Anthers of flowers at anthesis were sampled from herbarium specimens or collected in the field and fixed in FAA 50 (Johansen 1940) for 25 genera and 26 species of Malpighiaceae (Table 1). Herbarium samples were submitted to a rehydration protocol, being boiled in distilled water for 15 minutes, treated with 2% potassium hydroxide for two hours at room temperature and washed in tap water, being posteriorly dehydrated in an alcoholic series and stored in ethanol 70% (modified from Smith and Smith 1942).

Fixed or rehydrated samples were embedded using standard methods for Technovit historesin and sectioned at 2 µm thickness (Arévalo-Rodrigues et al. 2020). Sections were stained with toluidine blue/p-phenylenediamine (1% aqueous/1% isopropanol: methanol, 1:1) for identification of metachromasy/phenolic compounds and lipid identification, respectively (Feder and O’Brien 1968; Kivimäenpää et al. 2004), and subsequently mounted in water slides for structural analyses under light microscopy.

Regarding herbarium samples, it is possible to perform histochemical analysis of rehydrated samples, except for identifying low-weight lipophilic molecules, including phenolic-based molecules, essential oils, and alkaloids. Connective glands were histochemically characterised in ten genera, representing nine of the ten informal phylogenetic clades currently accepted for Malpighiaceae (Table 1). Additionally, not all rehydrated materials resulted in suitable samples for anatomical/histochemical analyses, aside from the reduced number of flowers in this study. To detect the main classes of compounds in the glands’ exudate, we used copper acetate/rubeanic acid for fatty acids (Ganter and Jolles 1969), PAS reaction for total polysaccharides (McManus 1948), and ruthenium red or tannic acid and ferric chloride for mucilage (Gregory and Baas 1989; Pizzolato 1977). The digital images were acquired with an Olympus BX53 compound microscope equipped with an Olympus I-Colour 5 digital camera and Image Pro Express 6.3 software.

Fixed anther samples from all 25 genera and 26 species of Malpighiaceae sampled (Table 1) were fully dehydrated using 100% ethanol, rinsed in a hexamethyldisilane (HMDS) series (33.3, 50.0, and 66.6% v/v in 100% ethanol), and then rinsed three times in 100% HMDS for 1 min each to dry the material (Jeger et al. 2009). Chemically dried and herbarium samples were mounted on stubs, sputter-coated with gold using a Leica ACE200 system, and viewed using a JEOL JSM 741F scanning electron microscope at 10 kV. All SEM-analysed specimens were imaged and compared with the micromorphology of the genera analysed for anatomy and histochemistry under light microscopy.

Character coding followed the recommendations of Sereno (2007) for morphological analyses. Primary homology hypotheses (i.e., Ad doc hypothesis; De Pinna 1991; Ochoterena et al. 2019) were proposed for a total of 46 genera and 59 species of Malpighiaceae: 25 genera and 26 species sampled in this study, and 21 genera and 33 species of Malpighiaceae and nine families of Malpighiales sampled from the specialised literature (Tables 1, 2). Three micromorphological characters were scored, coded, and optimised (i.e. ancestral state reconstruction analysis) using the Maximum Likelihood criterium implemented on Mesquite 2.73 (Maddison and Maddison 2009) on a trimmed consensus tree from the molecular phylogeny of Malpighiaceae published by Davis and Anderson (2010; TreeBase accession 10998), evidencing a single tip per genus and the ten informal clades proposed by de Almeida and van den Berg (2021).

The glandular connectives observed in this study in all analysed species of Malpighiaceae were characterised as elaiophores, occurring as epidermal cells or papillae (i.e., trichomal elaiophores). The connective elaiophores described in this study were mainly formed by globose cells, with distinct anatomical features from those found in the sepals, as reported by Arévalo-Rodrigues et al. (2020; Fig. 6). Sepal elaiophores are composed of a palisade epidermis bearing a thick cuticle with vascularised parenchyma (Vogel 1974; Castro et al. 2001; Possobom et al. 2015; Araújo and Meira 2016; Possobom and Machado 2017; Aliscioni et al. 2022). Alternatively, connective elaiophores of Malpighiaceae comprise only one or two layers of globose epidermal cells or papillae (Possobom et al. 2015; Arévalo-Rodrigues et al. 2020; Avalos et al. 2020; Fig. 6). Nonetheless, trichomal elaiophores (i.e., papillae) in the connectives of Malpighiaceae have only recently been reported (Arévalo-Rodrigues et al. 2020). Despite epidermal and trichomal elaiophores sometimes co-occurring in the same plant organ (Pansarin et al. 2009), their occurrence in the same family is uncommon in flowering plants, being only reported for Malpighiaceae, Iridaceae, Orchidaceae, Plantaginaceae and Scrophulariaceae so far (Vogel 1974, 1984; Vogel and Machado 1991; Cosacov et al. 2009; Renner and Schaefer 2010).

Figure 6. 

