Sexual compatibility of the olive cultivar ‘Kalamata’ assessed by paternity analysis

Paternity analysis was used to assess the self-incompatibility of the olive (Olea europaea L.) cultivar ‘Kalamata’ and to identify some compatible pollenisers under a Mediterranean-type climate. Eight microsatellite markers were used for genotyping three ‘Kalamata’ mother trees, 120 embryos, and all potential pollen donors. The identified alleles were analysed using FaMoz software and showed that ‘Kalamata’ was highly self-incompatible. Only three ‘Kalamata’ embryos were assigned to ‘Kalamata’ self-fertilisation, even though it was the most available pollen donor. The alleles were also analysed using NTSYS-pc (version 2.02 k) software and identified 54 potential pollen donors in the study site; however, not all of them were located within the effective pollination distance of the mother trees (30 m in olive). According to the results of this study, ‘Kalamata’ (as a host) was compatible with ‘Barnea’, ‘Benito’, and ‘Katsourela’ (six ‘Kalamata’ embryos assigned in each) but incompatible with ‘Arbequina’, ‘Azapa’, and ‘Picual’ (zero ‘Kalamata’ embryos assigned in each). The olive growers could use some of these compatible pollenisers with ‘Kalamata’ to guarantee good fruit set. Additional key words: Olea europaea L.; polleniser; self-compatible; self-incompatibility.


Introduction
Self-incompatibility (SI) is a mechanism to prevent self-fertilisation in plants.Most olive (Olea europaea L.) cultivars are self-incompatible or show some level of SI and need to be fertilised by other cultivars for successful fruit set (Fabbri et al., 2004;Conner & Fereres, 2005).As a result, SI obliges olive growers to plant more than one cultivar in their orchards to ensure sufficient crosspollination (Martin et al., 2005;Mookerjee et al., 2005).Climatic conditions, especially air temperature, have a significant effect on the degree of SI; thus, it can change from environment to environment and from year to year (Androulakis & Loupassaki, 1990;Lavee et al., 2002).On the other hand, the olive pollen grain can be carried by wind as far as 12 km (Fabbri et al., 2004); however the effective pollination distance (EPD) has been reported to be 30 m in normal conditions (Ayerza & Coates, 2004;Fabbri et al., 2004).
Different methods have been used to study the SI of olive: measurement of fruit set (Cuevas et al., 2001) and pollen tube observation (Cuevas et al., 2001;Wu et al., 2002) after controlled crossing, in vitro pollen germination and pollen tube growth in a culture medium sometimes containing pistil extracts of other cultivars (Lavee & Datt, 1978;Fernandez-Escobar et al., 1983), and paternity analysis (de la Rosa et al., 2004;Mookerjee et al., 2005;Diaz et al., 2006).In paternity analysis, the genotype of the mother plant is compared to the genotype of offspring to distinguish the father.Microsatellite markers are codominant and highly polymorphic, two characteristics that make them especially useful for paternity analysis (Queller et al., 1993).
The aim of this study was to assess the SI of 'Kalamata' and the cross-incompatibility between 'Kalamata' (as a host) and other cultivars using eight microsatellite markers in order to select good pollenisers for a Mediterranean-type climate.'Kalamata' is one of the most popular table olives grown worldwide and in Australia (Kailis & Davies, 2004) and can also be used for oil extraction.

Plant materials
The study was conducted in 2004 on three flowering trees of 'Kalamata' (trees 30IIA, 30IIB, and 30IIIB, as mother trees 1, 2, and 3, respectively) at the National Sexual compatibility of cv.'Kalamata' assessed by paternity analysis the quality was calculated by the ratio of absorbance at 260 and 280 nm.DNA samples with absorbance ratios more than 1.8 were used for further analysis and stored at -20 °C.

Genotyping
Eight microsatellite primers were used for genotyping the mother trees, embryos, and potential pollen donors (Table 1).FAM-and HEX-labelled primers were obtained from GeneWorks Pty Ltd, Adelaide, SA, Australia, and NED-labelled primers were obtained from Applied Biosystems, USA.The primers were used for amplification in three groups: 1) UDO8, EMO2, and DCA9, 2) UDO24, DCA4, and DCA14, and 3) UDO6 and DCA3.The segregation of the microsatellite primers used had been previously tested by Mookerjee et al. (2005).The only new marker was DCA9, which was tested on the progeny of a cross between 'Frantoio' and 'Kalamata', and the segregation of the amplification products fitted the expected 1:1:1:1 (χ 2 = 7.000, df 3).
Polymerase chain reaction (PCR) was performed in a volume of 6 µL containing 60 ng DNA of parents or 0.1 µL DNA of embryos (measurement not performed due to the small quantity of the DNA extracted for each embryo), 0.5 mM of each dNTP, 0.15 U of Immolase TM DNA Polymerase (Bioline), 1 × ImmoBuffer [16 mM (NH 4 ) 2 SO 4 , 670 mM Tris-HCl pH 8.3, 0.1% Tween-20], 2 mM MgCl 2 , and 0.5 µM of each forward and reverse primer using a MJ Research Tetrad thermal cycler (MJ Research).The PCR program included an initial denaturation at 95 °C for 7 min, 35 cycles of 45 s at 95 °C, 45 s at 55 °C, 45 s at 72 °C, and a final extension at 72 °C for 20 min.The PCR products were diluted 1:100 and 3 µL was separated on an ABI Prism 3730 DNA Analyser (Applied Biosystems) using LIZ 500 standard.The alleles were scored using GeneMapper version 3.7 (Applied Biosystems).

