Crop Science 42:700-704 (2002)
© 2002 Crop Science Society of America
CROP BREEDING, GENETICS & CYTOLOGY
Relative Performance of Soybean in End-Trimmed and Plant-to-Length Plots
S. J. Meisa,
W. T. Schapaugh, Jr.*,a and
G. A. Millikenb
a Dep. of Agronomy, Kansas State Univ., Manhattan, KS 66506
b Dep. of Statistics, Kansas State Univ., Manhattan, KS 66506
* Corresponding author (scha0035{at}ksu.edu)
 |
ABSTRACT
|
|---|
A common method to reduce or eliminate potential end-row effects in yield trials is to remove a portion of the plants from the end of the row prior to harvest. Consistent evidence supporting or rejecting the need to end-trim field plots is not available. This study was conducted to determine if plots planted to harvest length and not end-trimmed could be substituted for end-trimmed plots without affecting adversely the relative yield performance of entries in a trial. During 1997 and 1998, separate trials of soybean [Glycine max (L.) Merr.] genotypes in maturity groups III, IV, and V were evaluated in four-row plots at three locations with three replications per location. In each trial, 20 entries were evaluated using three plot treatments: (i) a plot planted to a harvest length of 4.6 m, (ii) a 4.6 m plot end-trimmed to a harvest length of 3.8 m during early vegetative development, and (iii) a 4.6-m plot end-trimmed to a harvest length of 3.8 m following maturity. Seed yield and row length were determined for each plot. Seed yields were significantly different among entries for all three trials. The interactions of plot treatment by entry within year and of plot treatment by entry by location within year were not significant for seed yield in all three trials. Plots planted to harvest length provided an unbiased estimate of the relative performance of yield in soybean.
|
INTRODUCTION
|
|---|
AN IMPORTANT ASPECT of plant breeding is to determine an efficient method to manage yield plots with the least amount of time, labor, and equipment and still maintain accuracy in selections. One time-consuming and labor-intensive aspect of soybean plot maintenance involves end-trimming yield plots to a uniform length. End-trimming at maturity helps ensure a uniform harvest length and removes those plants whose seed yields have been inflated because of less competition for sunlight, water, and nutrients. Boerma et al. (1976) found that the relative yield of genotypes changed when plots were not end-trimmed at maturity. Thus, end-trimming at maturity eliminated the potential of the inflated yields from the end of the row to impact the relative performance among entries.
Another approach used by breeders is to end-trim during vegetative development. The procedure generally can be accomplished faster than when plants are mature, and the timing does not interfere with harvest. The disadvantage of trimming early is that the end-row plants possessing inflated seed yields are not removed prior to harvest. This potential loss in accuracy or precision of selection may be considered an acceptable compromise under certain situations, such as in preliminary yield trials. However, inflated yields from the end of the rows do not necessarily impact the relative performance of the entries. Wilcox (1970) found that the relative seed yield of genotypes within a maturity group was not affected by the growth stage at which plots were end-trimmed. Philbrook and Oplinger (1988) reported that in two of three years the relative yield performance of genotypes was unaffected by end-trimming treatment, but in one season the relative performance was significantly influenced by the time of end-trimming.
An alternative to end-trimming is to plant plots to harvest length and eliminate the end-trimming procedure altogether. If reasonable accuracy of selections can be maintained, resources devoted to end-trimming could be allocated to other aspects of the evaluation process. This study was conducted to determine if soybean plots planted to harvest length and not end-trimmed could be substituted for end-trimmed plots without affecting adversely the relative yield performance of entries in a trial.
|
MATERIALS AND METHODS
|
|---|
Separate trials of soybean genotypes in maturity groups III, IV, and V were evaluated. Each maturity group was tested at three locations in Kansas in 1997 and 1998 (Table 1). Each trial consisted of F4:6 genotypes from crosses randomly selected for all characteristics, except maturity, along with two or three checks for a total of 20 entries. Most entries evaluated in the maturity group III and IV trials possessed the indeterminate growth habit, while most group V entries were determinates.
