Zornia J.F.Gmel. is a genus within the Adesmia clade (tribe Dalbergieae) of the legume family (Leguminosae; Lavin et al. 2001). It currently comprises approximately 90 species, primarily distributed across pantropical regions (POWO 2025). The genus is characterised by its herbaceous or shrubby habit, flowers arranged in spiciform inflorescences, paired peltate bracteoles protecting each flower, and stipules resembling bracteoles (Fortuna-Perez et al. 2013; Zeferino et al. 2025). Zornia was validly published by Gmelin in 1791, based on Zornia bracteata J.F.Gmel., and has traditionally been divided into two subgenera based on inflorescence type and three sections according to leaflet morphology (Mohlenbrock 1961).

The genus was recently resolved as monophyletic and closely related to Poiretia Vent. and Amicia Kunth, forming the Zornia-Amicia-Poiretia clade (ZAP clade). This clade is characterised by secretory cavities in leaflets, with anatomical and morphological functional traits adapted to colonising dry environments (Fortuna-Perez et al. 2021). However, molecular studies do not support the monophyly of the subgenera and sections of Zornia (Lavin et al. 2001; Fortuna-Perez et al. 2013, 2021). Fortuna-Perez et al. (2013) instead recognised two major clades and one subclade within Zornia. Clade A includes most species with 4-foliolate leaves, 1–4 articles per loment, and non-bristled articles, predominantly occurring in north-eastern Brazil in dry forests and savannas. Clade B comprises species with bifoliolate leaves, up to 15 articles per loment, and bristled articles, distributed across both the New and Old Worlds. Within Clade B, Subclade B1 is restricted to Africa, Asia, and Australia, although it lacks support from any morphological synapomorphy.

Previous studies describe Zornia pollen grains as having a distinctive morphology within the Adesmia clade, characterised by a markedly prolate shape and 3-colpate (rarely 3-colporate) apertures (Melhem 1966; Salgado-Labouriau 1973; Pire 1974; Carreira et al. 1996; Silvestre-Capelato and Melhem 1997; Moreti et al. 2007; Buril et al. 2011; Luz et al. 2013; Silva et al. 2016). However, there are inconsistencies among these studies regarding aperture type, specifically, colpate versus colporate apertures in some species. Over the last three years, the palynology of the Adesmia clade has been extensively studied, with only Zornia remaining to be described in a comprehensive and standardised way to support a pollen evolutionary study (Antonio-Domingues et al. 2022a, 2022b, 2023, 2024). We analysed the pollen morphology, ultrasculpture, and ultrastructure of Zornia species to (1) complete the palynotaxonomic study on the Adesmia clade, (2) confirm the aperture morphology, and (3) test if pollen morphology recovers any apomorphy for Clades A and B.

Unopened mature floral buds were obtained from 18 herbarium specimens belonging to 15 species of Zornia (Z. adenophora (Domin) Mohlenbr., Z. areolata Mohlenbr., Z. bracteata J.F.Gmel., Z. brasiliensis Vogel, Z. capensis Pers., Z. contorta Mohlenbr., Z. crinita (Mohlenbr.) Vanni, Z. gardneriana Moric., Z. glochidiata Rchb. ex DC., Z. harmsiana Standl., Z. myriadena Benth., Z. pardina Mohlenbr., Z. setosa Baker f., Z. thymifolia Kunth, and Z. virgata Moric.). Specimens were sampled from the K, MBM, SP, and UEC herbaria (acronyms follow Thiers 2025). We follow the guidelines for herbarium sampling recently proposed by Almeida et al. (2023), with all specimens examined being listed in the Suppl. material 7. Samples were labelled with a three-letter abbreviation of the species epithet plus the last two digits of the herbarium barcode or of the collection number (see Suppl. materials 1, 7). The selected Zornia species represent the full extent of the genus’s morphological and geographical diversity, including taxa from both the New and Old Worlds, as well as representatives from each of the clades sampled in the phylogeny of the genus proposed by Fortuna-Perez et al. (2013).

Pollen grains were processed for acetolysis analysis (Erdtman 1960), following the modifications proposed by Melhem et al. (2003). Measurements were performed under light microscopy (LM) using an Olympus OSM-4 (10×) micrometre drum coupled to the eyepiece of an Olympus BX50 binocular microscope. The polar and equatorial axes in equatorial views were measured on 25 randomly selected pollen grains from each taxon. These measurements were statistically analysed for the arithmetic mean (x), average standard deviation (Sx), sample standard deviation (s), coefficient of variability V%, and 95% confidence interval (CI) (Zar 2010; Vieira 2016). Ten measurements of the prominent pollen morphometric parameters were also made: length and width of the colpus and the thickness of the nexine and sexine layers. Exine measurements were performed in the mesocolpium region. The arithmetic means and the range were calculated for each pollen grain size class and exine (sexine + nexine). Photomicrographs were obtained using an Olympus BX-50 light microscope equipped with an Olympus U-CMAD-2 video camera and Olympus CellSens 1.5 software. Permanent LM slides were deposited in the pollen reference collection of Laboratório de Palinologia PALINO-IPA, Instituto de Pesquisas Ambientais, São Paulo State, Brazil.

