reference values will be generated by drying and grinding fresh samples after scanning and determining nutritive value using appropriate laboratory chemical methods. Calibration accuracy will be evaluated using standard error of prediction (SEP) and coefficient of determination (??2). These equations will be loaded into the portable equipment and made available to producers.Task 3.3. Conduct a regional workshop on the use of remote sensing in alfalfa and forage production. In the final year of the project, we will organize and conduct a regional workshop on the use of on-farm optical devices and regional remote sensing in alfalfa and forage production.Efforts employed to transfer knowledge gained to stakeholders will include scientific papers and presentation, field days, conferences, producer workshops, classroom and laboratory instruction, and extension publications. Efforts will be evaluated using science citations, extension program evaluation metrics, and class evaluations and testing scores.

Progress 09/01/18 to 08/31/23

Outputs
Target Audience:Forage growers, livestock producers, extension educators, ag industry professionals, consumers, undergraduate students, graduate students, post-docs Changes/Problems:The planned in-person conference/workshop was not held due to covid-driven disruptions and changes in the way people consume information. Funds that would have been used to support this task are being returned to NIFA. As peer reviewed publications are finalized, the research team still plans to present results in a webinar series planned for 2024 outside the reporting time of the project. What opportunities for training and professional development has the project provided?Throughout the project, it has provided training and educational opportunities for numerous undergraduate and graduate students to learn about remote sensing technology. How have the results been disseminated to communities of interest?In its final year, results were presented by an undergraduate at a research forum and a report. Journal papers are being prepared for future publication. What do you plan to do during the next reporting period to accomplish the goals? Nothing Reported

Impacts
What was accomplished under these goals? Objective 1. Identify spectral signatures for pre-harvest alfalfa and its primary companion grasses. Key outcomes or other accomplishments realized: Objective 1 was complete in 2021. Objective 2. Use spectral unmixing algorithms to determine pre-harvest yield, nutritive composition, and abundance of alfalfa and grass in a mixture. Data collected: Data and samples were collected in project year 4 and analyzed during 2022/2023. Each spectrum was then visually assigned to fractional cover of species in plots. Spectra were also compared to ancillary data including nutritional composition of the forages (crude protein, neutral and acid detergent fiber, and cell wall digestibility). Summary statistics and discussion of results: We are continuing to work on classifying and unmixing the spectra collected over individual vegetation types and mixes. Initial results suggest that separating individual spectra for alfalfa versus grasses is straightforward and highly accurate using machine learning approaches like random forests. Separating different varieties of alfalfa and species of grass depends on the amount of noise in the spectral data and on the growth phase of the plants. Unmixing similarly shows promising initial results for separating alfalfa from grasses, while separating out more subtle variations and identifying unusual endmembers (e.g. weeds, flowers) is more challenging. We anticipate producing a publication with these findings in 2024. Key outcomes or other accomplishments realized: Initial results indicate that hyperspectral analysis will be accurate for determining the abundance of plant functional groups such as grass versus legume in forage mixtures but that differentiating grass species will be more difficult. Objective 3. Develop extension materials to assist in adoption of precision technologies in alfalfa production. Major activities completed / experiments conducted: A report/white paper on comparison of handheld versus benchtop NIRS instruments was prepared. Key outcomes or other accomplishments realized: The handheld NIR spectrophometer did not produce comparable values to the standard benchtop unit for forage moisture or fiber, but hand-held predictions were similar to the benchtop NIRS for crude protein. These results indicate hand-held NIRS units may not be accurate enough for ration formulation.

Publications

  • Type: Other Status: Published Year Published: 2023 Citation: Bontrager, J., J. Paling, and K. Cassida. 2023. Analyzing yield and forage quality impacts of alfalfa grass forage crop mixtures. MSU University Undergraduate Research & Arts Forum, April 14, 2023, East Lansing, MI.


