Review. Flower biology in apricot and its implications for breeding

Different aspects related to flower biology have a close link to fruit set failures in apricot and other fruit trees. In this work, studies on pollen viability and germinability, stigma receptivity, ovule development and longevity, the different factors affecting the effective pollination period (EPP), are reviewed. The concept of EPP is based on biological parameters that are the successive steps that take place during the reproductive process and it is the frame within which the factors limiting an appropriate fruit set can be studied. The definition of this concept and its detailed study have allowed determination of the different limiting factors and the design of specif ic treatments to improve it. Knowledge of the incompatibility phenotype for many apricot cultivars has allowed advising about the planting of singlecultivar orchards. The study of the inheritance of this and other traits in apricot and other fruit trees has allowed planning of hybridisations to minimise or eliminate the production of undesirable seedlings, increasing the efficiency of the breeding programme. Studies on the flower biology of apricot have provided valuable information to help select the appropriate parent cultivars for breeding programmes, also this information is transferred to farmers to avoid losses produced by an inadequate cultivar selection. In this review we intend to give an updated overview of the state of the art in the research and the achievements thus far, as well as considering the implications of these studies for fruit breeding in general, with special attention to apricot breeding.


Introduction
Different aspects related to flower biology have a close link to fruit set failures in apricot and other fruit trees.Studies on the flower biology of apricot have provided valuable information to help select the appropriate parent cultivars for breeding programmes and transfer to farmers to avoid losses produced by an inadequate cultivar selection.In this review we intend to give an updated overview of the state of the art in the research and the achievements thus far, as well as considering the implications of these studies for apricot breeding.
The review covers the importance of an enough production of flower buds as well as their permanence on the branches until they develop to flowers, are pollinated and become fruits, for the suitability of apricot cultivars to give adequate yields.
Different aspects related to the pollen, the stigma and the ovules are considered that influence greatly the possibilities of the flowers to set fruits.The effective pollination period (EPP) is considered to be an expression of the likelihood that the flowers set fruit and therefore it links female fertility and pollination by considering in a global form all parameters related to reproductive biology.
The need of pollinators, with overlapping blooming times, as well as pollinating insects to transfer the pollen, make self-incompatible cultivars unsuitable to modern horticultural practises.Knowledge on the inheritance of this trait and methodologies to determine the genotypes of different cultivars as soon as possible, have allowed the planification of hybridisations so that the number of self-incompatible seedlings is minimised in the progenies from controlled crosses.A review of the main aspects of this trait in fruit trees and future prospect of the current research are outlined in this work.

Flower bud density and drop
Depending on the intensity, flower bud drop may negatively influence final yield.Several factors are considered common causes of flower bud drops (water stress, lack of chilling, high temperatures during autumn or winter, etc.).
Important losses of flower buds have been associated with deficit irrigation treatments in apricot (Hendrickson and Veihmeyer, 1950;Brown, 1952;Brown, 1953;Uriu, 1964).However, other authors did not find an influence of different irrigation treatments on flower bud drop in apricot (Alburquerque et al., 2003).
Warm temperatures during autumn and winter have been considered responsible for incorrect flower development and, therefore, large flower bud drops in peach (Brown, 1958;Monet and Bastard, 1971).Unsatisf ied chilling requirements have also been related to flower bud drop in apricot (Legave, 1978).However, other authors did not observe an influence of chilling on flower bud drop in apricot cultivars (Viti and Monteleone, 1991;Alburquerque et al., 2003).
The different results found could be explained if flower bud density (number of flower buds per branch section) and flower bud drop were genetically conditioned traits.A strong influence of the cultivar on flower bud drop in apricot has been found (Legave, 1975;Legave et al., 1982).Also in apricot, when nine different cultivars were studied during three consecutive years, flower bud density and flower bud drop were not affected by the climatic conditions of the different years but there were large differences between cultivars.Flower bud densities ranged from 63 to 180 buds cm -2 and percentages of flower bud drops were over 50% in many cultivars and ranged from 13% to 72%, expressed as averages from the three years (Alburquerque et al., unpublished results).It has been found also, in peach and nectarine, that flower bud density (Bellini and Gianelli, 1975;Okie and Werner, 1996) is highly dependent on the cultivar studied.
A scarce flower bud production and/or high flower bud drop is indicative of poor productivity.Since these characters seem to be cultivar-dependent, they may be inherited and therefore the use of such cultivars as parents within a breeding programme will not be advisable.

