Euphytica 36 (1987) 711-718

A COMPARISON BETWEEN SINGLE PLANT PLOTS AND FIVE PLANT PLOTS FOR THE INITIAL

SELECTION STAGE OF A POTATO BREEDING PROGRAMME

J. BROWN

Scottish Crop Research Institute, Pentlandtield, Roslin, Midlothian, Scotland

Received 13 June 1986

KEY WORDS

Solunum tuberosum, potato, single plant selection, potato breeding, selection efficiency.

SUMMARY

The efficiency of single plant selection in the initial selection stage of a potato breeding programme was examined. A random sample of potato clones was grown in single plant plots and also in two replicates of five plant plots. After harvest, each plot was visually assessed by three potato breeders; total tuber weight and number of tubers per plant were also recorded, from which mean tuber weight was calculated. The error variances of the single plant plots were found to be significantly greater than from the five plant plots for total tuber weight, mean tuber weight and number oftubers per plant, but not significantly different for breeders’ preference. Coefficients of correlation between single and five plants plots were significantly greater than zero for all traits examined, but they were lower than the corresponding correlations between the two replicates of five plant plots. Although a random sample of clones was examined, each clone had been assessed for breeders’ preference the previous year. When the clones were grouped according to the previous years preference scores, it was found that the correlation coefficients between single and five plant plots for breeders’ preference were inversely related, in magnitude, to the mean preference score of each group. It was concluded that single plant selection was generally ineffective, particularly when only the ‘better’ clones, from the previous years assessment, are examined.

INTRODUCTION

All traditional potato breeding programmes begin by growing potato seedlings from true seeds. Most researchers agree that no selection of individual genotypes for agron- omic traits should be carried out at this seedling stage (HOWARD, 1962; FPEFPER, 1963; ANDERSON & HOWARD, 1981; TAI & YOUNG, 1984 and BROWN et al., 1984), and that this stage should only provide tubers for planting in the field the following year.

The plot size used in the early generations of a breeding programme tends to be small because of the limited number of tubers available for planting. Selection in the first clonal year is often carried out on single plant plots, usually by visual appraisal, to retain the ‘best’ clones (TAI & YOUNG, 1984; FITSC~IEN, 1984; SCHOLZ, 1986). Such selection has been shown to be ineffective (TAI, 1975; MARIS, 1964), and it has been suggested that only negative selection (i.e. discarding the poorest phenotypes) should be carried out. BROWN et al., (1986) found that positive selection at the seedling and

711

I. BROWN

first clonal year stages results at best in a random reduction in the number of genotypes, and that selection in these early stages may have a negative effect.

Selection at the first clonal year stage may be ineffective for a number of reasons. If evaluation is carried out on the phenotypic performance of a single, unreplicated, plant plot, micro-environmental variation may hamper effective selection (DAVIES & JOHNSTON, 1965, 1968 and 1974). The size of tubers used to plant the first clonal year can have a large effect on yield produced (BROWN & CALIGARI, 1986). The size of tuber plant will also influence breeders’ assessments, larger seed ‘:ubers resulting in a higher probability of selection (BLOMQUIST & LAUER, 1962 and BROWN et al., 1984). Thus the inefficiency of selection in the first clonal year is also influenced by the size of seed tuber produced from the seedlings.

If the inefficiency of selection at the first clonal year stage is due primarly to the effects of planting tubers which have been produced from seedlings the efficiency might be improved by delaying selection until field grown seed tubers are available. If selec- tion is carried out on a single plant basis then many more genotypes can be evaluated than with larger plots and greater replication.

Most previous work on the efficiency of single plant selection in potato breeding has been carried out by comparing selection of single plants at the first clonal year stage with selections on larger plots in a later clonal generation. This paper compares the efficiency of single plant selection with selection on the mean of two randomised live plant plots where both plot sizes use common origin, field grown, seed tubers.

MATERIALANDMETHODS

The progeny from eight potato crosses were examined. These crosses were chosen to be representative of the type that are normally included in the potato breeding scheme at the Scottish Crop Research Institute (HOLDEN, 1977). Seedlings from each of the crosses were grown under glass in 10 square pots. In 1982, a single tuber from each seedling was planted at a high grade seed site. Three tubers were kept from the produce of each plant and were grown in randomised three plant plots, under seed production conditions in 1983. The produce from each of these second clonal genera- tion plots was visually assessed by four potato breeders on a l-9 scale of increasing preference. The preference scores took account of a number of tuber characteristics to indicate commercial worth. A preference score of greater than 5 indicated that, under the normal breeding scheme, the particular plot would have been selected for re-evaluation in a later season. At this stage in the experiment however all the tubers from all the plots were retained.

