HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Free Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in ISI Web of Science Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via ISI Web of Science (1) Citing Articles via Google Scholar Google Scholar Articles by Esgar, R.W. Articles by Bullock, D.G. Search for Related Content PubMed Articles by Esgar, R.W. Articles by Bullock, D.G. Agricola Articles by Esgar, R.W. Articles by Bullock, D.G.

CROP ECOLOGY, PRODUCTION & MANAGEMENT

Thinning Border Rows Differentially Affects Hybrids in Corn Yield Trials

^a Dep. of Crop Sciences, Univ. of Illinois, Urbana, IL 61801-4798 USA

dbullock{at}uiuc.edu

	ABSTRACT

TOP ABSTRACT INTRODUCTION Materials and methods Results and discussion REFERENCES

 
Border rows are used in yield trials to negate the effect of neighboring plots. Border rows are often not thinned to a final stand as are the harvest rows and it is assumed differential density in the border rows has a trivial effect on harvest row yield. In this study, 10 corn (Zea mays L.) hybrids were planted in 1994, 1995, and 1996 at Urbana, IL, in a randomized complete block design with 10 replications. Experimental units were four row plots with the middle two harvest rows thinned to a final stand of 68000 plants/ha while the outer two border rows were either thinned as the harvest rows or not thinned. Seed weight (g/100 seed) was significantly and negatively correlated with plants per row and significantly and positively correlated with grain yield . Thinning border rows increased mean grain yield 0.2 Mg/ha and a significant hybrid x treatment interaction was detected. Thinning of border rows decreased the mean square error as much as 26%. Thinning of border rows affected the membership of the highest yielding (elite) group. This is particularly important because it is among this elite group that new hybrids are chosen. We conclude that both border rows and harvest rows should be thinned to a uniform final stand in corn yield trials and failure to do so may result in erroneous conclusions and wrong decisions.

Abbreviations: ANOVA, analysis of variance • CV, coefficient of variation • LSD, least significant difference • MSE, mean square error

	INTRODUCTION

TOP ABSTRACT INTRODUCTION Materials and methods Results and discussion REFERENCES

 
BORDER ROWS are used in yield trials to negate the effect of differing levels of competition from neighboring experimental units. Border rows have been shown to be critical for crops such as soybean [Glycine max (L.) Merr.] (Hartwig et al., 1951), sunflower (Helianthus annuus L.) (Fick and Swallers, 1975), and even navy beans (Phaseolus vulgaris L.) (Down and Thayer, 1942). Border rows are less necessary for crop such as grain sorghum [Sorghum bicolor (L.) Moench] (Ross, 1958) and corn (Federer and Sprague, 1947). Additional protection from the competition of neighboring experimental units has been suggested in the form of blocking by plant height (Pendleton and Seif, 1962) or maturity (Keng and Hall, 1987).

The use of border rows for corn yield trials may be unnecessary or at least inefficient. Bowman (1989) indicated that corn yield trials would be more efficient if a given area of land replication was maximized by abandoning border rows. In particular, they recommended the use of non-bordered two row plots with a doubling of replications rather than four row plots with only the inner two rows being harvested as an estimate of yield.

While the arguments for abandoning border rows are well made, the fact remains that border rows are still used in many corn yield trials. In many cases, land area is not as large a determining factor as is total number of replications. In these cases, the maximum number of replications which can be handled is determined with little regard to land area. Usually border rows are then used, but are not managed at the same level as the harvest rows. In particular, it is often seen that the harvest rows are over-planted and hand-thinned to a final stand to avoid differential stands among hybrids while border rows are not thinned and are often at a higher density than that of the harvest rows. This lack of thinning represents a great savings in labor and time and has long been thought to be of trivial consequences, but we have found no documentation in the literature to support that assumption. The objective of this study was to investigate the effect of thinning border rows in corn yield trials. We were working with the initial hypothesis that hybrids react differently to border row competition and this difference affects the relative ranking of the hybrids by mean separation and thus affects the conclusions made from the data of the corn yield trial.

	Materials and methods

TOP ABSTRACT INTRODUCTION Materials and methods Results and discussion REFERENCES

 
This field experiment was conducted at the University of Illinois Agronomy South Farm at Urbana, IL. On 17 May 1994, 3 May 1995, and 18 April 1996, 10 corn hybrids (Table 1) were planted into a Thorp/Cullo silt loam soil (fine-silt, mixed, mesic Argiaquic Argialbolls). Irrigation was not used. Weeds were controlled by a combination of herbicides and mechanical cultivation. Soil fertility was not yield limiting. The previous crop each year was soybean.