Entries evaluated during 1997 in the maturity group III trial originated from 14 crosses with one or two entries derived from each cross. Parents of the crosses were C1842, C1843, ‘Corsica’, K1200, K1212, K1213, ‘KS4694’, LN86-3357, LN88-10534, LS86-1922, and P9268-009. Two checks, ‘Iroquis’ and ‘Macon’, were evaluated in the trial. Entries evaluated during 1998 in the maturity group III trial originated from six crosses. From one to six entries were derived from each cross. Parents of the crosses were ‘Asgrow 4715’, Corsica, ‘Flyer’, ‘Stressland’, ‘Hutcheson’, K1230, K88-22-42, LG89-6661, ‘Manokin’, and P6929-004. Three checks, ‘KS3494’, Macon, and Stressland, were evaluated in the trial.
Entries evaluated during 1997 in the maturity group IV trial originated from nine crosses. From one to four entries came from each cross. Parents of the crosses were C1842, Corsica, K1191, K1200, K1212, LN86-3357, LN88-10534, and LS86-1922. Three checks, K1253, KS4694, and Stressland, were included in the trial. Entries evaluated during 1998 in the maturity group IV trial originated from nine crosses. From one to four entries were derived from each cross. Parents of the crosses were Asgrow 4715, Flyer, Stressland, HS90-3487, Hutcheson, KS4694, K1230, K87-7-95, LG89-6661, LN89-295, Manokin, and P6929-004. KS4694 and Stressland served as checks in the trial.
Entries evaluated during 1997 in the maturity group V trial originated from 14 crosses. From one to three entries were derived from each cross. Parents of the crosses were Asgrow 4715, ‘Asgrow 5403’, ‘Hartwig’, Hutcheson, K1196, LS87-1123, Manokin, N86-7682, ‘Pioneer 9591’, R89-332, and S88-1934. Entries evaluated during 1998 in the maturity group V trial originated from nine crosses. From one to seven entries were derived from each cross. Parents of the crosses were Stressland, HS90-3487, Hutcheson, K1192, K1230, LS89-1527, Manokin, N86-7682, S88-1934. Manokin and Hutcheson served as checks in both the 1997 and 1998 trials.
The 20 entries from each trial were evaluated by means of three plot treatments: (i) a plot planted to a harvest length of 4.6 m and not end-trimmed (plant-to-length treatment); (ii) a plot planted to 4.6 m and then end-trimmed to a harvest length of 3.8 m during vegetative development (early treatment); and (iii) a plot planted to 4.6 m and then end-trimmed to a harvest length of 3.8 m after the plants had reached the R8 growth stage (maturity treatment). Early treatment plots were end-trimmed after all trials were planted. Average growth stage at the time of trimming the early treatments was V5, but ranged from the V2 to V6 (Fehr et al., 1971). All plots consisted of four rows spaced 0.76 m apart and planted at a seeding rate of approximately 23 seeds m-1 of row with the same four-row cone plot planter for all trials. A chain driven gearbox on the planter controlled row length. Using an average of a four-person crew per location, end-trimmed plots were staked to a length of 3.8 m during early vegetative development, and at the appropriate growth stage, the outermost 0.4-m sections from each end of the two middle rows were removed by hoeing the plants outside of the staked plot length.
A split-plot design was used with plot treatments serving as whole plots and entries serving as subplots. Each treatment was replicated three times at each location. Data were analyzed with the Statistical Analysis System using the Mixed procedure (Littell et al., 1996) and the General Linear Model procedure (SAS Institute, 1990). End-trimming treatment was the only fixed source of variation. The variables measured on the center two rows of each plot were (i) seed yield (kg ha-1) adjusted to 130 g kg-1 moisture and based on the target row length; (ii) row length (m), the distance from end-plant to end-plant of the center two rows ready for harvest; (iii) maturity, the date when 95% of the pods reached mature color; and (iv) lodging, a 1 = good to 5 = poor, score. Lodging scores are not discussed because few genotypes were leaning more than 45 at maturity.
Variances in seed yield for each of the three end-trimming treatments were calculated. These were measures of the variability of seed yield for an end-trimming treatment within an entry x location combination and were computed by the Mixed procedure in SAS. The least significant difference (LSD) in seed yield for each end-trimming treatment also was computed. The LSD denoted the minimum difference in mean seed yield among entries required for entries to be found significantly different from one another. Phenotypic correlations among end-trimming treatments for mean seed yield of entries were calculated using Pearson's correlation coefficient. Variances in row length for each of the three end-trimming treatments were calculated. These were measures of the variability in row length for an end-trimming treatment within an entry x location combination and were computed by the Mixed procedure in SAS. Row length least squares means were calculated for each end-trimming treatment.
|
RESULTS AND DISCUSSION
|
|---|
In 1997, all three locations for each of the three trials were considered for analysis. However, in 1998, the Powhattan data were discarded because of poor seedling emergence. In the combined analyses across years, significant differences in row length were found among treatments for all three trials, because row lengths of plots in the plant-to-length plot treatment were designed to be longer than those of plots in the two end-trimmed treatments (Table 2). The interaction of location x treatment within year was also significant for row length in all three trials. This was a result of location-to-location variation in the plant-to-length plots. No significant differences in row length were found among entries within year for all three trials. Row length was a function of the planter and/or person performing the end-trimming, and neither of the two factors should have discriminated among entries.