Progress 09/01/21 to 08/31/22

Outputs
Target Audience:Forage growers, livestock producers, extension educators, agricultural industry professionals, consumers, undergraduate students, graduate students, post-docs Changes/Problems:Based on the success of online programming discovered by the scientific community by necessity during covid and the rapid development of remote sensing technology during that period, the team decided to present results in a webinar series published online instead of holding an in-person event. We anticipate this will be better received than an in-person workshop and ultimately reach a bigger audience. What opportunities for training and professional development has the project provided?The project provided training and educational opportunities for undergraduate and graduate student workers. How have the results been disseminated to communities of interest?Results were presented in both presentation and poster format at the North American Alfalfa Improvement Conference attended by the international community of alfalfa scientists. Results were also presented at the Great Lakes Forage and Grazing Conference. Concepts and results were included in college courses at both universities. What do you plan to do during the next reporting period to accomplish the goals? The unmixing algorithm will be completed and we will conduct a webinar series in lieu of an in-person conference.

Impacts
What was accomplished under these goals? Objective 1. Objective 1 was completed in 2021. Objective 2. Use spectral unmixing algorithms to determine pre-harvest yield, nutritive composition, and abundance of alfalfa and grass in a mixture. Data collected: Spectral and ancillary data (yield, nutrient value, species abundance, and soil moisture) were collected from forage in first and second cuttings of field plots at two locations for a total of 120 plot observations. Summary statistics and discussion of results: Analysis of spectral data will be completed in 2023. Key outcomes or other accomplishments realized: Nothing to report. Objective 3. Develop extension materials to assist in adoption of precision technologies in alfalfa production. Major activities completed / experiments conducted: Spectra were collected from dry monoculture and mixed-species samples from two cuttings at each Michigan location and compared to benchtop NIRS results. Key outcomes or other accomplishments realized: Analysis of the instrument comparison data is ongoing.

Publications

  • Type: Conference Papers and Presentations Status: Published Year Published: 2022 Citation: Cassida K, AP Nejadhashemi, K Dahlin, Y Newman, and *B Saravi. 2022. Precision agriculture tools for optimizing alfalfa production and marketing. Proc. North Amer. Alfalfa Improvement Conf. Abstract online https://naaic.org/Meetings/National/2022meeting/LP1%20-%20Cassida%20-%20Precision%20Agriculture%20Tools%20for%20Optimizing%20Alfalfa%20Production%20&%20Marketing.pdf/
  • Type: Other Status: Published Year Published: 2022 Citation: Cassida K, AP Nejadhashemi, K Dahlin, Y Newman, and *B Saravi. 2022. Precision agriculture tools for optimizing alfalfa production and marketing. Online https://naaic.org/Meetings/National/2022meeting/Videos/Cassida.mp4/ North Amer. Alfalfa Improvement Conference, Lansing, Michigan, June 7-9, 2022. (video)
  • Type: Conference Papers and Presentations Status: Published Year Published: 2022 Citation: Cassida, K. 2022. Forage Research Update: Grass-Legume Mixtures. Great Lakes Forage & Grazing Conference, St. John, MI. Mar. 17, 2022.


Progress 09/01/20 to 08/31/21

Outputs
Target Audience:Forage growers, livestock producers, extension educators, ag industry professionals, consumers, undergraduate students, graduate students, post-docs Changes/Problems:This project requested and received an additional NCE. Due to covid-related travel trestrictions, we were unable to complete field work originally planned for 2021 but this was rescheduled to 2022 when we will resume the field work portion of the project. What opportunities for training and professional development has the project provided? Nothing Reported How have the results been disseminated to communities of interest?Results were presented at an online producer meeting in Michigan. Concepts and results were included in college courses at both universities. What do you plan to do during the next reporting period to accomplish the goals?We will proceed as planned with the project, but now two years behind schedule because of Covid disruptions.

