‘Comment on Saumitou et al. (2017): Elucidation of the genetic architecture of self‐incompatibility in olive: evolutionary consequences and perspectives for orchard management’

Abstract The new self‐incompatibility system (SI) was presented by Saumitou‐Laprade, Vernet, Vekemans et al. (2017). Evolutionary Applications based on 89 crosses between varieties in the olive tree. Four main points are not clear. We are examining here as follows: (i) the assertion that the self‐incompatibility system is sporophytic was not sustained by pollen germination data; (ii) surprisingly, the new model does not explain that about one‐third of pairwise combinations of olive varieties leads to asymmetric fruit setting; (iii) DNA preparation from one seed may contain two embryos, and thus, embryos should be separated before seed extraction; (iv) although effective self‐fertility in olive varieties was reported by many studies, the DSI model fails to explain self‐fertility in some olive varieties. Moreover, we cannot discuss result data, as science cannot be verified because variety names were encoded, this does not allow comparison of data with previous works. The DSI model on olive self‐incompatibility should explain more features than the model based on four dominance levels shared by six S‐alleles. Perspectives for orchard management based on this model may face serious limitations. An olive variety does not have a fifty percent chance of cross‐incompatibility, but surely fewer, and thus, the sporophytic system limits fruit production. Evolutionary perspectives of self‐incompatibility in Oleaceae should include data from the Jasmineae tribe that displays heterostyly SI.


| INTRODUCTION
presented a new self-incompatibility system (SI) based on 89 crosses between varieties in the olive tree.
The authors have indicated that there are doubts on the sporophytic SI (SSI) in the olive, published by Breton and Bervillé (2012). Predictions from the SSI model have always matched experimental data based on fruit setting Farinelli, Breton, Famiani, & Bervillé, 2015); moreover, they showed the scale of dominance shared between 6 S-alleles. The Breton et al.'s model was sustained by all cross data and some diagnostics for SI based on pollen germination (Bradley & Griggs, 1963;Ouksili, 1983), and so far, in more than three thousand crosses (see references in Farinelli et al., 2015;Koubouris, Breton, Metzidakis, & Vasilakakis, 2014). No ambiguity has appeared to identify the sporophytic system, because all these authors displayed reciprocal crosses with opposite fruit sets (Gerstel, 1950). Gerstel (1950) based his studies on Guayule (Parthenium argentatum Gray).
It is unclear to us how Saumitou-Laprade et al. (2017) could conclude on SSI after pollen germination tests and a few controlled crosses under pollination bags although they only observed symmetric compatibility or incompatibility for each pairwise combination of olive varieties. They observed 1:1 segregation for self-fertility in pseudobackcross progenies Oit64xOit27 (which denomination is Oit64?). We have to believe because the cross remains unidentified. Such a genetic structure progenies is not common in genetic analysis. Some of the authors in Saumitou-Laprade et al. (2017) also handle the offspring Picholine marocaine x Picholine in which the 1:1 segregation of selffertility should be checked. Breton, Farinelli, Koubouris, and Bervillé (2016) have shown than the self-fertility level depends on the S-allele pair and on modifiers which co-segregated with the S-loci.
Also Saumitou-Laprade et al. (2017) provided the Collani et al. (2012) reference to sustain SSI. Until now, a SSI system resulted from cross results data, based on fruit set or on pollen germination tests.
It is very rare that sporophytic SI is verified through molecular data, except in Brassiceae (Chookajorn, Kachroo, Ripol, Clark, & Nasrallah, 2004). Thus, it remains to be given, which crosses between identified varieties show in Saumitou-Laprade et al. (2017) that SI is sporophytic yet? Saumitou-Laprade et al. (2017) introduced 2 S-alleles (S1 and S2) and displayed symmetric diagnostics for compatibility in G1 and G2 groups (Table 1). Symmetry in compatibility or in incompatibility for pairwise combinations of olive varieties is observed and verified based on fruit setting in about half of pairwise combinations of olive varieties (Table 3 in Breton et al., 2014). Furthermore, several authors have reported that for the (more or less) other half of olive crosses, pairwise combinations of olive varieties show asymmetric fruit set (Musho, 1977;Ouksili, 1983;Villemur, Musho, Delmas, Maamar, & Ouksili, 1984;Moutier, Terrien, Pécout, Hostalnou, & Margier, 2006;Farinelli et al. 2008;Spinardi & Bassi, 2012;Farinelli et al., 2015). The proportion is more or less 50% depending on the set of varieties sampled for the study by each team.
The explanation for asymmetry of fruit setting has been given in Breton et al. (2014) and asymmetry in fruit setting proved the This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. sporophytic model as pointed out by Gerstel (1950). In one direction, the cross leads to fruit set because of compatibility, and in the other direction, the cross fails because of incompatibility. However, in some peculiar situations, asymmetry may occur because of male sterility of one olive variety (Besnard, Khadari, Villemur, & Bervillé, 2000;Villemur et al., 1984), and because one variety is self-fertile (Table 1) (Breton et al., 2016;Spinardi & Bassi, 2012).
Such pairs of asymmetric crosses are due to the differential Sallele expression in the female and the male parts (see references in Saumitou-Laprade et al., 2017). This has been shown in crosses for several species harboring a SSI system in guayule, sunflower, and hazelnut (Gerstel, 1950;Mehlenbacher, 1997;Ségala, Ségala, & Piquemal, 1980). Recently, Breton et al. (2016) have shown a correlation between the S-allele pair and the level of self-fertility. In some pairwise combinations of varieties, when the host variety is self-fertile enough, fruit setting occurs under the bag and the final diagnostic for cross compatibility could be erroneous. Indeed, compatibility or incompatibility diagnostic has not been given to the pair Frantoio and Leccino (Spinardi & Bassi, 2012), because the origin of fruit remains inexplicable. Consequently, a column was added in Table 1 of the present study to show whether fruit setting in the host may be attributed to selfpollination and not to foreign pollen. Controls by paternity tests have not been performed at this time. Thus, some examples of pairs of varieties that lead to asymmetric crosses are given in Table 1, with reference to Breton et al. (2014) and Farinelli et al. (2015), based on the list of varieties (#105) deciphered for the S-allele pair, so far (C. M. Breton, D. Farinelli, G. Koubouris, F. Famiani, A. J. Bervillé, Unpublished). (Table 1) Consequently, the DSI model is probably useful to draw continuity between the SI systems in Oleaceae-Phillyrea-Fraxinus-Olea -this is an important opened question. However, no comment was given on the Jasmineae tribe of this family that displays architecture SI (Olesen, Dupont, Ehlers, Valido, & Hansen, 2005 Tanche is partially male sterile (Besnard et al., 2000). Symmetric crosses occur when the two varieties do not share the same S-allele pair, and when both S-determinants are present on the pollen coat.  Table 4) have shown that each host variety has a different probability to match with a pollen donor, that renders complex future orchard composition to equilibrate host varieties and pollinizers. For those orchards, already in production, the recommendation to enhance pollination is to introduce new pollen sources by grafting or planting different pollinizers to ensure correct pollination.

