Introduction
Research on African diatoms has a long tradition, with the first works dating back to the 19th century (i.e. Ehrenberg 1845; Cleve 1881). The dynamic development of diatomological research took place mainly in the second half of the 20th century. Research from diverse habitats was carried out by four scientists: B.J. Cholnoky, M.H. Giffen, F.R Schoeman and R.E.M. Archibald, whose efforts are the foundation of more recent studies (i.e. Cocquyt et al. 2017; Cocquyt and Taylor 2019; Ridder and Taylor 2020). The general diatom diversity of inland waters from southern Africa was studied by Cholnoky (e.g. Cholnoky 1960, 1962, 1966), who also described many new species (Cholnoky 1954, 1955, 1956, 1957, 1966). On the other hand, Giffen, in his works from the mid-1960s, dealt mainly with marine and estuarine diatoms (i.e. Giffen 1963, 1967, 1970, 1973, 1976), as well as some freshwater species from the Eastern Cape Region (Giffen 1966). Undoubtedly, the greatest contribution to understanding both the taxonomy and ecology of diatoms was made by Schoeman and Archibald, who, for over 15 years, published together over 20 works on diatom flora in southern Africa, supplemented by hand-drawings (i.e. Schoeman 1969, 1970; Archibald 1966) and, in subsequent years, also with microphotographs of the observed species (i.e. Schoeman and Archibald 1986a, 1986b, 1986c, 1986d; Archibald and Schoeman 1984, 1985). More recent researchers continued to work on taxonomy (i.e. Cocquyt 2006; Cocquyt et al. 2013, 2014, 2016; Taylor et al. 2014a, b), but also concentrated on ecological monitoring (i.e. Bellinger et al. 2006; Taylor et al. 2007b, c). There are two elaborate reports of the most common diatom species from the Congo and Zambezi Basins (Taylor and Cocquyt 2016) and South Africa (Taylor et al. 2007a). In recent years, many new species were described from different parts of Africa. Various studies confirm the presence of a unique diatom diversity, composed of rare and endemic diatoms species (Cocquyt et al. 2013, 2014, 2017; Cocquyt and Ryken 2017; Taylor et al. 2014b).
Despite a long history of diatom research in southern Africa, the knowledge about terrestrial diatom assemblages is rather limited (Taylor et al. 2010, 2014a; Van de Vijver et al. 2010). Although most of the studies focused on marine, freshwater and brackish diatoms, there are still some unexplored aerial habitats. A study of terrestrial moss-inhabiting diatom communities, conducted by Van de Vijver et al. (2010), resulted in the description of one species, Muelleria taylorii Van de Vijver & Cocquyt from Drakensburg, Freestate Province, South Africa.
Two genera are mainly considered to occur in terrestrial habitats, like mosses, rocks or soils, the genus Luticola and genus Microcostatus (Round et al. 1990; Johansen and Sray 1998). The genus Luticola D.G. Mann (Round et al. 1990) was distinguished from Navicula to accommodate species included in Naviculae sect. Punctatae with Luticola mutica (Kütz.) D.G. Mann as generitype. So far, over 200 species have been documented worldwide (Guiry and Guiry 2021). A monograph, summarising the entire genus (Levkov et al. 2013), presented a detailed and extensive revision of the Luticola genus, including more than 20 species observed from Africa. Still, more recent studies show that the diversity of the genus Luticola is underestimated and many new species have been described from Europe (Levkov et al. 2017), Asia (Liu et al. 2017; Glushchenko et al. 2017; Lokhande et al. 2020), South America (Bąk et al. 2017; Straube et al. 2017; Simonato et al. 2020; Peszek et al. 2021), Madagascar (Bąk et al. 2019) and Antarctic Region (Zidarova et al. 2014; Kohler et al. 2015; Chattová et al. 2017; Kochman-Kędziora et al. 2020). The genus Luticola is highly diverse in valve morphology. Species of this genus are also widespread in brackish, freshwater and terrestrial ecosystems. However, the genus is ecologically characterised as being aerophilous, often noted from sites in splash zones, soil and amongst mosses (Round et al. 1990; Kociolek et al. 2017). It also shows a high degree of endemism (Kociolek et al. 2017; Lokhande et al. 2020).
