Source: CORNELL UNIVERSITY submitted to NRP
DEVELOPMENT OF A WEED EMERGENCE MODEL FOR THE NORTHEASTERN UNITED STATES
Sponsoring Institution
National Institute of Food and Agriculture
Project Status
COMPLETE
Funding Source
Reporting Frequency
Annual
Accession No.
1018540
Grant No.
(N/A)
Cumulative Award Amt.
(N/A)
Proposal No.
(N/A)
Multistate No.
NE-1838
Project Start Date
Dec 12, 2018
Project End Date
Sep 30, 2022
Grant Year
(N/A)
Program Code
[(N/A)]- (N/A)
Recipient Organization
CORNELL UNIVERSITY
(N/A)
ITHACA,NY 14853
Performing Department
Crop & Soil Sciences
Non Technical Summary
Unlike crops, which have been selected for uniform emergence, weed species have evolved variability in timing of their emergence; even seeds maturing on the same plant may germinate at different times. This "bet-hedging" strategy, with which a weed avoids putting all its "seed in one basket" of emergence timing, enables weeds to escape control measures that are applied at the "wrong" time. Post-emergence management carried out too early, i.e. before most problem weeds have emerged, will yield low returns for the effort, investment, and ecological cost of the management (herbicide off-target effects, soil compaction, etc.), as weed seeds that have yet to germinate are often unaffected. Providing seedling emergence information so that farmers can effectively time their weed management operations can increase efficacy of control, reduce labor costs, and minimize any negative environmental impacts (e.g. reduce the likelihood that repeat applications of an herbicide or cultivation may be required for late germinating/emerging weeds). There is, therefore, an urgent need for the development of time-specific weed management tools to help address the frequently asked, yet to be answered, question of when is the "right" time to control weeds? Weed seedling emergence is a complex process regulated by a multitude of internal (e.g. species-specific parameters such as base temperature, base water potential) and environmental (e.g. soil temperature and moisture) factors. A range of modeling approaches, varying from simple empirical to advanced mechanistic models, have therefore been adopted to quantify the extent and time of emergence for a significant number of weeds. These can be used to produce weed management decision support tools, which enable farmers to determine the percent emergence of a specific weed species by a given date, taking into account the weather, management actions, and field conditions to that point. Populations of weeds respond differently in different regions to climate and habitat, requiring that emergence models be modified for a particular region. No weed management decision support tool exists for the Northeastern region of the United States, despite recent advances in our understanding of regional weed emergence patterns and developments in fine-scale weather prediction and soil moisture modeling. Data exist to create a weed forecasting product similar to those available for insect and disease threats to Northeastern agriculture, which would enable farmers to approach weed management with more precision and planning. In the past decade, decision support tools have been developed to help farmers manage weeds effectively in the Midwestern United States and Europe; these would serve as a road map for the Northeastern decision support tool. Recent advancements in climate and weather models and computational power have generated detailed weather data that are available to the general public free of charge. In the Northeast, daily weather data are now available on a 4 × 4 km grid across the region using the Applied Climate Information System (ACIS) Web Services (DeGaetano et al. 2014). These databases provide an unprecedented opportunity to estimate parameters directly relevant to seedling emergence such as growing degree day and hydrothermal time, from soil temperature and moisture data at very fine spatial resolution.The overarching goal of this project is to work collaboratively across the northeast region to optimize farmers' ability to manage weeds in agricultural systems, in the face of challenges posed from a changing climate and increased prevalence of herbicide resistant weeds.In this proposal, our goal is to develop and validate a user-friendly, online decision support tool for the real time prediction of weed emergence in the northeastern US. The decision support tool takes GPS location, soil type, tillage, crop data, and accesses weather history to provide percent emergence of the farmer's problem weeds at that location.
Animal Health Component
20%
Research Effort Categories
Basic
0%
Applied
20%
Developmental
80%
Classification

Knowledge Area (KA)Subject of Investigation (SOI)Field of Science (FOS)Percent
21352401140100%
Knowledge Area
213 - Weeds Affecting Plants;

Subject Of Investigation
5240 - Seeds and other plant propagules;