Data analysis
For each locus allele, frequency and the following genetic parameters in the parent population were calculated: -A O : the observed number of alleles.
-A E : the effective number of alleles, which is a measure of diversity, was calculated according to the  (Morgante et al., 1994): , where p i is the frequency of the i th allele.
-H O : the observed heterozygosity was calculated as the proportion of heterozygotes over genotypes for each locus.
-H E : the expected heterozygosity or gene diversity reflects the level of polymorphism and was estimated using the formula (Nei, 1973) , where p i is the frequency of the i th allele.
-PD: The power of discrimination was calculated using PowerStats (version 12) software (Promega Corporation) as (Kloosterman et al., 1993): , where pi is the frequency of the i th genotype.
-IP: Identity probability represents the probability that two individuals drawn from a population will have the same genotype (Jamieson & Taylor, 1997;Waits et al., 2001).It was computed by FaMoz software (Gerber et al., 2003) to show the probability of wrongly assigning a genotype as the pollen donor.
-EP: Exclusion probability was computed by FaMoz software (Gerber et al., 2003) for paternity and shows the capability of the marker system to exclude any given relationship (Jamieson & Taylor, 1997) and in this experiment any unlikely pollen donor.
-NP: Null allele probability was estimated according to the formula (Brookfield, 1996): Genotyping data were used for paternity analysis using FaMoz (http://www.pierroton.inra.fr/genetics/labo/Software/Famoz/index.html) a software generated by Gerber et al. (2003).FaMoz uses the genotypes of offspring, mother and potential pollen donors to calculate the log of the odds ratio (LOD) scores for any po-tential parentage relationship.The genotype with the highest LOD score is considered as the most likely pollen donor (Gerber et al., 2003).To determine the threshold value of the LOD score to choose a genotype as a true pollen donor, simulation was done using 1,000 generated offspring from the genotyped parents.Possible genotyping error rate for both simulation and LOD score calculation was considered 0.01 (Gerber et al., 2003).Genotyping errors include scoring errors, false homozygotes owing to null alleles or weak amplifications, and mishandled samples (Blouin, 2003), and it is normally in the range of 0.25% to 2% for microsatellites (Ewen et al., 2000).The r × c Fisher exact test, twotailed (also called Fisher-Freeman-Halton test) was conducted using StatsDirect statistical software (version 2.5.7) to analyse the difference among the pollen donors with widespread use in Australia.
The index of SI (ISI) (Zapata & Arroyo, 1978), which is the ratio of fruit or seed set after self-pollination to fruit or seed set after open-pollination, as a potential compatible cross, was calculated to assess the level of SI.A ratio equal to or lower than 0.2 indicates an incompatible cross, between 0.2 and 1 a partially compatible cross, and equal to or higher than 1 a compatible cross.

Results
Genetic parameters of the eight microsatellite loci used are reported in Table 2.The number of alleles per locus (A O ) ranged from four (EMO2) to eight (UDO24), with a mean of 6.3.Expected heterozygosity (H E ), also called gene diversity, varied from 0.643 (UDO24) to 0.838 (DCA3).The cumulative identity probability (IP < 0.00001) showed that the probability of assigning a wrong genotype as the pollen donor was very low.The high cumulative exclusion probability (EP = 0.9968) showed that the marker system was able to exclude almost all (99.68%) unlikely pollen donors for any given offspring.Null allele probability (NP) showed that expected heterozygosity (H E ) in DCA3, DCA9, and DCA14 was not different from observed heterozygosity (H O ) statistically; thus, homozygosity was used instead of null heterozygosity.
The LOD scores for the most likely pollen donors ranged from 2.08 to 4.77.There were some embryos that had LOD scores lower than the threshold calculated in simulation (2.00) (unassigned embryos).There were also some assigned embryos, which had more than one possible pollen donor with the same LOD score.The pollen donors with the same LOD score were those with close genetic distance like 'Frantoio' and 'Mission (WA)', 'Picual' and 'Azapa', and 'Verdale' and 'Benito'.Such embryos were not used to select the compatible pollenisers.Table 3 shows the number of unassigned and assigned embryos.In 'Kalamata' 1, for example, 10 out of 40 embryos did not assign to any pollen donor, 6 assigned embryos had more than one possible pollen donor (with the same LOD score), and 24 assigned embryos were used to select the compatible pollenisers.
NTSYS-pc was used to generate a dendrogram including all 95 genotypes of the NOVA collection and showed the presence of 54 different cultivars (Fig. 2) in the study site (the NOVA collection).Since only some of them are grown commercially in Australia, they were classified into two groups.One group includes 22 cultivars with widespread use in Australia, and the other group includes the other 32 cultivars, with limited use in Australia (Table 4).The Fisher exact test was performed for the pollen donors with widespread use in Australia and showed a highly significant difference among them (p < 0.001).
The ISI was calculated to assess the level of SI in Kalamata.The number of 'Kalamata' embryos assigned to 'Kalamata' itself (3) was used as the fruit set after self-pollination, and the number of embryos assigned to all other pollen donors (69) used as the fruit set after open-pollination.The ISI of 0.04 showed that 'Kalamata' is a self-incompatible cultivar.
Table 5 shows the selected good and poor pollenisers.'Barnea' was a good polleniser for 'Kalamata', and is widely planted in Australia.'Mission (WA)', 'Benito', and 'Katsourela' were three good pollenisers but have limited use in Australia.The number of embryos assigned to 'Mission (WA)' and 'Benito' were 14 and 6, even though one and none of the mother trees were in the 'Kalamata' EPD, respectively.A pollen donor was considered as a poor polleniser only when all of the mother trees were in the EPD.'Frantoio', the most widespread cultivar at the study site, was a poor polleniser for 'Kalamata', even though 'Frantoio' trees were abundantly located around all of the mother trees, some very close to them (< 10 m).

