In breeding table grapes (Vitis vinifera), seedlessness is considered a trait of increasing importance to the consumer. A simple sounding concept, seedlessness at closer look, turns out to be a complex trait, its very definition being difficult because many factors are involved. Thus it became necessary to explore this trait in three complementary aspects constituting the aim of this study:
(I) Identification and definition of subtraits to seedlessness.
(II) Evaluation of effects of environmental factors such as gibberellin and growth retardents on the seed development, and on seed components.
(III) Developing molecular genetic markers linked to the subtraits of seedlessness by use of the RAPD technique as tools for preselection of seedless offsprings.
Stenospermocarpic seedlessness in grapes was discussed since Pearson (1932) and Stout (1936), and lately in detail by Ramming et al., (1990), Spiegel-Roy et al., (1990), and by Ledbetter and Burgos (1993). Partially formed seeds in grapes after fertilization were termed `stenospermic' (Stout, 1936), and this term was applied to table grapes that were neither parthenocarpic nor seeded. A whole range of sizes of seeds and seed-traces was found in progenies from crosses of seeded X seedless cultivars, as well as in crosses between seedless X seedless cultivars. Furthermore, the classification of seeded vs. seedless individuals is influenced and biased by several factors. Perceptibility of the less- developed seed-traces may be affected by factors such as the firmness and crispness of the berry (Ledbetter and Shonnard, 1991), as well as the degree of development of the seed, its size and the sclerification of integuments (Ledbetter and Ramming, 1989).
The need to divide grape progenies into seeded and seedless offsprings, led investigators to look for parameters correlated to the seed content of the berries (Perl et al., 1989). The continuous nature of the parameters used for quantifying the seed content of grape berries (e.g. fresh weight of seeds per berry), led to the need for a differentiating tool: 25mg per aborted seed was used as threshold for seedlessness (Ramming et al., 1990).
Numerous theories for the inheritance of the stenospermocarpic character have been proposed, after analysing the "seeded to seedless" ratio in progenies. These attempts resulted in the assumptions that the trait of seedlessness is controlled (as a qualitative trait) by more than one recessive gene, or by dependence on two complementary recessive genes. (Stout, 1936; Weinberger and Harmon, 1964; Loomis and Weinberger, 1979; Ledbetter and Ramming, 1989; Spiegel-Roy et al., 1990a; Ramming et al., 1990). Contradictory conclusions have also been drawn, when the seedless character was treated as a quantitative trait with continuous variation of the fresh weight of the seeds (Sandhu et al., 1984). No wholly adequate inheritance scheme, taking into account the variable degrees of development of the seed traces, has yet been proposed.
Although few genetic factors are assumed to be involved, it is hard to overlook the quantitative nature of seed development and its response to environmental effects. I observed that the perceptibility of the seed traces is not necessarily correlated to the size of the seed, but mainly to the hardness of the seed-coat. Furthermore, in progenies of crosses between `seeded X seedless' cultivars, I found several combinations of the hardness of the seed-coats with different degrees of development of the endosperm. These observations led me to conduct the present study, to examine whether the assessment of the degrees of development of the seed components (i.e. seed-coat, endosperm and, possibly, embryo) as separate traits, would contribute to a better understanding of the seedless character in stenospermocarpic grapes, rather than using one single parameter called "seedlessness".
The seeds in seedless and in seeded berries were classified visually into four categories of size: normal seeds, large traces, medium traces and small traces. Seeds with fully developed and sclerified seed-coats were observed to bear endosperm at various developmental stages, and in seeds with soft and less developed seed-coats also fully developed endosperms were observed.
I found that the hardness of the seed-coat and the degree of development of the endosperm were transmitted as separate traits to the progenies. Two seeded cultivars (Oz, Early-Muscat) were chosen because they differed in seed hardness and in degree of development of their endosperm. The normal seeds of Oz are comparatively harder and contain less developed endosperm than those of Early-Muscat. Each cultivar was crossed with the same pollen donor parent (Flame-Seedless).
I demonstrated that the specific subtraits of stenospermocarpic seedlessness (hardness of seed coat and degree of development of endosperm) were transmitted to the progeny (Striem et al., 1992). 23.7% of the Oz progeny were classified as normally seeded offsprings bearing undeveloped or partially developed endosperm, while only 1.2% of the Early-Muscat progeny had a similar composition of seed components. This suggests that seedlessness in grapes could be more precisely analysed using the hardness of the seed-coat and the degree of development of the endosperm as subtraits of seedlessness.
