**4.1** *V. planifolia* **in Reunion Island**

The species *V. planifolia* originated in Mesoamerica (Portères, 1954). Some of the history of vanilla follows the history of chocolate because vanilla was gathered from the wild for use in flavoring chocolate beverages in the pre-Columbian Maya and Aztec cultures of southeastern Mexico and Central America. However, the Totonac people of Papantla in north-central Veracruz (Mexico) were probably the first group to cultivate *V. planifolia*  (Lubinsky et al., 2011). The species *V. planifolia* has an interesting history of dispersal to other tropical regions between 27° N and 27° S latitudes (Lubinsky et al., 2008a). After the discovery of the Americas by C. Colombus, the whole history of *V. planifolia* dissemination, following the discoveries of manual pollination by the slave Edmond Albius in 1841 and curing process by E. Loupy and D. De Floris is intimately linked to Reunion Island (Kahane et al., 2008). From then, *V. planifolia* was renowned as 'Bourbon Vanilla' since it was produced originally from Reunion Island (from 1848) and later from a cartel of Indian Ocean Island producers (Madagascar, Reunion, Comoros and Seychelles).

The true origin of cultivated vanilla outside of Mexico was unclear until AFLP and microsatellite markers were used to elucidate the patterns of introduction of *V. planifolia.*  These studies showed that most of the accessions cultivated today in the islands of the Indian Ocean and worldwide (Reunion Island, Madagascar, French Polynesia, French West

population genetic estimates, will be essential (Escuderoa et al., 2003) to provide new insights in the understanding of the mechanisms of maintenance and dynamics of *Vanilla*

*Vanilla* is the only orchid with a significant economic importance in food industry. It is cultivated for its aromatic fruit, a character restricted to some species from the American continent (Soto Arenas & Cameron, 2003). Only two species are grown to produce commercial vanilla: *V. planifolia* and *V. ×tahitensis*; with *V. planifolia* providing 95% of the world production, mainly originating from Madagascar, Indonesia, Comoros, Uganda and India (Roux-Cuvelier & Grisoni, 2010). Biodiversity in cultivated conditions depends on the level of diversity originally introduced and on cultivation practices used in different countries during domestication. Vanilla crops are established from stem cuttings of 8–12 nodes, collected from healthy and vigorous vines (Bory et al., 2008b; Bouriquet, 1954a; Purseglove et al., 1981; Soto Arenas & Cameron, 2003; Stehlé, 1952). As natural pollinators are absent in the areas of vanilla production, pollination is performed by hand following a simple method discovered by the slave Edmond Albius in Reunion Island in 1841 (Kahane et al., 2008). Given these cultivation practices, low levels of genetic diversity are expected in cultivation areas. However, for both species, different varieties, showing recognized but poorly defined morphological, agronomical and aromatic properties, are often cultivated by growers (Duval et al., 2006). Given the vegetative mode of propagation and the absence of pollinators, five hypotheses have been proposed to explain these variations (Bory et al., 2008b): (i) multiple introduction events, (ii) somatic mutations, (iii) sexual reproduction, (iv) polyploidy and (v) epigenetic modifications. In recent years, these hypotheses were explored, giving new insights on the processes involved during the dispersion and domestication of the two main cultivated *Vanilla* species. These results also give important

clues to the understanding of *Vanilla* evolutionary processes in natural conditions.

Island producers (Madagascar, Reunion, Comoros and Seychelles).

