Cucurbit Genetics Cooperative
Other Crop Genetics Cooperatives
Home About Membership Reports Gene Lists Conferences Links Search NCSU
Cucurbit Genetics Cooperative Report 21:18-20 (article 7) 1998

Stand Correction Methods for Cucumber Fruit Yield

Christopher S. Cramer

Department of Agronomy and Horticulture, Box 30003, New Mexico State University, Las Cruces, NM 88003-0003

Todd C. Wehner

Department of Horticultural Science, North Carolina State University, Raleigh, NC 27695-7609

Introduction. Field germination of cucumber (Cucumis sativus L.) seeds is highly dependent on environmental conditions. Cool, wet conditions during the spring planting season slows cucumber seed germination and increases seedling losses. Poor seed germination results in uneven seedling stands within and between plots and within and between fields. Yield comparisons are difficult to make when large differences in stand exist among plots. Several studies have determined fruit yield at different planting densities (5, 8). However, a correction for yield based on plant spacing has not been published for cucumber. Cramer and Wehner (1,2,3) corrected cucumber plant stand for 15 to 30 plants per 3.1 m test plot. They assumed a linear increase in fruit yield per plot as stand increased linearly. That correction method may not be the best, since increases in fruit yield may not be linear at all plant densities. The objective of this research was to evaluate several methods for stand correction that could be used in studies of cucumber yield. The methods were judged on their ability to increase the amount of variance explained by the model (coefficient of determination values), and decrease the amount of variation in the data (coefficient of variability).

Methods. In previous studies (1,2,3,4), eight cucumber populations were evaluated for fruit yield per se, and for combining ability for yield when lines were crossed to elite testers. For the studies of yield per se, 1 m plots were thinned to 30 plants while for the combining ability studies, 1.2 m plots were thinned to 16 plants. In this study, we examined several stand correction methods as follows: 1) no stand correction, 2) analysis of covariance using stand as a covariate, 3) elimination of data from plots with 50% or less stand from the analysis, 4) elimination of data from plots with 50% or less stand from the analysis combined with analysis using stand as a covariate, and 5) elimination of data from plots with 50% or less stand from the analysis combined with linear correction of yield to a full stand based on yield per plant at the current density. Traits studied were total and early fruit number per plot, percentage of marketable fruits per plot, fruit shape rating (6) and a simple weighted index, SWI (7). The analysis of variance model assumed replications as random effects and populations and cycles as fixed effects. Analyses were conducted using the different stand correction methods and the coefficient of determination values (R2 ) and coefficient of variability values (CV) were compared for each correction method.

Results. For both studies and all traits, stand correction method increased the coefficient of determination values (R2 ) and decreased the coefficient of variability (CV) values when compared to the same values for no correction (Tables 1 and 2). The correction method which gave the highest R2 values and lowest CV values varied depending upon the trait for the fruit yield per se data (Table 1). Analyzing total yield using stand as a covariate increased R2 values and decreased CV values when compared to other correction methods. The method of eliminating data from plots with low stand counts and conducting a linear correction for yield based on 30 plants per plot, R2 and CV values for the percentage of marketable fruits per plot. Correction methods which involved eliminating data from plots with low stand counts increased the R2 values for fruit shape, but had little effect on the CV values.

When the number of replications was significantly increased ( as in the studies where populations were crossed with a tester), a stand correction method which eliminated data from plots having low stand counts and analyzed data using stand as a covariate provided the highest R2 values and the lowest CV values for all yield traits (Table 2). In addition, this correction method may give a more accurate representation of trait means since low stand plots eliminated from the analysis.

Stand correction methods helped to explain further the variance observed for particular traits while accounting for variation based solely on plant stand. Researchers should also examine the trait means using each correction method to determine which method accurately represents the means observed. A correction method based on a linear increase in fruit yield as stand increases linearly may inflate yields at certain planting densities. Further research is needed to determine the regression response of fruit yield to planting density.

