Taxonomic revision of Saxicolella (Podostemaceae), African waterfall plants highly threatened by Hydro-Electric projects

Species of the genus Saxicolella Engl. (Podostemaceae) are African rheophytes, restricted to rapids and waterfalls as are all members of the family. Previously, Saxicolella sensu lato was shown to be polyphyletic with two separate clades. The name Pohliella Engl. was recently resurrected for one clade that is sister to the American genera Ceratolacis (Tul.) Wedd., Podostemum Michx. and all Old World Podostemoideae. Pohliella has distichous phyllotaxy, bilocular ovaries, filiform roots with paired holdfasts, and rootcaps. The second clade, Saxicolella sensu stricto, including the type of the generic name, has spiral phyllotaxy, unilocular ovaries, ribbon-like or crustose roots that lack both holdfasts and rootcaps. Saxicolella sensu stricto, sampled from the type species, S. nana Engl. of Cameroon, is embedded within and near the base of the major clade of African podostemoids and is sister to all other African genera apart from Inversodicraea R.E.Fr. and Monandriella Engl. Recently reduced to three species in Cameroon and SE Nigeria by the resurrection of Pohliella, Saxicolella sensu stricto is expanded to eight species in this monograph by description of five new taxa. Saxicolella futa Cheek and S. deniseae Cheek are newly described from Guinea, S. ijim Cheek from Cameroon, the informally named S. sp. A from Gabon, and S. angola Cheek from Angola. The known geographic range of the genus is thus expanded c. 2,500 km westwards to Guinea from eastern Nigeria and c.1,500 km southeastwards from near Yaoundé to Cuanza do Sul, Angola. The greatest concentration of species occurs in the Cross-Sanaga interval of western Cameroon and eastern Nigeria, with three species. Cameroon (3 species) followed by Nigeria and Guinea (2 species each) are the countries with highest species diversity. A classification is proposed grouping the species into three subgenera (Saxicolella, Butumia (G.Taylor) Cheek comb. et stat. nov. and Kinkonia Cheek subgen. nov.) based on root morphology and shoot position and morphology. The discovery, morphology, circumscription, distribution and ecology of Saxicolella is reviewed, an identification key to the species is presented, together with descriptions, synonymy and links to illustrations. All of the species are provisionally assessed as either Endangered or Critically Endangered using the IUCN 2012 Red List Criteria. The major threats, above all, are hydro-electric projects. Saxicolella deniseae may already be globally extinct, and two of the four known locations of S. angola appear lost, S. sp. A of Gabon is threatened at at least one of its three locations, while S. futa is threatened at all three locations. Contamination of watercourses by increased turbidity from silt-load due to anthropic changes and by eutrophication from pollution are also threats for the majority of the species.


Introduction
Podostemaceae are a pantropical family of annual or perennial herbs placed in Malpighiales in a sister relationship with Hypericaceae (Ruhfel et al. 2011). There are about 300 species globally, in c. 54 genera (Koi et al. 2012). Species numbers are highest in tropical America, followed by Asia, with Africa having c. 106 species (Cheek & Lebbie 2018). All species of the family are restricted to rocks in rapids and waterfalls of clear-water rivers (rheophytes) or occur in the spray zones of waterfalls (this paper). However, waterfalls are being increasingly exploited for hydro-power at risk to the survival of the Podostemaceae they contain (Schenk et al. 2015;Cheek et al. 2015;Cheek & Ameka 2016;Cheek et al. 2017a;Cheek et al. 2017b). Most of the African species of Podostemaceae are narrow endemics, many being known from only a single waterfall. New discoveries of species are still made frequently, in addition to those studies above (Rial 2002;Cheek 2003;Schenk & Thomas 2004;Beentje 2005; Cheek & Ameka 2008; Kita et al. 2008;Cheek & Haba 2016;Cheek et al. 2019a;Cheek et al. 2020a), including new genera (Cheek & Lebbie 2018).
Three subfamilies are recognised. Tristichoideae, sister to the rest of the family, have three foliose tepals that protect the developing flower, and are tricarpellate. Weddellinioideae, with a single genus are Neotropical and have two foliose tepals and bilocular ovaries. Podostemoideae, which is pantropical, is the most genus-and species-rich subfamily. It has flowers protected in a spathellum, a balloon-like sac in which the flower develops while the plant is underwater, and tepals reduced to vestigial, filiform structures (Cheek 2020). African Podostemoideae, or podostemoids, are the main focus of this paper.
Important characters in defining genera in African podostemoids are the position of the flower in the unruptured spathellum, and the number of locules, shape, and sculpture of the ovary. At species level, important characters are the shape and relative proportions of spathellae, stigmas, anthers, filaments, gynophores, pedicels, and leaves.
