Crop Science Journal of Natural Resources and Life Sciences Education
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Published online 2 December 2005
Published in Crop Sci 46:90-97 (2006)
© 2005 Crop Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via ISI Web of Science (1)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Tokatlidis, I. S.
Right arrow Articles by Papadopoulos, I. I.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Tokatlidis, I. S.
Right arrow Articles by Papadopoulos, I. I.
Agricola
Right arrow Articles by Tokatlidis, I. S.
Right arrow Articles by Papadopoulos, I. I.
Related Collections
Right arrow Crop Growth and Development
Right arrow Wheat
Right arrow Data Management

CROP BREEDING, GENETICS & CYTOLOGY

Single-Plant Selection at Ultra-Low Density to Improve Stability of a Bread Wheat Cultivar

Ioannis S. Tokatlidis*,a, Ioannis N. Xyniasb, John T. Tsialtasc and Ioannis I. Papadopoulosd

a Dep. of Agricultural Development, Democritus Univ. of Thrace, Orestiada, 68200, Greece
b Technological Education Inst. of Kalamata, 24100, Greece
c Hellenic Sugar Industry SA, Larissa, 41110, Greece
d Technological Education Inst. of W. Macedonia, Florina, 53100 Greece

* Corresponding author (itokatl{at}agro.duth.gr; itokatl{at}hotmail.com)


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
The goal of the study was to assess within bread wheat (Triticum aestivum L.) cultivar variation through honeycomb selection, under the ultra-low density (ULD) of 1.2 plants/m2. Divergent selection of individual plants characterized as providing high (H) and low (L) yield led to 10 H and 10 L first generation families, respectively. Further selection of high yielding plants within H families resulted in 20 second generation families. Progeny evaluation was conducted in two locations, under ULD and the typical crop density (TCD) of 500 plants/m2. Six of the first generation families were also tested, in two locations for two years, across four densities (100, 300, 500, and 700 plants/m2). Intra-cultivar selection improved yield potential per plant (i.e., expressed under low competition conditions), and there was an indication of overall crop yield potential improvement (i.e., maximum yield per unit area). Compared to the original cultivar at ULD conditions, five of the H first generation and 15 of the second generation families had significantly higher grain yield per plant (by 18 to 53%). Two of the H first generation and four of the second generation families significantly outperformed the original cultivar by 17 to 22% under TCD. Experimentation across the four densities showed that derived families exhibited less density dependence than their original cultivar, a determinant parameter for stability of performance. Results constituted evidence of low densities being more suitable for breeder's seed maintenance, so that any existing or newly developed variation is beneficially exploited.

Abbreviations: ULD, ultra low density of 1.2 plants/m2 • TCD, typical crop density of 500 plants/m2 • LSD, least significance difference


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
ELITE CULTIVARS of both cross-pollinated and predominantly self-pollinated crops are assumed to be genetically homogenous. To maintain uniformity, ear-to-row increases are made for production of breeder's seed. Thus, head rows with any "off-type" plants are eliminated, whereas the best plants within each row are selected. Nevertheless, significant genetic variation within cultivars has been reported either on the basis of phenotypic or molecular data. McClintock (1984) pointed out that the genome is dynamic and can modify itself in response to environmental stresses. Rasmusson and Philips (1997) noted that elite gene pools have inherent mechanisms to provide a continuing source of new genetic variability, thanks to the genome's plasticity.

Significant inbred line variation for several maize (Zea mays L.) traits was reported by Fleming et al. (1964) and Bogenschutz and Russell (1986). Sprague et al. (1960) found significant variation for yield and eight other agronomic traits within doubled monoploid lines. According to Peterson (1997), various factors such as gene polymorphism, duplication of DNA sequences, and transposon elements, can contribute to the vast genetic heterogeneity in maize. Gethi et al. (2002) estimated the level of genetic diversity among and within six inbred maize lines using SSR markers, and found that 7.6 and 4.6% of the total variation observed in gene frequency was among and within sources, respectively. The researchers argued that breeders should take into account the within inbred variation, especially when sampling earlier released materials.