	Results and discussion

TOP ABSTRACT INTRODUCTION Materials and methods Results and discussion REFERENCES

 
Numerous type of planters are used in corn yield trials. The majority are either cone planters or air planters. The planter used in this study is an air planter. This type of planter offers the advantage of very uniform and rapid seeding, but suffers from the fact that optimum performance requires adjustment for seed size. Thus if seed size is not uniform across hybrids in a trial, optimum planter performance would require adjustments each time seed size changed. Time constraints prevent such action. Some programs, such as the University of Illinois, have suggested a specific seed size for hybrid entries, but we know of no testing program that mandates a specific seed size. Seed weight of the hybrids used in this study did not differ significantly between years and was normally distributed with a mean of 30 g/100 seed and a standard deviation of 4.8 g/100 seed.

The common approach to this problem of non-uniform seed size is to select settings which perform well with the majority of the seed sizes present in a study. A ramification of this approach is that small-seeded hybrids are planted at a higher density than medium- or large-seeded hybrids. This comes about from multiple seeds adhering to a single position on the seed disk on the air planters and this results in two seeds (doubles) or even three seeds (triples) being planted in the same position rather than a single seed. For cone planters, doubles and triples can also appear as a result of seed size if seed is measured volumetrically or simply as an artifact of the need to overseed the harvest rows resulting in multiple seeds per cell of the cone. Thus, this problem of non-uniform stands in border rows can exist for most types of planters.

In this study, mean plant density for thinned rows was 68000 plants/ha, while the mean plant density for non-thinned rows was 78 500 plants/ha with a maximum of 87700 plants/ha (Table 1). Seed weight was significantly and negatively correlated with plant density . Since only the harvest rows were thinned, this correlation indicates that border row plant density decreased with increasing seed size. The final stand of the harvest rows were unaffected by seed size since they were over-seeded and thinned. We hypothesized the increased plant density in the border rows would present greater competition to the plants in the harvest rows and this did occur since seed size was significantly and positively correlated with grain yield for the non-thinned plots . Similarly, non-thinned grain yield was significantly and negatively correlated with non-thinned row plant density . In addition, analysis of variance (ANOVA) indicated that thinning border rows led to a significant increase of approximately 0.2 Mg/ha (12.1 Mg/ha thinned; 11.9 Mg/ha non-thinned) for the mean grain yield across all three years. ANOVA also indicated a highly significant (P < 0.0001) hybrid x treatment interaction indicating that not all hybrids responded to thinned border rows in the same manner.

The occurrence of these significant correlations, mean effects, and interaction are of concern, but we propose the pragmatic question before us is if the relative rank of hybrids changes due to border row competition and if so does the change in relative rank lead to a difference in conclusion as to which hybrid should be chosen by a producer for planting? In particular, we are asking if the interaction between hybrids and thinning treatment represents a major shuffling of the hybrids or simply a minor reordering. On first glance, it appears the latter may be the case since hybrid rank (i.e., 1 through 10) in the thinned and non-thinned treatments were significantly correlated ; however, this preliminary conclusion does not stand up to closer scrutiny.

We propose that the appropriate way to look at this question of rank change is to determine what we refer to as the elite group of hybrids. For the purpose of this discussion we define the elite group of hybrids as all hybrids which produce a mean grain yield which is not significantly different from that of the hybrid with the numerically largest yield in a study for a given year.

Note that this experiment was a randomized complete block design with 10 replications. For purposes of this portion of the analysis we ran separate analyses for the thinned and non-thinned treatments as though they were separate experiments. That approach, in theory, would overestimate the experimental mean square error since the area used for an individual block would be twice as large as necessary if we were to have conducted them separately; however, the increase in variance would be small and thus acceptable.

Theory would also suggest that non-uniform border row competition seen in non-thinned plots would increase the mean square error (MSE) and coefficient of variation (CV) above that seen for the thinned experiment and that is exactly what we saw. The MSE and CV were significantly larger for the non-thinned approach than the thinned (Table 1). This decrease in MSE with thinning of the border rows is responsible for the reduction in the least significant difference (LSD) and thus a decrease in the type II error rate. In 1994, the difference was small, but in 1995 and 1996 the differences were substantial with the MSE being reduced by 26 and 23%, respectively. While these reductions in MSE alone would serve well in any argument for thinning of border rows in corn yield trials, further analysis indicates an even more convincing point.

In Table 1, we have designated the elite group within each year by placing an upper case letter A after the mean grain yield. Note, that the mean separation for the purposes of determining the elite group was conducted with an LSD calculated on the basis of a type I error rate of 0.30 as suggested by Carmer (1976) and Carmer and Walker (1988) to reduce the type II error rate. The LSD values based upon the more traditional alpha levels of 0.10 and 0.05 are provided for those readers interested in either of those alternative, although similar, analyses.