The species *V. planifolia* originated in Mesoamerica (Portères, 1954). Some of the history of vanilla follows the history of chocolate because vanilla was gathered from the wild for use in flavoring chocolate beverages in the pre-Columbian Maya and Aztec cultures of southeastern Mexico and Central America. However, the Totonac people of Papantla in north-central Veracruz (Mexico) were probably the first group to cultivate *V. planifolia*  (Lubinsky et al., 2011). The species *V. planifolia* has an interesting history of dispersal to other tropical regions between 27° N and 27° S latitudes (Lubinsky et al., 2008a). After the discovery of the Americas by C. Colombus, the whole history of *V. planifolia* dissemination, following the discoveries of manual pollination by the slave Edmond Albius in 1841 and curing process by E. Loupy and D. De Floris is intimately linked to Reunion Island (Kahane et al., 2008). From then, *V. planifolia* was renowned as 'Bourbon Vanilla' since it was produced originally from Reunion Island (from 1848) and later from a cartel of Indian Ocean

The true origin of cultivated vanilla outside of Mexico was unclear until AFLP and microsatellite markers were used to elucidate the patterns of introduction of *V. planifolia.*  These studies showed that most of the accessions cultivated today in the islands of the Indian Ocean and worldwide (Reunion Island, Madagascar, French Polynesia, French West

populations and to provide guidelines for their preservation.

**4. Vanilla biodiversity in cultivated conditions** 

**4.1** *V. planifolia* **in Reunion Island** 

Indies, Mexico) and of different morphotypes (from Reunion 'Classique', 'Mexique', 'Sterile', 'Grosse Vanille' (table 3) and from Mexico 'Mansa', 'Acamaya', 'Mestiza') (Bory et al., 2008c; Lubinsky et al., 2008a) derive from a single introduced genotype. It could correspond to the lectotype that was introduced, early in the nineteenth century, by the Marquis of Blandford into the collection of Charles Greville at Paddington (UK) (Portères, 1954). Cuttings were sent to the botanical gardens of Paris (France) and Antwerp (Belgium) from where these specimens were disseminated to Reunion Island (by the ordinance officer of Bourbon, Marchant) and then worldwide (Bory et al., 2008b; Kahane et al., 2008).

Consequently, cultivated accessions in Reunion Island exhibit extremely low levels of genetic diversity and have evolved by the accumulation of point mutations through vegetative multiplication (Bory et al., 2008c) (table 3). Maximum genetic distance (Dmax) was 0.106 and the majority of the polymorphic AFLP bands revealed had frequencies in the extreme (0-10% and 90-100%) ranges, therefore corresponding to rare AFLP alleles (presence or absence) a pattern typical of point mutations (Bory et al., 2008c). One peculiar and rare phenotype 'Aiguille' found in Reunion Island was shown to result from sexual reproduction (selfing) (Bory et al., 2008c) (table 3) as its AFLP pattern fell within a group of selfed progeny with Dmax=0.140 and showed a strong pattern of segregation bands. The hypothesis was that it resulted from manual self-pollination and subsequent seed germination from a forgotten pod (Bory et al., 2008c). Flow cytometry, microdensitometry, chromosome counts and stomatal lengths showed that polyploidization has been actively involved in the diversification of *V. planifolia* in Reunion Island (Bory et al., 2008a). Three ploidy levels (2x, 3x, 4x) were revealed that allowed to explain the features of the 'Sterile' type which is autotriploid and of the 'Grosse Vanille' type, auto-tetraploid (Bory et al., 2008a). It was suggested that these resulted from the production of non-reduced gametes during the course of manual self-pollination performed by growers (Bory et al., 2010; Bory et al., 2008a).

As the particular phenotype 'Mexique' encountered in Reunion could not be explained by genetic or cytogenetic variations, we tested whether it could have resulted from epigenetic modifications as some studies showed that morphological variations in clonal populations could be explained by a combination of genetic and epigenetic factors (Imazio et al., 2002). Epigenetics corresponds to reversible but heritable modifications of gene expression without changes in the nucleotidic sequence (Mathieu et al., 2007; Wu & Morris, 2001), such as DNA methylation (Finnegan et al., 1998). Epigenetic modifications are heritable (Akimoto et al., 2007; Finnegan et al., 1996; Grant-Downton & Dickinson, 2006; Martienssen & Colot, 2001) and transmitted as well as by asexual propagation (Peraza-Echevarria et al., 2001).