Comparison between staminal and sepal elaiophores of a Malpighiaceae flower A transversion section of an anther of Stigmaphyllon blanchetii C.E.Anderson B longitudinal section of sepal elaiophore of Mcvaughia sergipana Amorim & R.F.Almeida. Flower of Heteropterys oberdanii Amorim is shown in a side view, evidencing the location of elaiophores in different organs (A modified from Arévalo-Rodrigues et al. 2020; B modified from Almeida et al. 2019; photograph of H. oberdanii by R.F.Almeida).

The heterogeneous connective elaiophore secretion produced in Malpighiaceae comprises a mixture of lipids, polysaccharides and phenolic compounds. The same exudate composition was found in the glandular connectives of Diplopterys pubipetala, Stigmaphyllon bonariense (Hook. & Arn.) C.E.Anderson, and S. jatrophifolium A.Juss. (Possobom et al. 2015; Avalos et al. 2020). A heterogeneous secretion mainly comprising non-volatile oils and polysaccharides is also produced by epidermal cells in the sepal elaiophores of Malpighiaceae (Vogel 1974; Lobreau-Callen 1989; Vinson et al. 1997; Castro et al. 2001; Possobom and Machado 2017; Aliscioni et al. 2022). This heterogeneous composition grants the exudate better fluidity, facilitating its collection by pollinators (Pansarin et al. 2009) being also observed in the nectar and stigmatic secretions of other flowering plant families, including Malpighiaceae (Endress 1994; Fahn 2000; Aliscioni et al. 2018). This fluid exudate of connective glands in Malpighiaceae resembles the fluid resin secreted by the connective glands of Calophyllaceae and the filaments of Clusiaceae and future in depth histochemistry studies should shed some light into their homology (Amaral et al. 2017; Cabral et al. 2021). Additionally, phenolic compounds show a central role as antioxidants and are secondarily astringent and toxic (Bruneton 1999; Simões et al. 2004), protecting floral resources from pathogens. Monoeca bees have been reported to collect pollen with the anthers’ exudate (Possobom et al. 2015). The connective elaiophores have the primary function of adhering pollen grains to the pollinator’s body, but they might also show a secondary chemical attraction function for these oil-collecting bees (Possobom and Machado 2017; Avalos et al. 2020).

The character-mapping analysis recovered unicellular globose cells (i.e., epidermal elaiophores) producing mainly non-volatile oil in the connectives of Malpighiaceae as a new synapomorphy for this family. This result was only possible due to our comprehensive analysis sampling 46 genera (out of 75) from nine of the ten informal phylogenetic clades currently recognised in Malpighiaceae (see Almeida and van den Berg 2021; Almeida et al. 2023). In fact, unicellular 2-branched hairs, conspicuous sepal elaiophores and clawed petals are already recognised as morphological synapomorphies for Malpighiaceae (APG IV 2016; Stevens 2001, onwards), with connective elaiophores proposed here as a fourth morphological synapomorphy for this family. Nonetheless, additional anatomic, histochemical and SEM studies sampling all the remaining genera of Malpighiaceae (see Table 2 for additional information) are still needed to better explore the evolutionary patterns of connective elaiophores structure and exudate composition in this family.

Based on our sampling, the character-mapping analysis recovered glandular connectives as a possible synapomorphy for Malpighiales, one of Rosids’ major extant orders, with 42 families and ~16,000 species with a mostly pantropical distribution, exhibiting remarkable morphological and ecological diversity (Xi et al. 2012). However, further studies focusing on characterising glandular tissue in the stamens of most families comprising Malpighiales are still needed. In fact, we were only able to sample nine out of the 42 families currently accepted in this order (see Table 1). Although strongly monophyletic, relationships between Malpighiales families remain elusive, and no consensus has been reached even in the phylogenomic era (Davis et al. 2005; Wurdack and Davis 2009; Xi et al. 2012; Cai et al. 2019). Even though unambiguous morphological synapomorphies have not been proposed yet for Malpighiales, the order is generally characterised by a combination of homoplastic characters (i.e., paracytic stomata, leaf-blades with toothed margins, diverse extrafloral glands, ovules with thin, slender nucelli and endothelium, and dry stigmas; Stevens 2001, onwards), possibly including staminal glands.

The character-mapping analysis also recovered different types of staminal gland exudate as possible synapomorphies for some major clades of Malpighiales. This order currently comprises four major clades (i.e., Clusioids, Euphorbioids, Malpighioids, and Salicoids) recognised by different studies (Xi et al. 2012; Cai et al. 2019). However, no synapomorphies have ever been proposed for these four major clades of Malpighiales (Xi et al. 2012; Stevens 2001, onwards). The most recent common ancestor for the Clusioids had resin connective glands recovered in our analysis as a possible synapomorphy for this clade. Several types of exudates are secreted by the connective glands of Clusioids, such as wax and resins in Calophyllaceae (Crockett 2010), Clusiaceae (Crockett 2010; Amaral et al. 2017), and Hypericaceae (Crockett 2010). Currently, the Clusioids only show basifixed anthers as a homoplastic morphological character uniting its members (Silva-Batista et al. 2021).