In 1994 Asgrow RX 775 and DeKalb DK 604 constituted the elite group in the thinned treatment while Whisnand 88 combined with Asgrow RX 775 in the non-thinned treatment. Thus, DeKalb DK 604 replaced Whisnand 88 when thinning was used. The non-thinned plant density of DeKalb DK 604 was 78500 plants/ha and thus substantially larger than the 68000 plants/ha density of the thinned treatment and presumably benefitted from reduced competition after thinning. Whisnand 88, however, had a non-thinned plant density of only 72600 plants/ha and presumably benefit less from thinning. We propose that by thinning the border rows, and arguably more closely simulating actual field conditions, DeKalb 604 and Whisnand 88 exchanged positions as members of the elite group.

In 1995, Kruger K9514 joined Asgrow RX 775 and DeKalb DK 604 in the elite group of the thinned treatment while Payco 903 joined DeKalb DK 604 as replacements for Asgrow RX 775 and Kruger K9514 in the non-thinned treatment. Note that the none-thinned plant density of Asgrow RX 777 (87100) and Kruger K9514 (86200) were substantially greater than the 68000 plants/ha density of the thinned treatment and presumably benefitted from reduced competition after thinning.

In 1996, the hybrids planted changed, but the story remained the same. Golden Harvest H-2529 and Burrus BX-65 were the sole members of the elite group of the thinned treatment while the non-thinned treatment elite group was composed of those same two hybrids with the additional membership of Growmark FS-6890, Becks 5305, Trisler T-5315, and Terra TR-1126. Again, the non-thinned plant density of both Golden Harvest H-2529 and Burrus Bx-65 were substantially larger than 68000 plants/ha and presumably benefitted from reduced competition after thinning. We propose the reduction in competition provided a more realistic estimate of the yield differences among the hybrids and provided a far more useful test where only two hybrids were among the elite group rather than six hybrids.

From these analyses, we conclude that it is appropriate to accept our initial hypothesis and conclude that hybrids react differently to border row competition and these differences affect the relative ranking of the hybrids by mean separation and thus affect the conclusions made from the data of corn yield trials. A decision maker would come up with a decidedly different set of members of the elite group if thinned rather than non-thinned plots were used. This is particularly important, because it is from this elite group that new hybrids are chosen. We conclude that both harvest rows and border rows should be thinned to a uniform final stand in corn yield trials and failure to do so may result in erroneous conclusions and wrong decisions.

Received for publication October 14, 1998.

	REFERENCES

TOP ABSTRACT INTRODUCTION Materials and methods Results and discussion REFERENCES

 

• Bowman D.T. Plot configuration in corn yield trials. Crop Sci. 1989;29:1202-1206.[Abstract/Free Full Text]
• Carmer S.G. Optimal significance levels for application of the least significant difference in crop performance trials. Crop Sci. 1976;16:95-99.
• Carmer S.G., Walker W.M. Significance from a statistician's viewpoint. J. Prod. Agric. 1988;1:27-33.
• Down E.E., Thayer J.W., Jr. Plot technic studies with navy beans. J. Am. Soc. Agron. 1942;34:919-922.
• Federer W.T., Sprague G.F. A comparison of variance components in corn yield trials. I. Error, test x line, and line components in top-cross experiments. Agron. J. 1947;39:453-463.[Free Full Text]
• Fick G.N., Swallers C.M. Intercultivar competition in sunflower test plots. Agron. J. 1975;67:743-745.[Abstract/Free Full Text]
• Hartwig E.E., Johnson H.W., Carr R.B. Border effects in soybean test plots. Agron. J. 1951;43:443-445.[Free Full Text]
• Keng J.C.W., Hall J.W. The effects of guard rows on variety test results of maize. Agric Sys. 1987;23:187-195.
• Pendleton J.W., Seif R.D. Role of height in corn competition. Crop Sci. 1962;2:154-156.
• Ritichie, S.W., J.J. Hanway, and G.O. Benson. 1993. How a corn plant develops. Iowa State Univ. of Sci. and Technol.—Coop. Ext. Serv. Spec. Rep. 48.
• Ross W.M. A comparison of grain sorghum varieties in plots with and without border rows. Agron. J. 1958;50:344-345.[Abstract/Free Full Text]
• SAS Institute. SAS uses's guide: Statistics, 6th ed Cary, NC: SAS Inst, 1989.

This article has been cited by other articles:

				O. David, H. Monod, J. Lorgeou, and G. Philippeau Control of Interplot Interference in Grain Maize: A Multi-Site Comparison Crop Sci., March 1, 2001; 41(2): 406 - 414. [Abstract] [Full Text] [PDF]

This Article Abstract Full Text (PDF) Free Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in ISI Web of Science Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via ISI Web of Science (1) Citing Articles via Google Scholar Google Scholar Articles by Esgar, R.W. Articles by Bullock, D.G. Search for Related Content PubMed Articles by Esgar, R.W. Articles by Bullock, D.G. Agricola Articles by Esgar, R.W. Articles by Bullock, D.G.