According to the more recent phylogeny inferred from DNA sequences, Cervantesiaceae Nickrent & Der includes 8 genera that formerly belonged to Santalaceae R. Br. s.l. (Der & Nickrent, 2008; Nickrent & al., 2010). However, the classical concept of Santalales R. Br. ex Bercht. & J. Presl and its families based on morphology (v.gr., Cronquist, 1981; Takhtajan, 1997) are followed by other researchers (v.gr., Kuijt & Hansen, 2015). Cervantesiaceae is formed by hemiparasitic trees or shrubs with thorns, bisexual or unisexual flowers, solitary or grouped in inflorescences, 4-5-merous campanulate or urceolate corollas, hairs on the petals that are opposite to and in equal number with the stamens, and a nectary disk often with lobes alternating between the petals (Nickrent & al., 2010).

Currently this family is formed by genera Acanthosyris (Eichler) Griseb., Cervantesia Ruiz & Pav., Jodina Hook. & Arn. ex Meisn., Okoubaka Pellegrin & Normand, Pilgerina Z.S. Rogers & al., Pyrularia Michx., Scleropyrum Arn., and Staufferia Z.S. Rogers & al., all them mainly from template and Andean South America, tropical Africa and Madagascar, template and tropical Asia reaching Malesia, and one species from the southeastern North America. It was claimed that the intergeneric relationship within this family was totally resolved, and it was composed of two monophyletic groups. One of them, the Cervantesia group, includes all South American genera, Acanthosyris, Cervantesia, and Jodina. These genera show shared features such as an unilocular ovary with a twisted placental column, although the position of the ovary varies, being half-inferior —Acanthosyris—, superior —Cervantesia—, or inferior —Jodina— (Nickrent & al., 2010).

Among them, Jodina is the southernmost genus, endemic to central and southeastern South America, where it grows from Bolivia, extreme southern Brazil, Uruguay, and Argentina, reaching there its southernmost distribution range, approximately in the north of Colorado River in Buenos Aires and La Pampa Provinces (Luna & De la Sota, 2003; Dettke & Caires, 2016). This is a monotypic genus with the species J. rhombifolia (Hook. & Arn.) Reiss., called “sombra de toro” or “peje”. Jodina rhombifolia is a tree up to 8 m tall, with rhombic and glabrous leaves, and 3 spines on the free ends (Luna & de la Sota, 2003; Kuijt & Hansen, 2015). One of its more interesting characteristics is to be a hemiparasite plant, having been encountered haustorial connections with roots of Celtis tala Gillies ex Planch. and Scutia buxifolia Reissek (Luna & Giudice, 2005). In relation to the reproductive features, Jodina has bisexual flowers, a nectary disk that alternates with the petals, and a fleshy fruit with a single seed (Kuijt & Hansen, 2015).

The fruits of Cervantesiaceae and Santalaceae s.l. were mentioned in general terms as drupaceous with a stony pit (Der & Nickrent, 2008; Rogers & al., 2008) or as true drupes (Nickrent & al., 2010; Kuijt & Hansen, 2015). With regard to Jodina, different types of fruit have been described. Following Dawson (1944) and Rodríguez Mattos (1967), the fruit of J. rhombifolia is a drupaceous capsule because the “tepals” are deciduous at maturity. Then, Bhatnagar & Sabharwal (1969) defined this fruit as a pseudodrupe, since the stony layer of the fruit wall is the mesocarp. According to Kuijt & Hansen (2015), Jodina has a drupe-like fruit with a stony endocarp, and a fleshy rugose exocarp that detaches into 5 parts at maturity. The drupe has been broadly defined as an indehiscent, and singled seeded fruit with a fleshy-stony pericarp, being the stony portion constituted by the endocarp (Roth, 1977, 1987). Transitional types may be found, giving rise to concepts such as pseudodrupe or drupaceous fruit for those with fleshy exocarp and undifferentiated endocarp, or with stony mesocarp (Spujt, 1994). On the development and anatomy of J. rhombifolia seed, the only published works were those of Bhatnagar & Sabharwal (1966, 1969), which emphasized on endospermogenesis and the early stages of embryo development.

Because the type of fruit is an important feature used in Systematics, and the available information on the development and structure of the fruit of Jodina is scarce and controversial, the main objective of this work was to study its fruit structure and seed morphology at different developmental stages.

Flowers, fruits, and seeds of J. rhombifolia were collected at Partido de Magdalena —35º 11’ S, 57º 17’ W, northeastern Buenos Aires Province, Argentina— from May to December during the years 2011-2013. Thus, different stages of fruit and seed development were covered, from immature fruits with greenish persistent petals enclosing the pericarp and developing seed —about 3-4 months after anthesis—, to more mature stages of fruit and seed formation, with the reddish persistent petals or when they were detached exposing the white fleshy covering —5-6 months after anthesis—. Ripe seeds, considered at this stage once the fleshy covering has disorganized —near 7 months after anthesis—, were also collected. The number of trees sampled during the three years was around 60. The voucher specimens were deposited at herbarium LP —ARGENTINA. Buenos Aires, Pdo Magdalena, Estancia San Isidro, 23-V-2011, M.L. Luna 114 & al.; ibídem, 7-IX-2011, M.L. Luna 135 & al.; ibídem, 16-VII-2012, M.L. Luna 142 & al.; ibídem, 21-X-2012, M.L. Luna 150 & al.; ibídem, 15-XII-2012, M.L. Luna 154 & al.; ibídem, 5-VI-2013, M.L. Luna 161 & al.; ibídem, 14-VIII-2013, M.L. Luna 167 & al.