For scanning electron micrographs (SEM), non-acetolysed pollen grains were rinsed in an ethanol series (50–100%), placed on a metal stub, and sputter-coated (Leica EM ACE 600, Vienna, Austria) with gold (80 nm). Samples were observed under a JEOL JSMIT300LV microscope (Tokyo, Japan) operating at a 20 kV electron beam, and the images were digitised.

For transmission electron microscopy (TEM), mature anthers were removed from flower buds of five species (Zornia brasiliensis, Z. bracteata, Z. contorta, Z. crinita, Z. myriadena) and fixed in a modified Karnovsky solution (2% glutaraldehyde, 2% paraformaldehyde, 5 mM calcium chloride in 0.05 M sodium cacodylate buffer, pH 7.2, Karnovsky 1965) for at least 24 h, under refrigeration. They were rinsed with sodium cacodylate buffer (0.1 M) and immediately post-fixed in osmium tetroxide (1% in 0.1 M sodium cacodylate) for 1 h at room temperature. Samples were then rinsed in a saline solution (0.9% sodium chloride) and pre-stained with uranyl acetate (2.5%), followed by dehydration in an acetone series (35–100%) for 20 min each. Samples were then gradually infiltrated with Spurr resin (Spurr 1969) and embedded in pure resin for 48 h at 70°C. Ultrathin sections (60 nm) were obtained with an ultramicrotome (Sorvallter Blum MT2, Norwalk, CT, USA), deposited on copper grids and poststained in aqueous uranyl acetate (2.5%), followed by lead citrate (0.1%) (Reynolds 1963). Images were obtained using a transmission electron microscope (JEOL JEM-1400, Tokyo, Japan) at an acceleration voltage of 80 kV.

Pollen terminology was based on Punt et al. (2007) and Halbritter et al. (2018). Pollen shape classes and amb types follow Erdtman (1952), and colpus length categories follow Mark et al. (2012). The endoaperture classes and index of sexine and nexine thickness were proposed by Antonio-Domingues et al. (2022a). TEM descriptions of the exine infratectum followed the terminology proposed by Halbritter et al. (2018) and Ferguson and Skvarla (1983).

Principal component analysis (PCA) was performed to investigate whether pollen grain characteristics enabled the grouping of genera and species. Seven metric variables (polar axis [PA], equatorial axis [EA], colpus length [CL], colpus width [CW], nexine thickness [N], sexine thickness [S = infratectum + tectum], exine thickness [Ex = N + S]) and two classes/indices (shape class [P/E = PA/EA], index of sexine and nexine thickness [S/N]) were analysed using FITOPAC (Shepherd 1996) and PC-ORD v.7 (McCune and Mefford 2016).

Sequences of the Internal Transcribed Spacer (ITS, including ITS1 and ITS2 spacer regions and the 5.8S ribosomal subunit) were downloaded from GenBank for all ingroup and outgroup species sampled by Fortuna-Perez et al. (2013). A maximum likelihood analysis was run using Raxml-GUI (Edler et al. 2021). Character coding followed the recommendations of Sereno (2007) for morphological analyses. Primary homology hypotheses (De Pinna 1991) were proposed for pollen shape, size, ornamentation, exine structure, and apertures. A total of 13 pollen characters were scored for Zornia and outgroups (Suppl. material 5). All characters were optimised on the concatenated tree using the Maximum Likelihood function (mk1 model) using Mesquite v.2.73 (Maddison and Maddison 2006) and visualised with Winclada (Nixon 1999).

According to palynological standards of pollen morphology variation, Zornia can be categorised as stenopalynous, as all species exhibit the same pollen type, with some subtle differences between pollen grains, such as ornamentation, shape, size, and thickness of the exine. However, the presence of colpate apertures is a unifying pollen character of the genus Zornia. In addition, the thickness of the pollen nexine (> 0.5 μm) is a synapomorphy for Zornia, while Zornia Clade A was supported by two homoplasies correlated with ultrasculpture of the apocolpium (psilate-perforate) and sexine thickness (< 0.5 μm) and Zornia Clade B by the exoaperture width (5–10 μm). Finally, the stenopalynous nature of Zornia pollen grains is corroborated here by the large number of homoplasies recovered.