| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Dep. of Crop and Soil Sci., Pennsylvania State Univ., University Park, PA 16802
b Dep. of Agric. Econ. and Rural Sociology, Pennsylvania State Univ., University Park, PA 16802
* Corresponding author (jap5{at}psu.edu)
Received for publication December 3, 2003.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Abbreviations: CMA, crop management association • PAAN, potentially available added nitrogen
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Potentially available added nitrogen is an estimate of available N from previous legume crops and manure and fertilizer applications. There is often poor agreement between PAAN required by the crop and applied PAAN. To make a PAAN recommendation, the crop N requirements, indigenous inorganic N, and the quantity of organic N mineralized must be estimated. Thus, properly crediting manure and other N sources, when making fertilizer N recommendations, is an obvious goal to improve N management practices (Schepers et al., 1991).
Mineralization of organic N depends on external factors such as the size of the organic N pool, the availability of organic N to microorganisms, soil water content, temperature and aeration, and substrate quality (Rice and Havlin, 1994; Westerman et al., 1999). Some researchers cite deficiencies in farming systems due to N recommendations that are made based on normal weather for the coming season (Power et al., 2001). Because of variability in soil properties and uncertainties in weather, it is difficult to make N management recommendations and decisions. Historic farm-derived data on PAAN, soil series, and crop yields are used in this study to provide insight into the success of N management recommendations and practices. The actual regional findings may not be transferable to other regions of the state or country, but the process of data utilization should be of broad interest.
The objectives of this paper were to demonstrate how a farm-derived database can be employed to determine whether (i) yields associated with soil series were similar to soil series yield potentials when variations due to PAAN and rainfall were removed, (ii) the amount of PAAN needed to produce a unit of corn was consistent with state recommendations, and (iii) PAAN was in balance with PAAN recommended for observed yields.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Most of the soils in the region are categorized as Class II and III soils by the Natural Resources Conservation Service (NRCS), which established eight soil capability classes to define the suitability of soils for the production of field crops (Helms, 1992). Class I soils have few limitations that restrict their use; Classes II, III, and IV have moderate, severe, and very severe limitations, respectively, that reduce the choice of crops.
Soil capability classes do not intrinsically reflect yield capability but generally describe the suitability of soils for most field crops. In this study, the Class I soil series—Hagerstown (Typic Hapludalfs) and Washington (Ultic Hapludalfs)—have finer textures, are deep, have moderate to high water-holding capacities, and are well suited to crop production. The Class II and III soil series—Edom (Typic Hapludalfs), Morrison (Utic Hapludalfs), and Berks (Typic Dystrudepts)—are coarser textured, moderate to deep, and have moderate to very low water-holding capacity. The Class IV soil series—Klinesville (Lithic Dystrudepts) and Weikert (Lithic Dystrudepts)—also have coarser textures and are shallow, with very low water-holding capacity. Twenty-three percent of fields had soil series classified in Class I, 60% had soil series classified in Class II, and III, and 17% had soil series classified in Class IV.
Corn grain yield capabilities for the soil series in this study, as determined from soil capability classes and regional experience, range from 6.3 to 9.4 Mg ha–1 (Duiker, 2004). Fields averaged 1.4 ha, and the mean soil P and K levels were 120 and 151 mg kg–1, respectively, using the Mehlich 3 (Wolf and Beegle, 1995) or Bray P1 (Frank et al., 1998) extractants. In Pennsylvania, the optimum soil P and K ranges (Mehlich 3 soil test) for agronomic crops are 30 to 50 mg P kg–1 and 100 to 150 mg K kg–1 for grain crops and 100 to 200 mg K kg–1 for forage crops (Beegle, 2004a). Soil P levels are relatively high in fields studied because growers produce animal products and apply significant amounts of manures.
Data Collection
In Pennsylvania, farmer members of crop management associations (CMAs) actively participate as on-farm researchers by managing, interpreting, and assessing field-specific information collected on an average of 48 fields per farm. Members of CMAs pay salaries and wages of crop management personnel hired to provide an array of consulting services, which include crop scouting, soil sampling, record keeping, and crop, nutrient, and pest management recommendations. Producers paying for crop management services make every effort to plant and harvest on time, plant high-yielding cultivars, and work with their crop management personnel to develop strategies, and implement practices, for sound crop, nutrient, and pest management.