Field Of Science
1140 - Weed science;
Goals / Objectives
Link Northeastern weed emergence timing data to existing weed emergence models and modern weather prediction models to create an online tool for farmers that will help them plan their weed management for optimal weed control. This tool will include three weeds that are problematic across the region: common lambsquarters (Chenopodium album), redroot pigweed (Amaranthus retroflexus) and large crabgrass (Digitaria sanguinalis). Common ragweed (Ambrosia artemisiifolia) will also be included in the northern portion of the Northeast and morningglory species (Ipomoea spp.) in the southern portion of the region. Individual participating states may also include one additional species of particular interest to their state. Collect weed emergence data across the region to validate and refine the existing weed emergence models to fit Northeastern data, and refine the decision support tool through testing by select farmers and extension staff.
Project Methods
Objective 1. The emergence model equations published in Myers et al. (2005) were used to create a preliminary emergence prediction tool by fitting existing weed emergence data to precipitation, temperature and soil data for two research farm locations central in New York State. This prediction tool uses the soil temperature model of Bittelli et al. (2015) as presented in their book [Soil Physics with Python, 1st Edition Oxford University Press]. The model is linked to gridded daily temperature and precipitation data via the Applied Climate Information System (DeGaetano et al. 2014).The temperature grid is based on the methods ofDeGaetano & Belcher (2007) and the precipitation grid uses the procedure outlined in DeGaetano & Wilks (2009). The model also requires daily evapotranspiration for which we use DeGaetano et al. (1994) to compute pan evaporation, which is then adjusted to bare soil evaporation using Allen et al. (2006). The forecast data are extracted from the National Digital Forecast Database (NDFD) (Glahn & Ruth 2003). The resulting pilot model is available at https://alexsinfarosa.github.io/weed-modelV2/.While the pilot lacks the ability to alter soil characteristics and soil moisture cutoff mechanisms, it serves as a proof-of-concept for the proposed tool. We have tested the resulting tool against data collected by DiTommaso's research group in two extreme precipitation years (2016, 2017) in New York State (DiTommaso et al. 2018), and incorporated soil moisture cutoffs from WeedCast (Forcella et al. 1998). At the end of Year 1, data from the Multistate project will be used to test both the emergence equations and assumptions made in the decision support tool and those on which WeedCast was based. We will use the equations for each species with the closest fit and modify them as needed to fit the emergence patterns observed across the Northeast. Data from Year 2 will be used to validate the resulting tool and refine its fit across the region. In addition, a select subset of cooperating farmers and extension educators from each participating state will test a draft of the online decision support tool in year 2. Their qualitative feedback will be used to refine the online interface. In Year 3, data from research and farmer input will be used to further validate and refine the model, and a fully functional version will be installed on the NEWA site (http://newa.cornell.edu/).Objective 2. All regional collaborators will establish research plots to validate/refine the emergence model. This proposal focuses on warm season annual weeds, which are difficult to control and are projected to become more problematic with climate change. Data will be collected from earliest available crop planting date until the regional date of soybean canopy closure. Dates will vary by site, as the study area extends from Virginia to Maine and annual weather patterns influence planting date. Each site will include two treatments, one with initial tillage in a field with a history of tillage and one with no tillage in a field with a history of no-till agriculture. Eight 1 m2 plots for each treatment will be established, for a total of sixteen plots per site. Either 0.25 or 0.5 m2 subplots will be sampled per plot, depending on the density of weed seedlings. Data will be collected weekly; all emerged weeds of the selected species of interest will be identified and counted, and all emerged plants removed without disrupting the soil (clipped or pinched). Each participating institution will plant a mixture of three species of common interest across the region, selected by the collaborators: common lambsquarters (Chenopodium album), redroot pigweed (Amaranthus retroflexus) and large crabgrass (Digitaria sanguinalis). In addition, southern states will also incorporate morningglories (Ipomoea spp.) into the weed mix, while northern states will plant common ragweed (Ambrosia artemisiifolia). These species were selected by requesting the top five most problematic weeds from the collaborating weed scientists, and selecting the species that appeared most frequently in their lists. All species except large crabgrass are among the Weed Science Society of America's (WSSA) top ten most troublesome or common weeds, and three species (common ragweed, common lambsquarters, and redroot pigweed) currently have herbicide resistant biotypes in the participating states. Finally, each location will add an additional species critical for their state that is not part of the larger study; these species will be added to the decision support tool for that state.As the lead institution, Cornell will establish all weed species included in any trial, with the exception of weeds such as Palmer amaranth (Amaranthus palmeri) that are not yet found in locations where the trials will be established. We will conduct site visits to ensure that comparable data are collected across sites. Cornell will repeat the experiment at three sites with a range of soil types, ensuring data coverage for different soil conditions. In Year 2, we will solicit feedback on a draft decision support tool from a select group of cooperating farmers and extension educators to assess whether the model is providing accurate information across the region. This feedback will inform research and model modifications in Year 3.The cooperating researchers from across the Northeast will meet annually to share data and experiences, refine the study design as necessary, and plan for the upcoming year.