View this table:
[in this window]
[in a new window]
|
Table 2. Combined 2-yr analyses of variance for row lengths of soybean entries evaluated in maturity groups III, IV, and V trials.
|
|
Mean row lengths of end-trimmed early and at maturity treatments were similar to each other (Table 3). However, for most trials, the ranges and variances in row length of plots end-trimmed at maturity tended to be greater than plots trimmed early. The plant-to-length plots tended to have the greatest variation in row length among the three treatments, but only significantly greater than the end-trimmed at maturity treatment in two of six trials. Row lengths of plots that were planted to harvest length were expected to be more variable than the row lengths of plots that were end-trimmed. The accuracy of the planter's row-length gearing system was affected by ground surface variability because it was wheel driven. Changes in the ground surface influenced wheel travel, which impacted the variability in row lengths within and across locations. However, since end-trimming early and at maturity were both staked and trimmed manually, it was expected the degree of precision for these two treatments would be similar, but this was observed in only two trials. A possible explanation for the end-trimmed at maturity plots being more variable in length than the plots trimmed early may have been related to the timing of the trimming procedure and measurement of the row lengths. All end-trimmed plots were staked to a length of 3.8 m during early vegetative development. End-trimmed early plots were then trimmed immediately after marking the row length to be harvested. Row lengths were recorded on the end-trimmed early plots and plant-to-length plots as soon as the former were trimmed to the appropriate harvest length. Row lengths were not recorded on the end-trimmed at maturity plots until after end-trimming which occurred 4 mo later. During this period, a small percentage of plants in each of the plots could have died. Plant mortality at the ends of the plots could have resulted in gaps between the plant at the end of the plot and the end-trimming stake placed in the plot early in the season to designate the harvest area. Thus, row lengths could have been closer to the target length more frequently in the early-trimmed plots, whereas the gaps from plant mortality could have resulted in the increased variability observed for the plots trimmed at maturity.
While the seed yields of the plots trimmed at maturity tended to be smaller than the yield of the other two treatments, the mean seed yields of the trimmed treatments were not statistically different over locations and years for all three maturity groups (Table 4). However, the year by treatment interaction for seed yield was significant for the maturity group V trials. The mean seed yield for the end-trimmed at maturity plots in relation to the other plot treatments was considerably higher in 1997 than in 1998 (Fig. 1)
.
View this table:
[in this window]
[in a new window]
|
Table 4. Combined 2-yr analyses of variance for seed yield of soybean entries evaluated in maturity group III, IV, and V trials.
|
|
In the combined analyses across years, significant differences in seed yield were found among entries within years for all three trials (Table 4). These significant differences were helpful in determining whether the relative performance among entries was affected by end-trimming treatment. The interactions of end-trimming treatment x entry within year, end-trimming treatment x entry x location within year, and location x end-trimming treatment within year were not significant for seed yield in all three trials. Because the interactions of end-trimming treatment x entry within year and end-trimming treatment x entry x location within year were not significant, the relative seed yield performance among entries was unaffected by end-trimming treatment. Wilcox (1970) also reported that the relative performance among genotypes within a maturity group was unaffected by the growth stage at which field plots were end-trimmed. However, these results contradicted the findings of Boerma et al. (1976) who concluded that end-trimming field plots at growth stages other than maturity would result in a change in relative performance among genotypes. Philbrook and Oplinger (1988) also observed a significant genotype x time of end-trimming interaction for yield in one out of three years. The longer growing season of the maturity group V to VIII entries studied by Boerma et al. (1976) may have influenced the response of the genotypes to end-trimming. Philbrook and Oplinger (1988) evaluated entries ranging in maturity from groups 0 through II. This probably represents a greater range in maturity among entries than the difference noted from the earliest to latest entry in these evaluations, which averaged about 10 days per maturity group trial.