The genus Microcostatus, described by Johansen and Sray (1998), encompasses small naviculoid species with two longitudinal depressions next to the central sternum, a simple raphe system and external microcostae. Until now, this small genus has only 24 species documented worldwide (Guiry and Guiry 2021). Taylor et al. (2010) listed 23 taxa, including three described from South Africa. More recently, papers described additional six species: Microcostatus edaphicus Stanek-Tarkowska, Noga, Wetzel & Ector and M. aerophilus Stanek-Tarkowska, Noga, Wetzel & Ector from Europe (Stanek-Tarkowska et al. 2016), M. salinus Li & Suzuki (Li et al. 2016) and M. werumii Metzeltin, Lange-Bertalot & Soninkhishig (Metzeltin et al. 2009) from Asia and two other species: M. australoshetlandicus Van de Vijver, Kopalová, Zidarova & E.J. Cox and M. elisabethianus Van de Vijver & Ector from Antarctic and Sub-Antarctic Region (Van de Vijver et al. 2013; Van de Vijver and Ector 2019).
The present paper aims to contribute information on the distribution and environmental preferences of terrestrial diatoms in South Africa. This is the first paper providing information about moss-inhabiting diatom assemblages from Western Cape Province, South Africa. Three taxa from the genus Luticola and one from Microcostatus, which cannot be identified using currently available literature, were observed during the study. Additional analysis, based on detailed light and scanning electron microscopy, showed a set of unique features allowing us to describe them as new species. Additional comparison with the most similar taxa is also provided.
Discussion
The vast majority of research on diatoms in Africa was conducted in the second half of the twentieth century and many species descriptions are based on sketches and the classification of species is based on broad approaches to genera (Kociolek and Williams 2015). Many of them still wait for their verification using modern-day taxonomy focusing on detailed microscopic analysis.
Originally, many species of the genus Luticola were described as Navicula of section Punctatea. The genus verification made by Levkov et al. (2013) transferred most of them to Luticola, However, some taxa are still waiting for verification and potential transfer. Probably, based on sketches, the following species potentially are belonging to genus Luticola: Navicula lagerheimii f. rotundataCholnoky (1954, p. 218, pl. 3, fig. 81), Navicula inserata var. ellipticaCholnoky (1960, p. 65, pl. 5, figs. 204–205), Navicula submuticaFusey (1964, p. 27, pl. 3, fig. 38), N. submutica var. capitataFusey (1964, p. 27, pl. 3, fig. 41), N. submutica var. ellipticaFusey (1964, p. 27, pl. 3, fig. 40), N. submutica var. rectangularisFusey (1964, p. 27, pl. 3, fig. 39), Navicula guluensisGiffen (1966, p. 238, fig. 70). However, their transfer require detailed study of type materials.
The three new Luticola species have comparable valve outline with protracted apices and striae density. They also share the presence of several shallow depressions in the central area visible in SEM. Despite their unique set of features, the taxa show some similarities to other species of this genus, especially under the light microscope. Luticola microcephala sp. nov. shows the highest degree of similarity to two aerophytic taxa described from India: Luticola jogensis (H.P. Gandhi) Kale, Levkov & Karthick (Kale et al. 2017, p. 30, figs 2–26) and L. gandhii Kale, Levkov & Karthick (Kale et al. 2017, p. 33, figs 27–51). However, Luticola microcephala sp. nov. can be easily separated from both Indian taxa. Luticola jogensis has much larger valves (28.5–36.0 µm length and 7.0–9.0 µm width versus 14.0–24.0 µm length and 4.5–6.6 µm width in L. microcephala sp. nov.). Additionally, L. jogensis has 5–6 areolae per striae, whereas striae in L. microcephala sp. nov. are composed of 3 to 4 areolae. The second of the Indian species L. gandhii differs from L. microcephala sp. nov. in having rostrate, not capitate apices. The most pronounced difference amongst all three species is the number of rows of areolae on the valve mantle. Luticola microcephala sp. nov. has only one row of elongated areolae, becoming more rounded in the central part of the valve and near the apices. On the contrary, L. jogensis has 2 or 3 rows (Kale et al. 2017, p. 34, fig. 22) and L. gandhii has always three rows of areolae (Kale et al. 2017, p. 37, fig. 48).