However, Saxicolella Engl. was one of the few African genera that Cusset did not revise. Yet, in her Flore du Cameroun account (Cusset 1987), in addition to the type species Saxicolella nana Engl., she included in Saxicolella the genus Pohliella Engl. with two species P. laciniata Engl. and P. flabellata G. Taylor. Taylor (1953) had already expressed his doubts about Pohliella "Apart from differences in habit and shape of the stigmas, I am not satisfied that the key characters used by Engler to distinguish Pohliella from Saxicolella are sufficiently diagnostic." (Taylor 1953: 55). The two genera were subsequently treated as synonymous under the name Saxicolella (e.g. Cook & Rutishauser 2001. It is not difficult to see why this was the case. Both genera are unusual among African Podostemoideae in that the ovary is not inverted in the spathellum but erect. Further, at anthesis, the ovary is not held on a long pedicel that exceeds the length of the spathellum by 3 -4 times but is either held inside the ruptured spathellum with only the styles and stamen emerging, or the pedicel is only as long or at most twice as long as the spathellum. Further, both genera have unistaminate flowers, which are not common in African Podostemoideae, where two stamens per flower are usual. However, the molecular phylogeny of Koi et al. (2012) showed that Saxicolella in this broad sense is polyphyletic, with two clades arising at different points from the family tree. It was shown by Cheek (2020) that these two clades differ from each other in several important characters, sufficient to merit generic separation (Table 1 below, reproduced from Cheek  2020).
Therefore, Pohliella was proposed for resurrection (Cheek 2020), leaving Saxicolella in the narrow sense, with two species in Cameroon, one of which extended to Nigeria and one endemic to Nigeria. In this paper we provide a taxonomic revision of the genus Saxicolella including both newly collected and previously overlooked material that conforms to the delimitation of the genus as represented in Table 1. We also review what information is available about the genus.

Material and Methods
Four of the eight species accepted in this paper have been studied in the wild by the authors. Fieldwork to collect data on Saxicolella for this paper began as part of general botanical surveys in Cameroon for conservation management. The methodology used was as reported in Cheek & Cable (1997), and specimen data storage by Gosline, in Cheek et al. (2004). Fieldwork on the genus recommenced in Guinea nearly 20 years later as part of targeted surveys partly focussed on waterfalls, where Google Earth was used to target and navigate to waterfalls. A blade was used to remove plants from their rocks when exposed in the dry season, rehydrating them first if already desiccated. Conventional herbarium and silica gel specimens were made and photos were also taken where possible. The most complete set of material was deposited in the National Herbarium of the country concerned.
Herbarium material was examined with a Leica Wild M8 dissecting binocular microscope. This was fitted with an eyepiece graticule measuring in units of 0.025 mm at maximum magnification. Botanical line-drawings were made using the same equipment, fitted with a camera lucida. The morphological species concept was followed in defining species (each species being separated from its congeners by several, usually qualitative, morphological disjunctions), and the overall morphology of species was described and illustrated based on herbarium specimens following standard botanical procedures as documented in Davis & Heywood (1963). All specimens cited have been seen by the first author unless indicated`n.v.' Herbarium citations follow Index Herbariorum (Thiers, continuously updated) and authors of plant names IPNI (continuously updated). Material or images were studied from, and checks made for specimens at B, BM, COI, EA, GC, HNG, K, L, LISC, MO, P, SCA and YA. Key online specimen databases searched included GBIF.org, Tropicos.org, MNHN collections website https://science.mnhn.fr/institution/mnhn/collection/p/item/search/form?lang=en_US and those of COI: https://coicatalogue.uc.pt/?Collector= gossweiler&t=results&orderby=relevance&orderdire-ction=DESC&size=10 and LISC: https://actd.iict.pt/ list/?cat=quick_filter&search_keys We were not able to inspect the DNA voucher specimen listed by Koi et al. (2012) as at TNS, but this is thought to be from the type location where standard herbarium specimens have already been made, and have been viewed. In total 16 unique herbarium records of the genus were studied not counting their duplicates. Nomenclatural changes were made according to the Code (Turland et al. 2018).
Conservation assessments were either taken from the recent literature (see citations) or made using the categories and criteria of IUCN (2012). The cell-size used for calculating area of occupancy was 4 km 2 , as advocated by IUCN (https://www.iucnredlist.org/ resources/redlistguidelines). Geocat (Bachman et al. 2011) was used to calculate extent of occurrence in those few species with more than three locations.
Rheophytic herbs. Roots ribbon-like and/or disc-like, crustose, highly dorsiventrally flattened, to at least five times as wide as thick, adhering to the substrate by root hairs on the ventral surface, rootcaps and haptera absent. Shoots erect, minute and supporting sessile leaf rosettes, the stem then not visible, then arising from the margins of the radiating ribbon-like part of the root or, (S. futa) the sinuses of the bifurcations of the ribbon-like root OR up to several cm long, unbranched or branched, arising from either the centre of the disc-like crustose part of the root.

Discovery
The first published and type species of Saxicolella, S. nana, was collected in Kamerun, then a German colony, now Cameroon, in January 1914 by the renowned botanist Mildbraed (Engler 1926 Rutishauser (2001Rutishauser ( , 2007. Characters taken from Engler (1926), Hall (1971), Ameka et al. (2002), Cheek (pers. obs. in Ghana, 1995 andCameroon, 2008) and Koi et al. (2012: 473,  author. The second and third authors then went on in Feb. 2018 to collect the species described as S. deniseae Cheek sp. nov. Saxicolella sp. A only came to our attention as this paper was being concluded in mid-2021, thanks to photos via GBIF.org of recent collections by the LBV-MO botanical team.