Progress 10/01/20 to 09/30/21

Outputs
Target Audience:In this reporting season, we did not conduct any extension around our research. We restarted our work after a one-year hiatus due to COVID-19. Once we have products to share, our target audience will be extension agents and growers. Changes/Problems: Nothing Reported What opportunities for training and professional development has the project provided?Students at all of the participating institutions participated in data collection, developing weed seedling identification skills, a useful and difficult skill set. 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?In the next year of the grant, we intend to use this year's data to refine our weed emergence model and develop a decision tool using the results of that work. The model will be posted on the Northeast Environment and Weather Applications website (newa.cornell.edu/crop-and-pest-management). We will also collect a third year of weed emergence data for further refinement of our models, andcomplete the scoping review of weed emergence literature.

Impacts
What was accomplished under these goals? We collected our second year of field data, and re-started our scoping review of weed emergence modeling literature. Our contributors deployed the field sensors purchased before the COVID lockdown in 2020. Our partnership collected data in New York (3 locations), New Jersey, Pennsylvania, Delaware, and Virginia (2 locations).

Publications


    Progress 10/01/19 to 09/30/20

    Outputs
    Target Audience:We educated undergraduate students in the Cornell Integrated Pest Management course about the changes expected with climate change on weed seedling emergence and the complexities ofmodeling and predicting weedemergence. We reached the weed science community with an update on our project and the changes to the models that our data suggested from the first year of data collection. Changes/Problems:We determined that sensor systems for soil moisture and temperature were needed to develop a more accurate model for the predictive tool. As a result, we submitted and were approved for a budget modification to reallocate funds to the purchase of the sensor systems. The coronavirus pandemic halted this project, and all but one of our field sites collected no data in 2020. We paused both work and spending on this project from March 2020, and will collect our second year of data in 2021 and the third in 2022. We received a no-cost extension to support this change. What opportunities for training and professional development has the project provided?Our visiting scholar, Carlos Santos of the University of Seville, had the opportunity for substantial training in weed seedling emergence modeling with Mohsen Mesgaren of UC Davis, one of our mulitstate collaborators. How have the results been disseminated to communities of interest?Our progress was presented by Caroline Marschner at the 2020 NEPPSC conference in Philadelphia, PA, and to the Cornell University Integrated Pest Management course through the lens of weed emergence and climate change. What do you plan to do during the next reporting period to accomplish the goals?We plan to conduct our winter meetings, send the new sensor systems to our collaborators, and conduct our second field season this year (2021). We will continue to work on the review paper for the project. The next round of revisions for the model will fall in the 2022 fiscal year, as will development of the public-facing prediction model.

    Impacts
    What was accomplished under these goals? From October of 2019 through February of 2020, we planned and executed our first year of data collection, compiled the resulting data, and updated our emergence models with assistance from all of our partners. Mohsen Mesgaren of UC Davis and Carlos Santos of the University of Seville were key to this first years' work, especially in the model development area. Our progress were presented by Caroline Marschner at the 2020 NEPPSC conference in Philadelphia, PA, and we held two winter collaboratormeetings to update our research plans and organize the upcoming field season. As our first season of data were collected, it became clear that we needed more reliable in-field data collection for soil moisture and temperature. We submitted a revision to our budget to purchase field sensor equipment for this project, which was accepted, and received the sensors in March of 2020 for the 2020 field season. In March 2020, the novel coronavirus pandemic and the resultinglockdowns at universities across the US precluded most of our partners, including our own lab, from initiating our 2020 field season in time to capture weed seedling emergence for our target species (March-April emergence initiation, depending on the state).In order to retain the ability to collect the three years of data necessary for this project, we stopped all spending and work on this project and received a one-year no cost extension, resulting in a grant end date of September 2022.

    Publications

    • Type: Conference Papers and Presentations Status: Published Year Published: 2020 Citation: Marschner, C., Degaetano, A., Sousa-Ortega, C., VanGessel, M.J., & DiTommaso, A. (2020). Comparing models for weed seedling emergence in the Northeast. Proceedings of the Northeastern Plant, Pest and Soils Conference.