Amongst other species described from South Africa, Navicula submutica var. capitata also shows some degree of similarity to L. microcephala sp. nov., but there is a lack of detailed microscopic pictures of this species. However, based on the description and the line drawing of this taxon (Fusey 1964, p. 27, pl. 3, fig. 41), Navicula submutica var. capitata differs from L. microcephala sp. nov. in having more lanceolate valve outline, wider central area and lower striae density (16 in 10 µm versus 19–22 in L. microcephala sp. nov.).
Luticola asymmetrica sp. nov. shows morphological similarity to five Luticola species reported from the African continent: L. imbricatiformis Levkov, Metzeltin & Pavlov (2013, p. 134, pl. 28, figs 1–11), L. falknerorum Metzeltin & Lange-Bertalot (Metzeltin and Lange-Bertalot 2007, p. 156, figs 1–9), L. fuhrmannii Metzeltin & Levkov in Levkov et al. (2013, p. 116, pl. 23, figs 1–20), L. gesierichiae Levkov, Metzeltin & Pavlov in Levkov et al. (2013, p.119, pl. 48, figs 17–28 and pl. 49, figs 1–5) and L. frickei Levkov, Metzeltin & Pavlov in Levkov et al. (2013, p. 114, pl. 48, figs 29–42, pl. 49, figs 6–8 and pl. 52, fig. 1). However, Luticola asymmetrica sp. nov. can be distinguished, based on its asymmetrical, weakly dorsiventral valve margin. Additionally, Luticola imbricatiformis has slightly elongated stigma positioned close to the centre of the valve (Levkov et al. 2013, p. 337, fig. 11), whereas L. asymmetrica sp. nov. has a round isolated pore located in the central area, halfway between the valve margin and central raphe. Luticola falknerorum has a higher striae density (21–24 in 10 μm in comparison to 17–20 in 10 µm in Luticola asymmetrica sp. nov.) and more areolae per striae (5–6 versus 3–5 in Luticola asymmetrica sp. nov.). Both species can also be separated, based on external raphe structure. Raphe branches of L. falknerorum are bordered by silica ridges and have short proximal raphe endings (Levkov et al. 2013, p. 341 figs 2, 3), whereas L. asymmetrica sp. nov. does not have silica ridges and has asymmetrical, long proximal raphe endings, almost reaching the first row of areolae. Similar silica ridges are also present in L. gesierichiae (Levkov et al. 2013, p. 379, figs 2, 3). The characteristic asymmetrical valve outline with broadly-rounded apices is the main feature distinguishing L. asymmetrica sp. nov. from the mentioned L. gesierichiae, as well as from L. frickei. Both species have a regular linear-lanceolate valve shape with narrowly-rounded apices (Levkov et al. 2013). The last of the similar species – Luticola fuhrmannii is widely distributed in tropical areas of South America and Africa. Nonetheless, it can be separated, based on its elongated stigma positioned almost on the valve mantle (Levkov et. al. 2013, p. 385, fig. 4), whereas stigma of Luticola asymmetrica sp. nov. are round and located on the central area in the middle between the valve margin and proximal raphe endings.
The third of described Luticola species – Luticola terrestris sp. nov shows a high degree of similarity with several species from two informal morphological groups proposed by Levkov et al. (2013). Luticola terrestris sp. nov resembles L. tenuis Levkov, Metzeltin & Pavlov (2013, p. 236, pl. 35, figs 1–18) and L. micra Levkov, Metzeltin & Pavlov (2013, p. 156, pl. 35, figs 19–37) from the group A and L. incana Levkov, Metzeltin & Pavlov from the group B. Luticola tenuis and Luticola terrestris sp. nov. are observed from terrestrial habitats. Both species overlap dimensions; however, Luticola terrestris sp. nov. has higher striae density (20–23 in 10 µm versus 18–20 in L. tenuis). Moreover, L. tenuis has more linear valve outline and slightly deflected proximal raphe endings (Levkov et al. 2013, p. 35, figs 17, 18) in contrast to more curved proximal raphe fissures in L. terrestris sp. nov. Luticola micra differs from Luticola terrestris sp. nov. in range of valve dimensions (8–18 µm length and 4–5 µm width versus 8.1–28.3 µm length and 4.4–6.1 µm width) and the lack of ghost areolae in the central area. Additionally, L. micra has elongated stigma (Levkov et al. 2013, p. 35, fig. 19), whereas stigma in L. terrestris sp. nov. are regularly rounded. In addition, Luticola incana shows similarity with Luticola terrestris sp. nov., especially in the valve outline; however, it has a smaller range of valve dimensions (12.5–20.5 µm