The new species published in this paper are unlikely to be the last added to the genus. It is expected that botanical survey of the many rapids and waterfalls of Africa that have never been inspected for Podostemaceae will produce additional species new to science if this can be done before they are modified by the hydro-electric projects which are likely to result in their extinction.

Morphology
While species of several other genera of African Podostemaceae have been investigated in detail for their morphology and anatomy in such studies as Moline et al. (2007) and Thiv et al. (2009), this has not been the case for any of the species of the genus Saxicolella Engl. as delimited here (the Ghanaian species previously referred to as Saxicolella have been transferred to Pohliella, see Cheek 2020). None of the species appear to have been investigated anatomically, nor has their micromorphology been investigated under the electron microscope. The overview presented here is partly based on the protologues of the species already published by Engler (1926) & Taylor (1953, but mainly from the observations of the authors of the four new species described below. Root. The root (thallus) is either crustose and/or with several, ± broadly ribbon-like arms radiating from a central crustose area (rarely absent/not detected e.g. Saxicolella futa). It is usually several times wider than thick, and is closely appressed to the substrate of smooth rock to which is firmly fastened by numerous short root hairs. A faint raised ridge running along the midline of the rootribbon of S. futa suggests that as in Inversodicraea (Cheek et al. 2017b), a single, central vascular bundle is present. Photosynthesis seems to be mainly performed by the ribbon-like roots since these make up most of the surface area of the plants, in fact >90% of the area in almost all species. Root-caps have not been reported nor observed, but are in any case not usual in those Podostemaceae genera with crustose and broad ribbon-like roots. Nor are haptera, also known as hold-fasts, present. Roots are neither recorded nor preserved in the available material of S. angola, and are incompletely known in S. nana and S. flabellata. The ribbon-like roots of individuals appear to radiate out from the central point of establishment, presumably where a seed has germinated and established. In contrast, in S. nana the radial growth appears to be "crustose", that is, not in the form of distinct separate ribbon-like structures, but a solid mass which extends outwards more or less evenly along the circumference, with only slight lobing at the margins.
In most species, e.g. Saxicolella deniseae, and S. marginalis, the root is intermediate: it has both a central crustose part several centimetres in diameter, but also the margins are well-developed into radiating ribbon-roots. In S. futa the central crustose part if developed at all, must be small and only a short-lived stage which is lost by fruiting time (if it is developed in the first place), leaving the radiating roots disconnected from each other at the centre.
In most species the ribbon-like, radiating roots rarely (Saxicolella deniseae, S. marginalis and S. ijim) branch, when they bifurcate into two equal branches. However, in S. futa the branching is frequent and regular and the roots form a distinctive pattern. In fact, each species of Saxicolella can be identified by the architecture and grossmorphology of its root alone (where known), although this can be difficult to convey in words.
Shoots. The origins of the shoots from the roots and their development, follows one of three patterns: 1. the shoots arise only from the central, more or less disc-like, crustose part of the root, and not from the radiating ribbon-like roots. The shoots form visible stems with measurable internodes. Saxicolella nana, S. flabellata, S. ijim and, possibly, (root unknown but visible stems present) S. angola. 2. The shoots arise only from the margins of the radiating, ribbon-like roots. The shoots are sessile, not forming visible stems but supporting an inconspicuous rosette of reduced leaves and a terminal spathellum. Saxicolella marginalis, S. deniseae, S. sp. A 3. The shoots arise only from the synusiae of the bifurcations of the radiating, ribbon-like roots. As in 2, the shoots are sessile, not forming visible stems. Saxicolella futa.
The shoot patterns appear to correlate with the three root patterns (see Roots, above). The taxonomic significance is discussed below.
In those species where visible stems are developed, they are erect, terete, and in those species where they exceed more than 5 mm long, sparingly branched. In Saxicolella ijim, the stems are robust and free-standing at anthesis. This species was found in the spray zone of a waterfall (Cheek pers. obs.) and is not supported by water as appears to be the case of the more laxly stemmed S. flabellata which has the longest (21 cm) stems in the genus, described as flowing in the protologue (Taylor 1953).
Leaves. The phyllotaxy is consistently spiral. The leaves are best developed in the species with pattern 1 shoot position, where visible stems are developed. The largest leaves are those of Saxicolella flabellata which are flabellate (dorsiventrally flattened with radiating lobes) and up to 3 cm long, 2 cm wide. Each leaf bifurcates or trifurcates up to four times, the ultimate segments are capillary. The base is sheathing. Stipules are inconspicuous.
Leaves in Saxicolella angola are poorly preserved, smaller, but otherwise similar, with fewer bifurcations and with stipules conspicuous. In S. nana the leaves are filiform-capillary and trifurcate, while those of S. ijim are unbranched and laterally compressed.