The pollen
When studying pollen from apricot cultivars it was found that, with the exception of some male sterile cultivars like 'Colorao' or 'Arrogante', most of the apricot cultivars produce pollen in quantities that range from 2,000 to 4,000 grains per anther, which is more than 90,000 pollen grains per flower (Egea and Burgos, 1993).Furthermore, this pollen has a high percentage of viability and germinates, emitting a pollen tube, in a wide range of temperatures (Vachun, 1981;Egea et al., 1992).

Male sterility
Male sterility is defined as the deviant condition in normally bisexual plants when no viable pollen is formed (Frankel and Galun, 1977).Male sterility has been exploited as an effective tool to aid hybrid seed production in many crops.However, male sterility is an undesirable characteristic in scion cultivars of Prunus to be used for fruit production, because this trait would restrict yield in large monoculture production blocks.Male-sterile cultivars need cross-pollination and production would depend on an adequate pollen transfer from other cultivars.
A review of apricot pollen fertility (Burgos, 1991) indicated that only three male sterile cultivars, 'Arrogante', 'Colorao de Moxó' and 'Colorao', have been described.Male-sterile anthers can be distinguished visually from normal fertile anthers during the bloom period.Shrunken, discoloured anthers are indicative of male sterility and provide a sharp contrast to the swollen, yellow appearance of normal, pollen-fertile anthers (Burgos and Ledbetter, 1994).A relatively high number of male-sterile trees were observed by these authors in progenies from controlled hybridisations among fertile cultivars in apricot, and they proposed a preliminary model for the inheritance of the trait.Later, it was confirmed (Burgos and Egea, 2001) that the trait is controlled by one recessive gene (Table 1).Five cultivars or selections included in this study were heterozygous for this trait and, since all hybridisations among them were performed to combine fruit quality attributes and the heterozygous status was unknown, this trait can be of economic importance in the eff iciency of an apricot breeding programme, since hybridisation among heterozygous cultivars would produce 25% of male-sterile progeny.
Recent research in peach has described a different type of male sterility, that has been proposed to be due to cytoplasmic inheritance (Werner and Creller, 1997).These authors found that all crosses between the male-sterile parent and normal cultivars resulted in a completely male-sterile offspring.Furthermore, when these F 1 seedlings were open-pollinated or backcrossed with the fertile parent all progenies were male-sterile.
The knowledge on the inheritance of this trait will help to plan hybridisations, so that production of male-sterile progeny is avoided through selection of homozygous fertile parents.Also, this information and the progenies generated to obtain it, have helped the search for molecular markers for this trait, that will allow detection and elimination of male-sterile plants at the seedling stage (Badenes et al., 2000).

The pistil
It has been demonstrated that fruit set is determined by numerous factors that affect different processes occurring in the pistil during pollination, pollen tubes germination and growth through the stiles and ovule fertilisation.For instance, it has been found that high temperatures during the pre-blossom weeks produce abnormal flowers and diminish fruit set in apricot (Rodrigo and Herrero, 2002) as well as the ovule viability in almond (Egea and Burgos, 1995a).Stigma receptivity (Egea et al., 1991a;Egea and Burgos, 1992;González et al., 1995), the role of the pistil in controlling pollen tubes growth (Herrero, 1992;Herrero and Hormaza, 1996), ovule maturity at anthesis (Egea and Burgos, 1994;Egea and Burgos, 1998;Alburquerque et al., 2000 and2002a) and its subsequent evolution (Burgos and Egea, 1993;Burgos et al., 1995) have been studied widely in apricot and other fruit trees.

