Selection in segregating generations of autogamous species II. Response to selection in two stages

Summary In autogamous species the optimum number of families is calculated for a selection program in two stages conducted in segregating generations. The influence of epistasis is not large. In case of much epistasis the number of families at the first stage must be reduced. In comparison with selection in later generations the second stage is relatively important. Two stage selection is more efficient than combined selection.     

Comparison of selection procedures in breeding for seed yield in segregating sesame populations

Summary Four hundred sesame (Sesamum indicum L.) lines selected by five methods from five crosses were tested for yield and other agronomic characters in F₆ and F₇. More high yielding lines were isolated by the bulk method and in the cross involving two local cultivars when two seasons means and performance in each individual season were considered. The cross involving two exotic cultivars did not produce a single line superior to local recommended variety. The 80 highest yielding lines recorded a 10.8% increase in yield compared to the mean of 400 lines. Thirty five lines were among the top 20% both seasons with a 13.2% increase in yield. Highly significant seasonal and population effects were recorded. There were no significant differences among the methods of selection although the yield differences among lines were highly significant. Interactions of seasons × populations, seasons × methods, populations × methods and seasons × populations × methods were highly significant. The results indicate that simple and less expensive bulk method may be used with equal success as the pedigree, single seed descent or early generation testing methods when selecting for yield in segregating sesame populations. The importance of inclusion of a high yielding, locally adapted cultivar in the crossing programmes for yield breeding of sesame has also been revealed.Abbreviations BM Bulk method - EGT Early generation testing - G × E Genotype × environment - MSS Modified single seed descent - PM Pedigree method - SSD Single seed descent     

Selection for yield in early generations of self-fertilizing crops

Summary If a breeder is to apply selection to a cross-population of a self-fertilizing crop and wishes to maximize his chances of finding a productive genotype, he will have to introduce yield tests in the F₃ lines. Should he not do so because of his intention to work on many populations simultaneously, his tests would be considerably less accurate and the finding of a favourable combination more or less a matter of mere chance.Bulk breeding will result in genetic drift or numbers of plants too large to manage, or in a combination of both.Application of the Single Seed Descent for quantitative characteristics causes such a genetic drift, that the method is not to be recommended.Multiple crosses (double-crosses) to improve quantitative characteristics require populations of a size, that cannot be realized and/or selected. This type of crosses is more suited for cases in which qualitative characteristics form the breeding goals and when it is impossible to obtain the favourable combination using only two varieties.Genic recessive male sterility may be a means to raise large progenies. In the F₃ it confuses the issue of yield of fruits or seeds, which is highly vulnerable in any case. In the F₂ its significance for finding the proper recombinant is very low.     

Honeycomb selection for yield in early generations of spring wheat

Summary The effectiveness of the honeycomb selection method for yield in spring wheat (Triticum aestivum L.) was evaluated using progenies from two wheat crosses, Glenlea x NB131 and Glenlea x Era. Honeycomb selection was carried out in the F₂ and F₃ generations, grown at the University of Manitoba in the summers of 1980 and 1981, respectively. In both generations, divergent selection was made for both high and low yield. Plants selected in the F₃ generation were entered in an F₄ yield test in the summer of 1982. Results of the experiment showed that honeycomb selection for yield in the F₂ and F₃ generations was effective in identifying parents of high- and low-yielding lines. F₃ plants from highyielding F₂ selections gave higher yields than those from low-yielding F₂ selections by 11.5% and 13.0% for Glenlea x NB131 and Glenlea x Era crosses, respectively. The F₄ yield test showed that high yielding selections from both crosses significantly outyielded by 8.9% low yielding selections and by 14.4% the unselected composite lines. It is concluded that the honeycomb selection method can be used for early generation selection in spring wheat.     

Number and size of cross progenies from a constant total number of plants manageable in a breeding program

Summary This short paper discusses the optimum number of crosses and size of progenies for a given total population size to minimize the risk of not finding any favourable genotype in the breeding material. It is shown that the number of crosses must generally be as large as possible. In special situations it may be advantageous to include also less promising crosses instead of enlarging the progenies of more promising crosses.     