In pattern 2 species, Saxicolella marginalis and S. deniseae, whilst the shoots are reduced and visible stems are not formed, the leaves appear reduced to the sheathing, stipulate base with only a rudimentary blade, while in S. sp. A, the linear blade is as long as the flower In pattern 3 Saxicolella futa, the leaves are reduced further, to inconspicuous, minute 0.3 mm long concave sheaths with stipules and blade not developed.
Leaves of the type usual in African podostemoids are absentthat is, those which are filiform, terete and bifurcate repeatedly in the distal half, and which are shed before anthesis.
Inflorescences. In all species flowers occur singly at the apex of shoots except in Saxicolella flabellata and S. angola where they are in terminal clusters. The developing spathellae are protected by the subtending leaves in the earliest stages. In fact, the leaves appear to function primarily as protective bracts in most of the remaining species of the genus. The spathellum varies from globose (S. ijim) to narrowly ellipsoid, sometimes with a small apiculus. It lacks a distinct stipe.
The flower is erect and held within the opened spathellum at anthesis. Generally, only the styles and anthers are exserted from the ruptured spathellum but sometimes all or part of the ovary is projected from the spathellum. However, in Saxicolella nana and S. sp. A. the ovary can be projected on a naked pedicel as long as itself. A short pedicel, two filiform (rarely spatulate) tepals that flank the single stamen, and a short gynophore are present (absent in Saxicolella sp. A), all concealed within the ruptured spathellum at anthesis. The anther-thecae often face away from each other (latrorse). Pollen is dyad (where available for study).
The free-central axile placenta in the unilocular ovary is either narrowly spindle-shaped e.g. Saxicolella angola, S. nana, S. futa or broadly so, occupying about half the radius of the locular cavity in S. ijim. The seeds are all ellipsoid, completely covering the placenta, where known.
The fruit, as the ovary, is fully erect, and isolobous (the two valves are equal). The fruit is carried further out of the spathellum remains by the extension of the pedicel post-anthesis. The two valves dehisce but usually persist in the fruit. The seeds are mucilaginous where known as usual in the family.
DISTRIBUTION. Tropical West Africa: Guinea, Nigeria, Cameroon, Gabon and Angola.
Eight species. Saxicolella species are restricted to Africa and extend from the Guinea Highlands of Guinea-Conakry in west Africa (newly recorded here) to Angola (newly reported here) in western southcentral Africa. They are not recorded from the Congo basin, nor eastern Africa.
The geographic range of the genera Talbotiella Baker f., (Leguminosae, nine species of evergreen tree) recently also extended to Guinea (van der Burgt et al. 2018), is similar to that of Saxicolella although that genus does not extend to Angola (Mackinder et al. 2010). Mischogyne Exell (Annonaceae trees, five species, Gosline et al. 2019) also has a similar distribution but has an outlying species in Tanzania and one in DRC. The highest species diversity of Saxicolella is the Cross-Sanaga River interval of eastern Nigeria-western Cameroon which has three of the eight species: S. marginalis, S. flabellata, and S. ijim. The Cross-Sanaga River interval area contains the highest species and generic diversity of flowering plants per degree square in Tropical Africa according to several studies (Barthlott et al. 1996;Dagallier et al. 2020) possibly in part because it corresponds with the Cameroon Highlands (Cheek et al. 2001). Many of the species and some genera (e.g. Medusandra Brenan (Peridiscaceae, Breteler et al. 2015;Soltis et al. 2007) are both endemic and threatened. Saxicolella species are known only from the five countries mentioned but are likely to be found in intervening areas such as Sierra Leone, Liberia, Ivory Coast and Congo-Brazzaville. Of the eight known species, four are point endemics.
HABITAT. Saxicolella only grows, as with most Podostemaceae, in sites with seasonally or permanently, fast-flowing, well aerated, silt-free fresh water. They are always associated with waterfalls and rapids. Unusually among African Podostemaceae genera, about half the known species of Saxicolella occur mainly in the 700 -1400 m altitudinal range, when other genera are predominantly restricted to lowland elevations. However, S. nana, S. flabellata, S. deniseae and S. sp. A all occur in the (100 -) 400 -700 m altitudinal band. The species of the genus appear to avoid coastal areas. Although Cameroon has the highest species diversity of both Saxicolella and Podostemaceae in Africa, Saxicolella is absent from the most species-diverse Podostemaceae site (which has 10 species) the Lobé Falls near Kribi, at the coast in the lowland evergreen forest belt (Cheek et al. 2017b). However, all but one of the eight species of Saxicolella co-occur at least once with one or several other species of Podostemaceae (see individual species accounts). The exceptions is S. ijim, which is was not observed to occur with other Podostemaceae, possibly because of its unusual ecological niche. Saxicolella ijim is unique in the genus in that it flowers in the sprayzone of a waterfall, and it is not immersed in water immediately before flowering as is usual in other species. However, Ledermanniella prasina J.J.Schenck & D.W.Thomas of the Korup has the same ecology (Schenk & Thomas 2004) and L. letouzeyi C.Cusset of the Bakossi Mts can also occur in the spray-zone of waterfalls although not exclusively as does Saxicolella ijim (Cheek et al. 2004).