    Progress 12/12/18 to 09/30/19

    Outputs
    Target Audience:Information from the project was presented to weed and plantscientists at the 2019 Northeast Plant, Pests, and Soils conference in Hunt Valley, Maryland from January 8-10. ~20 scientists attending. The project was presented to ~30 Cornell University undergraduate students as part of the Weed Ecology and Management course August 28 & 29, 2019. The project was presented to ~8 Cornell University graduate students as part of the graduate level Weed Management seminar on March 28, 2019. Information from the project was presented to growers onMay 8, 2019, at the North Jersey Commercial Fruit Grower Twilight Meeting II (Rutgers University Snyder Research and Extension Farm, Pittstown, NJ) Information from the project was presented to growers on August 7, 2019, at the RAREC Vegetable Twilight Meeting and Research Tour (Rutgers Agricultural Research & Extension Center, Bridgeton, NJ) Preliminary results from the project were shared as part of a talk on weed identification resources in New York to extension educators and related personnel on November 14th at the Cornell Cooperative Extension 2018 Agricultural In-Service. Results were shared in talks in boththe horticultureand agriculture tracks. Changes/Problems:Bill Phillips, our collaborator in Maryland, had to remove himself from the project for health reasons. Happily, Theresa Pisckacova joined the project, adding a state and maintaining our state count at 8 while expanding our geographic range. Our project requires soil moisture sensors and data loggers, which we did not anticipate when writing the grant. We will be taking funds from travel and other pieces of the budget and reallocating them to the purchase of these sensors and data loggers. What opportunities for training and professional development has the project provided?Most states used undergraduate summer helpers to collect data, which is an excellent training opportunity for advanced weed identification skills. As this experiment checks weekly emergence, students develop skills identifying weeds at the cotyledon and first-leaf stage, which is critical for effective weed management in agriculture. These students receive one-on-one field instruction from weed science technical staff and scientists. Carlos Sousa was a visiting graduate student from the University of Seville, Spain. He brought modeling expertise to the project, and received close assistance and instruction from Dr. Mesgaren on how to fine-tune emergence models. Theresa Pisckacova is a graduate student atNorth Carolina State University, who singelhandedly added North Carolina to our data set. She is developing expertise in project management and interstate project collaboration. Caroline Marschner is developing project management skills while leading this project. She attended Cornell Cooperative Extension's Program Management Leadership Cohort course, which has supported both this project and Smith-Lever Project #2018-19-268: Development of a Weed Identification Network for New York State. 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?In year 2, we will present our current models to the Northeast Plant, Pest, and Soils Conference in Philadelphia, PA in January of 2020. We will hold our annual meeting at that conference, discuss changes to field plans for the next year, and recruit new researchers if possible. We will purchase and deploy soil moisture sensors during the 2020 field season. We will continue to work on model refinement, including testing of equation component values from existing literature, experimentation with biphasic equations, and modeling sub-regions of our trial separately to account for regional phenotypic variability. We will continue to work on the scoping review of weed emergence literature, and present our findings to undergraduate students and growers.

    Impacts
    What was accomplished under these goals? 1) We worked with scientists from Spain, California, and New York to improve the models for emergence in our region. We worked with six site-years of preliminary data collected in New York and Delaware, and explored model variables such as the kind of equation used, the start of growing degree day accumulation, the lower threshholds for growing degree day accumulation, moisture threshholds for emergence of each species, and lag periods before emergence initiation. Working with Carlos Sousa from the University of Seville, Spain, Dr. Mohsen Mesgaren of the University of California at Davis, and Art DiGaetano of Cornell University's Department of Atmospheric Sciences, we compared model methods and explored resources for geospatial data the online model could use for soil texture information and temperature, precipitation, and soil mosture data. We also experimented with soil moisture probes for our field sites. We developed a set of equations that include hydrothermal time andsoil type, and validated them on several states of the first year of data collection from across the region. Model fit needs improvement; we look forward to a second year of data collection and further model refinements around soil moisture and regional differences in weed phenology. We also initiated a scoping review of weed emergence literature; this project will continue into years two and three. 2) We collected our first of three years of emergence data in eight states: Maine, New Hampshire, New York, New Jersey, Pennsylvania, Delaware, Virginia, and North Carolina. These data included the target weeds listed above, as well as additional species collected in various states. Delaware collected data on fifteen species, and New York collected data on at least twenty species.

    Publications