Table 1. Coefficient of determination ( R2) and coefficient of variability (CV)P values from an analysis of variance of five yield traits from eight cucumber populations using five methods of stand correction on data (1,2).z

Correction method
Total fruits/plot
Early fruits/plot
% marketable fruits/plot
Fruit shape
SWI
Coefficient of determination ( R2)
None
0.67
0.60
0.82
0.67
0.64
Stand covariate
0.70
0.62
0.82
0.67
0.67
Low stand plots eliminated
0.68
0.62

0.83

0.70

0.65
Low stand plots eliminated and stand covariate
0.70.
0.62
0.83
0.70
0.66
Low stand plots eliminated and stand corrected
0.67
0.65
0.83
.0.70
0.71
Coefficient of variation (CV)
None
34
80
21
15
32
Stand covariate
33
78
21
15
31
Low stand plots eliminated
33
78
22
15
31
Low stand plots eliminated and covariate
32
77
22
15
31
Low stand plots eliminated and stand corrected
33
73
22
15
22

z The model included replications (16), as whole plots, and cycles of selection (3) as sub plots. For stand correction methods 1 and 2, the number of observations per traits were at 381 (total and early fruits/plot), 379 (fruit shape), and 378 (percentage of marketable fruits/plot, SWI). For stand correction methods 3, 4 and 5, the number of observations per trails were 353 (total and early fruits/plot). 352 (fruit shape), and 351 (percentage of marketable fruits/plot, SWI).

Table 2. Coefficient of determination (R2) and coefficient of variability (CV) values from an analysis of variance of five yield traits from eight cucumber populations using five methods of stand correction on data (3, 4).z

Correction method
Total fruits/plot
Early fruits/plot
% marketable fruits/plot
Fruit shape
SWI
Coefficient of determination ( R2)
None
0.70
0.72
0.57
0.52
Stand covariate
0.76
0.73
0.57
0.52
Low stand plots eliminated
0.70
0.72
0.57
0.52
Low stand plots eliminated and stand covariate
0.77
0.75
0.63
0.54
Low stand plots eliminated and stand corrected
0.59
0.62
0.57
0.52
Coefficient of variation (CV)
None
34
56
15
16
Stand covariate
31
55
15
16
Low stand plots eliminated
34
56
15
16
Low stand plots eliminated and stand covariate
28
52
13
16
Low stand plots eliminated and stand corrected
41
70
15
16

z The model included replications (88), populations (8) as whole plots, and cycles of selection (3) as sub plots. For stand correction methods 1 and 2, the number of observations per traits were 747 (total and early fruits/plot), 738 (percentage of marketable fruits/plot), fruit shape), and 737 (SWI). For stand correction methods 3, 4 and 5, the number of observations per traits were 702 (total and early fruits/plot), 696 (fruit shape) and 695 (percentage of marketable fruits/plot, SWI).

Literature Cited

  1. Cramer, C.S. and T.C. Wehner. 1998. Fruit yield and yield component means and correlations of four pickling cucumber populations improved through recurrent selection. J. Amer. Soc. Hort. Sci. (in press).
  2. Cramer, C.S. and T.C.Wehner. 1998. Fruit yield and yield component means and correlations of four slicing cucumber populations improved through six to ten cycles of recurrent selection. J. Amer. Soc. Hort. Sci. (in press).
  3. Cramer, C.S. and T.C. Wehner. 1998. Performance of three selection cycles from four pickling cucumber populations hybridized to a tester. J. Amer. Soc.Hort. Sci. (in press).
  4. Cramer, C.S. and T.C. Wehner. 1998. Performance of three selection cycles from four slicing cucumber populations hybridized to a tester. J. Amer. Soc. Hort. Sci. (in press).
  5. Lower, R.L., O.S. Smith and A. Ghaderi. 1983. Effects of plant density, arrangement, and genotype on stability of sex expression in cucumbers. HortScience 18:737-738.
  6. Strefeler, M.S.nd T.C. Wehner. 1986. Estimates of heritabilities and genetic variances of three yield and five quality traits in three fresh-market cucumber populations. J. Amer. Soc. Hort. Sci. 111:599-605.
  7. Wehner, T.C. 1982. Weighted selection indices for trials and segregating populations. Cucurbit Genet. Coop. Rpt. 5: 18-20.
  8. Wehner, T.C. 1986. Efficacy of 3 single-harvest tests for evaluation of yield in pickling cucumber. Euphytica 35:493-501.
Home About Membership Reports Gene Lists Conferences Links Search NCSU
Department of Horticultural Science Box 7609North Carolina State UniversityRaleigh, NC 27695-7609919-515-5363
Page citation: Wehner, T.C., Cucurbit Genetics Cooperative;
Created by T.C. Wehner and T. Ng, 1 June 2005; design by C.T. Glenn;
send questions to T.C. Wehner; last revised on 23 October, 2009