Pollination & Hybridisation
Although we suspect that pollination is by flying insects such as bees, as reported in other African podostemoids (Cheek et al. 2017b), no floral visitors have been reported or observed thus far for any Saxicolella species. Hybridisation, reported for the first time in African Podostemaceae in Cheek et al. (2017b) is not known in Saxicolella. Since none of the species is sympatric, this is not unexpected.

Habitat partitioning
In those four species of Saxicolella that co-occur at sites with other Podostemaceae species, it has not been possible to study habitat partitioning except for Saxicolella futa at one site in Guinea: Case study: Salaa Falls, Futa Djalon, Guinea-Conakry At this site four species of Podostemaceae occur in close proximity some tens of metres downstream from the main tourist falls. At one point, all four can be found within a 1 m by 1 m square. When observed by MC and DM in Jan. 2018, midway through the dry season, Stonesia heterospathella G.Taylor was in fullflower, having been exposed by the slowly falling water in recent weeks, while Ledermanniella guineense C.Cusset, growing deeper in the water than any other species, was just beginning to flower. Highest up the gradually sloping flat rock surfaces were colonies of Tristicha trifaria Spreng., long dead. Saxicolella futa, also long dead, grew on the rock surface 30 -100 cm above the level of the water surface, in a band below the Tristicha and above the Stonesia, plants of the two species intermingling at the interfaces. This same zonation, with Tristicha (above) and Stonesia (below) the Saxicolella futa was also seen just above the main Kambadga Falls near Pita, where Saxicolella futa was much rarer. Saxicolella futa appears to grow or compete better in slightly deeper water, than the Tristicha, and needs a shorter growing season (a shorter period underwater) than both the Stonesia and Ledermanniella.

Conservation status
The principal threats to Saxicolella species apply to Podostemoideae species as a whole, especially in Africa. Because they are restricted to habitats with clean, non-turbid, aerated water, with a rock substrate, degradation of any these environmental factors pose threats. Given that so many Podostemaceae species, including Saxicolella, are restricted to only one or two locations, they are especially at risk. A threat at even one location is likely to pose a high extinction risk for any Saxicolella present. All of the species are provisionally assessed as either Endangered or Critically Endangered using the IUCN 2012 standard.

Turbidity & eutrophication threats
Turbidity in the water, indicating that silt is present, can reduce establishment of seedlings (Philbrick & Novelo 1995). It can also reduce photosynthesis during the main growth period, when plants are under water in the wet season (Cheek et al. 2015).
Algal growth can blanket Podostemaceae plants at some sites and reduce their ability to photosynthesise. Such growth appears to be associated with nutrient addition to rivers from human populations that may use water courses for processing crops, cleaning, and removal of waste-products. Saxicolella futa (this paper) appears to be threatened in this way.

Hydroelectric Power Project threats
The greatest threats of global extinction for species of Podostemaceae such as those of the genus Saxicolella are from hydroelectric projects which have been growing rapidly in number in recent years as a source of cheap, greener energy in Africa. They are attractive to investors and governments being seen as sustainable and a good alternative to hydrocarbon-sourced energy. While hydroelectric projects have many environmental benefits compared with alternative options, all too often they threaten Podostemaceae species with extinction, and there are now many documented cases of local and global extinctions of Podostemaceae species resulting from such projects (Cheek et al. 2015;Cheek et al. 2017b;Couch et al. 2019).
Dams for hydro-electricity generation are constructed just above rapids or falls so as to benefit from the vertical drop in water levels (the "head") at these sites. The construction of the dams may directly impact upon the falls and the species that they contain. More usually dams negatively affect populations of Podostemaceae through changes in water flow by four different threats: Cascade systems are steadily being developed in the Cuanza of Angola, where two of the four global locations of Saxicolella angola are thought to have been lost even before the species is formally named (see that species), in the Ogooué of Gabon which will threaten S. sp. A (see that species), and also in the Konkouré of Guinea which may already have destroyed the only known global population of S. deniseae.

Difficulties with EIAs for Podostemaceae
It is extremely rare that competent Environmental Impact studies (EIA) are requested and conducted in advance of planning for such hydro projects in our experience. If EIA studies are conducted, they usually do not take into account the possible presence of Podostemaceae at these sites: many botanists mistake these flowering plants for mosses or algae (groups of plants usually regarded as non-threatened) and do not collect samples for identification so that dam construction goes ahead in ignorance of the presence of these often highly threatened species. Even if such studies have been done in advance, and samples collected from which Podostemaceae can be identified, two further obstacles exist 1) many Podostemaceae have out-of-date Red List assessments which often misrepresent the species as being of low or unthreatened status when they may be highly threatened and 2) most Podostemaceae species remain without a Red List assessment. Unless species can be shown to have a published Red List assessment of EN or CR on iucnredlist.org, or an extent of occurrence of <50,000 km 2 , they are generally not considered to merit concerted conservation action in projects by the International Finance Corporation of the World Bank Group that often supports finance of such projects (IFC 2019).
