Ann. appl. Biol. (1987), 110, 357-363 Printed in Great Britain

357

The efficiency of visual selection in early generations of a potato breeding programme

BY J. BROWN, P. D. S. CALIGARI*, G. R. MACKAY AND G. E. L. SWAN Scottish Crop Research Institute, Pentlandfield, Roslin,

Midlothian, Scotland, EH25 9RF

(Accepted 1 September 1986)

S U M M A R Y

The efficiency of visual selection in the early generations of a potato breeding programme is examined. Tubers from 571 potato genotypes were scored by four breeders after being grown, from true seed, in a glasshouse and in the field for three consecutive years. The four breeders were in good agreement as to which clones would be selected in any environment. The association between breeders was greatest when the clones were grown in plots in the field. All correlations of breeders’ preferences between different environments were significantly greater than zero, but only accounted for a small percentage of the total variation. Selection in both the glasshouse and first clonal year produced a desirable response. However, such selection carried a high cost in terms of losing clones with commercial potential. Comparison of a random sample of clones with ones from the same crosses which had been selected, indicated that selection in the glasshouse and first clonal year was at best random with some suggestion, however, of a negative effect.

INTRODUCTION

Many potato breeders regard the first year of a breeding scheme, the seedling year (grown from true botanic seed in a glasshouse or under an aphid-proof screen), only as a means of producing healthy tubers to be planted, grown and assessed in the first clonal year (Howard, 1962; Maris, 1966; Pfeffer, 1963; Krug, Wriedt & Weber, 1974a, b ; Anderson & Howard, 1981; Tai & Young, 1984). The efficiency of visual selection at the seedling stage was examined by Brown, Caligari, Mackay & Swan (1984) with reference to the first clonal year. They found that selection had some effect but pointed out that this was likely to have been overestimated due to mother tuber effects. Such effects have also been noted by Blomquist & Lauer (1962).

Previous studies on selection in the first clonal year have given equivocal answers as to the desirability of carrying out such selection (for instance, Tai, 1975; Maris, 1962). Both these authors suggested that the correlations found did merit positive selection being practised, but suggested that only the very poorest genotypes should be discarded. This is, however, somewhat imprecise when it comes to its practical application in a potato breeding programme. The present study examines a number of potato clones grown from true seed in the glasshouse and subsequently, from tubers, in the field for three subsequent years. The genotypes were a random sample from each of the eight crosses referred to by Brown et al. (1984). The results of selection upon these genotypes were also compared with those on genotypes from the same progenies surviving selection in the normal commercial breeding

Present address : Department of Agricultural Botany, The University, Whiteknights, Reading, U.K. RG6 2AS

0 1987 Association of Applied Biologists

358 J . B R O W N , P . D . S . C A L I G A R I , G . R . MACKAY A N D G . E . L . S W A N

scheme at the Scottish Crop Research Institute (SCRI), the principal features of which are common to many potato breeding schemes (Brown et al., 1984).

MATERIALS A N D METHODS

The genotypes examined were derived from eight crosses chosen to be representative of the material that would normally be screened by the Potato Breeding Department at SCRI. Details of the potato breeding scheme used at SCRI are described by Mackay (1982) and Holden (1977).

The growing conditions in the glasshouse and first clonal year are detailed by Brown et al. (1984) and will only be briefly described here. Two hundred seedlings were grown to maturity in an aphid-proof glasshouse. Each genotype was harvested individually and assessed on a 1-9 scale of increasing preference by four potato breeders (Bl-B4). All genotypes which produced tubers were retained and planted at two locations, Blythbank farm, Peebleshire (BB) and the Murrays farm, East Lothian (MURR) in the first clonal year. At each location the trial was completely randomised, each plot being a single plant, and where sufficient tubers were available each clone was replicated twice. At both sites tubers were planted 47 cm apart with a gap of 98 cm after every tenth plant. Drills were 78 cm apart. At harvest the same four breeders who had assessed the seedlings, scored each piant, again on a 1-9 scale of increasing preference depending on the commercial appearance of the tubers, thus taking into account all tuber characters that each breeder perceived as important.