length and 5–6 width vs. 8.1–28.3 µm length and 44.0–6.1 µm width). Moreover, L. incana does not show a ghost areolae in the central area and has short, slightly deflected distal raphe endings (Levkov et al. 2013, p. 135) in contrast to L. terrestris sp. nov., where distal raphe endings are distinctly hooked and expanded on to the valve mantle. Finally, all three compared species are described from other continents: both L. micra and L. tenuis from Europe and L. incana from South America. Amongst species described from Africa, two Navicula taxa (which probably should be placed in the genus Luticola) show some degree of similarity to L. terrestris sp. nov. Navicula guluensis can be easily distinguished, based on striae density (15 in 10 µm versus 20–23 in 10 µm) and shape of raphe endings. Navicula guluensis has shortly deflected distal raphe endings (Giffen 1963, fig. 60; Giffen 1966, pl 3, fig. 59), whereas distal raphe endings of L. terrestris sp. nov are elongated and reach the valve mantle. The second of the African species Navicula submutica var. rectangularis shares with L. terrestris sp. nov. the valve outline, but has less dense striae with only 15–20 in 10 µm (Fusey 1964, p. 27, pl. 3, fig. 39).
Based on light microscopy observations, the Microcostatus meridionalis sp. nov. is the most similar to Microcostatus egregius (Hustedt) Lange-Bertalot (Stanek-Tarkowska et al. 2016), Microcostatus werumii Metzeltin, Lange-Bertalot & Soninkhishig (Metzeltin et al. 2009), Microcostatus edaphicus C.E. Wetzel, Noga, Ector & Stanek-Tarkowska (Stanek-Tarkowska et al. 2016), Microcostatus aerophilus Stanek-Tarkowska, Noga, C.E. Wetzel & Ector (Stanek-Tarkowska et al. 2016) and Microcostatus krasskei (Hustedt) J.R. Johansen & Sray (Johansen and Sray 1998). All species share similar valve outline and invisible or very difficult to observed striae in LM. Only in two species – M. meridionalis sp. nov. and M. werumii, apart from a raphe and valve outline, no other morphological features are discernible in LM. In other species, some additional features (for example, depressions on both sides of the sternum, visible as arch-shaped shadows or striation pattern) can be observed in LM. Microcostatus meridionalis sp. nov can be separated, based on its cuneate valve ends, whereas M. werumii has slightly rostrate valve ends. Additionally, proximal raphe endings are different in both species – in M. werumii, they are are poorly separated and located close to each other (Metzeltin et al. 2009, p. 225, figs 1–16). The drop-like proximal raphe endings of M. meridionalis are clearly visible, located on the asymmetrically constructed sternum. Under SEM, M. meridionalis sp. nov. poses a set of unique features that allows it to be easily distinguished from other representatives of the genus Microcostatus. Microcostatus meridionalis sp. nov. has areolae occluded only in the central area, not on the entire valve face. Moreover, it has no pseudoconopeum/conopeum in contrast to M. aerophilus and M. egregius (Stanek-Tarkowska et al. 2016, p. 166, figs 13–15). Internally, in the aspect of striae pattern and the shape of the central area, the described species is similar to Microcostatus schoemanii Taylor, Levanets, S. Blanco & Ector (Taylor et al. 2010, p. 180, fig. 8) and M. cholnokyi Taylor, Levanets, S. Blanco & Ector (Taylor et al. 2010, p. 182, fig. 25), but, based on the different shape of raphe endings and valve outline, are easy to distinguish. Regarding valve dimensions and striae density, Microcostatus meridionalis sp. nov. resembles M. krasskei. Both taxa overlap their dimensions (5–14 µm length, 3–4 µm width and 35–45 striae in 10 µm in M. krasskei versus 7.5–14 µm length, 3.5–4.5 µm width and 36–42 striae in 10 µm in M. meridionalis), but can be separated, based on the striae pattern and present of central area in. M. meridionalis (Johansen and Sray 1998, p. 99, figs 17, 19). The others of the aforementioned most similar species, although they have a similar shape, differ in striae density. Microcostatus egregius has 33–36 striae in 10 µm, whereas Microcostatus aerophilus up to 40–50 in 10 µm (Metzeltin et al. 2009; Stanek-Tarkowska et al. 2016), unlike to Microcostatus meridionalis which has 36–42 striae in 10 µm. Microcostatus aerophilus (Stanek-Tarkowska et al. 2016) fully matches in the aspect of valve dimensions, but usually has higher striae density (40–50 in 10 µm).