PHENOLOGY. Species of the genus generally flower as water levels drop after the rainy season, exposing the plants that have developed underwater in previous months, and triggering flowering, and seed set, and if the plants become dried out, death. Saxicolella futa is thought to complete its life-cycle in 6 months or less (see that species), but other species, such as S. ijim, may prove to be perennial if they are kept moist by waterfall spray throughout the dry season. ETYMOLOGY. The name Saxicolella is compounded of saxicole, meaning 'dweller on rock' andella a diminuitive. The whole signifies "little dweller on rock". However, almost all Podostemaceae always grow on a rock substrate, and many are diminuitive. VERNACULAR NAMES. None have been recorded. Generally, e.g. in Guinea, local communities do not have terms for different species of the genus at a location, but one term, treating the family as one entity (Cheek pers. obs, Guinea 2018, 2019).
Infrageneric classification. The species fall into three groups strongly supported by root and shoot type and pattern, and by leaf characters. It can be argued that the three groups of species within Saxicolella sensu stricto could each be recognised as separate genera since in Asia, the same grouping characters (shoot position) have been found useful for this purpose (Koi et al. 2012, see above under shoots). Moreover, this argument is strengthened by the correlation of root characteristics with these same groups (see above), but with no correlated floral or fruit characters. However, we are reluctant to increase the number of genera if there is an alternative option. Therefore, we have opted to recognise these three groups at subgeneric rather than generic level. Should molecular phylogenetic work support generic recognition (e.g., by long branches with high support values), consideration might then be given to elevate these subgenera to generic level. This would necessitate resurrection of the generic name Butumia G.Taylor (here adopted as a subgeneric name) and elevating the subgeneric name Kinkonia (proposed below for the eccentric Saxicolella futa species of the Futa) to generic level. However, it is possible that these groupings are the result of convergence and have no phylogenetic value.
Saxicolella sensu stricto, sampled from Cameroonian material, is embedded within and near the base of the major clade of African podostemoids and is sister to all other African genera apart from Inversodicraea R.E.Fr. and Monandriella Engl. (Koi et al. 2012). The sister relationship of Saxicolella and Monandriella shown by Koi et al. (2012) was foreshadowed by Engler who in his global treatment of Podostemaceae placed these two genera consecutively (Engler 1930: 29).
Identification Key to the species of Saxicolella Subgenus 1. Saxicolella Type species: Saxicolella nana Engl.
PHENOLOGY. Flowering and fruiting in mid-December. ETYMOLOGY. Referring to the shape of the leaves, flabellate meaning fan-shaped. VERNACULAR NAMES. None are recorded.
NOTES. Distinct from all other known species of the genus in the very long stems, and in bearing spathellae from dimorphic leaved shoots, those arising from the roots being different from those from the long stems. Similar to Saxicolella angola in the spathellae in terminal clusters, not single, and subtended by more or less flabellate leaf rosettes.
Originally described by Taylor as a Pohliella, he explained that he was in a quandary as to placement in this genus or in Saxicolella as described by Engler (1926) which work he criticised ("I am not satisfied that the key characters used by Engler are sufficiently diagnostic"). In fact, Engler had separated these two genera in his key (Engler 1930: 29) based on locule number, and fruit rib number although they differ in other features. Taylor based his placement on features other than those in Engler's key ("I have placed it in this genus due to the subulate stigmas and dichotomous leaves"), despite the first being discordant ("it deviates from the generic description in having a unilocular ovary") (Taylor 1953: 55). It seems from the molecular phylogenetic evidence of Koi et al. (2012) that in this case locularity is indeed a better indication of relationships than leaf habit and stigma shape. Saxicolella submersa (J.B.Hall) C.D.K.Cook & Rutish. is superficially very similar to Pohliella laciniata which grows in the same area at the forest border of Cameroon and Nigeria. Both species flower from long stems when these reach the water surface as the levels drop in the drier season. That both species have clusters of several flowers surrounded by rosettes of flabellate leaves that form a protective funnel, borne on long stems may be convergence to this scenario. These rosettes may function to float on the surface and protect the flowers they contain from water droplets (observed by the first author for Pohliella laciniata in Cameroon). The same trait (several flowers surrounded by a rosette of flabellate leaves) is seen otherwise seen only in Saxicolella angola where the ecology is unreported and the specimens fragmentary. Other species of Saxicolella have single flowers borne terminally in rosettes of leaves that are linear or highly reduced and can have no protective function during flowering (although they are likely to protect the developing flower buds).