It was not possible to retain all the genotypes for planting in the second clonal year; therefore a random sample of at least 70 clones was taken from each of the eight crosses and these were grown at both BB and MURR in 1983. In the second clonal year each clone was represented by two three-plant plots at each location, subject to the availability of tubers produced in the first clonal year. Each trial consisted of two completely randomised blocks, At harvest the produce from each plot was again visually assessed in the field by the four breeders.

In general a preference score of 5 or more indicated that a particular clone would have been selected in the normal breeding scheme. Between successive years no clones were, in fact, discarded from the experiment other than on a random basis. The data presented are from observations on 571 clones which were grown and assessed in each of the three years.

The growing conditions and plot sizes used in this experiment were those normally used in the potato breeding scheme at SCRI. The experiment differed however in that normally the first clonal year and the second clonal year plots are not replicated and would only be grown at BB. Furthermore, only selected genotypes are retained for further evaluation the following year in the normal breeding scheme.

RESULTS

A comparison of the scores for the four breeders are presented as correlation coefficients in Table 1 for the glasshouse (GH), the first clonal year (FCY) and the second clonal year (SCY). Each correlation is based on observations of 571 clones and all the coefficients are highly significantly different from zero. The average correlation coefficient increases from glasshouse through first clonal year to second clonal year, accounting for 42%, 54%, and 63% of the total variation respectively. Clearly, there is considerable agreement amongst the breeders in their assessment of the clones, this being best in the second clonal year and poorest in the glasshouse.

The data from the 571 clones can also be compared over years. The comparisons for each breeder, together with the means over breeders, are presented in Table 2 as the correlation of

Visual selection in potato breeding 359

Table 1. Correlations between the preference scores offoour potato breeders (Bl-B4) assessing the same genotypes in the glasshouse (GH), first clonal year (FCY) and second clonal year

(SC Y)

B2 GH FCY SCY

FCY SCY

FCY SCY

B3 GH

B4 GH

0.6 1 0.77 0.80 0.55 0.74 0.76 0.76 0.84 0.83 0.5 1 0.72 0.74 Average GH = 0.645 0.69 0.68 0-77 FCY = 0.738 0.76 0.75 0.79 SCY = 0.795 B1 B2 B3

Table 2. Correlations between glasshouse (GH), jirst clonal year (FCY) and second clonal year (SC r) for four breeders’ (BI-B4) preference scores

GH u FCY GH u SCY FCY u SCY

B1 B2 B3 B4

0.313 0.283 0.560 0.369 0.304 0.507 0.487 0.367 0.610 0444 0.338 0.577

Mean (four breeders) 0.483 0.363 0.640

glasshouse with first clonal year, glasshouse with second clonal year and first clonal year with second clonal year. The ranking of the three correlation coefficients is consistent over breeders, and hence over their means, and shows

~ F C Y v SCY> rcn v FCY> ~ G H v SCY.

Thus these correlations, like those in Table 1, show a general trend of increasing correlation with increasing years. Such a trend could be a reflection of increasing numbers of clonal generations after the growth from true botanic seed or the increasing ‘plot’ sizes. Although both explanations are possible, if this trend was simply due to plot size changes then it would be expected that the correlation of glasshouse with first clonal year would be smaller than glasshouse with second clonal year whereas, in fact, the reverse is true.

The preference scores of the four breeders were averaged and used to examine which clones would have been selected over the years of the experiment. As previously noted, the 1-9 scoring scale was originally defined so that a score of 5 or more would indicate that a clone would have been selected in a normal breeding scheme. When the scores of four breeders are averaged, however, the point of delineation between selection and rejection becomes less clear. For this reason it was decided to choose cut-off values on the 1-9 scale which would give approximately the same percentage of clones selected in each of the years. Thus in the glasshouse a mean preference of 2 5.0 was taken as indicating a clone was selected (giving approximately 24% of clones selected), while in the first clonal year and second clonal year means 2 4.0 were taken (giving approximately 23% and 20% respectively).