Sometimes, a phenotypic reversion correlated with demethylation of the epi-mutated gene can occur and its expression is restored (Jaligot et al., 2004). These epigenetic mutations have important phenotypic as well as evolutionary consequences, this representing a current field of investigation (Finnegan, 2001; Kalisz & Purugganan, 2004; B. Liu & Wendel, 2003). DNA methylation proceeds by the addition in a newly replicated DNA of a methyl group by a DNA methlytransferase (Finnegan et al., 1998; Martienssen & Colot, 2001). Cytosine is the most frequently methlylated base, resulting in 5-methylcytosine formation (5mC) (Martienssen & Colot, 2001). Plant methylation is restricted to the nuclear genome and is concentrated in repeated sequence regions (Finnegan et al., 1998). Methylation is implied in many biological processes such as 'gene silencing', mobile DNA elements control, DNA replication duration, chromosome structure determination, and mutation frequency increase (Finnegan et al., 1998; Paszkowski & Whitham, 2001). Many spontaneous or induced epimutations are known in maize, Arabidopsis and other plant species and are responsible

Biodiversity and Evolution in the *Vanilla* Genus 15

collected in vanilla plantations in Reunion Island (St-Philippe or Ste-Rose) and are

We used the restriction enzyme *Eco*RI as well as *Msp*I and *Hpa*II, isochizomers that cut the same restriction site CCGG but show different sensitivity to methylation (table 5). The MSAP methodology used was as described in (Reyna-López et al., 1997). *Hpa*II digests were repeated twice. The adaptators used are presented in table 6 and 8 *Eco/Hpa* primer

Case number 1 1 1 CCGG Case number 2 1 0 5hmCCGG Case number 3 0 1 C5mCGG Case number 4 0 0 5mC5mCGG or 5mCCGG Table 5. Methylation sensitivity of *Hpa*II and *Msp*I (m : methylation; hm : hemimethylation). The comparison of the profiles from the amplification after DNA digestion with *Eco*RI/*Hpa*II and *Eco*RI/*Msp*I gives informations on the methylation status of the internal cytosine in sequence CCGG (table 5). For example a band present in the *Msp*I profile and absent in *Hpa*II indicates a methylation of the internal cytosine, whereas the opposite situation indicates an hemimethylation of the external cytosine. A methylation event was considered as polymorphic when at least one accession differed from the others in its

> Name Sequence (5'-3') Double strand adaptators

> > Pre-amplification primers

Selective amplification primers

Between 48 and 70 fragments were revealed by primer combination. On the 483 CCGG sites observed, 188 were non methylated (38.9%), 36 were methylated (7.45%), with 5 sites only presenting methylation polymorphisms (1.03%) in 4 accessions. Accessions CR0340 and CR0341 were hypomethylated, they showed bands in both their *Hpa*II and *Msp*I profiles whereas the other accessions only presented these bands with *Msp*I. CR0340 was

Table 6. Adaptator and primer sequences used in MSAP analysis.

Ad EcoRI1 CTC GTA GAC TGC GTA CC Ad EcoRI2 AAT TGG TAC GCA GTC Ad HpaII1 GAT CAT GAG TCC TGC T Ad HpaII2 CGA GCA GGA CTC ATG A

Eco-A GAC TGC GTA CCA ATT CA Hpa-A TCA TGA GTC CTG CTC GGA

Eco-AC GAC TGC GTA CCA ATT CAC Eco-AG GAC TGC GTA CCA ATT CAG Hpa-ATT ATC ATG AGT CCT GT CGG ATT Hpa-ATG ATC ATG AGT CCT GT CGG ATG Hpa-AAC ATC ATG AGT CCT GT CGG AAC Hpa-AAG ATC ATG AGT CCT GT CGG AAG

*Eco*RI/*Hpa*II *Eco*RI/*Msp*I CCGG methylation

maintained in the BRC Vatel collection.

profile.

combinations were used for selective amplification.


