Intra-cultivar variation

At crop density, the magnitude of intra-cultivar variation (i.e. the relative superiority of the top over the bottom genotype) for seedcotton yield across sites and years averaged 0·22 for the 3GPLs of Christina (Chr-6 v. control), and 0·10 for the 3GPLs of Corona (Crn-17 v. Crn-5 and control) (Table 1). Significant differences among the 3GPLs at crop density were also discovered for fibre micronaire within the three cultivars, for fibre length within Corona and Flora, and for fibre strength within Corona (Table 2), but not for any of the three physiological traits.

There are important comparisons between these findings for the 3GPLs which were tested at crop density and the findings obtained at low density using 1GPLs and 2GPLs (Tokatlidis et al. Reference Tokatlidis, Tsikrikoni, Tsialtas, Lithourgidis and Bebeli2008). At low density, greater variability was observed. For instance, the magnitude of intra-cultivar variation for seedcotton yield averaged 1·25 (1GPLs) and 1·28 (2GPLs) for cvar Christina and 1·14 (1GPLs) and 1·26 (2GPLs) for cvar Corona. Moreover, at low density, significant heterogeneity was found for seedcotton yield within cvar Flora, for fibre length within cvar Christina, and for three physiological traits within the three cultivars (Tokatlidis et al. Reference Tokatlidis, Tsikrikoni, Tsialtas, Lithourgidis and Bebeli2008). This supports the hypothesis that very low densities accentuate phenotypic differentiation (Fasoula & Fasoula Reference Fasoula and Fasoula1997, Reference Fasoula and Fasoula2002; Tokatlidis et al. Reference Tokatlidis, Xynias, Tsialtas and Papadopoulos2006, Reference Tokatlidis, Has, Mylonas, Has, Evgenidis, Melidis, Copandean and Ninou2010), thus allowing even limited genetic variability to be identified. These findings are in agreement with those found in other studies (Fasoula Reference Fasoula1990; Fasoulas Reference Fasoulas and Gillham2000; Tokatlidis Reference Tokatlidis2000; Tokatlidis et al. Reference Tokatlidis, Xynias, Tsialtas and Papadopoulos2006; Fasoula & Boerma Reference Fasoula and Boerma2007). Because of the accumulated reports of intra-cultivar variation, continued selection under very low density has been proposed (Fasoula & Fasoula Reference Fasoula and Fasoula2002). In contrast to the typical maintenance stewardship, in which, elite cultivars are treated as permanent records and usually plants of ‘incorrect’ type are removed (Parlevliet Reference Parlevliet2007), the suggested technique is more a selection rather than a rouging process. It may progressively lead to new registration by the improvement of important agronomic traits, the required time for which is hard to be answered depending on both crop and cultivar.

Response to intra-cultivar selection

Averaged over sites, the cvar Christina- and cvar Corona-derived 1GPLs outperformed the original cultivars by 16 and 2·6%, respectively (Fig. 1). Even though more favourable environmental conditions prevailed when the 1GPLs were evaluated (Tokatlidis et al. Reference Tokatlidis, Tsikrikoni, Tsialtas, Lithourgidis and Bebeli2008), the advanced self-pollination of 2GPLs achieved greater increase in seedcotton yield over the controls (19% for Christina and 3·7% for Corona). With a few exceptions, both 1GPLs and 2GPLs exhibited lower CV for single-plant yields than their original parents (Tokatlidis et al. Reference Tokatlidis, Tsikrikoni, Tsialtas, Lithourgidis and Bebeli2008), suggesting homozygosity or homogeneity within the selected lines, and indicating that the selection imposed narrowed existing genetic variation within these two gene pools which had been assumed to be fairly homogeneous. By contrast, selection within cvar Flora resulted in 1GPLs with 10% lower average yield than the original cultivar but this decreased to 3·3% in the second generation (Fig. 1), and one possible cause was a considerable degree of heterozygosity in the original cultivar (Tokatlidis et al. Reference Tokatlidis, Tsikrikoni, Tsialtas, Lithourgidis and Bebeli2008). Across sites, the 1GPLs averaged a CV value 11% higher than cvar Flora, but the 2GPLs and the control averaged a similar CV value, suggesting variability due to heterozygosity. Whether the hypothesis is valid, repeated selfing is required to fix homozygosity and find out whether this variability is exploitable.