In the maturity group III and V trials in 1997, seed yield variance was greatest for the plant-to-length plot treatment, which indicated that the two end-trimming treatments were more precise for detecting seed yield differences among entries (Table 5). For example, in the maturity group III trial in 1997, the LSD among entries were 271 and 292 kg ha-1 for the end-trimmed at maturity and end-trimmed early plots, respectively, whereas the LSD among entries was 475 kg ha-1 in the plant-to-length plots. However, this trend was not repeated in the other trials. For the 1997 maturity group IV trial and all 1998 trials, no significant differences occurred in seed yield variances among the three plot treatments. Thus, in four out of six trials, the plant-to-length plot treatment was as precise as the other two end-trimming treatments for detecting seed yield differences among entries.
View this table:
[in this window]
[in a new window]
|
Table 5. Variance (s2) and least significant difference (LSD) of soybean yield measurements for three treatments in 1997 and 1998.
|
|
Correlations among end-trimming treatments for mean seed yield of entries were low and mostly not significantly different from zero in 1997 for the maturity group IV trial and also in 1998 for the maturity group III trial (Table 6). No significant differences in seed yield were noted among entries for these two trials. In the remaining trials, where significant differences in yield among entries were noted, phenotypic correlations were positive and significantly different from zero among end-trimming treatments for mean seed yield of entries. Correlations between the early and plant-to-length plots with the end-trimmed at maturity plots were positive and similar in magnitude. The largest correlations among trimming treatments were observed in the group V trials. These trials also exhibited the greatest genetic variability in seed yield among entries.
View this table:
[in this window]
[in a new window]
|
Table 6. Maturity group III, IV, and V phenotypic correlations among mean seed yields of entries evaluated with each end-trimming treatment during 1997 and 1998 (n = 20).
|
|
|
CONCLUSIONS
|
|---|
The ability to establish a specific row length for all yield plots by the plant-to-length approach was not as precise as the end-trimmed early treatment, but was similar in precision to the end-trimmed at maturity treatment in a majority of the trials. In addition, the precision of the plant-to-length plot treatment in detecting differences in seed yield among entries was poorer in some trials than the treatments involving end-trimming. However, seed yield of entries among end-trimming treatments tended to be correlated positively, and most importantly, the interactions of entry by end-trimming treatment within year, and entry x location x end-trimming treatment within year were not statistically significant for any of the six trials. Thus, the relative performance among entries for seed yield was not statistically different for the plant-to-length, end-trimmed early, and end-trimmed at maturity treatments. Consequently, plots planted to harvest length and not end-trimmed may provide a viable alternative to end-trimming plots in evaluating soybean entries grouped by maturity, enabling resources devoted to end-trimming to be allocated to other aspects of the evaluation process.
Received for publication November 2, 2000.
|
REFERENCES
|
|---|
- Boerma, H.R., W.H. Marchant, and M.B. Parker. 1976. Response of soybeans in maturity groups V, VI, VII, and VIII to end-trimming. Agron. J. 68:723–725.[Abstract/Free Full Text]
- Fehr, W.R., C.E. Caviness, D.T. Burmood, and J.S. Pennington. 1971. Stage of development descriptions for soybeans, Glycine max (L.) Merrill. Crop Sci. 11:929–931.[Abstract/Free Full Text]
- Littell, R.C., G.A. Milliken, W.W. Stroup, and R.D. Wolfinger. 1996. SAS system for mixed models. SAS Inst., Cary, NC.
- Philbrook, B.D., and E.S. Oplinger. 1988. Spacing pattern and end-trimming effects on solid-seeded soybean plot comparisons. Agron. J. 80:727–733.[Abstract/Free Full Text]
- SAS Institute. 1990. SAS user's guide: Statistics. 4th ed. SAS Inst., Cary, NC.
- Wilcox, J.R. 1970. Response of soybeans to end-trimming at various growth stages. Crop Sci. 10:555–557.[Abstract/Free Full Text]
This article has been cited by other articles:
|
|
|
|
 
T. E. Carter Jr., J. W. Burton, P. E. Rzewnicki, M. R. Villagarcia, M. O. Fountain, D. T. Bowman, and E. Taliercio
Registration of 'N8101' Small-Seeded Soybean
Journal of Plant Registrations,
January 1, 2009;
3(1):
22 - 27.
[Abstract]
[Full Text]
[PDF]
|
|
|
TOP
ABSTRACT
INTRODUCTION