Studies in other crops also suggest substantial intracultivar variation. Byth and Weber (1968) reported genetic variability for various agronomic traits within F5–derived soybean [Glycine max (L) Merr.] lines considered to be relatively homozygous. Gordon and Byth (1972) found significant differentiation within a tobacco (Nicotiana tabacum L.) cultivar for several agronomic traits. Molecular analysis by RFLPs in inbred lines of sunflower (Helianthus annuus L.) revealed the existence of within-cultivar variation (Zhang et al., 1995). Olufowote et al. (1997) also found within-cultivar variation in rice (Oryza sativa L.) by using microsatellite and RFLP analysis.

Undoubtedly, the magnitude of genetic variation within a cultivar is expected to be limited, and thus phenotypically recognizable under optimal conditions that maximize phenotypic expression and differentiation. According to Fasoula and Fasoula (1997, 2002), optimal conditions include low plant density, which reduces competition for environmental resources. The researchers claimed that planting at very low density reduces plant stress, allows maximum grain production per plant, and optimizes single-plant heritability by increasing genetic variance in proportion to environmental variance. Single-plant selection under very low density was effectively used to reveal the within-cultivar variation in several crops, such as grain yield of bread wheat (Fasoula, 1990), snap bean (Phaseolus vulgaris L.) pod number and yield (Traka-Mavrona et al., 2000), cotton (Gossypium hirsutum L.) yield, and tolerance to Verticillium wilt, caused by Verticillium dahliae Kleb., (Fasoulas, 2000), and soybean seed protein, oil, and fatty acid composition (Fasoula and Boerma, 2005). In recent work (Tokatlidis et al., 2004), single-generation divergent selection for individual plant yields within a bread wheat cultivar, at an ultra-low density of 1.2 plants/m2, revealed significant variation for yield, protein content, carbon isotope discrimination, and ash content. Thus, the objective of this study was to evaluate yield potential of the progeny of first and second generation selections, aiming to investigate whether variation is exploitable to upgrade the cultivar's productivity, and to assess optimal conditions for breeder's seed conservation.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Source Material and Experimental Treatments
Source material was a bread wheat cultivar (cv. Nestos), developed by the Cereal Institute of the National Agricultural Research Foundation, Greece, via pedigree selection for four successive generations (F5 to F8) within the INIA66R//Hbgn/drc material of Oregon State University. The cultivar was published in the Greek National catalog and the catalog of the European Union in 1988. The "ear to row" procedure is used to conserve the breeder's seed, and a plant population of 500 plants/m2 is suggested as an optimum density for maximum grain yield per unit area (S.N. Stratilakis, personal communication). This cultivar was chosen because it is recommended for regions that suffer from very low temperatures, and frost is a common problem every winter in the first location where this study was conducted (temperatures down to –20°C).

Experimentation was conducted during five successive growing seasons (1998–99 to 2002–03), in two locations of northern Greece, under rainfed conditions. The first location (Site 1) was the Technological Education Institute Farm of Florina (40°46'7N latitude, 21°22' 8 E longitude, 705m altitude). The second location (Site 2) was the Agricultural Research Station Farm of Nea Zoi, National Agricultural Research Foundation, (40°40'8N latitude, 22°31' 9 E longitude, 10m altitude). The experiments were treated with identical fertilization and weed control. Fertilizers corresponding to 64 kg/ha N and 80 kg/ha P2O5 were applied at planting as phosphate ammonium (16–20–0), while additional nitrogen (104 kg/ha) was applied early in April as calcium nitrate ammonium (26–0–0). Complete weed control was obtained by pre-emergent herbicide [isoproturon, 3-(4-isopropylphenyl)-1,1-dimethylurea] and hand-weeding application. Figure 1 illustrates the experimentation process for single-plant selection and progeny evaluation.



View larger version (24K):
[in this window]
[in a new window]
  		Fig. 1. Experimental procedure across two sites and five growing seasons. Pedigree single-plant selection was conducted for two generations under the ultra-low density (ULD) of 1.2 plants/m2 in honeycomb trials (NR0 and R21). Progeny families were tested under ULD in R21 trials, typical crop density (TCD) of 500 plants/m2 in RCB trials, and four densities (100, 300, 500, 700 plants/m2) in SPLIT-PLOT trials.

 
Single-Plant Selection
Pedigree selection was performed strictly under the ULD of 1.2 plants/m2, at honeycomb experimental trials (Fasoulas and Fasoula, 1995). Honeycomb designs sample effectively for spatial heterogeneity by allocating the plants of each entry across the entire experimental area, and in such a way that each plant lies in the center of concentric rings containing plants of all other entries.

