Beta-testing the Adapt-N Tool in On-farm Strip Trials

This project aims to increase adoption of adaptive nitrogen (N) management for corn production using better rates and timing of application. The new Adapt-N tool provides N fertilizer recommendations adapted to the spring rainfall and temperature conditions of the current season, using high-resolution weather data, a sophisticated computer model, and field-specific information on soil properties and soil and crop management.


Justification

This project will lead to adoption of adaptive nitrogen (N) management for corn production using better rates and timing of application. The new Adapt-N tool provides precise N fertilizer recommendations that account for the effects of seasonal conditions using high-resolution climate data, a sophisticated computer model, and field-specific information on crop and soil management.
Appropriate N management is of key importance in corn production systems because of the high cost of N fertilizer, the relatively large N inputs that are used, and concerns over N pollution of the environment. Soil N available to the crop from SOM mineralization and early fertilizer applications varies from year to year due to dynamic and complex interactions among weather, soil type, hydrology, crop growth, and management practices. Therefore, corn response to N application is highly variable, so optimum fertilizer N applications may vary by 100 lbs/ac or more from year to year (van Es et al., 2007; Mullen et al., 2010).

A number of recent studies show that this variability in economic optimum N rate (EONR) is highly influenced by early-season weather, especially rainfall quantities and timing (Sogbedji et al., 2001c; Kay et al., 2006). In humid regions, crop N requirements for corn thus cannot be accurately predicted at the beginning of the growing season and even less so during the previous fall, because N contributions from SOM mineralization and early-season losses aren’t accounted for (See Fig 1.). Therefore, fall or early-spring N applications cannot be precise – even with slow-release or nitrification-inhibition technology – and early season soil testing can only achieve limited accuracy.
When corn N fertilizer recommendations are based on average or modal crop response using methods like the static mass balance approach (Stanford et al., 1973) or maximum net economic return to N approach (MRTN, Sawyer et al., 2006), this generally results in excessive fertilization in years with dry springs, and inadequate fertilization in wet years with high early season N losses. Organic N (e.g., manure) applications result in even higher soil nitrate accumulations in the late spring and a greater potential for loss from excessive soil wetness. This uncertainty creates an incentive for growers to apply “insurance N” in excess of the amount needed, as over-application is not visibly obvious, while under-application is. This reduces farm profits, and at the same time losses via nitrate leaching and denitrification are high - with significant amounts of N2O being produced. The largest losses happen when fertilizer inputs exceed crop N needs, or are present before the crop is able to make use of them (van Es et al., 2002; Snyder et al., 2007; van Es et al., 2007; Hoben et al., 2011).
The above concerns have increased with the documented greater occurrence of high-rainfall events in the early growing season in the humid regions of North America over the past 30 years (Iowa Environmental Mesonet, Iowa State University), and other possible impacts arising from climate change. Therefore research to address the need for climate change adaptation should strongly focus on efforts to better incorporate weather effects into N management, which increases the precision of N fertilizer management and provides incentives for better timing of N applications (sidedress vs. preplant). Resurgence in the willingness to sidedress, rather than apply all N prior to planting will be facilitated by increasing awareness and availability of appropriate tools for decision making. Recent availability of high-clearance N sidedress equipment (e.g., by Miller-St Nazianz and Hagie corporations) and improved manure-sidedress equipment (using drag-lines; Manders Equipment LLC, Weston, OH) will be beneficial for on-farm application as well, because these tools increase the window of opportunity for sidedressing, and provide greater methodological flexibility for manure application, respectively.
Adaptive, and therefore more precise (right rate), in-season (right timing and source) management of N on a field-by-field basis (right place) will lead to higher profits while benefiting the environment by reducing fertilizer when less is needed, but maintaining yields with higher inputs when more is needed.
Computer simulation models offer an alternative that can account for these dynamic soil and crop N processes, thus allowing the farmer to better manage the “4R’s”, as is encouraged by organizations like The Fertilizer Institute and IPNI. One of the potential outcomes of broader adoption of Adapt-N will be a move towards better timing, placement, and formulation. Farmers and consultants will realize the benefits of the 4Rs and more efficient nutrient use as Adapt-N explicitly accounts for these factors in its rate recommendations. Also, Adapt-N provides an estimate of the environmental impact of N management practices, which is of interest to conservation and regulatory agencies.

Adapt-N Tool for N Application Decisions
Adapt-N is a web–based tool (http://adapt-n.cals.cornell.edu) that is designed to provide adaptive N recommendations for corn growers. The interface allows users to input crop, soil, and N management data and obtain N recommendations. There are three key components of this model-based N management decision support tool:
1.) A well-calibrated and tested dynamic simulation model of soil N processes and crop N uptake: We developed the Precison Nitrogen Management (PNM) model (Melkonian et al., 2007 to track soil and crop N flows in corn cropping systems in the Northeast U.S. The PNM model has two components: LEACHMN (Hutson et al., 2003), which simulates key C and N transformation processes, and soil water storage and movement; and a corn N uptake, growth and yield model (Muchow and Sinclair, 1995). Both models have been well tested (Muchow and Sinclair, 1995; Sogbedji et al., 2001a; Sogbedji et al., 2001b; Sogbedji et al., 2006).
2.) Model access to weather data (key drivers of soil/crop processes) at the appropriate scale for farm-level applications: The conventional weather station network is not sufficiently dense to pick up local variation in temperature and precipitation in humid regions. The Northeast Regional Climate Center (NRCC) and the Cornell University Center for Advanced Computing (CAC) have collaborated to produce and distribute near real-time high resolution temperature data and precipitation data (5 X 5 km gridded) for the Northeast US (DeGaetano and Belcher, 2007; DeGaetano and Wilks, 2009), and starting in 2012 for the entire Eastern US. These data are updated daily and can be automatically accessed by Adapt- N through a web service.