The studied assemblages consisted of diatoms with various geographic distributions. Cosmopolitan species, such as: Humidophila contenta, Hantzschia amphioxys, Nitzschia brevissima, Pinnularia borealis (Krammer and Lange-Bertalot 1986; Krammer 2000; Lange-Bertalot et al. 2017) occurred together with species with a wide distribution in the tropics – Luticola distinguenda and L. intermedia (Levkov et al. 2013; Glushchenko et al. 2017; Straube et al. 2017; Da Silva-Lehmkuhl et al. 2019). On the other hand, several of the noted species have not been reported from Africa so far or their occurrence in Africa has not been confirmed. Luticola permuticoides was originally described by Metzeltin and Lange-Bertalot (2007) from South America. So far, the occurrence of this species in Africa has not been confirmed. However, in the original description, the authors indicated the presence of a similar taxon (Metzeltin and Lange-Bertalot 2002, plate 27, fig. 17) in Madagascar, which probably is, indeed, Luticola permuticoides. Another species that has not been recorded in Africa so far is Orthoseira circularis, a taxon originally described from South America, but observed also in materials from Java (Houk et al. 2017). Together with other species identified in our study, O. circularis is probably a typical inhabitant of the terrestrial environment. Based on the presented results and literature data, both mentioned taxa can be considered as having pantropical distribution in terrestrial habitats.
In the studied material, some species were scarce and their exact identification was not entirely possible. In the literature data, there is a lack of information about valve dimensions of Stauroneis pygmaea f. undulata – the species originally described from Asia (Hustedt 1942). For this reason, despite the high level of morphological similarity in the valve outline and striae pattern, the observed species were noted as Stauroneis cf. pygmaea f. undulata. Two populations of Eunotia, observed in samples 2018/424 and 2018/426, show a highly similar pattern to Eunotia pseudominor Pavlov & Levkov. Both of our populations have similar dimensions: 12–35.6 µm length, 3.6–5.3 µm width and 11–15 striae in 10 µm in E. pseudominor (Pavlov and Levkov 2013) and 9.9–33.4 µm length, 3.2–4.1 µm width and 13–16 striae in 10 µm in our populations of Eunotia aff. Pseudominor; however, smaller valves were observed in our material. Moreover, the taxa observed in this study have less straight valve outline and more bent apices. Both taxa are also associated with moss vegetation. All of the observed differences in morphology can result from both the possible presence of two distinct species or only morphological differences within a single widely-distributed species. Therefore, we decided to identify it as similar to E. pseudominor (Eunotia aff. pseudominor).
Many diatom taxa develop various adaptations to changes in humidity of aerial habitats, such as: production of thickened valves, reduction of areolae number and occlusion of areolae with silica (Lowe 2011; Lowe et al. 2007; Round et al. 1990; Dodd and Stoermer 1962; Main 2003). Newly-described species represent genera commonly noted in various terrestrial habitats (Johansen and Sray 1998; Levkov et al. 2013; Lange-Bertalot et al. 2017). Amongst them, the highest level of adaptation is shown in M. meridionalis, which has fully covered areolae in the central part of the valve. Accompanying Pinnularia and Hantzschia taxa also show the presence of external areolae occlusions. This type of structure is also present on the surface of Eunotia cf. pseudominor, whereas the genus Eunotia does not create hymen or other occlusions (Round et. al 1990). The development of diatoms as epiphytes on mosses, which are able to accumulate water to provide a favourable environment for many organisms, can by also considered as form of adaptation to local environmental conditions (Round 1957; Lindo and Gonzales 2010; Glime 2017).
The investigated moss samples were characterised by a small diversity of diatom species (from 9 to 15 species per sample). This low species richness is quite typical for terrestrial environments, such as soils, rock crevices or clumps of terrestrial bryophytes. Especially in the Southern Hemisphere, terrestrial diatom assemblages are still poorly known, both in taxonomic and ecological aspects. The present study showed that these environments are often “hot spots” for the occurrence of potentially new and rare taxa.