Divergent single-plant selection within source material produced the first generation families, whereas the second generation families were selected for high yield. During the 1998–1999 growing season, foundation seed of cv. Nestos was used to establish at Site 1 a nonreplicated (NR–0) honeycomb trial, with a total of 1054 plants spaced 100 x 100 cm in a triangular grid. Moving-circle selection (Fasoulas and Fasoula, 1995) was applied to select 10 high-yielding plants, resulting in 10 H families. From each circle that gave a high-yielding plant, a low-yielding plant was also selected, giving 10 L families. Selected L plants had to produce at least 10 g to give enough seed for progeny evaluation. The following season H and L families, along with their original cv. Nestos, were arranged according to the R-21 replicated honeycomb design, with 55 plants per entry, at both sites. These experiments were used for both progeny evaluation and further selection. The moving-circle method was used to select four high-yielding plants within each of the best five families (significantly superior over cv. Nestos), resulting in 20 s generation families.

Progeny Evaluation
Progeny families of both generations were evaluated under either ULD or the TCD of 500 plants/m2. The first step was to test them under ULD in R21 replicated honeycomb trials (described in the previous paragraph), in both sites for a single year. Seed obtained from all plants of each entry, being mixed and screened, constituted the material that was evaluated under TCD in both sites for a single year. The experimental design was the one-factor randomized complete block (RCB) design, with three replications per entry. Each plot consisted of eight rows of 5 m long with 17 cm between rows. A total of 105 g of seed was sowed in each plot, since the thousand kernel weight measured in a few representative seed samples had been found to be 30 g. Additionally, six of the first generation families that performed the best under ULD (i.e., five H significantly differing, and an L not significantly differing from cv. Nestos), along with the original cultivar, were tested under four densities (100, 300, 500, 700 plants/m2), with four replications per family and density, in both sites for 2 yr. Entries were arranged according to a two-factor split-plot design, where densities constituted the main plots and families the subplots. Subplot construction was similar to that of plots in RCB trials.

Grain Yield Computation and Statistical Analysis
Under ULD, mean grain yield per plant was determined by hand harvesting and threshing individual plants. Entries' mean values were compared at the 5% level by z-test for independent samples and different standard deviations. Coefficients of variation (CV) of single-plant yields were also determined. A small combine was used to harvest and thresh the six internal rows of each plot (plot area of 5.1 m2) of the RCB and split-plot trials, and mean grain yield per hectare (ha) was measured after adjusting to 12% grain moisture. Comparison of means was conducted by least significance difference (LSD) after analysis of variance (ANOVA), for the one-factor RCB design combined over locations and for a two-factor split-plot design over locations and years. To compare the two means, LSD was computed by the formula: LSD = t0.05[EMS(1/n1 + 1/n2)]1/2, where EMS was the error mean square for estimating the significance, t0.05 was the theoretical t value for EMS degrees of freedom, and n1 and n2 were the number of values used to compute the means.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Source Material
Single-plant yield of the source material cv. Nestos ranged from 1 to 90 g, and averaged 30.5 g with 17.1 g standard deviation (Fig. 2a). Kurtosis of single-plant yield frequency distribution was not significant, but positive skewness of 0.582 was significant (P < 0.001), showing a leftward departure from normality. Yield of the 10 single-plants selected as high yielders ranged from 51 to 83 g, whereas yield of those selected as low yielders was between 10 and 23 g.



View larger version (24K):
[in this window]
[in a new window]
  		Fig. 2. Single-plant yield frequency distribution of the source material (a), and of the first generation H (continues line) and L (dotted line) families in Site 1 (b) and Site 2 (c), under the ultra-low density (ULD) of 1.2 plants/m2. (X = mean yield, CV = coefficient of variation, Sk. = skewness, Kur. = kurtosis, n = number of harvested plants). {dagger} P < 0.05, {ddagger} P < 0.01, {dagger}{dagger} P < 0.001.