The numbers of clones which would have been selected and rejected at the different stages are given in Table 3. The results clearly show that in both sets, as many clones are selected in the second year from tile group that was previously rejected as from the one that was previously selected. However, if frequencies are considered, then a higher frequency of clones

360 J . B R O W N , P. D . S . CALIGARI, G . R . MACKAY A N D G . E . L . S W A N

Table 3. Number of clones selected or rejected in glasshouse (GH) followed byJirst clonal year (FCY) and infirst clonal year (FCY) followed hy second clonal year (SCY)

Selected 65 50

Rejected 37 1 85 FCY

Rejected Selected GH

Selected 51 64

Rejected 390 66 SCY

Rejected Selected FCY

Table 4. The observed and expected number of clones selected (+) or rejected (-) in glasshouse (GH), first clonal year (FCY) and second clonal year (SCY). The expected numbers were obtained as the product of the overallfiequencies of selection or rejection at each

stage multiplied by the total number of clones (a) All possible categories GH FCY SCY

+ + + + +

+ + + t

+ +

-

- -

- - - - + - -

Observed

31 30 19 33 55 36 32

335

Expected

6.19 24.55 2 1 .oo 19.99 83.27 79.27 67.82

268.92

(b) Grouped according to number of years in which a clone was selected Observed Expected (0) (E) (OHE)

Selected in 3 years = 31 6.19 24.8 1 Selected in 2 years = 82 65.54 16.46 Selected in 1 year = 123 230.36 - 107.36 Selected in 0 year = 335 268.92 66.08

is selected in the second year from among the ones that had previously been selected than the ones rejected. Indeed the ratio of these frequencies is 3.36 between glasshouse and first clonal year and 7.4 between first clonal year and second clonal year. In other words a genotype is 3.36 times more likely to be selected in the first clonal year if it was selected in the glasshouse than if it was rejected and 7.4 times more likely to be selected in the second clonal year if it was selected in the first clonal year.

It would appear that the frequency of clones considered acceptable is increased by selection in the previous year. Unfortunately a considerable number of desirable clones are also lost in such a process. Taking the examination a stage further, the number of clones which were selected or rejected in each of the three years are combined in Table 4(a) to show the observed numbers that fell into the eight possible categories. Also shown are the expected numbers for each category which were estimated from the frequency of selection or rejection, multiplying over years (ie. assuming independence) to give the expected frequency and finally multiplying by the total number of clones to give the expected number, e.g. for the category ‘selected in the

Visual selection in potato breeding 36 1

glasshouse, rejected in the first clonal year and selected in the second clonal year’, the expected number is:

23.641100 x 77.231100 x 20.14/100 x 571 = 20.9956 clones.

A comparison of observed and expected numbers gives a highly significant x 2 indicating that selection has had some effect. Examination of Table4(b) shows the departures are produced by an excess of observed clones falling particularly into the categories ‘selected in all years’ and ‘rejected in all years’, indeed only the clones falling into the category ‘selected only in one year’ apear to be under-represented. Clearly, selection is producing a deviation in the desired direction. However, inspection of the numbers involved shows that selection has not been particularly effective. Also, if it is considered, as would seem reasonable, that the scores in the second clonal year are the most accurate assessment of the worth of a clone, then 11 5 (31 + 19 + 33 + 32) clones were deemed worthy of selection by the breeders but only 31 of these were selected in both the previous years. In other words, of the 115 clones, 84 would have been rejected in the normal breeding scheme and would not be available for assessment at all in the second clonal year (i.e. 73%). Even if selection had not been practised in the glasshouse, 51 of the 115 clones would still have been rejected.