Impacts
What was accomplished under these goals? Objective 1. Identify spectral signatures for pre-harvest alfalfa and its primary companion grasses. Results: A website was designed and published to house the spectral data collected in 2019. Summary statistics and discussion of results: Spectral signatures for monocultures were unique for alfalfa and grasses, supporting our hypothesis that we will be able to distinguish them in mixtures in the subsequent work. Key outcomes or other accomplishments realized: This completes Objective 1. Objective 2. Use spectral unmixing algorithms to determine pre-harvest yield, nutritive composition, and abundance of alfalfa and grass in a mixture. Data collected: Spectral data collection was delayed until 2022 because of covid-related travel restrictions in fall 2020 and a university freeze on hiring graduate students through 2021. Nevertheless, forage yield, nutritive value, and species abundance in mixtures was measured according to protocol at the East Lansing Site in fall 2020 and both sites for the entire 2021 growing season. Summary statistics and discussion of results: Binary mixtures of alfalfa and orchardgrass, tall fescue, or meadow fescue yielded similarly to alfalfa monocultures but 70, 81, and 67% more than monocultures of each grass, respectively. Mixtures improved fiber digestibility over alfalfa monocultures, confirming the importance of grass admixtures for improving forage nutritive value. Key outcomes or other accomplishments realized: Maintenance of plots with standard harvest intervals was important even with delayed spectral data collection because it maintained the normal productive potential of the plots for later spectral measurements. Objective 3. Develop extension materials to assist in adoption of precision technologies in alfalfa production. Major activities completed /experiments conducted: At MSU, goals of the project and 2020 forage yield and nutritive value of the mixtures were reported in an online producer presentation in January 2021, beginning the process of disseminating results to stakeholders. At UW-RF, a key effort to develop a fresh forage-forage NIRS calibration failed to produce a valid calibration. This is consistent with results from other research groups because high water content interferes with NIRS. Consequently, we ended our efforts to make a fresh-forage calibration. Instead we will focus on comparing hand-held and benchtop NIRS instruments for dry forage measurements. Key outcomes or other accomplishments realized: Fresh-forage NIRS calibrations are not accurate enough to meet producer needs.

Publications

  • Type: Websites Status: Published Year Published: 2020 Citation: Saravi, B., and A.P. Nejadhashemi. 2020. Forage Spectral Library. Online Sept. 2020, https://dsiweb.cse.msu.edu/fsd/
  • Type: Other Status: Published Year Published: 2021 Citation: 2) Cassida, K. 2021. Mix It Up in the Hayfield. Virtual Crop & IPM Update for the Upper Peninsula, Jan. 15, 2021. Online 1/15/21, https://www.facebook.com/1477467478999687/videos/728482124456555/?__so__=watchlist&__rv__=video_home_www_playlist_video_list


Progress 09/01/19 to 08/31/20

Outputs
Target Audience: Nothing Reported Changes/Problems:COVID-19-related university closures and restrictions have delayed all scheduled work. NCE have been requested for all funded projects. What opportunities for training and professional development has the project provided? Nothing Reported 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?Work will continue on all goals as allowed while following all state and university COVID-19 protocols

Impacts
What was accomplished under these goals? Title: Precision Agriculture Tools for Optimizing Alfalfa Production and Marketing. Collaborators: Y. Newman, P. Nejadhashemi, K. Dahlin. Data Collected: Data collection in 2020 was paused due to COVID-19 travel restrictions. An NCE has been requested so that data collection can resume in 2021. The conference planned for 2021 will be delayed to 2022. During Summary statistics and discussion of results: An online database (URL) was built to house the spectral data collected in 2019 and is currently being beta-tested. Key Outcomes and other accomplishments realized: Not yet available.

Publications


    Progress 09/01/18 to 08/31/19

    Outputs
    Target Audience:Forage growers, livestock producers, extension educators, ag industry professionals, consumers, undergraduate students, graduate students, post-docs Changes/Problems: Nothing Reported What opportunities for training and professional development has the project provided? Nothing Reported How have the results been disseminated to communities of interest?The project was included in a research field day (see other products). What do you plan to do during the next reporting period to accomplish the goals?We will proceed as planned with the project.

    Impacts
    What was accomplished under these goals? Objective 1. Identify spectral signatures for pre-harvest alfalfa and its primary companion grasses. Data collected: Spectral data and accompanying stand characteristics (yield, quality, soil moisture, light interception) were collected from alfalfa, orchardgrass, and tall fescue plots. Summary statistics and discussion of results: Too early to report. Key outcomes or other accomplishments realized: Too early to report. Objective 2. Use spectral unmixing algorithms to determine pre-harvest yield, nutritive composition, and abundance of alfalfa and grass in a mixture. Data collected: Plots with mixtures of alfalfa and grass were established at two sites. Adverse weather delayed planting, so collection of spectra from these plots is delayed until 2020. Summary statistics and discussion of results: Too early to report. Key outcomes or other accomplishments realized: Too early to report. Objective 3. Develop extension materials to assist in adoption of precision technologies in alfalfa production. Major activities completed / experiments conducted: Educational activities were conducted at a research field day. Key outcomes or other accomplishments realized: Too early to report.

    Publications