 
First Generation
Under ULD, mean experimental yield was 25.1 g/plant in Site 1 and 30.2 g/plant in Site 2, with different entry rank. When the plants of all the H or all the L families were considered together, the single-plant frequency distributions, depicted in Fig. 2 for Site 1 (b) and Site 2 (c), revealed a significant departure from normality according to skewness and kurtosis values, with mean and mode to be transposed leftwards. In both sites, overall mean yield of H families was significantly higher than overall mean yield of L families (P < 0.002). Pooled data from both sites are given in Table 1, and summarized. Five out of the 10 H families gave significantly higher yield than cv. Nestos (by 25–38%). The average mean yield of the 10 H families was also significantly greater than that of cv. Nestos (by 20%). On the contrary, selection failed to isolate any of the L families with significantly lower yield than the original cultivar. The average CV value of the 10 L families was similar to CV value of cv. Nestos, and by 9.4% higher compared with that of the 10 H families.


View this table:
[in this window]
[in a new window]
  		Table 1. Mean grain yield of the first generation families, their original cv. Nestos, the 10 H families on average (H.), and the 10 L families on average (L.), under the ultra-low density (ULD) of 1.2 plants/m2 and the typical crop density (TCD) of 500 plants/m2. (n = number of plants, CV = coefficient of variation of single-plant yields). Data obtained from two locations (Tokatlidis et al., 2004).

 
Under TCD, mean productivities of the two experiments were similar (3094 kg/ha in Site 1, and 3098 kg/ha in Site 2). The ANOVA revealed significant F value for families (P < 0.008), but not for the family by location interaction. Two out of the 10 H families gave significantly higher yield than the original cultivar by 19 and 22% (Table 1). Compared to the mean yield of cv. Nestos, the average mean yield of the 10 H families was 3.7% higher, and the yield of the 10 L families was 3.1% lower.

The ANOVA for the best six families under ULD showed significant differences between densities (Table 2). The highest yield was obtained at the density of 500 plants/m2 (3696 kg/ha), followed by those at 700 plants/m2 (3416 kg/ha), 300 plants/m2 (3306 kg/ha), and 100 plants/m2 (2749 kg/ha). Significant differences were also found between families (P < 0.03). Compared to cv. Nestos, five out of the six families (including L8), and the five H families on average as well, were significantly superior (Table 2). The density by family interaction was also significant (P < 0.04). In relevance to original cultivar, the density that most favored the families was the lowest (100 plants/m2), since three of them exhibited significantly higher yield, and average yield of the five H families was also significantly greater. Even though at higher densities, overall families' means were also higher than those of cv. Nestos, and none of them was significantly superior at 300 plants/m2, whereas one family was significantly superior at 500 and 700 plants/m2, respectively.


View this table:
[in this window]
[in a new window]
  		Table 2. Mean grain yield (kg/ha) of the six first generation families, the original cv. Nestos, and the five H families on average (H.), under four densities. Data obtained from the SPLIT PLOT trials (two locations and 2 yr).

 
Second Generation
Under ULD, mean experimental yield was 22.6 g/plant in Site 1 and 24.3 in Site 2, with different family rank. Results obtained from both sites (Table 3) showed that 15 out of the 20 families gave significantly higher yield than cv. Nestos (by 18–53%). The 20 families averaged 23.7 g/plant, being significantly superior over cv. Nestos (by 24%). When families were grouped according to their origin and therefore could be assumed as replicates of the respective initial family, average mean of the four families originated from H2 was found to be significantly superior over cv. Nestos, and the same was observed for groups H7 and H10. Family CV values ranged from 43.3 to 60.8% (averaged 50.3%), and CV of cv. Nestos was 58%.


View this table:
[in this window]
[in a new window]
  		Table 3. Mean grain yield of the 20 second generation families, and their original cv. Nestos, under the ultra-low density (ULD) of 1.2 plants/m2 and the typical crop density (TCD) of 500 plants/m2. The average yield of families with the same origin (e.g., H2. represents the mean performance of the four families extracted from line H2), and the overall family yield (H.) are also given (n = number of plants, CV = coefficient of variation of single-plant yields). Data obtained from two locations.

 
Under TCD, mean productivity of the two experiments was 3329 and 3133 kg/ha in Site 1 and Site 2, respectively. Families differed significantly (P < 0.03), but the family by location interaction was not significant. Pooled data from the two experiments (Table 3) showed that four out of the 20 families gave significantly higher yield than the original cultivar (by 17 to 20%). Mean yield of the 20 families, on average, was 9% higher compared to that of cv. Nestos, but not significantly superior. The average yields of groups H2 and H10 were significantly greater compared to the mean yield of cv. Nestos.