Other species are also both confined to the forest of the Afi River Forest Reserve and adjoining Cross River forests in SE Nigeria and known from only one or two collections, e.g. Anchomanes nigritianus Rendle (Moxon-Holt & Cheek 2021) and Talbotiella eketensis Baker f. (Mackinder et al. 2010).   Perennial or annual herb, rosette-like, 7 -8 cm diam. Root crustose in the central part of the plant, at the edge radiating and divided into separate, free, lobes, 0.5 -0.8 cm wide ( Fig. 2A). Stems (1 -) 2 -6, arising from the crustose centre of the root rosette (absent from the radiating lobes), erect, free-standing, branched from the base, (1 -) 4 -7 cm tall, spreading as wide as the root rosette, terete, each c. 2 mm diam. at base, with 4 -8, ± evenly spaced, short leafy side-branches (Fig. 2B), proximal branches up to 1.5 cm long, phyllotaxy spiral, diam. slightly more slender than the principal axes, leaves with axillary rosette shoots. Rosette (spur) shoots axillary with stems inconspicuous, 1 -3 per axil, each bearing 3 -8 leaves and a single terminal spathellum.
Saxicolella ijim occurs within a few kilometres of S. marginalis. These two species, together with S. angola, occur at the highest altitudes known for the genus (1300 -1400 m alt.). The two Bamenda Highland species are easily separated since while the first has long stems, which are only produced from the centre of the radiating root rosette, the second lacks long stems completely, and instead bears numerous sessile, rosette-like stems along the margins of the radiating ribbon-like roots.
Saxicolella ijim is similar to Saxicolella sp. A in the unusual feature of the ovary being sessile (the gynophore being absent).
A summary of the rare, high altitude plant species of Kilum-Ijim (Mt Oku) is given by Maisels et al. (2000). Additional narrowly endemic species discovered from the Kilum-Ijim area are: Annual (probably) herb, 1.2 -1.9 cm tall. Root incompletely seen, probably crustose. Portion at base of stem (Fig. 3A) shield-like c. 1.1 mm diam., irregularly convex. Stems unbranched, erect, stout, selfsupporting 9 -15 mm long, terete, c. 0.5 mm wide at base, increasing to c. 1 mm diam. at apex. Proximal 1.8 -5.5 mm lacking leaf-scars, distal portion with 5 -7 leaf scars, internodes 0.5 mm long (proximal-most internodes) to 2 mm long (more distal internodes). Scars ± amplexicaul, distal nodes with leaf bases persistent, sheathing (Fig. 3B); stem apex with a head of flowers surrounded by clusters of heteromorphic leaves, phyllotaxy spiral. Leaves of outermost (proximal) part of apical cluster ligulate, 2.5 -3 × 0.25 -0.3 mm, or spatulate, that is with the distal end broader, elliptic, 0.8 × 0.4 mm, apex rounded or obtuse, base slightly sheathing, stipules absent (Fig.  3C); innermost, more distal leaves of apical cluster broadly ovate, or ovate in outline 0.6 -1 (-1.5) × 1 -1.5 mm, apex entire or slightly or deeply bifid, lobes equal or unequal, base broad, with or without marginal stipules. Stipules sometimes exceeding blades, subulate 0.5 × 0.2 mm. Spathellae in terminal cluster of (2 -) 5 -8, pre-dehiscence narrowly ellipsoid c. 2.5 × 0.8 mm, dehiscing into (2 -) 3 -5 subequal lobes, overall c. 2.5 × 1.5 mm. Flower partly concealed in spathellum at anthesis (Fig. 3D). Pedicel 0 -0.5 mm Known from four collections, each at a different location of the Cuanza or an affluent. The collection site of "Punta Filomeno da Camera" has not been found but its altitude of 100 m on the Cuanza corresponds with the hydroelectric dam at Cambambe and so it is likely that the species has been lost at this location. However, the identification of the specimen is not completely certain since it was sterile and only viewed online. The site at "Cuene or Cuno, bridge on river", since it is at 650 m alt. on the Cuanza, corresponds with the newly constructed Lauca Hydroelectric project, Angola's largest. Here again the species is unlikely to survive due to the loss of its habitat and hydrological change. This leaves two locations upstream where the species is likely to survive. These two locations are c. 200 km apart. Gossweiler 9428 (type specimen) near Mussende was recorded 31 June 1930 (presumably an error by Gossweiler for 1 July). The exact site was not given, but is most likely to be the ford across the Gango River 16.5 km from Mussende on the road to Quibala (10°36'07.7" S, 15°52'49. 54"E observed on Google Earth), since this set of rapids is closest to Mussende (the town and river mentioned on the label) and is most readily accessible from that town. That vehicles are likely to drive over the plants at the type locality is highly possible, and this would constitute a threat. The second locality, the Kalandula Falls (also known as the Calandula falls and formerly Duque de Bragança falls) on the Lucuala (or Licuala) River (Gossweiler 8858, 29 Aug. 1922), is now a major tourist attraction, probably because it is one of the major falls by volume in Africa and is only 360 km by road from the capital, Luanda. It is evident from the numerous posts of photographs by tourists on the internet (https://en.wikipedia.org/wiki/ Kalandula_Falls accessed 30 May 2021) that trampling by visitors occurs, which can destroy plants of Podostemaceae as at the Lobe Falls in Cameroon (Cheek et al. 2017b). Both locations, are here ascribed an area of occupancy of 4 km 2 as preferred by IUCN (2012). Therefore, Saxicolella angola is here assessed as Endangered EN B2ab(iii) using the categories and criteria of IUCN (2012). Despite this assessment, and the fact that this species has not been recorded in the wild for 78 years, and despite the fact that Hess (1953), a Podostemaceae specialist who visited the Kalandula Falls in 1950 did not find the species there, we doubt that this species is extinct, although it is probably not common. This is because at both known locations, further apparently suitable habitat can be seen on Google Earth immediately upstream and/or downstream. However, there is no guarantee that the species occurs at these sites because it is rare and infrequent (only 2 of the 35 Angolan Podostemaceae held at COI are this taxon, and one of these is only doubtfully identified). Finally, there is no cause for complacency about the security of this species since the surviving two sites are both at risk of future new hydroelectric projects in which there is currently an upsurge in Africa and which are inimicable to the survival of Podostemaceae (Cheek et al. 2017b).