Macro styles
The length of some pistils places the stigmas above the anthers when their natural position should be at the same or a lower height.Macro styles are a cultivar characteristic that is inherited, although climatic conditions, especially temperatures before or after anthesis, play an important role in regulating the manifestation of the trait.In apricot cultivars with the stigma 2 to 3 mm above the anthers, at anthesis, a much lower number of pollen grains has been found on the stigmas than in flowers of cultivars with the anthers at the same height as or above the stigmas, when those flowers were within bagged branches and cross-pollination was absent (Egea and Burgos, 1993).Macro styles may produce important crop failures when there are few bees or when climatic conditions do not allow the activity of these insects.Self-compatible cultivars with long styles may behave as incompatible in these conditions.Also, since stigmas project out of the flower, the risk of quick desiccation and subsequent loss of receptivity is high.

The stigma
Stigma receptivity is fundamental, in many instances, for explanation of phenomena observed during fruit setting.In some cases, the stigma has been considered responsible for the success of some cultivars like the pear 'Decana del Comizio' (Bini and Bellini, 1971;Bini, 1972).Other papers have reported immature stigmas at anthesis in the pear 'Agua de Aranjuez' (Herrero, 1983;Sanzol et al., 2003) or the apple 'Cox's Orange Pippin' (Williams et al., 1984).In apricot, immature stigmas at anthesis have been found also in some apricot cultivars, reaching an optimum receptivity two to four days after anthesis and losing it very quickly thereafter (Burgos et al., 1991;Egea et al., 1991a).In the Southeast of Spain, many apricot cultivars have an extremely short period in which stigmas are receptive (Egea and Burgos, 1992).
Frequently, more than two ovules have been found in apricot.However, extra ovules are generally mal-formed or they degenerate quickly (Burgos and Egea, 1993;Egea and Burgos, 1995b).
Figure 1 shows the different embryo sac developmental stages in apricot.At anthesis, apricot ovules are not mature and frequently they are in a very immature stage (Egea and Burgos, 1994;Alburquerque et al., 2000 and2002a).Most ovules examined were within the first three stages of development in our classification (i.e. from ovules without embryo sac to fournuclei embryo sacs), with high percentages of ovules without a differentiated embryo sac (Table 2).Lillecrap et al. (1999) found small and delayed embryo sacs at anthesis in an apricot cultivar with frequent low yields, whereas most embryo sacs had eight nuclei in two other cultivars which produced good yields generally.
In South-Eastern Spain, apricot cultivars with immature ovules at anthesis (embryo sacs with four nuclei) produced normal crops (Egea and Burgos, 1998).Therefore, those ovules with, at least, a four-nuclei embryo sac at anthesis have been considered as functional (Alburquerque et al., 2002a).In Table 3, the percentages of functional ovules and fructification of nine apricot cultivars are reported.Cultivars with more than 50% fruit set had also high percentages of functional ovules, suggesting that a certain degree of megagametophyte development at anthesis is necessary for fertilisation to be successful, although it may not be enough to ensure a good crop since some cultivars with high percentages of functional ovules had low fruit set.
Both the ovary and the ovule provide signals that orient and direct pollen tube growth to the right course (Herrero, 2001).In peach, particular secretions from ovary cells along the pollen tube pathway are required for the pollen tube to proceed towards the embryo sac (Arbeloa and Herrero, 1987;Herrero, 2000).