It is worthwhile to describe the progenies, as well as parents, in terms of seed-coat hardness and endosperm development, and possibly also embryo development. Thus it would seem to be worthwhile to include the degree of development of the seed components in the matrix of data for genetic analysis in breeding programs, along with data on organoleptic evaluation, and fresh weight (mg) of seeds per berry.
In order to evaluate the response of the seed content in berries to one of the main environmental effects on grapes, I conducted experiments with different plant growth regulators (gibberellin, ancymidol and XE-1019). The chemicals were applied by dipping the cluster at prebloom, at full bloom and at anthesis, using several seeded and seedless grape cultivars).
Results indicated a significant decrease in number and size of seeds and seed traces, in response to gibberellin treatments. Almost all normal seeds of Black-Rose and Seeded-Sultanina were eliminated. Mean fresh weight of large seed traces, medium and small seed traces increased non- significantly. A reduction in the degree of development of seed coat and endosperm was observed. Pre-bloom treatments affected berry set in arresting the development of seed, much more than did full bloom treatments with gibberellin. Treatments at anthesis had minor effects on the size of the developing seed, and some effect in reducing the degree of development of the seed-coat and endosperm was found.
Growth retardant chemicals (ancymidol, XE-1019) affected mainly Sultanina berries and seeds, while the berries and seeds of Superior showed no significant response. Regarding Sultanina, a significant reduction in the size of berries (from 1.8 gr. to 1.5 gr. per berry) was found. A reduction in the total fresh weight of seeds per berry (from 5.2 mg. to 3.9 mg) was found, and accompanied by an increase in the number and mean weight of the small seed traces, indicating a shift from large seeds to small seed traces. A significant increase of 8% in the germination percentage was observed in Sultanina and Flame-Seedless, in response to the treatments with growth retardant chemicals. No significant change in the degree of development of the seed components was found.
The response of grape seeds to gibberellin and growth retardants was found to be limited mainly to the size and not to the degree of development of their seed components. Early (pre-bloom) treatments were found to have a stronger effect on the seeds, than applications at late bloom or at fruitset. The trait of seedlessness in grapes wasfound to be sensitive to environmental effects, tending to influence the final size of the seeds as a continuous trait (measured as fresh weight of the seeds). In conclusion we may say that the differential response to growth regulators supports the claim to consider the degree of development of seed components as subtraits of seedlessness.
The introduction of restriction fragment length polymorphism (RFLP) methodologies to plants has significantly improved our capability to develop genetic molecular markers (Beckmann and Soller 1986, Paterson et al 1988). A major development in this area was the recent discovery of minisatellite DNA sequences (Jeffreys et al 1985). The M13 DNA (Vassart et al 1987) and the human minisatellite probes (Jeffreys et al 1985) were found to detect DNA fingerprints in a wide range of organisms. This was demonstrated first in the human and animals genomes (Jeffreys et al 1985; Vassart et al 1987) and later in plants (Dallas 1988; Ryskov et al 1988, Nybom et al 1990).
In order to uncover DNA polymorphism in Vitis vinifera, I subjected seven vegetatively propagated cultivars to Southern blot analysis with different multi-loci minisatellite probes (Striem et al, 1990). Highly polymorphic patterns were detected when the DNA was hybridized with the M13 probe. When Hinf-I restriction enzyme was used only 3 bands out of 42 were common to all cultivars. Similarly, using the Hae-III digest, only 6 bands out of 37 were common to all cultivars. Bands common to all cultivars were generally smaller than 2 Kb in size. The total number of bands per cultivar ranged between 13-27 and 15-21 in the Hinf-I and Hae-III digest, respectively. Less variations between cultivars were observed when the same blot was rehybridized with the 33.6 probe.
So far, ongoing breeding programs in grapes utilized mainly morphological (Spiegel-Roy 1980), biochemical (Bachmann and Blaich 1988), and isozyme (Weeden et al 1988, Parfitt and Arulsekar 1989) markers. The use of new molecular genetic methodologies to generate markers for a wide range of applications in plants has been discussed (Roose 1988, Beckmann and Soller 1986). Cultivar identification can be made in each probe/enzyme combination. However, using this technique no polymorphism could be detected between Sultanina and the mutant cultivar `Seeded Sultanina', or between `Flame-Tokay' and the mutant cultivar `Seedless-Tokay'.