In 1984,276 clones from among those grown in the second clonal year were taken at random, but with the constraint that each of the eight crosses was represented by at least 25 clones. These clones were grown in single plant plots and also in two five plant plots at a ware growing site. Within the block of single plant plots and both replicates of five plant plots the clones were completely randomised. Amongst these clones, six commercial cultivars were also grown as controls. Each control cultivar was represented by four plots in the single plant block and also in each replicate of the five plant plots. Both the single and live plant plots were planted in drulls 75 cm apart. Each five plant plot was arranged in a single drill with 45 cm between plants

712 Euphytica 36 (1987)

SINGLE PLANT SELECTION IN POTATO BREEDING

and a 120 cm gap between plots. The single plant plots were arranged with 55 cm between plants to allow the harvest of individual plants.

The single plant plots were harvested by hand and the tubers from each plant were exposed on the soil surface. At this stage the produce from each plant was visually assessed for commercial preference by three potato breeders on a l-9 scale, as pre- viously noted. Tubers were then bagged and taken into store. The five plant plots were mechanically dug, the produce placed in wooden boxes, and taken into store. In the store, the produce from each of the live plant plots was also visually assessed by the same three breeders. The produce from each of the single plant and five plant plots were weighed, the number of tubers per plot counted and mean tuber weight was calculated. All yield data were converted to yield per plant. Breeders’ preference data are the mean score of the three breeders (four breeders in 1983) who assessed each plot.

RESULTS

The error variances from the analysis of variance of the data recorded on the six control cultivars, grown in single plant plots and in five plant plots are shown in Table 1. The error variances from the two plot sizes were compared using Bartlett’s test (SNEDE- COR & COCHRAN, 1974). There was no significant difference between the error variance from single plant plots and that from the five plant plots for breeders’ preference. However, the error variance from the single plant plots was significantly larger than in the five plant plots for total tuber weight, mean tuber weight and number of tubers per plant.

Correlations for breeders’ preference, total tuber weight, mean tuber weight and number of tubers per plant between the single plant plots and the five plants plots were all found to be significantly greater (P < 0.001) than zero (Table 2). The largest correlation coefficient (r = 0.64) was for total tuber weight while the lowest correlation coefficient (r = 0.54) was found for mean tuber weight. The correlation coefficients for the four variates examined were consistantly larger between the two replicates of the five plant plots than between the single and the mean of two five plant plots.

It would appear from the correlation coefficients between the single and five plant plots, that it should be possible to select on a single plant basis. However, the simulta- neous distribution of breeders’ preference scores on the single plant plots and the mean score of two five plant plots (Table 3) shows that many clones given high scores in the five plant plots were assessed as having a poor preference in the single plant plots,

Table 1. Error variances of breeders’ preference (Pref), total tuber weight in kg per plant (TTW), mean tuber weight in gram per plant (MTW) and number of tubers per plants (TN) from the analysis of variance of six control cultivars grown in a single replicate of one plant plots and two replicates of five plant plots.

df Pref TTW MTW TN

2 x five plant plots 36 0.7309 0.0361 479 4.04 Single plant plot 18 0.6013 0.2019 7764 14.88 x2 (Bartletts test) I 0.18 ns 15.56*** 38.34*** 8.96**

ns = not significant; ** = 0.01 > P > 0.001; *** = P < 0.001.

Euphytica36 (1987) 713

J. BROWN

Table 2. Coefftcients of correlation between breeders’ preference, total tuber weight, mean tuber weight and number of tubers per plant recorded on a single plant basis (Single) and the mean, averaged over two replicates, of five plant plots (2 x St). The correlation between the two replicates of the five plant plots (Rep. 1 and Rep. 2) is also shown.