Relationships between Yields and CVs
On the basis of the correlation coefficients (r values), a positive association was found between yield under ULD and yield under TCD, either in the first or in the second generation. The respective r values were 0.53 (P < 0.02) and 0.48 (P < 0.03). Grain yield under both ULD and TCD was negatively related to CV of the single-plant yields obtained under ULD. The exception was in case of yield under TCD in the first generation, where a nonsignificant r value was computed (r = –0.35, P < 0.12). Correlation coefficients between CV and yield per plant under ULD were –0.81 (P < 0.001) in the first, and –0.66 (P < 0.002) in the second generation, while r value between CV and yield per hectare under TCD in the second generation was –0.63 (P < 0.003).


   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
CONCLUSIONS
 REFERENCES
 
Considering all the data, one could postulate that cv. Nestos yielded lower than expected under the experimental conditions. After data from TCD, including the split plot trials, were adjusted to grain yield per hectare and averaged across single experiments (eight environments), the overall experimental mean yield was computed to be around 3500 kg/ha, more or less equal to average yield normally obtained by farmers in Greece. In dense stand, Fasoula (1990) reported much higher yield in cv. Siete Cerros (up to 8000 kg/ha), whereas Stratilakis and Goulas (2003) found mean yield of 13 experimental varieties to be around 4000 kg/ha, with both studies being conducted in Greece. On the other hand, in our experiments under ULD, the highest experimental mean yield per plant was 30.5 g (source material), and much lower than those reported by the other researchers. Under similar conditions the mean yield per plant of cv. Siete Cerros reached up to 179 g (Fasoula, 1990), and that of cv. Pinios was 78 g with 25 g standard deviation (Gouli-Vavdinoudi and Roupakias, 2000). Genetic buffering of cv. Nestos at the individual plant level to cope with environmental influences could be a reasonable explanation.

Although yield distribution of a cultivar is expected to be normal, source material did not confirm this expectation, and its yield distribution shifted leftwards (Fig. 2a). This might not be obtained by chance, since the same impact was observed in the first (Fig. 2b,c), as well as in the second generation (not shown). According to Fasoula and Fasoula (1997), yield distribution pattern of a cultivar reflects the degree of genetic buffering for less environmental influence on phenotypic expression. Normal yield distribution is achieved when CV values are around 33%. Increased CVs are associated with increased percentages of low-yielding plants, as distributions of Fig. 2 depict (c vs. a and b). Positive skewness results from negative intra-genome gene interactions (Fasoula and Fasoula, 1997) or high frequency of unfavorable alleles (Traka- Mavrona et al., 2000) that impair individual stability. Fasoula and Fasoula (2000; 2002) speculated that the genetic background of a given genotype for high and stable yield across environments is reflected by three yield components determined in the absence of competition. These are (i) yield potential per plant, (ii) tolerance to stresses, determined by standardized mean (that is, the reverse value of CV), and (iii) responsiveness to inputs, determined by standardized selection differential. As far as the cv. Nestos is concerned, data obtained under ULD for the first and the second yield factors (i.e., low mean yield per plant, and leftward yield distribution) are indicative that cv. Nestos lacks genetic buffering at the individual plant level, and thus strong environmental influence make mean and mode of the yield distribution to transpose leftwards. According to data obtained from experimental maize inbred lines and their single-crosses under the low density of 0.74 plants/m2, Tokatlidis et al. (1999) demonstrated that the larger the CV of the single plant yields, the greater the departure of the yield distribution from normality and the more reduced the yield and stability. Coefficients of variation of materials characterized as better buffered were around 39%, whereas those of materials lacking genetic buffering ranged from 62 to 88%. The researchers pointed out that the degree of genetic buffering is responsible for a negative relationship between yield potential per plant and CV values, being consistent with the inverse association between CV and yield found in this study. Fasoulas (1993) speculated that CV of individual plant yields depicts the genetic background of a species; with regard to stability, CVs must be taken into account when the selection process at low densities is under way (Fasoula and Fasoula, 2000; 2002). For example, low CV is assumed to reflect genotype's stability for this particular trait (Taylor et al., 1999).