Table 3. *V. planifolia* morphotypes encountered in Reunion Island and their description.


Table 4. *V. planifolia* Reunion Island accessions surveyed in the MSAP analysis (StP: Saint Philippe; SteR: Ste Rose).

for the generation of mutant phenotypes (Finnegan et al., 1996; Martienssen & Colot, 2001). To assess whether 'Mexique' morphotypes might have resulted from epigenetic modifications, we selected the MSAP (Methylation-sensitive amplified polymorphism) method (Reyna-López et al., 1997), an AFLP-derived methodology which allows the visualization of a large number of markers revealing cytosine methylation state at each digestion site, without any *a priori* knowledge of genomic sequences. MSAP analyses were performed on a sample of 'Classique' and 'Mexique' accessions (table 4). Twenty-four accessions were collected in the collection of Provanille in Bras-Panon (Reunion Island), corresponding to 8 varieties with three cuttings. This was to verify if genetic or methylation polymorphism, if existing, is transmitted through vegetative multiplication. Others were

Point mutations Dmax = 0.106

As self progenies

Same AFLP profile as 'Classique', autotriploid (3x)

Same AFLP profile as 'Classique', autotetraploid (4x)

Same AFLP and MSAP profile as 'Classique'

Table 3. *V. planifolia* morphotypes encountered in Reunion Island and their description.

Accession Morphotype Collection Accession Morphotype Collection CR0217 'Classique' Provanille 3A11 CR0493 'Mexique' Provanille 15A8 CR0218 Provanille 3A11 CR0494 Provanille 15A8 CR0219 Provanille 3A11 CR0495 Provanille 15A8 CR0343 'Classique' Provanille 6A8 CR0334 'Mexique' Provanille 6A5 CR0344 Provanille 6A8 CR0335 Provanille 6A5 CR0345 Provanille 6A8 CR0336 Provanille 6A5 CR0457 'Classique' Provanille 15A6 CR0337 'Mexique' Provanille 6A6 CR0458 Provanille 15A6 CR0338 Provanille 6A6 CR0459 Provanille 15A6 CR0339 Provanille 6A6 CR0563 'Classique' Provanille 16B2 CR0001 'Mexique' BRC Vatel CR0564 Provanille 16B2 CR0002 'Mexique' BRC Vatel CR0565 Provanille 16B2 CR0627 'Mexique' BRC Vatel StP CR0340 'Classique' Provanille 6A7 CR0649 'Mexique' BRC Vatel StP CR0341 Provanille 6A7 CR0632 'Mexique' BRC Vatel StP CR0342 Provanille 6A7 CR0711 'Classique' BRC Vatel SteR CR0647 'Classique' BRC Vatel StP CR0714 'Classique' BRC Vatel SteR

Table 4. *V. planifolia* Reunion Island accessions surveyed in the MSAP analysis (StP: Saint

for the generation of mutant phenotypes (Finnegan et al., 1996; Martienssen & Colot, 2001). To assess whether 'Mexique' morphotypes might have resulted from epigenetic modifications, we selected the MSAP (Methylation-sensitive amplified polymorphism) method (Reyna-López et al., 1997), an AFLP-derived methodology which allows the visualization of a large number of markers revealing cytosine methylation state at each digestion site, without any *a priori* knowledge of genomic sequences. MSAP analyses were performed on a sample of 'Classique' and 'Mexique' accessions (table 4). Twenty-four accessions were collected in the collection of Provanille in Bras-Panon (Reunion Island), corresponding to 8 varieties with three cuttings. This was to verify if genetic or methylation polymorphism, if existing, is transmitted through vegetative multiplication. Others were

Mexico then Antwerp Botanical Gardens

Selfing of 'Classique', unreduced gamete (2n x n)

Selfing of 'Classique', unreduced gametes (2n x 2n)

Epigenetic or genetic single dominant mutation with pleiotropic effects

Dmax=0.140 Selfing of 'Classique'

Morphotypes Characteristics Diversity/genetics Origin

'Classique' The most cultivated

'Aiguille' Slender leaves and

'Sterile' 'Classique', but self-

CR0650 'Classique' BRC Vatel StP

Philippe; SteR: Ste Rose).