Discussion
In this study, the sexual compatibility of some olive cultivars with 'Kalamata' (as a host) was assessed using eight microsatellite markers.Previous studies used four (de la Rosa et al., 2004;Diaz et al., 2006) and eight microsatellite markers (Mookerjee et al., 2005) for paternity analysis in olive and four (Robledo-Arnuncio & Gil, 2005), five (Isagi et al.,, 2004), and six microsatellite markers (Oddou-Muratorio et al., 2003) for paternity analysis in other tree species.The high EP (0.9968) showed that the applied markers were able to exclude almost all (99.68%) unlikely pollen donors for any given offspring.
Null alleles are alleles with no detectable PCR product after electrophoresis.The identification of null  Crete between 1979 and1983 (ISI = 0.14, 0.14, 0.16, 0.13, and0.12, respectively) (Androulakis & Loupassaki, 1990) and at Roseworthy, SA, Australia (Wu et al., 2002).'Kalamata' was also completely selfincompatible at the much cooler Gumeracha site in SA, Australia in a study using paternity analysis (Mookerjee et al., 2005).The results presented here confirmed the presence of a high level of SI in 'Kalamata' at Roseworthy, SA, Australia (ISI = 0.04).Only three 'Kalamata' embryos were assigned to 'Kalamata' itself (as a pollen donor) from 74 embryos assigned and used for the analysis, and it is low considering the amount Boldface shows that the mother trees were within the effective pollination distance (EPD) of the pollen donor (30 m). 1 Also called 'Hardy's Mammoth' in Australia.
of 'Kalamata' pollen that must have been in contact with the stigmas on the mother trees.Under the conditions of this study the only good polleniser with widespread use in Australia was 'Barnea' (six assigned embryos).Good pollenisers with limited use in Australia were 'Mission (WA)', 'Benito',and 'Katsourela' (14,6,and 6 assigned embryos,respectively).
Olive pollen grains may be carried as far as 12 km (Fabbri et al., 2004), but a polleniser cannot be effective from that distance.The EPD in olive has been reported to be 30 m (Ayerza & Coates, 2004;Fabbri et al., 2004).All good and poor pollenisers in this study were within the EPD of all three 'Kalamata' mother trees, except for 'Mission (WA)' and 'Benito' that show high cross-compatibility with 'Kalamata' (as a host).
Wind is the primary agent of olive pollination, though insects often visit olive flowers to collect pollen (Martin et al., 2005).Long-term averages of climatic data from the Roseworthy Agricultural College Weather Station (34.51 S,138.68E) (Australian Bureau of Meteorology) shows that at the NOVA collection there are some air currents at 9 am during November (olive flowering time) in all directions and at 3 pm especially towards south west (calm days were less than 2%).

Figure 2 .
Figure 2. Genetic distance of the 95 genotypes present in the study site.

Table 1 .
Microsatellite loci used for genotyping and paternity analysis

Table 2 .
Genetic parameters of eight microsatellite loci in parental population O : Observed number of alleles.A E : Effective number of alleles.H O : Observed heterozygosity.H E : Expected heterozygosity or gene diversity.PD: Power of discrimination.IP: Identity probability.EP: Exclusion probability.NP: Null allele probability. 1 The values in bold mean that the test was significant.NA: Not applicable. 2 The values are cumulative for IP and EP and mean for the other parameters.

Table 3 .
Number of embryos unassigned, assigned, and used to select pollenisers

Table 4 .
Number of embryos assigned to putative pollen donors in Kalamata

Table 5 .
Good and poor pollenisers selected for Kalamata