Nevertheless, the entire results of this study indicated that productivity of cv. Nestos can be improved. It is interesting to notice that two families (H2 and H10) were significantly superior over cv. Nestos in both generations under either ULD or TCD, as well as averaged across the four densities in the split plot trials (Fig. 3). Data from the split plot trials (Table 2), however, deserve further consideration. Highest yields were recorded for the TCD of 500 plants/m2, which is an optimum density for maximum yield per unit area. At that density, even though family H10 expressed 11% superiority over cv. Nestos, the average superiority of the five H families was only 2%. But as density departs from the optimum upward or downward, families' superiority over cv. Nestos increases, being 3.1, 5, and 9.6% at 700, 300, and 100 plants/m2, respectively (Fig. 4). This finding is of paramount importance in relation to yield stability. According to Fasoula and Fasoula (2000; 2002), selection at low density facilitate incorporation into new cultivars the three yield components mentioned previously, which render them less density dependent and thus constitute key parameters for high and stable performance. Improvement of yield potential per plant extends the lower limit of the optimum density range, tolerance to stresses extends the upper limit of the optimum density range, and responsiveness to inputs enables the exploitation of favorable environments. Compared to the original cultivar, families expressing higher yield per plant by 20 and 24% under ULD in the first and second generations, and with average CVs lower by 10 and 13%, are very desirable. This implies that selection improved the first and second of the performance factors, the implication being the response of families to density changes to be distinct from that of cv. Nestos (Fig. 4). Consequently, families give the impression of being less density dependent than cv. Nestos. The capacity of the derived families to produce higher at lower densities ensures more stable performance across densities, which is an important implication for crop yield stability. This is because genotypes characterized by improved yield potential per plant are more likely to compensate for yield loss due to any missing or poorly developed plants, a situation that commonly occurs. The implications of strong density dependence on stability of performance of maize hybrids have been thoroughly discussed by Tokatlidis and Koutroubas (2004), whereas the beneficial role of improved yield potential per plant in developing density-independent maize hybrids was depicted by experimental data (Tokatlidis et al., 2001). Farmers have always favored the density-independent cultivar due to their stable and dependable performance (Fasoula and Fasoula, 2000).



View larger version (31K):
[in this window]
[in a new window]
  		Fig. 3. Mean experimental performance of the families H2 and H10 under ultra-low density (ULD), typical crop density (TCD), and four densities on average (SPLIT PLOT), being significantly superior over cv. Nestos in both generations.

 


View larger version (10K):
[in this window]
[in a new window]
  		Fig. 4. The average response of the five H first generation families to density changes, and in relevance to cv. Nestos. The equation of line is y = 4 x 10–5X2 – 0.04X + 113. Data obtained from the SPLIT PLOT trials (four densities, two locations, and 2 yr).

 
Apparently, the method of breeder's seed maintenance deserves reconsideration. A new approach that accentuates existing or newly developed genetic variation might be necessary to facilitate the within-cultivar selection. Rasmusson and Phillips (1997) reported that in barley (Hordeum vulgare L.), incremental gains for several traits were made in a very narrow gene pool, attributable to variation present in the original gene pool as well as to de novo variation. The continuous selection within a cultivar is necessary to exploit the potential existence of variation for either cultivar conservation or upgrading, and this target is feasible at the single-plant level in the absence of competition (Fasoulas, 1993). Christakis and Fasoulas (2002) found exploitable genetic variation for yield in tomato (Lycopersicon esculentum L.) that was uncovered in advanced generations, after the point of achieving theoretical homozygosity (F7 generation). The selection study for modified oil and protein in maize, with selection being practiced effectively for more than 90 generations (Dudley and Lambert, 1992), highlights the importance of continuous selection. Fasoula and Fasoula (2000) stated, "Continuous selection after the release of cultivars is imposed by the need to eliminate deleterious mutations and exploit any positive source of existing and newly derived variation, either genetic or epigenetic." Seed production under high densities may favor cultivar degeneration because the rate of genotypes characterized as strong competitors and low yielders increases, at the expense of weak competitors and high yielders, as a result of the inverse relationship between competitive and yielding ability (Fasoula and Fasoula, 2000). Peng et al. (1999) found that maximum yield of the rice cultivar named IR8, developed by International Rice Research Institute (Philippines), reduced by around 2 Mg/h (20%) during the past 30 yr. According to the results of this work, selection under ultra-low density seems to be an effective way of the breeder's seed conservation, targeting the avoidance of a cultivar's gradual degeneration.