The Euphorbioids had eglandular stamens recovered as a possible homoplasy shared with Elatinaceae and Podostemaceae in our analysis probably representing a reversal due to the specialised aquatic life form. The Euphorbioids are currently circumscribed by homoplastic characters such as plants often monoecious, flowers small, often imperfect, and 3-merous or not, ovules 1–2/carpel, inner integument usually thicker than outer, epitropous, fruit a part-septicidal + loculicidal capsule/schizocarp, cotyledons longer and broader than radicle (Stevens 2001, onwards). The Euphorbioids comprise five additional families (i.e., Euphorbiaceae, Ixonanthaceae, Linaceae, Peraceae, and Rafflesiaceae) that had never had their stamens anatomically studied, focusing on glandular tissues. In fact, most of these families are pantropically distributed or are difficult to analyse due to their rarity (i.e., Rafflesiaceae; POWO 2023; Stevens 2001, onwards). Further studies are urgently needed in these Euphorbioid families to corroborate the absence of staminal glands in this major clade of Malpighiales.

In our analysis, the Malpighioids (Malpighiales) had oil glandular stamens recovered as a possible synapomorphy. This major clade of Malpighiales had only three (out of 11) of its families anatomically explored regarding staminal glands. Malpighiaceae has already been the subject of staminal glands anatomical studies by several authors (Possobom et al. 2015; Arévalo-Rodrigues et al. 2020; Avalos et al. 2020; this study). On the other hand, Elatinaceae comprises only two genera, Bergia and Elatine, which were recently sampled in a comprehensive anatomical study by Bonifácio et al. (2023) and seemed to present, indeed, eglandular stamens in most of its species, except for Elatine lindbergii Rohrb. These authors performed histochemical tests for E. lindbergii indicating the presence of phenolic compounds, but they failed to test the exudate of these glands for lipids. This family is also an aquatic family, just like Podostemaceae, even though not that morphologically specialised and having cleistogamous (sometimes flowering underwater) and/or apomictic flowers. Caryocaraceae have also been anatomically studied by several authors regarding their staminal glands (Dickison 1990; Matthews and Endress 2011; Sousa-Paiva et al. 2019). In fact, the Malpighioids (Malpighiales) show few circumscribing homoplasies, such as gynoecium with longitudinal bulges above the placentae, outer integument 3–7 cells across, and inner integument 5–10 cells across (Stevens 2001, onwards). Nonetheless, further studies are still needed on the eight remaining families of the Malpighioids to test the relevance of staminal oil glands as possible synapomorphies or homoplasies circumscribing this major clade of Malpighiales.

Finally, the Salicoids had nectar connective glands recovered as a synapomorphy in our analysis. Nectar connective glands have recently been suggested as a new morphological synapomorphy for Humiriaceae (Wurdack and Zartman 2019). However, this exudate is also recorded in the filament or connective glands of Violaceae within the Salicoids (Feng 2005; Wurdack and Zartman 2019). In fact, only persistent endosperm has ever been proposed as a morphological homoplasy to circumscribe the Salicoids, making the nectar connective glands reported in this study a prominent character to further explore in this group (Stevens 2001, onwards). Classical morphological studies had long suggested that there was a group that included Salicaceae, Achariaceae, Violaceae, Flacourtiaceae, and Passifloraceae and its segregates, Malesherbiaceae and Turneraceae, in part because of their common possession of parietal placentation, some sort of corona or scales in the flower, nectaries outside the stamens, etc. (e.g. Cronquist 1981). However, additional studies in members of Achariaceae, Goupiaceae, Passifloraceae s.lat., and Salicaceae s.lat. are still needed to better explore the presence, structure, and nature of connective glands in these Salicoid families.

The authors have declared that no competing interests exist.

No ethical statement was reported.

We thank Smithsonian Institution for RFA’s Cuatrecasas Award, and Programa de Desenvolvimento Científico e Tecnológico Regional CNPq/FAPEG (grants #317720/2021-0 and #202110267000867) for RFA’s postdoctoral fellowship.

Conceptualization: RFA. Data curation: PCG, GAR. Formal analysis: PCG, RFA. Funding acquisition: PCG, ILM. Investigation: GAR. Methodology: GAR, RFA, PCG. Project administration: PCG. Resources: PCG, ILM. Supervision: PCG. Writing - original draft: RFA. Writing - review and editing: GAR, ILM, PCG.

Rafael Felipe de Almeida https://orcid.org/0000-0002-9562-9287

Gustavo Arévalo-Rodrigues https://orcid.org/0000-0002-1860-1707

Isa L. de Morais https://orcid.org/0000-0001-8748-9723

Poliana Cardoso-Gustavson https://orcid.org/0000-0003-0978-9389

All of the data that support the findings of this study are available in the main text.