The positive response to selection within cvars Christina and Corona was verified at crop density. Over sites and years, their 3GPLs yielded 12 and 5·2% higher than the original cultivars respectively (Fig. 1). It has been suggested that phenotypic differentiation decreases at high densities due to increased interplant competition (Fasoula Reference Fasoula1990; Fasoula & Fasoula Reference Fasoula and Fasoula1997, Reference Fasoula and Fasoula2002; Tokatlidis et al. Reference Tokatlidis, Xynias, Tsialtas and Papadopoulos2006, Reference Tokatlidis, Has, Mylonas, Has, Evgenidis, Melidis, Copandean and Ninou2010). This constitutes a reasonable explanation to the fact that in cvar Christina the 19% overall superiority for the 2GPLs at low density decreased to 12% for the 3GPLs at crop density (Fig. 1). A similar influence of low- v. commercial crop-density conditions on the magnitude of entry differentiation has been reported in bread wheat (Fasoula Reference Fasoula1990; Tokatlidis et al. Reference Tokatlidis, Xynias, Tsialtas and Papadopoulos2006). Selection for seedcotton yield may not exclude cultivar improvement for other important agronomic traits either, and the case of high yielding line Crn-16 (Table 1) with the strongest fibre (Table 2) is instanced. Line Chr-6 that out-yielded cvar Christina by 22% at crop density (Table 1) was also superior for fibre length at low density (Tokatlidis et al. Reference Tokatlidis, Tsikrikoni, Tsialtas, Lithourgidis and Bebeli2008). Leaf carbon isotope discrimination, Δ, which reflects plant discrimination against ¹³CO₂ and the assimilation of ¹²CO₂ during photosynthesis, has been suggested as an indirect assessment of water use efficiency, since the two traits are negatively associated (Saranga et al. Reference Saranga, Jiang, Wright, Yakir and Paterson2004; Tsialtas et al. Reference Tsialtas, Tokatlidis, Tsikrikoni and Lithourgidis2008). Lines Chr-2, Chr-3 and Chr-8, as well as Crn-20, had significantly lower Δ values than the controls at low density (Tokatlidis et al. Reference Tokatlidis, Tsikrikoni, Tsialtas, Lithourgidis and Bebeli2008), showing a possible advantage under drought stress conditions. Honeycomb single-plant selection for seedcotton yield within the susceptible cvar Sindos 80 led to the resistant cvar Macedonia in terms of Verticillium wilt (Fasoulas Reference Fasoulas and Gillham2000). No progress from intra-cultivar selection was found in cvar Flora based on the 3GPLs performance at crop density (Fig. 1). Although cotton is a self-pollinated species, natural crossing may range from 0 to almost 0·5 (Xanthopoulos & Kechagia Reference Xanthopoulos and Kechagia2000). If the cross-pollination hypothesis is real in cvar Flora, it cannot be excluded that continuous selection at low density may lead to new lines capable of out-yielding cvar Flora, and leading essentially to a new cultivar. In tomato (Lycopersicon esculentum L.), exploitable genetic variation for yield was uncovered in advanced generations, after the point of achieving theoretical homozygosity, i.e. F₇ generation (Christakis & Fasoulas Reference Christakis and Fasoulas2002).