The data for this study were collected by 19 West Branch CMA farmers from Juniata, Lycoming, Mifflin, Northumberland, Snyder, and Union counties. The 19 farmers grew corn on a total of 1119 fields, with 1595 ha, from 1997 through 2000. Data for each field for each year included measures of the following variables: crop history, soil series indicator, field size, current state of fertility (N, P, K), yield, and estimated growing season rainfall. From these data, the following variables were computed: yield capability, PAAN, N required to produce a unit of grain, and PAAN in excess or deficit of crop requirement. Other data associated with pest management, production costs, and dates of production activities were also recorded but are not part of the analysis reported here.
Crop Management Database
The Crop Management Database (CropMD) is a relational database that uses FileMaker Pro as its runtime engine (FileMaker, 2004). In 1989, the CropMD software was developed at Pennsylvania State University to help growers manage field-specific information. In 1990, the CropMD software was adopted by CMA consultants providing services to producers enrolled in a USDA cost-share program that encouraged the implementation of integrated pest and nutrient management practices. To maintain program compliance, producers were required to document their planned and actual use of pesticides and fertilizers. Thus, CropMD software was the tool CMA consultants used to help producers document compliance and maintain cost-share support. Since CropMD was first implemented, several hundred other non-CMA producers and crop management consultants in Pennsylvania, Iowa, Vermont, and New Jersey have used the software.
PAAN Calculations
CropMD uses Penn State Agronomy Guide factors (Beegle, 2004a) to calculate PAAN from manures, legumes, and fertilizers. Nitrogen available from current and previous manure applications is calculated based on manure application rate, type of manure, manure nutrient analysis, previous manure history, and whether manure was incorporated. Nitrogen available from a previous legume crop is calculated based on the previous legume type and estimate of the percentage of legume plants in the field. The CropMD calculates N available from commercial fertilizers based on the percentage of available N in the fertilizer material applied.
Corn Yield Measurements
Yields were measured and estimated by a variety of methods, ranging from weighing wagons to hand-weighing yields from samples and extrapolating to the entire field. Yield estimates were made on a field-specific basis. Since crop yields were an important measure of their production success, producers were sensitive to the need for accurately measuring and estimating yields.
In 1997, 1999, and 2000, there was less than 0.13 Mg ha–1 difference between CMA yield medians and the Pennsylvania Agricultural Statistics Service (PASS) state yield means. In 1998, the CMA yield median was 0.82 Mg ha–1 less than the state average. According to extension agents in the study region, dry conditions followed by significant rains caused poor seedling emergence (Murphy and Hostetter, 1998) and ultimately reduced yields.
In this analysis, to compare corn yields associated with all fields, tonnes per hectare of silage were converted to megagrams per hectare of grain by multiplying tonnes of silage times the empirically determined factor of 0.168. Kilograms of PAAN required by the crop were calculated by multiplying megagrams per hectare of corn grain times 19 (Beegle, 2004a).
Weather Data
Although weather data were not collected at farm sites over the 4-yr period, ZedX, Inc. (Bellefonte, PA) computed estimated daily precipitation and minimum/maximum temperatures for each farm location. ZedX, Inc. uses a computer program that ingests National Climatic Data Center (NCDC) data, selects weather stations near a target location, performs quality control checks on the selected NCDC data, and uses a two-dimensional optimal interpolation technique with a correction for elevation to generate data for the target location (J. Schlegel, personal communication, 2002).
The ZedX program used at least 50 nearby weather stations as input to create data for the targeted CMA producers. For the CMA daily data, the interpolation algorithms used a search radius of 2 degrees latitude and longitude from the target location. In addition, the program's output for the target locations went through a quality control process to identify and correct any problems with the data (J. Schlegel, personal communication, 2002). Only growing season rainfall—from 1 May through 31 August—was considered in this study.
Statistics
A PAAN and rainfall interaction term was used to examine whether yield was determined nonlinearly by PAAN and rainfall. Based on a t test, the interaction between PAAN and rainfall was not statistically significant for six of the seven soil series. On this basis, a linear model was used to describe yield as a function of PAAN and rainfall.
The individual soil series models and a joint soil series model were used to determine, through an F test, whether PAAN and rainfall coefficients varied across soil series. The test found a significant difference between the coefficients of the regression models and those of the joint model. Therefore, the individual soil series regression models were used to determine yield relationships among soil series.