| CONCLUSION
Our goal was to improve the clarity of SI in the olive -here olive means the cultivated form. Olive growers will probably not be interested by these exchanges unless they can identify the olive materials. To summarize, the pertinent points addressed in the letter, at least for us are, 1. 1:1 segregation of SI should be checked in two different pseudo-backcross offsprings.

The main progenies should be identified (which denomination is
Oit64?) as the materials given in tables to enable comparison with published data.

The work described in Saumitou-Laprade et al. (2017) is a verifica-
tion of the DSI model, and it is the first step. The second step is to predict, for chosen pairs of varieties after crosses in both directions, that fruit set is symmetric (they will succeed or they will fail in both directions), and the third step is to predict for some other crosses -in both directions -that they succeed in one direction and fail in the other direction. Then, we would see comparison of prediction based on pollen germination and experimental data for fruit setting.
We have gone in Breton et al. (2014) and Farinelli et al. (2015) through these steps successfully for more than 100 pairwise combinations of varieties.

4.
Fruit numbers were not referred to hermaphroditic flowers, thus the fruit number under a bag has no meaning when comparing fruit setting between varieties. This point is the key problem in most olive studies. Marchese et al. (2016) have eliminated most profiles (supposed to correspond to one embryo) because they have more than two SSR alleles at some loci. Consequently, using seedlings helps to avoid the problem to mix embryos, but delays the data for 1 year. The most probable father has never been verified by other independent method(s), and nobody has published the verification of the compatibility between the putative father and the host based on a controlled cross.

5.
Finally, Saumitou-Laprade et al. (2017) found differences between pollen germination tests (data are qualitative, all or nothing) leading to all their inferences, and the bag method, which provided fruit set quantitative data leading to other inferences (Table 1). This is sustained in senecio (Brennan, Harris, Tabah, & Hiscock, 2002), chicory (Gonthier et al., 2013), in sunflower (Nooryazdan, Serieys, David, Baciliéri, & Bervillé, 2010), and here for the olive. Consequently, inferences from pollination germination tests remain to be conciliated with those from fruit set data under bags.