The effective pollination period
Williams (1966) introduced the concept of «effective pollination period» (EPP) as the period during which pollination is effective to produce a fruit, and described in detail the approach used to estimate the EPP in orchard conditions, which basically consists of hand-pollinating flowers at time intervals from anthesis and later recording the initial and final fruit set in these flowers (Williams, 1970a).Microscopic examination of pollen tube kinetics and ovule viability can be useful as an indirect estimation of the EPP.Since   Egea andBurgos (1994 and1998).
the EPP is determined by the longevity of the ovule minus the time required by the pollen tube to reach the ovule, this indirect estimation will be valid whenever the EPP values do not exceed the stigmatic receptivity period (Williams, 1966).The microscopic approach provides additional information on the parameters that limit the EPP that is not obtained with the estimation in the orchard.
The EPP was defined as a function of pollen tube speed and ovule longevity.Therefore, it links female fertility and pollination and is an expression of the likelihood that the flowers set fruit.Flower fertility is the capability to produce fruits when flowers are pollinated, at the right time, with compatible pollen.Theoretically, each normally-developed flower is able to set a fruit if pollinated with the appropriate pollen just after anthesis.Its failure to do so is indicative of female sterility.However, under normal conditions, flowers are not always pollinated at anthesis and stigmas remain receptive for several days (Williams, 1970b;Williams et al., 1984).
Stigma receptivity, the speed of pollen tube growth and ovule longevity are three factors commonly-studied in the literature about EPP.Different studies report their relative importances, depending on the species and climatic conditions.There must be a good synchronisation between them, although genetic and environmental factors may unbalance the process and, therefore, decrease fruit setting (Thompson and Liu, 1973).
In fruit trees, including apricot, EPP duration has been estimated to be very variable, depending on the species, cultivar and environmental conditions, ranging from two days to more than a week (Sanzol and Herrero, 2001).When the limiting factor of EPP was determined, in the reviewed papers, a good correlation was found between the two period lengths.In kiwi, the short EPP found was attributed to a fast loss of pollen germinability due to high temperatures (Galimberti et al., 1987) or to lack of support of pollen germination by the stigma (González et al., 1995).Delays in stigma maturation (Martínez-Tellez and Crossa-Raynaud, 1982;Herrero, 1983) or a short receptivity period (Williams, 1965;Guerrero-Prieto et al., 1985;Burgos et al., 1991;Egea et al., 1991a) may limit the EPP.Williams (1970c) found that ovule development is affected by high temperatures, but with temperatures between 7 and 15ºC ovule development is normal while there is an increase in the speed of pollen tube growth.In these conditions, the EPP is improved.In the climatic conditions of South-Eastern Spain, the limited period of stigma receptivity has been found to be responsible for a short EPP in apricot.For many cultivars examined, high temperatures at bloom limit the stigma receptivity to only one to three days after anthesis, in the most extreme cases (Egea et al., 1991a;Egea and Burgos, 1992).Pollen tubes grow fast in these conditions but at least three days are necessary to reach the ovary.
The longevity of the ovule is related to its stage of development at anthesis.Ovules mature at anthesis will remain viable only a short time, limiting the EPP.On the other hand, if ovules are very immature at anthesis, there may be asynchronies between pollen tube arrival and the maturity of the ovules, which will affect fruit set.The most favourable condition for fructification would be when ovules are at intermediate stages of development (embryo sacs with four to eight nuclei) at anthesis (Alburquerque et al., 2002a).

Self-(in)compatibility
Incompatibility is the inability of a fertile seededplant to produce zygotes after self-or cross-pollination (self-or cross-incompatibility) (Heslop-Harrison, 1975).This reaction is an active, regulated constraint of pollen tube growth where, depending on the species and the system operating, the process may be blocked at the initial steps of pollen hydration and germination on the stigma (Dickinson, 1995), during pollen tube growth in the style (Matton et al., 1994) or further down in the ovary (Sage et al., 1994).
Recognising and rejecting their own pollen before fertilisation allows self-incompatible plants to promote outcrossing and improve genetic variability, which is considered to play an important role in the evolutionary success of the angiosperms.Outcrossing establishes a regulated degree of heterozygosity in the population.Incompatibility occurs in more than 3,000 species of 250 genera, that belong to about 70 families (Van Gastel, 1976).Although, traditionally, the European group of apricot (within which the apricots grown in Europe, North America, South Africa and Australia are included) has been described as self-compatible (Mehlenbacher et al., 1991), in the last two decades many widely-cultivated apricot cultivars have been described as self-incompatible (Tables 4 and 5).In fruit trees, incompatibility complicates horticultural practices because self-incompatible clones require the addition of polli- nators and the yield depends on abundant pollen transfer among the trees.