Single v. 2 x 5t Rep. 1 v.Rep. 2

Breeders’ preference 0.57 0.76 Total tuber weight 0.64 0.84 Mean tuber weight 0.54 0.74 Number of tubers per plant 0.61 0.69

and vice versa. As noted, a score of more than 5 indicates that that particular plot is considered to have ‘commerical’ worth, and hence, under a normal breeding scheme, would have been selected for re-trial. From this information 202 clones (67%) would have been discarded in both the single and five plant plots, while 26 clones (9%) would have been selected in both plot sizes. However, 27 clones (9%) would have been selected when grown in a single plant plot but discarded when grown in a five plant plot, and 45 clones (15%) would have been selected in the five plant plots but discarded as a single plant plot. It should be noted that a higher proportion of clones was selected from the live plant plots (24%) than was selected in the single plant plots (18%). Despite this it is obvous that a high proportion of clones that would have been selected (breeders’ preference > 5) in either plot size (26 selected in both plot sizes + 27 selected only in the single plant plots + 45 selected only in the five plant plots = 98 clones) would have been discarded on the basis of the score from either the single or live plant plots. From these 98 clones, 73% would have been discarded in one of the two plot sizes.

The similarity of total tuber weight recorded on a single plant and a five plant plot can be compared in a similar way (Table 4). If the top 20% of clones was retained, according to total tuber weight, then 218 clones (73%) would have been discarded on both the single and five plant plots, while 38 clones (13%) would have been selected in both plot sizes. Also 22 clones (7%) would have been selected for total tuber weight on the basis of the mean of two five tuber plots but discarded on the basis of a single

Table 3. Simultaneous distribution of breeders’ preference scores recorded on single plant plots and two five plant plots.

Single plant plot

8.1-9.0 7.1-8.0 6.1-7.0 5.16.0 4.1-5.0 3.14.0 2.1-3.0 1.1-2.0 1.0

Mean of two five plant plots

1.0 1.1-2.0 2.1-3.0 3.14.0 4.1-5.0 5.16.0 6.1-7.0 7.1-8.0 8.1-9.0

1 : i :

; 8 i 6 6 13 4 4 2 i 8 13 28 21 10

i 10 k 25 18 29 10 21 8 12 1 .

1 4 9 2 2 i : 3 4

714 Euphytica 36 (1987)

SINGLE PLANT SELECTION IN POTATO BREEDING

Table 4. Simultaneous distribution of total tuber weight recorded on a single plant plot and five plant plots. The classification is according to the rank. The 300 clones were ranked (I = highest) according to total tuber weight, thus l-10% comprises clones ranked 271st to 300th, l&20% comprises clones ranked 241st to 270th and so on. The category 96100% comprises of course the highest yielding clones, ranked 1st to 30th.

Single plant plot

Mean of two five plant plots

l-10 lo-20 2630 3WlO 40-50 XL-60 6G70 7G80 8&90 90-100

9&100% 1 2 ; i 4

2 1 9 17 8& 90%

i i 2 6 4 8

7& 80% ; 2

3 4 2 8 10 1 6s 70% . 2 3 10 4 2 2 1 50- 60% 2 4 5 3 3 5 6 1 1 40- 50% 3 4 5 3 6 2 2 2 2 1 3fL 40% 6 4 2 4 3 3 4 4 2s 30% 2 4 6 7 4 6 i : lf& 20% 6 7 3 3 6 1 2 i l- 10% 11 6 5 3 1 2 1 i

plant. Similarly 22 clones would have been selected as a single plant but discarded on the basis of the mean total tuber weight of two live plant plots. In summary, from the 60 clones which would have been selected on the basis of the mean total tuber weight from two five plant plots, only 63% of these would have been selected on their performance as a single plant plot.

Although the clones used in this study were a random sample, in that no deliberate selection pressure had been exerted on them, each clone had been assessed for breeders’ preference the year prior to this experiment (1983). If the clones are grouped into four classes (at most 3; greater than 3 but at moest 4; greater than 4 but at most 5; and greater than 5) according to the score in 1983, the similarity within each group between performance as a single plant and the mean of two five plant plot can be examined. The correlation coefficients for breeders’ preference (Table 5) are found to decrease in magnitude as the preference scores of the group in the previous year increased. For example, the correlation coefficient for breeders’ preference scores on the group which were assessed to be at most 3 in 1983 was 0.70, while the corresponding correlation coefficient of the group which were assessed as greater than 5 in 1983 was only 0.33. There was a suggestion of a similar relationship, although not so strong, for total tuber weight. The correlations for the control cultivars in the experiment

Table 5. Correlation coefficients between single plant plots and the mean of two five plant plots for breeders’ preference, total tuber weight, mean tuber weight and number of tubers per plant. Clones are grouped according to breeders’ preference scores in the previous year.