For light microscopy analyses —LM—, flowers, fruits, and seeds at different stages of development were fixed in formaldehyde-acetic acid-alcohol, dehydrated through an ethanol series and embedded in Paraplast. Sections —8-10 μm thick— were double stained with safranin-fast green (Johansen, 1940).

Other staining procedures were used, such as iodine-potassium iodide for amyloids (Kooiman, 1960), periodic acid-Schiff —PAS— for carbohydrates (O’Brien & Mc Cully, 1981), and the following fluorochromes: Nile blue for neutral lipids and 8-anilino-1- naphthalene-sulphonic acid for proteins —for both: BV filter, main length wave 463 µm— (Fulcher, 1982). Observations were made in a Nikon Microphot- FX epifluorescence microscope.

For scanning electron microscopy —SEM—, flowers were fixed in formalin: propionic acid: ethyl alcohol —0.5:0.5:9—, dehydrated in ethanol-acetone series and critical point dried using CO₂ (Sancho & Otegui, 2000). The seeds were analyzed under SEM without pre-treatment. In both cases, materials were mounted on double-sided tape on aluminum stubs, sputter coating with gold-palladium and examined with a Jeol JSM-T100 microscope.

Crude protein in endosperm was determined preliminary by the Macro-Kjeldahl method whereas fats were calculated by continuous extraction with a Soxhlet extractor, using benzene as solvent (Bullock & Moore, 1992).

The previous descriptions reported in the literature for the fruit type of J. rhombifolia are summarized in Table 1.

The fruit of Jodina is a pseudodrupe, since the fruit wall is constituted by the persistent reddish petals, along with the modified nectary disk —“white fleshy layer”— plus the sclerified mesocarp. Actually, this “white fleshy layer” of the fruit, which corresponds to the modified nectary disk, has not been mentioned previously for this genus. The development of the nectary disk after fertilization and its persistence in the mature fruit has been cited only for Quinchamalium Molina (Johri & Agarwal, 1965), a genus of Schoepfiaceae Blume (sensu Nickrent & al., 2010), a basal family of the Santalales. However the authors have not considered the modified nectary disk when described the fruit wall of Quinchamalium chilense Lam. Following Spujt (1994), pseudodrupes also include fruits with “accessory structures” —like the nectary disk—, as part of the fruit wall.

Description of the fruit at different stages of development leads to its classification in different types. In this sense, the fruit of Jodina has been described in a less mature stage when petals have not yet fallen off, or in a more mature stage after their detachment (Table 1). Bhatnagar & Sabharwal (1966) also classified the fruit of Jodina as a pseudodrupe, but in this case considering the persistent perianth as the fleshy layer. Furthermore, Heide-Jørgensen (2008) described as 5 carpels the modified and persistent reddish petals of J. rhombifolia —as already mentioned, the gynoecium is tricarpellary.

Taking into account the Cervantesiaceae, Der & Nickrent (2008) and Nickrent & al. (2010) mentioned that in Jodina the “outer exocarp” dehisces leaving exposed a white fleshy inner “exocarp”, whereas in other genera, such as Cervantesia and Staufferia, the 5 perianth valves remain on the mature fruit. As mentioned above, the white inner “exocarp” corresponds to the modified nectary disk. Detailed studies about fruit wall development are not recorded in the other related genera —Cervantesia and Staufferia—. In the case of Acanthosyris, Nee (1996) stated that the fleshy layer corresponded to the mesocarp, but without conducting developmental studies. Given what was found in the present work a re-analysis of fruit development in the other genera of Cervantesiaceae will provide additional information for their systematic relationships.

With respect to the development of the seed, and specifically for endospermogenesis, we registered different developmental stages of the cellular and haustorial portions of the endosperm, as described previously (Bhatnagar & Sabharwal, 1966, 1969). Since the ovule integument was consumed by the haustorial portion of the endosperm, the resulting seed became naked and the resulting structure should be called pyrene —i.e., mesocarp plus the enclosed seed—. As for reserves contained in the endosperm, we detected protein bodies and starch grains, besides lipids previously described (Bhatnagar & Sabharwal, 1969). Protein content of the seeds is similar to that found in Santalum acuminatum (R. Br.) A. DC. —Santalaceae s. str., Lott & Buttrose (1978)—. Chemical studies within Cervantesiaceae would contribute to clarify more aspects of endosperm development and the seed reserves present in the genera of this family.

As a corollary, information about the fruit type —fle-shy or dry— may help to understand evolutionary and ecological aspects in angiosperms. Since the kind of fruit has been associated with different habitats (Lorts & al., 2008), the misinterpretations of fruit types may lead to erroneous ecological or evolutionary associations. As the fruits of the Santalaceae s.l. have been interpreted as dry or fleshy by the different authors, studies like the one here presented will be fundamental to clarify such assumptions.