Evaluation of Pedigree and Single Seed Descent Selection Methods for Cultivar Development in Cowpea (Vigna unguiculata L. Walp)

Three crosses of cowpea (Vigna unguiculata [L.] Walp) were used to assess the effectiveness of two selection procedures in cowpea. The selection procedures were pedigree selection (PD) and single seed descent selection (SSD). Lines developed through each procedure were yield tested and selected at F₆ and F₇ for PD and SSD, respectively. The selected lines for both procedures were compared at F₈ for grain yield per plant (YLD), number of pods per plant (NPO), and pod development period (PDP), in two locations. Lines developed through both procedures differed more in their population means than in the performance of their top 10% or highest yielding lines. There was no consistent relationship between magnitude of genetic variability and percentage of superior lines obtained from each procedure. Both procedures have been effective in producing superior genotypes for yield and number of pods. The PD method produced superior transgressive segregants in two out of the three crosses for PDP. SSD allowed a more rapid generation than PD.     

Sample size required for marker assisted selection in improving quantitative traits of self-fertilizing species

The efficiency of selection for desired trait genotypes in a molecular marker assisted selection for a quantitative trait in self-fertilizing crop is considered. The QTLs controlling the trait were assumed to be unlinked. It was supposed that the selection starts in F₂, derived from a cross between inbred lines, and this selection will terminate if one or more plants with the desired trait genotype is found. If no plant with the desired trait genotype is found in F₂ then the selection is continued in the F₃ progeny that is derived from a single selected F₂ plant. Which F₂ plant is to be selected was determined according to the rank which is related to the marker genotype of the F₂ plants. And this rank was based on the expected frequency of the desired trait genotype in the progeny. The plant with the top rank among all F₂ plants is then selected with the first priority. Additionally the number of F₃ plants in the progeny was set to be equal to the number of plants that are required for detecting one or more plants with the desired trait genotype with a given probability. The probability of getting at least one plant with the desired trait genotype is expressed as a function of the number of F₂ plants (N).The required value for N and the total number of plants (T) in F₂ and F₃ for detecting at least one plant with the desired trait genotype were calculated for different situations. T was always smaller for a single marker than for flanking markers. The minimum of T and monotonous decrease of N can be observed when the cumulative-expected-frequency of selected marker genotypes of F₂ plants increased. This revised version was published online in July 2006 with corrections to the Cover Date.     

Recurrent selection for grain yield in early generations of two barley populations

Summary Two spring barley composites, one based on eight West-European two-rowed cultivars (A) and one, the predominantly six-rowed composite XXI, based on several thousands of barley cultivars (B), formed the starting point of a recurrent selection procedure. The aim was to study whether recurrent selection in early generations is an effective procedure to improve barley populations for agronomic characteristics, especially grain yield.After two cycles of recurrent selection in the two populations A and B separately, and consisting of single plant selection followed by line selection, one cycle of recurrent selection was applied to the population produced by intercrossing the selected A and selected B lines.The selection for grain yield in the F2 single plant stages and the F3 line stages was carried out by selecting the plants or lines with the largest mean head weight provided that they did not tiller too poorly. Also some selection was applied against extremes in heading date and against tall plants.After two cycles of recurrent selection the grain yield, tested over two years and two locations, had increased with 16.5% in population A and with 27.0% in population B. Both the single plant selection and the line selection had contributed significantly to this response.The population created by crossing the selected A with the selected B lines showed a significant reduction in yield compared to the mean yield of the A and B lines constituting this population. Single plant selection, followed by line selection did raise the yield level 5.7% above that of the mean of these A and B lines. Several lines produced from this intercross combined an excellent grain yield with outstanding resistance to barley leaf rust and powdery mildew. However, these lines like the entire intercross population suffered from susceptibility to lodging, a characteristic derived from population B.In five experimental situations mixtures of genotypes were compared with the mean of the monocultures of the constituting genotypes. The mixture yielded always more, the average mixture effect being 4.5%.     

Comparison of some selection procedures and objectives in faba beans (Vicia faba L.). I. direct and index selection for seed yield