The residual errors associated with five of the multiple regression equations were not normally distributed. Therefore, bootstrapping, a nonparametric procedure that makes no assumptions about the form of the distribution being sampled (Chernick, 1999), was used to estimate mean yields and confidence intervals when PAAN and rainfall conditions were average for the 4-yr period: 132 kg ha–1 and 36 cm, respectively (see last two columns, Table 1). Each soil series had ranges that encompassed the average rainfall and PAAN values.
|
Statistical analyses were performed using Minitab software (Minitab, 2001).
|
RESULTS AND DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Based on average 4-yr rainfall and PAAN conditions, the Hagerstown mean yield was statistically greater than all other soil series mean yields except Washington (Table 1). The Weikert soil series mean yield was significantly smaller than each of the other soil series mean yields except Klinesville. Of all the soil series, the Klinesville series responded least to PAAN and rainfall, as indicated by the small t values associated with the PAAN and rainfall coefficients (see Columns 7 and 9, Table 1).
The ranking order of soil series based on calculated means was similar to the Penn State Agronomy Guide soil series yield capability ranking (Duiker, 2004), validating the relative yield capability differences among them (Table 2). However, calculated mean yields were smaller than soil series yield capabilities since all soil series yield capabilities exceeded calculated mean yield confidence intervals. The average corn yield for the 4-yr period was 6.1 Mg ha–1 while the weighted average of yield capabilities associated with fields was 8.0 Mg ha–1. Thus, on this basis, only 76% of yield capabilities were achieved.
|
PAAN Grain Recommendation Compared with Calculated PAAN Required
To determine kilograms of PAAN required to produce 1 kg of corn grain, the average PAAN was divided by the calculated mean yield that was based on average PAAN and rainfall for the 4-yr period (last column, Table 2). Hagerstown and Washington required the least amount of PAAN, and Klinesville and Weikert the most, to produce a kilogram of corn, reflecting the inherent fertility of the better soils.
Pennsylvania researchers studying the economic optimum PAAN rates for corn production on nonorganic sites with Class I soils found that an average of 0.014 kg of PAAN is required to produce 1 kg of corn (Fox and Piekielek, 1995). However, these results were based solely on fields where maximum yields exceeded 70% or more of the yield potential for the site. Fields with poor plant population, poor weed control, severe drought, or other limiting factors were eliminated from the analysis. Earlier research, conducted on organic and nonorganic cornfields, supported the current Agronomy Guide recommendation of 0.0196 kg PAAN kg–1 corn (Fox and Piekielek, 1983) when only nonorganic fertilized fields were considered. Results from this current study, which includes organic and nonorganic sites under a variety of management and rainfall conditions, are consistent with the value of 0.0196 kg PAAN kg–1 corn for Class I soils but indicates more PAAN per unit of yield is required on poorer soils. This may indicate that yield-capability–based PAAN recommendations may not be valid.
Actual PAAN Compared with Recommended PAAN
There was a mean difference of 15 kg ha–1 between actual PAAN and PAAN recommended based on crop yield. Recommended PAAN was calculated using the Agronomy Guide recommendation of 0.0196 kg kg–1 of corn grain (Beegle, 2004a). While the mean difference was relatively small in relation to the average PAAN of 132 kg ha–1, the variation among fields was large (Fig. 1). The standard deviation of an observation was 61 kg ha–1, making the 95% confidence interval of a predicted observation plus or minus 133 kg ha–1. This wide variability in PAAN levels among fields indicates farmers could improve N management.
|
When only fields with actual PAAN in excess of recommended PAAN were considered, the mean excess PAAN was 55 kg ha–1, and the 95% confidence interval for the mean was 53 to 58 kg ha–1. There were 643 fields, with 924 ha that fell into this category. Based on only fields with excess PAAN, the total amount of excess PAAN for the 4-yr period was estimated to be 51 Mg. When the mean difference of 55 kg ha–1, associated with fields having excess N, was considered across 1595 ha, the mean excess was 28 kg ha–1.
In a 1988 Nebraska study, researchers found 58% of the irrigated corn land received more than 22 kg of N ha–1 in excess of the fertilizer recommendation (Schepers et al., 1991). They used 22 kg ha–1 because they estimated that the accuracy of anhydrous ammonia application equipment and related operator accuracy may not be better than ±22 kg ha–1. They also indicated a decline in excess available N over time. From 1980 to 1984, mean excess available N was 48 kg ha–1; in 1988, mean excess available N was about 20 kg ha–1. Researchers speculated that the reductions in excess N occurred from crediting more residual N in the soil and NO3–N in irrigation water without reducing yields. Crediting N sources in determining PAAN recommendations proves to be an obvious important practice that should be encouraged.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| The SCI Journals | Crop Science | Vadose Zone Journal | |||
| Journal of Plant Registrations | Soil Science Society of America Journal | ||||
| Journal of Natural Resources and Life Sciences Education |
Journal of Environmental Quality |
||||