Results from molecular studies indicate that the genome undergoes constant remodelling and restructuring, suggesting that it is more flexible and plastic than previously assumed (McClintock Reference McClintock1984; Zhang et al. Reference Zhang, Gentzbittel, Vear and Nicolas1995; Olufowote et al. Reference Olufowote, Xu, Chen, Park, Beachell, Dilday, Goto and McCouch1997; Rasmusson & Phillips Reference Rasmusson and Phillips1997; Gethi et al. Reference Gethi, Labate, Lamkey, Smith and Kresovich2002; Cullis Reference Cullis2005; Morgante et al. Reference Morgante, Brunner, Pea, Fengler, Zuccolo and Rafalski2005). If that is the case, then the observed response to selection within cotton cultivars may have important implications for cultivar maintenance and deserves further consideration. Laverack & Turner (Reference Laverack and Turner1995) stated that to maintain a cultivar true-to-type and healthy, removing off-type plants is an imprecise method. Gethi et al. (Reference Gethi, Labate, Lamkey, Smith and Kresovich2002) pointed out that breeders should take into account the within inbred line genetic variation in maize, particularly when sampling early released materials. The selection method of 1GPLs and 2GPLs at low density and the final performance test of the 3GPLs at crop density open up new prospects for similar studies with other cultivars and crops to determine how universal this rule may be. The practice of advancing generations or maintaining cultivars at commercial densities, may favour degeneration by increasing the relative contribution from strong competitor plants with low seed yields to the next generation compared to that of weak competitor plants with high seed yields (Fasoula Reference Fasoula1990; Fasoula & Fasoula Reference Fasoula and Fasoula1997, Reference Fasoula and Fasoula2002; Tokatlidis et al. Reference Tokatlidis, Xynias, Tsialtas and Papadopoulos2006, Reference Tokatlidis, Has, Mylonas, Has, Evgenidis, Melidis, Copandean and Ninou2010). On the other hand, cultivar improvement may result if selection continues at low density after the release of cultivars, covering an adequate number of sites. By this procedure, undesirable mutations or segregates may be removed, while exploiting those affecting yield, stability and other desirable traits. For instance, mote frequencies that represent a loss in yield and cause imperfections in yarn and cloth quality (Bolek Reference Bolek2006), or leaf morphology that affects yield, quality, maturity, water-use efficiency, etc. (Hao et al. Reference Hao, Yu, Dong, Fan, Ma, Song and Yu2008) might be improvable via intra-cultivar selection at very low density. By improving important agronomic traits and yield capacity for the farming-density conditions, a new registration may result. Intra-cultivar honeycomb selection at very low density resulted in progeny lines that, at farming-density, outperformed the cultivar by 8·0% (Fasoula Reference Fasoula1990) and 22% (Tokatlidis et al. Reference Tokatlidis, Xynias, Tsialtas and Papadopoulos2006) in bread wheat and by 10% in cotton (Fasoulas Reference Fasoulas and Gillham2000). Additionally, Tokatlidis et al. (Reference Tokatlidis, Xynias, Tsialtas and Papadopoulos2006) reported improvement of cultivar stability. Rasmusson & Phillips (Reference Rasmusson and Phillips1997) reported incremental genetic gains for several traits in a narrow gene pool of barley (Hordeum vulgare L.) and hypothesized that gain from selection within elite gene pools occurs because of variation present in the original gene pool as well as de novo generated variation. In maize, long-term selection studies for modified oil and protein showed that meaningful variation exists after 90 generations of selection (Dudley & Lambert Reference Dudley and Lambert2004).

In conclusion, the data reported herein suggest exploitable intra-cultivar variation in cotton. This variation may be due to latent variation, or epigenetic variation in response to environmental changes. Results of the present study and numerous supporting reports imply that the genomes of ‘pure line’ cultivars may not be as fixed as previously assumed. Continuous selection within elite cotton cultivars may prove to be a useful technique either to upgrade or to avoid gradual degeneration of genetic background and to maintain uniformity over longer periods of time. Low density, by accentuating the limited intra-cultivar variation, appears ideal to identify the existing and newly developed genetic differences, and to remove inferior plants and identify the superior ones. Hence, prospective analogous research in cotton and other crops may verify that the selection within elite cultivars at low densities as a maintenance of improvement process is a beneficial technique to secure optimal quality of breeder's seed.