Summary Starting with two heterogeneous F₂ populations, each based on 20 cross combinations, different selection procedures were applied for four consecutive selection cycles. Three of the selection procedures involved seed yield per plant (SYP): (i) direct selection for SYP (DSY); (ii) indexselection for SYP (ISY) with plant length, the ratio of the vegetative top to plant length, end of grain filling and duration of the generative stage as auxiliary characters and finally; (iii) independent culling levels selection (ICL) for the index described above and crude protein content. For ICL-selection the attention is focussed on SYP in this paper.For all three selection procedures, predicted and realized responses were calculated and expressed as percentage of a corresponding non-selected population. The intensities of selection differed over selection methods and generations, mostly due to practical constraints.As far as predicted responses are concerned, ISY-selection seemed to be slightly superior to DSY-selection and DSY-selection to ICL-selection. This order may partly be due to differences in selection intensity. However, none of the predictions was very accurate, but the predictions of ICL-selection seemed to be the most accurate. This has been ascribed also to the relative mild selection intensity applied in ICL-selection. It is concluded that a decision in favour of a complicated selection procedure like index selection based on predicted responses only cannot be defended. As far as realized responses are concerned, DSY-selection seems to be slightly superior to the other two selection procedures, as it is the only procedure that resulted in a more or less stable positive response to selection.The auxiliary characters, especially plant length, showed a marked correlated response to all three selection procedures. In case of direct selection for SYP this seemed paradoxical with the inefficiency of index selection, but, it may reflect the differences in the genuine genotypic correlations and the estimated genotypic correlations.     

Genetics of genetic conservation. II. Sample size when collecting seed of cross-pollinating species and the information that can be obtained from the evaluation of material held in gene banks

Summary Provided that the seed collected from each plant in natural populations of a species is kept separate from that of every other, a rough estimate of the heritability of a quantitative character can be obtained from the natural progenies raised from this seed in a randomised evaluation trial; a knowledge of the reproductive biology of the species can help to make this estimate of the heritability of the character more precise.The theory employed in this evaluation procedure can be used to investigate the possibility of reducing the number of plants visited in a population, when species set all of their seed by cross-fertilisation, by taking several seeds from each. We show that it is not worth taking more than eight seeds from each plant and that, in the absence of precise information about the paternity of this seed, it might be better to follow the general recommendation of Lawrence et al. (1995), by taking only one seed from each of the 172 plants.Having investigated the minimum sample size for genetic conservation in the narrow sense, we broaden discussion to consider sample size for evaluation and regeneration. It is pointed out that it is not necessary to take more than about ten seeds from each of 20–30 randomly chosen plants in each population visited and that it is possible to reduce this number of plants if material from a number of populations is evaluated in a single trial. Finally, we draw attention to the possibility of regenerating seed from a composite population founded by raising one plant from the seed taken from each plant of the original collection, as an alternative to regenerating the seed of each accession independently from that of every other.     

Inheritance in seed size of alfalfa: Quantitative analysis and response to selection

Seed size in alfalfa (Medicago saliva L.) has been positively correlated with seedling vigour and early growth, but there were few published reports on inheritance or selection for this trait. The objective of this research was to estimate components of genetic variance for inheritance of alfalfa seed size and determine the most efficient selection method. Components of genetic variance were estimated on seed and pollen plants of ‘BIC-7-WH’ and their progeny arranged in a North Carolina Design II mating design under controlled environmental conditions. Three selection methods, differing in parental control and selection pressure, were used to determine selection response. The seed parent genotype had a major role in determining alfalfa seed size, but the genotype of the seed had no influence. For genetic studies, pollen and seed parent effects on seed size should be measured on seed harvested from progeny plants. Seed size was controlled by additive and non-additive components of genetic variance. Heritability for seed size was 41.3%. Selection for seed size was effective and a significant shift for larger and smaller seed was attained after one cycle of selection.     

玉米轮回选择基础群体的评判

将６个玉米群体完全双列杂交的得到的群体及群要璋杂交种在３个地区进行试验，采用Ｇａｒｄｎｅｒ和Ｅｂｅｒｈａｒｔ的ＡＮＡＬＹＳＩＳⅡ和Ⅲ对性状进行了分析，结果表明，许白群产量ｖｊ最高，ｈｊ最低，均达显著水平，ｇｉ中等；洛孟综产量ｖｊ和ｈｊ较高，ｇｊ最高，二南２４群产量ｖｊ较高，ｈｊ较低，ｇｊ略高。这３个群综合农艺性状较好，应进行特殊配合力的轮回选择，洛孟综可用于选育自交系，ＢＳＳＳ产量ｖｊ最小，ｈｊ最     

Genetic response to index selection for grain yield, kernel weight and per cent protein in four wheat crosses