The demonstration of highly polymorphic patterns within the same Species, using minisatellite probes, encouraged further molecular investigations with a new genetic assay to detect nucleotide sequence polymorphisms by polymerase chain reaction (PCR) procedures, developed by Welsh and McClelland (1990) and Williams et al. (1990). Using the technique to produce molecular genetic markers, 185 different 10mer primers (Operon kits A, B, C, E, F, G, H, J and other primers) were tested. DNA was extracted from young grape leaves following the procedure given by Lodhi et al. (1993). As has been described by others (Gogorcena et al, 1993; Buschner et al, 1993; Jean-Jaques et al, 1993) genetic variability among grape cultivar was demonstrated. Some of the primers tested by us were highly polymorphic. Band patterns of two primers, successfully differentiated between 24 different grape cultivars.
110 of the primers tested, gave a distinct band pattern. Only 17 primers distinguished between two closely related cultivars, one being a mutant of the other (Sultanina vs. Seeded Sultanina). Bandsharing ratio of 96.1% was found in this pair, compared to 78.5% bandsharing ratio between two nonrelated cultivars (Early-Muscat and Flame-Seedless). 68 primers that uncovered polymorphism between `Early-Muscat' and `Flame- Seedless' were tested with 82 individuals of the progeny resulting from the cross between these cultivars. More than 400 polymorphic markers were identified.
In the present study I analyzed seven variables representing the trait of seedlessness: mean fresh weight of one seed (mg.), total fresh weight of seeds per berry (mg.); seed contents evaluated by perceptability (Spiegel-Roy, 1979); four seed size categories evaluated visually; degree of hardness of the seed coat; degree of development of the endosperm and degree of development of the embryo (Striem et al., 1992). 12 markers gave significant correlations with some of the seven subtraits of seedlessness, mainly with the quantitative ones. For example: primer OP-E10, band number 5 (size of about 900 bp), is present in Flame- Seedless, and also in a mixture of DNAs of 10 seedless offsprings, and absent in Early-Muscat band pattern. This marker gave significant correlation coefficients with the seven subtraits examined (for example: r=0.411, P=0.0017 with total fresh weight of seeds per berry).
By analyzing several markers I found that the quantitative traits, (concerning the fresh weight of seeds), had higher correlation coefficients and linear effects than the qualitative traits (evaluation of the seed content by taste and visually, as well as the degree of development of the seed components). Multiple linear regression analysis resulted in higher coefficients with the quantitative traits when seven markers were included in the model as independent variables. Anyway, a combination of four markers has proven adequate to distinguish between seeded and seedless offsprings. Based on their band pattern, 10 seeded offsprings had the same pattern, and thus could be excluded from the progeny. In the group of the opposite patterns, 18 out of 24 offsprings were seedless. This demonstrates the efficiency of using such markers as tools for pre-selection.
Significant correlations with muscat flavor were found with 11 markers (for example: r=0.575 with marker 62.12), and 16 markers showed significant correlations with berry color (for example:r=-0.508 with marker 91.7). In multiple linear regression analysis, when four markers were included in the model, correlations were R=0.935 for muscat flavor and R=0.898 for berry color.
Linkage analysis between markers can be performed to identify linkage groups. As soon as at least 200 markers for each parent (segregating in a 1:1 ratio in the progeny) will have accumulated, mapping can be commenced.
In conclusion one may say that the here presented investigation of the three aspects of the trait of stenospermocarpic seedlessness in grapes resulted in a better understanding of this trait:
1. The organoleptic evaluation, and also the visual classifiction of the size of the seeds were analysed, as was the degree of development of the seed components.
The degrees of development of the seed coat and of the endosperm were found to form recombinant individuals, bearing seed traces with no endosperm in hard seed coats and vice versa.
7 subtraits were defined, based also on the quantitative elements of the fresh weight of the seeds.
2. The response to gibberellin was found to emphasize the differential sensitivity of the seed components at different developmental stages to the treatment. However the quantitative factor of the seed size was found to play the main role in the response to growth regulators.
3. 13 Molecular genetic markers were identified, with a degree of linkage to factors involved with seedlessness. Highly significant coefficients were found mainly with the quantitative subtraits (i) mean fresh weight of one seed and (ii) total fresh weight of seeds per berry.
A combination of four markers was used to identify seeded offsprings among the progeny investigated. These results will be of value in breeding programs.
Biological and molecular aspects of seedlessness in grapes (Vitis vinifera)
Thesis Submitted for the Degree DOCTOR OF PHILOSOPHY
By Michael J. Striem
Submitted to the Senate of the Hebrew University of Jerusalem, January 1994