Genetic control
In Prunus, the incompatibility system operating in most of the studied species is controlled by one gene with several different alleles.Pollen is rejected when its S-allele is present in the genotype of the style.Hence, an incompatibility reaction will occur between two plants if their genotypes at the S locus do not differ in at least one allele (De Nettancourt, 1972;Heslop-Harrison, 1975).
Sweet cherry was the first Prunus species where this model was described (Crane and Brown, 1937).The same mechanism has been demonstrated in almond (Dicenta and García, 1993) and apricot (Burgos et al., 1997b).However, a different mode of inheritance was found in Japanese plum, for which it has been proposed that two genes with epistatic relationships control the trait (Arora and Singh, 1990).
In apricot, alleles for self-compatibility would allow pollen tube growth in any style (Table 6, reciprocal crosses of types I and II and crosses of types IV to VIII).Self-incompatibility alleles would stop pollen tube growth if the same allele was present in the pistil and the pollen grain (Table 6, crosses of types III, IV, VIII, X, XI and XII).
To determine the mode of inheritance of self-(in)compatibility in apricot, 19 families with a total of 948 seedlings, were evaluated (Table 7).Seedling segregation for the trait allowed it to be deduced that the parents used were heterozygous.Also, there were two families where segregation could only be explained if the parents shared one allele (Table 6, cross type IV).A similar situation had been found previously in almond when crossing 'Ferragnes' with the self-compatible cultivars  'Genco' and 'Tuono' (Dicenta and García, 1993).Further work on stylar proteins of almond (Boskovic et al., 1997) and apricot (Burgos et al., 1998) cultivars demonstrated the existence of a common S-allele.
Cross type XII in Table 6 could only happen if both self-incompatible parents have the same genotype.Two groups of cross-incompatible cultivars have been described after controlled pollinations.One of them includes three Hungarian apricot cultivars (Nyéki and Szabó, 1995) and the other the North American cultivars 'Lambertin', 'Goldrich' and 'Hargrand' (Egea and Burgos, 1996).

Molecular aspects of incompatibility
Within the Rosaceae, a correlation between known genotypes for self-(in)compatibility and bands resulting from electrophoresis of stylar extracts has been found in Japanese pear (Hiratsuka et al., 2001;Sassa et al., 1992) where the proteins have been characterised as glycoproteins with RNase activity (Hiratsuka, 1992;Sassa et al., 1993;Hiratsuka et al., 1995;Hiratsuka and Okada, 1995).Similar results have been found in apple (Sassa et al., 1994), and European and Chinese pears (Tomimoto et al., 1996).
In Prunus, similar studies have been carried out in sweet cherry (Mau et al., 1982;Boskovic and Tobutt, 1996;Boskovic and Tobutt, 2001) and almond (Tao et al., 1997;Boskovic et al., 1997;Certal et al., 2002).In our laboratory, a good correlation was established between RNases from stylar extracts and the available information on (in)compatibility genotypes of different apricot cultivars (Table 8).It was also demonstrated that these proteins were in-herited as if they were the products of the S gene (Burgos et al., 1998) and this methodology was used to genotype unknown cultivars and selections from the breeding programme (Alburquerque et al., 2002b).
A further step in the molecular research on S-alleles in fruit trees was the use of a combination of S-allele-specific primers, designed from non-conserved sequences from each allele in apple, and the digestion of PCR products with S-allele-specific restriction enzymes (Janssens et al., 1995).Results from this approach to the identification of S-alleles correlated perfectly with information on genotypes from phenotypic and RNases analyses and it is a rapid and useful method for determination of the genotype of different apple cultivars (Sakurai et al., 1997 and2000).A recent paper reports the identification of 15 different S-alleles in apple using this methodology (Broothaerts, 2003).
In apricot, the alleles S 1 and S 2 have been sequenced completely (Romero et al., 2003) by using a bacterial artificial chromosome (BAC) library from the cultivar Goldrich (Vilanova et al., 2003).This is a first step that will allow the design of primers from these sequences in order to amplify different S-alleles in apricot.The possibility of designing primers for the self-compatibility allele found in all self-compatible apricot cultivars tested to date (Alburquerque et al., 2002b) is especially interesting (Burgos et al., 1998).A similar strategy has allowed the design of molecular markers for this important trait in Japanese apricot (Tao et al., 2000(Tao et al., , 2002a(Tao et al., and 2002b)).