Quantitative genetic theory was used to investigate selection differentials, expected and observed direct and correlated responses to simultaneous improvement of grain yield, kernel weight, and grain protein content in F₃ and F₄ populations of four spring wheat (Triticum aestivum L.) crosses. Selection in the F₃ generation based on the Smith-Hazel index (SH) and yield was found to be superior to the other methods studied in identifying high-yielding lines, but resulted in substantial decrease in grain protein level. Consequently, a 1.0% increase in protein from selection for protein depressed grain yield as much as 536 kg/ha below the population mean, reducing the expected yield gains per generation selection cycle by 250%. The weight-free indices (EW), particularly the index involving only yield and kernel weight (EW3) and its linear approximation (EW4), and the desired gains indices were effective in improving protein content but were less efficient in selecting top-yielding lines. Selection in the opposite direction using truncation of the lowest 10% of the population based on weight-free-index involving yield, kernel weight and protein (EW2) reduced all traits significantly compared with the mean of the unselected population indicating the effectiveness of the unweighted selection index. The observed genetic gains (Ra) from selection based on yield, the SH, EW3 and EW4 indices were slightly lower than the expected advances per cycle (R) in populations from crosses ‘Sinton’בGlenlea’ (C1), ‘Glenlea’בNB505’ (C2), and ‘A2P5’בNB320’ (C6) (Ra/R = 70 to 85%) but were higher in cross ‘NB505’בA2P5’ (C5) (Ra/R = 126 to 143%). It was concluded that weight-free and the desired gains indices can be used to improve wheat grain yield and grain protein simultaneously in F₃ generation selection, as revealed by response to selection measured in the F₄ generation.     

Comparison of pedigree selection and single seed descent for oil yield in linseed (Linum usitatissimum L.)

Summary Four populations of linseed derived from five parents were advanced from the F₂ to the F₆ generation by SSD and to the F₅ by pedigree selection. These populations were used to compare the efficiency of the two methods in order to provide superior genotypes with respect to grain yield and oil content, i.e. oil yield. The results showed minimum differences between pedigree and SSD lines for grain yield, where in only one cross the SSD lines were significantly superior to the pedigree lines. Since pedigree selection was carried out for both, seed yield and oil content, a positive response to selection was expected. However, early selection for yield, a character with low heritability, was not successful. On the contrary, significant differences for oil content were detected between the two groups of lines in three of the four crosses studied. In these cases the pedigree lines were superior to the SSD lines. These results demonstrate that strict selection in early generations for oil content, a character with comparatively high heritability, is feasible and successful in linseed. However, selection for seed yield should be postponed to later inbred generations. Consequently, in breeding for maximum oil yield of linseed a two-step selection procedure is recommended.Abbreviations PS pedigree selection - SSD single seed descent - TGW thousand grain weight     

Methods of increasing short term response to full-sib family recurrent selection in small populations

Accelerated recurrent selection (ARS), in which selection is carried out on the predicted value of the progeny rather than on the observed performance of the parents, has been proposed as a method of increasing response to selection and of reducing cycle time. ARS schemes based on test cross evaluation of full-sib families have been compared by stochastic computer simulation. The difference in genetic and economic time scales is emphasised, with the economic long term (21 years) being only 21 or fewer cycles of selection. ARS schemes are shown frequently to offer improvements over standard recurrent selection methods under these circumstances, since they allow more cycles of selection in a given time frame. Schemes with very low effective population sizes often give the greatest response to selection over the time scales considered here. It is suggested that evaluation of cumulative responses to selection over defined periods of time, either by Monte Carlo simulation or by stochastic theory, is the best method of ranking alternative selection schemes. Evaluating response to selection by deterministic methods, or by attempting to take drift into account by evaluating schemes with identical effective population sizes can be misleading. This revised version was published online in August 2006 with corrections to the Cover Date.     

Influence of sample size on precision of heritability and expected selection response in red clover

Limited attention has been given to the effect of sample size on the precision of heritability estimates and expected selection responses. The objectives of this study were to evaluate the effect of sample size on the point estimate, its sampling error and interval width. Monte Carlo simulation and non-parametric bootstrap were used to estimate parameters in a red clover (Trifolium pratense L.) data set. We conducted separate analyses for two and four replications, and for combinations of six, 12, 24, 42 and 66 families, each having three, six, nine, 12 and 18 plants. Results showed that the effect of increasing number of families and replications on reduction of sampling error and interval width was greater than that of increasing plants per family. The sampling error and interval width of parameters decreased at a decreasing rate as sample size increased. To reduce the chances of having negative estimates of family variance components and to achieve a narrow confidence interval, an experiment of 42 families, each having six seedlings in four replications would be required.     