Apart from higher yields, the derived families exhibited lower CV values by around 10%, in comparison with the original cultivar, indicative of narrowing any existing genetic variation. On the other hand, divergent selection failed to isolate any low-yielding family. A possible explanation is that the yield of L selected plants was at least 10 g, despite the existence of plants with even lower yield (down to 1 g, as Fig. 2a shows). Therefore, it was unlikely that selection for low yield occurred. However, under ULD in both sites, average mean yield per plant of the 10 H families was significantly greater than that of the 10 L families, with lower CV as well (Fig. 2b,c). The performance of family L8, originally from a plant selected as a low yielder, when compared to cv. Nestos, was superior by 18 and 8% under ULD and TCD, respectively (Table 1), and in the split plot trials gave significantly higher yield (Table 2). But this finding does not undervalue the procedure usefulness, since confounding effects of genotype by environment interactions are always expected. In contrast, as density decreases environmental effects on phenotypic expression are reduced (Fasoula and Fasoula, 2002). Hence, optimal low-density conditions are required, so that within a cultivar differences become recognizable. Results that provide evidence of increased phenotypic expression and differentiation under lower densities were reported by Hamblin et al. (1978) in barley, Daynard and Muldoon (1983) in maize, Kyriakou and Fasoulas (1985) in winter rye (Secale cereale L.), and Fasoula (1990) in bread wheat. Several studies reported effective exploitation of intracultivar variation via selection under low densities. Divergent single-plant selection within bread wheat cv. Siette Cerros, at the same ULD of 1.2 plants/m2, led to three H families that in dense stand had 2 to 16% higher grain yield compared to the original cultivar (Fasoula, 1990). Pedigree selection for three generations within a snap bean cultivar, with plants being widely spaced, was successful in identifying families that under dense stand had 219 to 276% higher pod yield, compared to the source material (Traka-Mavrona et al., 2000). Tokatlidis (2000), after divergent single-plant selection within the maize inbred lines B73 and Mo17 at the density of 0.74 plants/m2, detected sublines with significant differences in grain yield that transmitted to their single-crosses. At that density, grain yield per plant of the highest yielding entry was greater than that of the lowest yielding by 65% (B73), 25% (Mo17), and 26% (B73 x Mo17). Cotton single-plant selection at the density of 0.74 plants/m2 over the years within the elite cv. Sindos 80 led to the release of the cv. Macedonia which, under experimental conditions across 32 environments, exhibited a 10% yield superiority over cv. Sindos 80. Additionally, two cotton families were identified with tolerance to Verticillium wilt, while the original cv. Sindos 80 was susceptible (Fasoulas, 2000).


   CONCLUSIONS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
REFERENCES
 
The results of this study open the possibility that selection within cultivars, especially for those released earlier, may prove to be a useful technique either to upgrade or to avoid gradual degeneration of genetic background. Results and accumulated reports of intracultivar variation suggest that selection should be a perpetual process, so that any existing or newly developed variation is exploited and optimal quality of breeder's seed is secured. Maximum phenotypic expression and differentiation, however, is a prerequisite to identify the within cultivar genetic differences, to remove inferior plants, and to select the superior ones. Optimal low density fulfills the condition, and seems to be the more suitable for breeder's seed conservation. Further study is needed to determine the ideal low density for each different crop that optimize breeder's seed conservation.


   ACKNOWLEDGMENTS
 
The authors are grateful to their anonymous reviewers as well as to Drs. Jodi Scheffler and Brent Godshalk for their critical review and helpful suggestions on this manuscript, contributing to the paper's improvement.

Received for publication February 6, 2006.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 





This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via ISI Web of Science (1)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Tokatlidis, I. S.
Right arrow Articles by Papadopoulos, I. I.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Tokatlidis, I. S.
Right arrow Articles by Papadopoulos, I. I.
Agricola
Right arrow Articles by Tokatlidis, I. S.
Right arrow Articles by Papadopoulos, I. I.
Related Collections
Right arrow Crop Growth and Development
Right arrow Wheat
Right arrow Data Management

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
The SCI Journals Agronomy Journal Vadose Zone Journal
Journal of Plant Registrations Soil Science Society of America Journal
Journal of Natural Resources
and Life Sciences Education		 Journal of
Environmental Quality