>354 >455 >5 Control (n = 106) (n = 64) (n = 31) (n = 24)

Breeders’ preference 0.70 0.50 0.43 0.33 0.46 Total tuber weight 0.74 0.57 0.50 0.65 0.13 Mean tuber weight 0.49 0.52 0.48 0.56 0.45 Numbers of tubers per plant 0.65 0.59 0.60 0.66 0.39

Euphytica36 (1987) 715

J. BROWN

were generally small in comparison to those for the random sample of clones. There was no evidence that the correlations differed in the different groups for mean tuber weight or number of tubers per plant.

DISCUSSIONANDCONCLUSIONS

The error variances of the control cultivars, grown in four plots in each block for the yield characters were significantly larger from the single plant plots than from the five plant plots. The error variances of the controls were not however significantly different for breeders’ preference. Also, there was greater agreement between the two replicates of the five plant plots than between the single plant plots with the mean of both five plant plots for all characters examined. Therefore, greater accuracy was achieved by assessing two five plant plots compared to assessing single plant plots.

Correlation coefficients obtained by correlating data recorded on a single plant plot with mean data recorded from two five plant plots were of greater magnitude than those reported for breeders’ preference (BROWN et al., 1986) and yield characters (CALI- GARI et al., 1986). Both these studies involved comparing single plant plot observations from first clonal year plots with two five plants grown in the third clonal year. In the experiment reported here, the correlations would be expected to be larger because the single and five plant plots were both grown in the same field and in the same year. However the greater correlations may also be due to using tubers produced from plants that had been grown in the field for both plot sizes, whereas for the correlations reported by BROWN et al., 1986) and CALIGARI et al., (1986) single plant plots were grown from tubers produced by seedlings that had been grown in pots in a glasshouse.

Despite the relatively high correlation coefficients obtained between single and live plant plots, examination of the distribution of scores from both plot sizes showed that many clones would have been discarded on the basis of a single plant observation, but would have been selected on the data recorded on two five plant plots. Examination of the selection ratio i.e. (the proportion of clones selected in single plant plots that were selected on the basis of two five plant plots divided by the proportion of clones discarded in single plant plots that were selected in live plant plots; see BROWN et al., 1986) showed that a genotype that was selected on a single plant basis, was 2.69 times more likely to be selected ([g 1 $&I) if g rown in two five plant plots, than a clone that was discarded on the basis of a single plant plot.

The size of plot needed to differentiate ‘good’ clones from ‘poor’ ones in likely to be associated with the heritability within the population that is being selected. In this experiment the population had not been selected, and the correlations of breeders’ preference and total tuber weight were positive and fairly high. When the total number of clones examined was partitioned according to breeders’ preference scores from the previous year, the correlation for breeders’ preference between single plant plots and the mean of two five plant plots was found to decrease as the previous years’ assessment increased. The correlations for total tuber yield decreased in a similar way. It should however be noted that total tuber yield and breeders’ preference scores are highly correlated (BROWN & CALIGARI, 1985). Also it is interesting that the control cultivars showed lower correlations, and it is reasonable to assume that these cultivars had previously been selected, not only for breeders’ preference, but also for yield charac-

716 Euphytica 36 (1987)

SINGLEPLANTSELECTIONINPOTATOBREEDING

ters. There is therefore a suggestion that a single plant plot is suitable for clonal selec- tion when the population is completely unselected and the plots are planted using field grown seed tubers.

Replication effectively raises the broad sense heritability (AIKMAN & LANGTON, 1983). However an increase in replication, or plot size, will result in either greater areas of land being needed in order to assess a set number of genotypes, or alternative- ly, a reduction in the total number of genotypes that can be screened in a fixed area of land. Obviously more genotypes can be evaluated in the same area of land if the plot size is kept to a minimum, without replication. AIKMAN & LANGTON (1983) were of the opinion that the most effective method, in the initial stages of selection, is to assess the maximum number of genotypes in single plant plots. Bos (1982) agreed that non-replicated trials are optimal if the broad sense heritability is at least 0.5. In the experiment reported here, approximately 3000 genotypes could have been grown in the same land area that was used to grow 300 clones in five plant plots, with two replicates. Also, if only 60 clones are to be retained for re-trial, then the ‘best’ 20% of the 300 clones would be selected but only the ‘best’ 2% of the 3000 would be kept. If the top 10% of clones were selected on the basis of total tuber weight of a single plant, then all but one of these clones were found within the top 60% of the two five plant plots (Table 4). It should however be noted that land availability may not be the limiting resource. Single plant plots are invariably harvested by hand digging while plots can be mechanically harvested. Also, if each individual plot requires to be weighed, the resources to handle 3000 weighings may not be available. It will take little more effort or time to weigh the produce from a five plant plot than it would to weigh the produce from a single plant.