'Grosse Vanille'

'Mexique'

type

thin pods

sterile

Bigger leaves, stems, flowers and fruits than 'Classique'

Darker bluish leaves with central gutter and curved sides, cylindrical pods

collected in vanilla plantations in Reunion Island (St-Philippe or Ste-Rose) and are maintained in the BRC Vatel collection.

We used the restriction enzyme *Eco*RI as well as *Msp*I and *Hpa*II, isochizomers that cut the same restriction site CCGG but show different sensitivity to methylation (table 5). The MSAP methodology used was as described in (Reyna-López et al., 1997). *Hpa*II digests were repeated twice. The adaptators used are presented in table 6 and 8 *Eco/Hpa* primer combinations were used for selective amplification.


Table 5. Methylation sensitivity of *Hpa*II and *Msp*I (m : methylation; hm : hemimethylation).

The comparison of the profiles from the amplification after DNA digestion with *Eco*RI/*Hpa*II and *Eco*RI/*Msp*I gives informations on the methylation status of the internal cytosine in sequence CCGG (table 5). For example a band present in the *Msp*I profile and absent in *Hpa*II indicates a methylation of the internal cytosine, whereas the opposite situation indicates an hemimethylation of the external cytosine. A methylation event was considered as polymorphic when at least one accession differed from the others in its profile.


Table 6. Adaptator and primer sequences used in MSAP analysis.

Between 48 and 70 fragments were revealed by primer combination. On the 483 CCGG sites observed, 188 were non methylated (38.9%), 36 were methylated (7.45%), with 5 sites only presenting methylation polymorphisms (1.03%) in 4 accessions. Accessions CR0340 and CR0341 were hypomethylated, they showed bands in both their *Hpa*II and *Msp*I profiles whereas the other accessions only presented these bands with *Msp*I. CR0340 was

Biodiversity and Evolution in the *Vanilla* Genus 17

counts (Lepers-Andrzejewski et al., 2011b). Similarly as in *V. planifolia* (Bory et al., 2008a), 'Sterile' morphotypes in *V. ×tahitensis* were also related to autotriploidy (Lepers-Andrzejewski et al., 2011b). It was hypothesized that they originated from a cross between the two most cultivated morphotypes 'Tahiti' (2x) and 'Haapape' (4x) (Lepers-Andrzejewski et al., 2011b). The remaining accessions showed a pattern related to 'Tahiti' but with 15 to 30 missing bands (Lepers-Andrzejewski et al., 2011a), a pattern consistent with segregation, as shown in *V. planifolia* for the 'Aiguille' morphotype or selfed progenies (Bory et al., 2010). For these accessions, graphical genotypes were constructed based an AFLP 'Tahiti' map and showed that morphotypes such as 'Parahurahu', 'Rearea', 'Oviri' and 'Tahiti court' displayed patterns consistent with an origin via self-pollination of 'Tahiti' (one single recombination event per bivalent) whereas others such as 'Popoti' and 'Paraauti' most probably resulted from a second generation of self-pollination (two recombinations events

These results therefore highlight two different domestication models. In both cases, the genetic base of the cultivated material is very narrow with obviously a single genotype introduced ('Classique' *V. planifolia* in Reunion Island and other cultivation areas; 'Tahiti' *V. ×tahitensis* in French Polynesia). Genetic variation revealed is however slightly higher in *V. ×tahitensis* than in *V. planifolia* because most of *V. ×tahitensis* morphotypes have resulted from selfing of the original 'Tahiti' (with sometimes more than one generation involved) (Lepers-Andrzejewski et al., 2011a). Only one rare case of self-pollination ('Aiguille') was detected in Reunion (Bory et al., 2010). This shows that deliberate or inadvertent seed germination has been strongly involved in the domestication of *V. ×tahitensis* in French Polynesia. In Reunion Island, the limited amount of variation revealed is more related to