Figure 2. The Adapt-N infrastructure that links high resolution climate data from the Northeast Regional Climate Center (NRCC), the Precision Nitrogen Management Model (PNM), and stakeholders via a user-friendly web service.

3.) A user-friendly interface: Adapt-N is designed for easy data entry, and uses information that is part of routine record keeping. Generating an N recommendation generally requires only several minutes once the needed information has been gathered, and requires less time with repeated use and experience (batch data entry will be available in 2012). N recommendations can be generated at any time and the tool therefore allows for continuous monitoring of N availability as the crop progresses in the growing season (daily text and email reports are planned for 2012). Soil testing, in contrast, is much more expensive and time-consuming, and offers lower precision. The Adapt-N interface can also be adapted to the use of finer scale, site-specific recommendations within fields for GPS-based applications (Graham et al., 2010).
Adapt-N simulates crop and soil N dynamics on a daily time step for the season, up to the date of the simulation (in most cases the time of sidedress). If organic inputs (manure, etc.) are part of the cropping system, then the simulations may start the previous season. The tool calculates supplemental N rates by considering (i) crop N uptake based on estimated yield (ii) early season N mineralization and losses, (iii) a probabilistic assessment of mid-to-late season net mineralization, and (iv) an economic factor (price ratio correction). Adapt-N also provides a large, valuable set of information, including graphs of the location’s precipitation and temperature, mineralization of N from OM, soil inorganic N content, and leaching losses, among others, on a daily time step. Adapt-N can also be used for end-of-season hindcasting, similar to a lower corn stalk N test (http://www.extension.iastate.edu/Publications/PM1584.pdf), but at much reduced cost - and based on our 2011 strip trial data likely with much higher accuracy. The Adapt-N tool is currently available for use by anyone. They can perform test simulations to better understand how the tool works. The basic process representation and calibration efforts are available from manuals and the journal articles. A challenge still lies in convincing people that a dynamic-adaptive approach is needed to develop accurate N recommendations, and that computer models are helpful in achieving that goal.
With funding from a National Conservation Innovation Grant from the NRCS (through Dec 2012) and from the New York Farm Viability Institute (through Dec 2011), we have been beta-testing the current version of the Adapt-N tool using 40 on-farm strip trials in Iowa and New York in 2011. These replicated strip trials compare growers' current N management with recommendations by Adapt-N. Compared to conventional practices used by NY growers in 2011, Adapt-N decreased unneeded fertilizer N inputs (by 57lb N/ac on average), and saved growers $42/acre on average, except in a few cases where corn followed a legume, which will be addressed in the 2012 version of Adapt-N. This winter we will use 2011 data and feedback to further refine and improve the model and interface.


Objectives

Our main hypothesis is that Adapt-N provides more accurate estimates of the seasonal EONR than conventional methods and tools, thereby benefitting farmers and the environment. Based on experiences from this year, it is clear that a larger number of replicated strip trials are needed for multiple growing seasons to

1. Further validate the Adapt-N tool for on-farm use, and
2. promote grower adoption of Adapt-N as part of their tool kit for adaptive N management.

Strip trials provide consultants and extension collaborators with the opportunity to learn how to use the tool in depth and understand its outputs enough to communicate with their growers about it, while growers are also learning about N dynamics, and receive help with safely testing this very new tool. While 2012 funding covers salary costs of the Cornell staff reasonably well, support is needed for soil and plant analyses for a large-enough number of strip trials in NY, and to pay collaborators for their efforts are a limiting factor.


Methodology

Our collaborators will work with growers who will host on-farm replicated strip trials. Collaborators will decide with their growers what level of strip trial replication and treatment inclusion is appropriate for them (See Figure 3 options). In as many cases as possibly we will encourage inclusion of check treatments, which are very beneficial for model validation, but can be costly for the grower. Some growers may decide that they would like to use Adapt-N in a 2-combine-width strip across multiple of their fields.


a) b) c)

Figure 3. Spatially balanced strip trial layouts where G = N application according to grower practice, A = N application according to Adapt-N recommendation, and Z = 0lb N applied at sidedress. Option a) has four replications and including a check treatment with 0lb N applied at sidedress, b) has four replications, and c) with two replications, is the simplest trial from which useful data could be obtained.

For each site, a Cornell Soil Health Test, and a 12” soil nitrate sample will be taken prior to planting, and a PSNT will be taken just before sidedress. Growers will decide on their rates, and then Adapt-N will be run by the collaborating extension and consulting staff, in collaboration with their growers, to determine the Adapt-N treatment just prior to sidedressing. Sidedress treatments will be implemented. Prior to harvest, 12” soil nitrate samples and corn stalk nitrate samples will be taken from each strip. Each collaborator will also report yield data by strip. Differences in yield, soil nitrate and corn stalk nitrate will be assessed by mixed model analysis. Nitrate values and yields will be compared to model simulated values by regression.

Field data will be used to evaluate production efficiencies and environmental benefits, and to further calibrate Adapt-N. Meetings and workshops involving growers and agricultural service providers will provide education, training on the use of Adapt-N, and opportunities for providing feedback for improvements. Adoption of adaptive N management will be measured by sign-up to use Adapt-N in the spring of 2013. One or two case studies can be developed in collaboration with IPNI.