In conclusion, single plant selection in potato breeding will result in the loss of many potentially valuable clones. Also, clonal selection on single plant plots will be most effective when the population has not been selected. However, once selection pressure has been applied, or should the initial population be less diverse than the one examined here, then more than a single plant (i.e. greater replication or larger plots) will need to be grown to select superior genotypes with any degree of certainty.

REFERENCES

AIKMAN, D. P. & F. A. LANGTON, 1983. Replication in initial selection trials of clonally propagated crops. Euphytica 32: 821-829.

ANDERSON, J. A. D. & H. W. HOWARD, 1981. Effectiveness of selection in the early stages of potato breeding programmes. Pot. Res. 24: 289299.

BLOMQUIST, A. W. & F. I. LAUER, 1962. First clonal generation potato progeny performance at two Minneso- ta locations. Am. Pot. J. 39: 46&463.

Bos, I., 1982. The optimum number of replicates when testing lines or families on a fixed number of plots. Euphytica 32: 3 1 l-3 18.

Brown, J., P. D. S. Caligari, G. R. Mackay & G. E. L. Swan, 1984. The efftciency of seedling selection by visual preference in a potato breeding programme. J. Agric. Sci. 103: 339-346.

BROWN, J. & P. D. S. CALIGARI, 1986. The efticiency of seedling selection for yield and yield components in a potato breeding programme. Z. Pflanzenziichtg 96: 5362.

BROWN, J. P. D. S. CALIGARI, G. R. MACKAY & G. E. L. SWAN, 1986. The efficiency of visual selection in early generations of a potato breeding programme. Ann. Appl. Biol. 110: (in press).

CALIGARI, P. D. S., J. BROWN & R. J. ABBOTT, 1986. Selection for yield and yield components in the early

Euphytica 36 (1987) 717

J.BROWN

generations of a potato breeding programme. Theor. Appl. Genet. 73: 218-222. DAVIES, H. T. & G. R. JOHNSTON, 1965. First clonal generation potato seedling selection at two locations.

Am. Pot. J. 42: 186189. DAVIES, H. T. & G. R. JOHNSTON, 1968. Second clonal generation potato seedling selection at two locations.

Am. Pot. J. 45: 15&153. DAVIES, H. T. & G. R. JOHNSTON, 1974. Reliability of potato selection in the first clonal generation. Am.

Pot. J. 51: 8-11. FITSCHEN, H. J., 1984. Ztichtung neuer kartoffelsorten. Kartoffelbau 35: 52-55. HOLDEN, J. H. W., 1977. Potato breeding at Pentlandfield. In: Scottish Plant Breeding Station Ann. Rep.

197677 pp. 6667. HOWARD, H. W., 1962. Some potato breeding problems. In: Plant Breeding Institute Ann. Rep. 196162.

pp. 5-21. MARIS, B., 1964. The modifiability ofcharacters important in potato breeding. Euphytica 15: 18-31. PFEFFER, C., 1963. Vergleichende Untersuchungen uber Auslesemdglichkeiten von im Freiland und in Top-

fen kultivierten kartoffelslmlingen. Ztichter 33: 611. SCHOLZ, M., 1986. General stratey ofpotato breeding in the German Democratic Republic. In: Proc. EAPR/

EUCARPIA breeding and variety assessment meeting. 1985, Camb. UK. SNEDECOR, G. W. & W. G. COCHRAN, 1974. Statistical methods. The Iowa State Univ. Press Aves, Iowa,

USA. p. 296297. TAI, G. C. C., 1975. Effectiveness of visual selection for early generation seedlings of potato. Crop. Sci.

15: 1518. TAI, G. C. C. & D. A. YOUNG, 1984. Early generation selection for important agronomic characters in

a potato breeding population. Am. Pot. J. 61: 419434.

718 Euphytica 36 (1987)