DISCUSSION A N D CONCLUSIONS

Brown et al. (1984) concluded that selection in the glasshouse would have some desirable effects as judged by the results in the first clonal year. They pointed out however, that selection in the glasshouse was influenced by the weight of tubers produced. This in turn bore a direct relationship to weight of the mother tuber planted and was significantly correlated with the probability of selection in the first clonal year. Thus, when taken overall, the observed correlation of glasshouse with first clonal year would be inflated and they concluded selection would have only marginal effects. The data presented here support this view in showing the correlation between the glasshouse and second clonal year to be lower than that with the first clonal year. Since plots in the second clonal year were larger than in the first clonal year it would be expected that the correlation would increase, or at worst stay constant, if no other effects were present. However, the estimated correlation was lower and indicates that the reducing carry-over effect of the mother tuber has lowered the apparent correlation between glasshouse and field.

The four breeders were in reasonable agreement on which clones would be selected in any environment particularly when the material was grown in the field. Indeed, it was highest in the second clonal year when the material was growing in the field and the plots were bigger than the first clonal year. Greater discrepancies between breeders in the earlier clonal generations therefore add to the inefficiency of selection.

The results presented here not only indicate that selection in the glasshouse is not very efficient but also raise the question of the efficiency of selection in the first clonal year. If selection had been practised in the glasshouse and first clonal year then of the 115 clones considered worthy of selection in the second clonal year, only 27% would have been retained to that stage. Had selection been carried out only in the first clonal year then still only 56% would have been retained. Thus selection in the first two generations would seem to carry a high cost in terms of losing clones which have, in fact, commercial potential.

The potential loss of material due to selection being carried out in the seedling and early clonal generations can be examined directly. At the same time as this experiment was set up, seeds of each of the same eight progenies were sown in the SCRI ‘Commercial Potato Breeding Programme’. The number of clones from one of these crosses as it passed through clonal generations is shown in Table 5 for the ‘commercial’ scheme and the experiment. In the

362 1. B R O W N , P . D . S . C A L I G A R I , G . R . MACKAY A N D G . E . L . S W A N

Table 5. Comparison of clones from a cross passing through the SCRI commercial breeding programme and in the experiment where they were reduced in number at random, when assesspd in glasshouse (GH) followed by first clonal year (FC Y), second clonal year (SCY) and

third -ckonal year (TC Y)

GH I98 I -seedlings

FCY 1982-single plant

SCY 1983-3 tuber plot

TCY 1984-2 x 5 tuber plots

1985-2 x 10 tuber plots

Commercial breeding programme Experiment

200

I I

158

(random) (random (selected) +selected)

I l l 17 8 17

l i 4 4 7

7- 1 1

normal breeding scheme where selection was carried out annually, 18 clones survived to the third clonal year and this reduced to five based on selection in the third clonal year. The experimental sample started with 200 seedlings but these were reduced at random to 70 by the second clonal year. On the basis of the second clonal year results, 25 clones were selected as having commercial merit and also 25 were taken at random, of which eight were common to the two groups. These were grown in the third clonal year alongside those from the breeding scheme. Of the 25 selected clones 1 1 were again selected while out of the ‘random group’ eight were selected. Therefore, if the breeding scheme had started with 70 which were all retained with no selection to the second clonal year, 1 I would have survived the next two rounds of selection, i.e. 1 in 6.4. However in the commercial scheme starting with 1000 seedlings and carrying out annual selection, only five clones have survived, i.e. 1 in 200. If it is assumed that the reduction from 1000 seedlings to 59 in the second clonal year is merely reducing the numbers at random (i.e. equivalent to the experiment where they were deliberately reduced at random) then of those 59, five were selected, i.e. 1 in 111.8, a ratio somewhat worse than that from the experiment. With such small numbers it is unwise to place too large an emphasis upon the apparent decrease but clearly there is no evidence of any positive effect for selection in these two early generations when examined in a practical breeding scheme.

