Assessing Spatial and Temporal Variability of Corn and Soybean Yields

Abstract

With the increased presence in the Corn Belt of combine‐mounted yield monitors generating yield maps, more producers are asking questions like “when do spatial yield differences become important?” and “should I change my production practices because of observed yield differences?” This study examined variations in corn ( Zea mays L.) and soybean [ Glycine max (L.) Merr.] yields over time. Yields were monitored from four continuous corn plots, four continuous soybean plots, and eight corn‐soybean rotation plots over a 10‐yr period from 1986 through 1995 at two Minnesota locations and one Wisconsin location. At each location, all plots were on a similar soil type in a uniform 2‐acre field. Plot size was at least 450 sq ft. At each location, each of the four continuous corn plots and each of the four continuous soybean plots produced the highest yield compared with the other three plots at least once during the 10 yr. In any single year for both continuous corn and continuous soybean, a yield range among the four plots at each location of more than 25% occurred in one‐fourth of the growing seasons studied. When averaged over 10 yr, however, the yield range among the four continuous corn plots and among the four continuous soybean plots at each location was less than 10%, and in five of six cases yield differences among the four plots were not significantly different. At each location, the yield range averaged at least 20 bu/acre for corn and 6 bu/acre for soybean. Yield variability among years was approximately three times greater for soybean and four times greater for corn than was variability among plots. Similar results were observed for the plots in the corn‐soybean rotation. These results demonstrate a relatively high amount of inherent yield variability, and suggest producers should not change management practices to influence yields when small yield differences occur (areas yielding up to 20 to 25% less than the field average), unless the differences are shown to be consistent over years. Research Question Combine mounted yield monitors and associated software enable producers to map crop yields of their fields, making variations in yield across the landscape more evident and quantifiable. This will lead more producers to ask questions like “when do spatial yield differences become important?” and “should I change my production practices because of observed yield differences across the field?” In this study, we determined the amount of spatial and temporal variability at three locations over a 10‐yr period from long‐term corn and soybean research plots with no apparent soil or topographical differences. Our hypothesis is that the yield differences observed from these plots should be small, especially relative to those observed with known soil and topographical differences, and that the magnitude of the yield differences observed may shed insight on when producers should consider changing production practices on portions of their fields to positively influence yields. Literature Summary Several researchers have suggested that caution must be used in interpreting yield results from any 1 yr, particularly when using yield information to modify chemical inputs. Long‐term field experiments conducted on small plot areas have been used to assess the potential viability of site‐specific management practices. When designing field research, care is usually taken to identify uniform areas with minimal soil and site variation. Study Description This study examined variations in corn and soybean yields from four continuous corn plots, four continuous soybean plots, and plots in a corn‐soybean rotation over a 10‐yr period from 1986 through 1995 at Lamberton and Waseca, MN, and Arlington, WI. Soils at each location were uniform with no visible topographical differences and were thought to have a uniform yield potential. At each location for continuous corn and continuous soybean, location average was the average yield of four plots each year, location range was the maximum minus minimum yield of the four plots each year, and location standard deviation was the standard deviation of the four plot yields each year. From 1986 through 1995 at each location, the 10‐yr average was the average yield from each plot, the 10‐yr range was the maximum minus minimum yield from each plot, and the 10‐yr standard deviation was the standard deviation in yield from each plot. Location averages, ranges, and standard deviations were averaged over the 10‐yr, and the 10‐yr averages, ranges, and standard deviations were averaged across the four plots. The average location standard deviation and the average 10‐yr standard deviation were measures of location and seasonal variability, respectively. Applied Questions How much yield variability could we expect across years at the same spot in a field, compared with variability among several areas located near one another on a uniform soil? Yield variability among years was approximately three times greater for soybean and four times greater for corn than was variability among plots (Table ). For continuous soybean yields, seasonal variability was 2.8, 2.2, and 3.6 times that of location variability at Lamberton, Waseca, and Arlington, respectively. For continuous corn yields, seasonal variability was 2.8, 4.3, and 4.0 times that of location variability at Lamberton, Waseca, and Arlington, respectively. Similar results were observed for corn and soybean grown in rotation (Table ). Yields were generally lower than normal in 1988 because of hot, dry conditions and in 1993 because of cool, wet conditions. In a relatively uniform field, what difference in yield in any one year or across several years would justify modifying future corn and soybean production? At each location, each of the four continuous corn plots and each of the four continuous soybean plots produced the highest yield compared with the other three plots at least once during the 10 yr. Over the 10 yr, the range in corn and soybean yield among the plots (location range) expressed as a percentage of the location average was 15 to 20% (Table ). In any single year for both continuous corn and continuous soybean, a yield range among the four plots at each location of more than 25% of the four plot average occurred in one‐fourth of the growing seasons studied. When averaged over 10 yr, however, the yield range among the four plots at each location was less than 10%, and in five of six cases actual yield differences between the four plots were not significantly different. These results demonstrate a relatively high amount of inherent yield variability, and suggest producers should not change management practices to influence yields when small yield differences occur (areas yielding 20 to 25% less than the field average), unless the differences are consistent over years. In evaluating yield variability across a field, does it make a difference whether we use data from good production years or poor production years? Yes. In these studies, the location range was large compared with the location average in seasons with poor growing conditions such as 1988 and 1993. Changing production practice in portions of a seemingly uniform field based only on yield differences after a poor growing season may lead to erroneous conclusions. These results underscore the necessity of in‐season field observations to aid yield map interpretation, especially if relatively large yield variations occur during poor growing seasons. Corn and soybean yields, ranges, and standard deviations from four continuous corn and four continuous soybean plots at three locations from 1986 through 1995. Avg. across 10 yr Avg. across four plots Crop Location Location 10‐yr 10‐yr Location Yield range std. dev. range std. dev. ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐ bu/acre ‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐ Continuous corn Lamberton 115 23.2 (20%) 9.9 87 27.9 Waseca 130 20.7 (16%) 9.0 114 39.2 Arlington 135 21.5 (16%) 9.4 116 37.3 Corn in rotation with soybean Lamberton 130 19.4 (15%) 8.6 97 31.6 Wasec

Abstract from the paper’s OpenAlex record, shown for indexing and research. Copyright remains with the authors or the publisher. Read at the source ↗

The text the model read, with the source record’s inline markup (italics, sub- and superscripts) rendered and its character entities decoded; quote offsets refer to the raw text.