Detecting Yield Differences From Alfalfa Plots of Various Widths

Bruce Anderson, Greg Cuomo, and Mike Trammell
Department of Agronomy
University of Nebraska, Lincoln, NE 68583-0910


Alfalfa variety trials currently are conducted with 2.5 to 8 foot plot widths and yield collected from 38 to 100 percent of the plot. Border effects influence yields of adjacent plots in some plot studies. Many investigators assume these effects are negligible in alfalfa trials, possibly because plots are harvested at nearly the same time and characteristics like plant height do not appear to favor one population over another. Thus, many alfalfa variety trials have no side borders for their plots, resulting in harvest of the entire plot for yield. We examined the effect of various plot widths on alfalfa variety yields and the statistical ability to detect differences among these varieties.

'Spredor II', 'Dawson', and 'Wrangler' alfalfa were seeded in 1988 near Mead, NE in plots containing 5, 10, or 15 rows spaced six inches apart with twelve inches between plots. Plot length was twelve feet with a three foot border of alfalfa on each end. Plots were arranged as a split-plot in a RCBD with six replicates. A three-foot wide strip was harvested from the center of each plot using a flail harvester from nineteen total harvests during five years. Thus, entire 5-row plots were harvested and alfalfa borders remained on 10-row and 15-row plots.

Spredor II, Dawson, and Wrangler yields averaged 5.40, 5.51, and 5.94 tons/acre annually, similar to historical yields at this location. Plot width did not affect yields even though six or seven rows were harvested from 10- and 15-row plots in contrast to just five rows harvested from the 5-row plots. Experiment wide variation (C.V.) and ability to detect differences in variety yields (F values) differed little among plot widths (Table 1). Plots with borders (10- and 15-rows) tended to have higher F values when several years of data were combined. This should result in greater ease in detecting significant differences in yield, but sensitivity analyses indicated there were no technique differences (p >0.05). Interactions of plot width and variety were insignificant (Table 2), suggesting that yield comparisons among varieties did not change due to plot width.

To conclude, plots with 5 rows spaced 6 inches apart and harvested with a threefoot wide flail-type plot harvester (commonly used for many alfalfa variety yield trials) were as effective at comparing yields of alfalfa varieties as wider plots that left an alfalfa border following harvest, indicating that plot width does not affect alfalfa variety trials.

Table 1. F values and C.V.'s of alfalfa variety yield tests using 5-row or 15-row plot widths.

5-row 15-row
Year F C.V. F C.V.
1988 8.87 9.12 9.54 8.06
1989 1.84 7.52 0.60 6.72
1990 20.05 5.75 15.58 5.70
1991 2.06 6.29 2.13 7.30
1992 3.77 8.04 9.64 6.37
1989-1992 3.34 4.56 8.40 4.19
1988-1992 4.27 4.25 9.33 3.98




Table 2. Interactionsl of plot width and variety.

Year F value Pr > F
1988 1.74 0.16
1989 0.71 0.59
1990 0.68 0.61
1991 0.50 0.74
1992 1.31 0.29
1989-1992 0.53 0.71
1988-1992 0.73 0.58

1Varieties differed (P<0.01) except 1989; width never differed (P>0.10) except 1988.



1995 Central Alfalfa Improvement Conference Proceedings