Conclusions
The study of the flower biology of apricot, described in this review, has had strong implications for the breeding programme of this species, which has been developed at the same time.First of all, the knowledge of the factors limiting fruit set in an important number of commercial cultivars has oriented the selection of parents.Some cultivar-dependent characteristics, like macro styles and flower bud density or drop, indicate that some cultivars would not be a good choice as parents in the breeding programme.Other factors, like ovule immaturity at anthesis, are signs of bad adaptation of the cultivars to local climatic conditions and these, therefore, would also be a wrong parental selection.In those cases when such parents must be used, the knowledge of these characteristics is important in order to evaluate the seedlings, paying much attention to the possible segregation of these traits within the progenies in order to select the ones that have not inherited the undesirable characters.
Determining the mode of inheritance of economically-important traits improves the eff iciency of breeding.For instance, male sterility may produce up to 25% of male-sterile seedlings from crosses between fertile heterozygous cultivars.The selection of the appropriate parents is, again, the solution.Also, determining the inheritance of self-(in)compatibility and the parents' genotypes for this trait allows hybridisations to be planned which minimise or eliminate the production of self-incompatible seedlings.The correlation between stylar RNases and different Salleles has been a great advance for determination of the genotype of a good number of cultivars.With this methodology, homozygous self-compatible cultivars can be easily identif ied, which will produce 100% self-compatible progeny regardless of the other parent's genotype.If the necessity of evaluating the progenies generated within the breeding programme, to discard the self-incompatible seedlings, is eliminated, the programme is speeded up, which greatly reduces its cost.
Self-incompatibility phenotype determination by controlled crosses and evaluation of fruit set or pollen tube growth as well as RNase analysis, to determine the genotype at the S locus, need mature trees with flowers, which, for fruit trees, means at least three years after seeds are obtained.Using PCR with S-allele-specif ic primers allows detection of the self-incompatible genotype in the f irst stages of plant development, and therefore allows roguing of undesirable seedlings straight after germination of the seeds.Specific primers to amplify selectively the allele (or alleles) that determine self-compatibility are molecular markers for this trait with 100% efficiency, since they are located within the S locus.In apricot, these primers have not yet been identified nor efficient molecular markers developed.However, some recent papers on this species, and methodologies developed in related Prunus species, indicate that they will soon be available.
The number of publications in recent years indicates the interest in the different aspects of reproductive biology.This interest is, possibly, closely linked to the fact that this knowledge may avoid production failures and also allows the efficiency of the fruit breeding programmes to be increased.

Figure 1 .
Figure 1.Developmental stages of the embryo sac.(A) No embryo sac.Mother cell of the embryo sac (m).Immature embryo sacs with two (B), four (C) and eight (D) nuclei.(E) Mature embryo sac with the egg cell (e), sinergids (e) and unfused polar nuclei (p).(F) Mature embryo sac with fused polar nuclei (fp).Bars represent 20 µm in A and 50 µm in B, C, D, E, and F.

Table 1 .
Number of male-fertile and male-sterile seedlings obtained from crosses among cultivars with different malesterile genotypes*

Table 3 .
Percentages of functional ovules and fruit set in different apricot cultivars

Table 4 .
Main self-compatible apricot cultivars

Table 6 .
Theoretical crosses and expected genotypes and phenotypes, depending on the self-(in)compatibility status of the parents * Reciprocal crosses are not included since the same genotypes are expected.

Table 7 .
Rate of self-compatible and self-incompatible seedlings obtained in crosses among apricot cultivars, depending on the type of cross*