The other seven progenies were also sown and assessed in the commercial scheme in parallel to their use in the experiment. However, no clone from any of these progenies survived commercial selection to be examined in the fourth clonal year. From the experiment, on the other hand, 45 clones were considered good enough to be trialled at this stage. Thus the results very clearly demonstrate that, at best, selection on the seedlings and in first clonal year is random, but there is an indication of a negative element involved. Indeed, when foliage maturity data on a 1-9 scale (where 1 = early, 9 = late) from the experiment was taken into account it was found to correlate negatively (r = --0.17, P < 0.05) at Blythbank with preference score. In the commercial scheme this is the only site used for the first clonal year

Visual selection in potato breeding 363

material and is a high-grade seed site. Such sites are generally planted later and harvested earlier than norma1 ware sites, where the potatoes for consumption are ultimately destined to be grown. The negative correlation is therefore perhaps not surprising since clones which mature very late will tend to produce small size tubers by the time the plants are harvested at a seed site. At a ware site, however, these later-maturing clones are likely to be acceptable, possibly even more acceptable than their earlier maturing counterparts.

The conclusion from this study is that in the early generations, all the time and effort spent on selection is producing an effect which is, at best, the same as random choice of the clones. There is, however, some indication that random choice is somewhat better and that selection is having a slight negative effect. Since selection is carried out in the seedlings in the glasshouse and the first clonal year at a ‘seed-site’ this is perhaps not surprising and may represent an effect similar that found by Hayward & Vivero (1984) where selection was carried out on spaced plants in a ryegrass breeding programme although ultimately they were to be grown in sward conditions where the density would be much higher.

R E F E R E N C E S ANDERSON, J. A. D. & HOWARD, H. w. (1981). Effectiveness of selection in the early stages of potato

breeding. Potato Research 24, 289-299. BLOMQUIST, A. W. & LAUER, F. I. (1962). First clonal generation potato progeny performance at two

Minnesota locations. American Potato Journal 39, 460-463. BROWN, J . , CALIGARI, P. D. s., MACKAY, G. R. SWAN, G. E. L. (1984). The efficiency of seedling selection

by visual preference in a potato breeding programme. Journal of Agricultural Science, Cambridge 103,

HAYWARD, M. D. & VIVERO, J . L. (1984). Selection for yield in hliumperenne. 11. Performance of spaced plant selection under competititve conditions. Euphytica 33, 787-800.

HOLDEN, J . H. W. (1977). Potato breeding at Pentlandfield. Report of the Scottish Plant Breeding Station for 1976, pp. 66-97.

HOWARD, H. W. (1963). Some potato breeding problems. Report of the Plant Breeding Institute, Cambridge for 1962, pp. 6-20.

KRUG, H., WRIEDT, G. & WEBER, w. E. (1974~). Untersuchungen zur Fruhselektion in der Kartoffelzuchtung I . Untersuchungen an der Samlingsgeneration. Zeitschrift f i r Ppanzenziichtung

339-346.

73, 141-162. KRUG, H., WRIEDT, G. & WEBER, W. E. (1974b). Untersuchungen zur Fruhselektion in der

Kartoffelzuchtung 11. Klongenerationen. Zeitschrgt fur Ppanzenziichtung 73, 163-193.

Scotland. Proceedings of the Scottish Society for Crop Research, Bulletin No. I , pp. 27-36.

Merkmalsbeziehungen zwischen den Generationen und innerhalb der

MACKAY, G. R. (1982). Breeding for resistance to pests and diseases. In Producing quality seedpotatoes in

MARIS, B. (1966). The modifiability of characters important in potato breeding. Euphytica 15, 18-31. MARIS, B. (1969). Studies on maturity, yield, under water weight and some other characters of potato

PFEFFER, c. (1963). Vergleichende untersuchungen uber auslese-moglichkeiten von im Freiland und in

TAI, G. C. C. (1975). Effectiveness of visual selection for early clonal generation seedlings of potato. Crop

TAI, G. c. c. &YOUNG, D. A. (1984). Early generation selection for important agronomic characteristics in

progenies. Euphytica 18, 297-319.

Topfen Kultivierten Kartoffelsamlingen. Zuchter 33, 6-1 1.

Science 15, 15-18.

a potato breeding population. American Potato Journal 61, 419-434. (Received 23 May 1986)