On the number and size of cross combinations in a breeding programme of self-fertilizing crops

Summary The number and size of crosses in a breeding programme were discussed to the conclusion that the number of crosses rather than the size of a cross should be increased with a given total population size. The advantage due to this manipulation, however, is relatively small when such genotypes as improved for many (twenty-five or more) loci are to be obtained, or when a high proportion (around 0.1 or more) of available combinations have the potentiality of releasing the objective genotypes. In such a case, the ease with which crossing and cultivation can be carried out becomes the decisive factor.     

Comparative efficiency of honeycomb and conventional pedigree methods of selection for yield and fiber quality in cotton (Gossypium spp.)

The effectiveness of honeycomb pedigree selection (HPS) as compared to conventional pedigree selection (CPS) was studied in one intraspecific (G. hirsutum L. × G. hirsutum L.) cross population (population I) and one interspecific (G. hirsutum L. × G. barbadense L.) cross population (population II). Combined selection for yield and lint quality traits was applied for four cycles at two locations for population I and atone location for population II. Finally, the best F₆ lines derived by each method, together with the unselected population derived by single seed descent (SSD) and three check cultivars, were tested in comparative experiments, separately for each cross, at the same locations. In both populations the analysis of variance indicated significant differences among the groups of the material tested for seed cotton yield, mean boll weight, micronaire reading, staple length, and uniformity ratio. No significant differences were found with respect to plant height, lint percentage, and fiber strength in population I and with respect to lint percentage and fiber strength in population II. In population I, on the basis of mean seed cottonyield and number of superior lines derived by each method as compared to the check varieties and the unselected SSD population, HPS-lines were superior to CPS-lines for seedcotton yield, fiber length and boll weight. HPS-lines, however, did not differ significantly in seedcotton yield from the best check cultivar Eva. Finally, the material derived by CPS was earlier in maturity than the material derived by HPS and SSD. Also in population II, on the basis of mean seedcotton yield and number of superior lines derived by each method as compared to the unselected SSD population, HPS-lines were superior to CPS-lines. No significant differences, however, were identified between the material of the two methods for lint quality traits. It was concluded that in both populations HPS was more effective than CPS in identifying lines with high yielding ability and good lint quality. This superiority of HPS is attributed, at least partially, to its effectiveness in early generation selection. This revised version was published online in August 2006 with corrections to the Cover Date.     

Effects of competition and selection pressure on yield response in winter rye (Secale cereale L.)

Summary Three intensities of mass honeycomb selection (14.3,5.3 and 1.6%) applied to an unselected rye population gave respectively an annual yield response of 0.28, -3.69 and -5.20% at 15 cm spacing, and of 4.07, 5.39 and 8.99% at 90 cm spacing. The negative response with competition was explained by strong negative correlation between competing and yielding ability which causes positive skewness because of transposition of low yielders and strong competitors from the left to the right tail of the distribution. The positive response in the absence of competition was mainly due to the increased genotypic differentiation which allowed effective discrimination between high and low yielding genotypes. The efficiency of the selection in the absence of competition was further improved by using the honeycomb designs which adjust soil heterogeneity and application of very high family and individual selection pressures.Two cycles of mass honeycomb selection increased the population yield by 29.4%, one cycle of mass plus one cycle of pedigree honeycomb selection did so by 34.5%. The results are discussed in relation to the selection response and to the efficiency of various breeding schemes.Part of senior author's doctoral thesis     

Genetic analysis of drought stress response in rapeseed (Brassica campestris and B. napus). I. Assessment of environments for maximum selection response in grain yield

Summary Expectations of yield improvement in environments where drought was the major environmental factor limiting yields were studied in two species of rapeseed. Selection for yield in a drought stressed environment was predicted to be a more efficient selection stategy for yield improvement in dryland situations than selection in a more optimal environment, or selection based on a drought response index. The results indicate that selection for yield in a stressed environment are expected to lead to genetic advances in yield under optimal conditions as well as in a drought index. Selection under well watered conditions, on the other hand, was also expected to lead to correlated increases in yield in droughted environments but to decreases in the drought index. These results were found in both species of rapeseed grown in different water stress situations.The genetic advance in a drought response index was predicted to be greater in B. napus and marginally less in B. campestris if selection was practised for yield in a stressed environment rather than direct selection for the drought index. This was due to the higher heritability estimates in the stressed environments and the positive genetic correlations with yield.