In both cases however, a noticeable diversification was achieved through polyploidy. Autotetraploidy generated varieties with bigger leaves and fruits, and autotriploidy generated self-sterile individuals. It is noteworthy that self-sterile *V. planifolia* varieties were also described in Mexico ('Oreja de Burro') (Castillo Martinez & Engleman, 1993; Soto Arenas & Dressler, 2010). It is most likely that these have resulted as well from autotriploidy. These results, as well as those that surveyed genome sizes in a wide range of *Vanilla* species (Bory et al., 2010) provide converging evidences for the importance of polyploidy and genome rearrangements during *Vanilla* evolution. Polyploidy can be of major importance in cultivation as well as in natural populations as triploidy and to a certain extent tetraploidy can be responsible for dramatic loss in fruit set. Further work is therefore needed to assess

Concordant data obtained on *V. planifolia* as well as *V. ×tahitensis* demonstrated an abnormal mitotic behaviour in the *Vanilla* genus, with a combination of somatic aneuploidy and

Most data in the literature give a basic number n=16 for *V. planifolia* with 2n= 32 (Chardard, 1963; Heim, 1954; Hoffmann, 1929, 1930; Martin, 1963). Hurel-Py (1938) was the first to show

partial endoreplication (Bory et al., 2008a; Lepers-Andrzejewski et al., 2011b).

vegetative propagation and the consecutive accumulation of point mutations.

polyploidization consequences on *Vanilla* reproductive biology.

**5.** *Vanilla* **genome dynamics** 

**5.1 Somatic aneuploidy** 

in the same bivalent) (Lepers-Andrzejewski et al., 2011a).

**4.3 Conclusions** 

hypomethylated at locus Eco-AG/Hpa-AAC/98bp and CR0341 at locus Eco-AC/Hpa-ATT/426bp. Accessions CR0632 and CR0711 were hypermethylated, they presented some bands in their *Msp*I profiles whereas the other accessions presented these bands in both their *Hpa*II and *Msp*I profiles. Accession CR0632 was hypermethylated at locus *Eco*-AG/*Hpa*-ATG/205bp and at locus *Eco*-AG/*Hpa*-AAC/382bp. Finally, accession CR0711 was hypermethylated at locus *Eco*-AG/*Hpa*-AAG/393bp.

These results showed that methylation is present in *V. planifolia* genome, with 7.45% of the fragments revealed being methylated. This value is in accordance with methylation rates reported in banana (7,5%, (Noyer et al., 2005)), but less than what is revealed in other conventionally propagated plant species such as rice (16.3%, (Xiong et al., 1999)), other bananas (18.4%, (Peraza-Echeverria et al., 2001)), apple (25%, (Xu et al., 2000)) and cotton (32%, (Keyte et al., 2006)).

A limited amount of methylation polymorphism (1%) was detected among 'Classique' and 'Mexique' accessions but the methylation patterns revealed were accession specific. Even for CR0340/0341/0342 which are three clones of the same accession, two different methylation polymorphisms were revealed in CR0340 and CR0341 and none in CR0342, showing that these methylation patterns are either not transmitted trough asexual propagation, or have appeared after clonal propagation. In all cases, no methylation marker was identified which could allow to specifically distinguish the 'Classique' and 'Mexique' morphotypes. A similar conclusion was obtained in studies performed on vegetatively propagated plants such as banana (Baurens et al., 2003; Noyer et al., 2005). Methylation polymorphisms were revealed but could not be correlated to morphological variations.

We therefore conclude that the 'Mexique' morphotype showing no detectable AFLP or MSAP polymorphism is most probably the result of a limited genetic or epigenetic dominant mutation event with pleiotropic effects.