Angola is currently going through a surge of development posing risks to its species-diverse Flora. Additional range-restricted newly described species endemic to Angola are Justicia cubangensis I.Darbysh. & Goyder (Darbyshire & Goyder 2019) and Stomatanthes tundavalaensis D.J.N.Hind (Hind & Goyder 2014).
PHENOLOGY. Only collected in flower (end June) and fruit (August) at the end of the dry season, the wet season being September to April ETYMOLOGY. Named for the country of Angola as a noun in apposition. This species is both unique to Angola and the only species of the genus currently known to occur in the country. NOTES. Saxicolella angola was first recognised as a distinct species, but informally, and not published, by C. Cusset in 1975. This is evident from her annotations of all four specimens of the species cited in this paper. She had annotated two specimens of what appears to be this taxon (Gossweiler 11353 and Gossweiler 12037 at COI, viewed online May 2021) as "cf. Saxicolella angolensis" dated 1975. Two other specimens, the basis of the description above, were sent on loan from K to P in 1982 (registered at K as H960/82) and were annotated as "Saxicolella gossweileri C. Cusset ined. 1998", on one of which she had deleted an earlier annotation she had made of "Pohliella angolensis C.Cusset ined.". Since the Code (Turland et al. 2018) advocates that such names should not be perpetuated without the permission of the author, and since that permission has not been obtained, an alternative name has been selected.

Saxicolella nana
RECOGNITION. Similar to Saxicolella sp. A (see below) but differing in the disc-like root, the shoots produced at its centre (not ribbon-like, the shoots at the margin), a distinct gynophore present, fruit 6-ribbed (vs ovary sessile, fruit 8-ribbed). DISTRIBUTION. Cameroon, Centrale province. Only known from the type locality near Mbalmayo, Nyong River.  (Engler 1930: 48)). Alt. c. 330 m. CONSERVATION STATUS. Saxicolella nana was assessed in 2007 as VU D2 (Ghogue 2010), citing a single location, the Nyong at Mbalmayo, with an AOO of <20 km 2 and major threats of water pollution, temperature extremes and sudden drought. Independently, a provisional assessment of CR B2ab(iii) was made in Onana & Cheek (2011: 254), unaware that at this point, Ghogue had rediscovered the species likely at the type locality in 2004, the samples all being sent to Z. Cameroon has been relatively well-surveyed for Podostemaceae thanks to pioneering collectors in the German colonial period (1884 -1916). More recently intensive surveys dedicated to finding sites for this family in Cameroon have been carried out above all by Ghogue, but also by dedicated Podostemaceae researchers from Switzerland, Ghana, USA, Japan and Britain but no further locations for Saxicolella nana have come to light. Much of the length of the Nyong S of Yaoundé is not suitable for Podostemaceae due to the absence of rapids. However, some other rapids can be detected there on Google Earth (viewed June 2021) and would be worth visiting at the correct season to establish if the species has more than a single site. We know of no hydro-electric dams planned for the Nyong at present but this could well change given the number of dams planned elsewhere in Cameroon. We contend that the data presented merits reassessment as Critically Endangered CR B1+B2ab(i-iv). PHENOLOGY. Flowering in January. ETYMOLOGY. The specific epithet refers to the small stature of the species. NOTES. The type at B was annotated by C. Cusset in 1974 as "Pohliella nana (Engl.) C. Cusset" suggesting that at that time she intended to make Saxicolella a synonym of Pohliella. She did not publish the combination and reversed the decision by the time she published her Flore du Cameroun account, when instead she sank Pohliella into Saxicolella (Cusset 1987).
The holotype at B bears three separate drawings. One of these appears to be made in the field by the collector, Mildbraed and shows the thalloid crustose root with a statement that the thallus firmly adheres for its entire surface. The second sketch (reproduced here as Fig. 4) is the most comprehensive and shows numerous welldelineated thumbnail sketches of the dissected specimen some of which do not figure in the published illustrations e.g. Engler (1926), presumably due to space limitations. The third drawing is a more roughly drawn outline of parts of the second with an ink caption that may have been a guide to the journal designer in assembling the published image. The second drawing shows that both the stem and root thallus are better developed than is figured in the protologue, perhaps to save space. We credit the illustrations to Josef Pohl (for whom Pohliella was named) who provided scientific illustrations for Engler for 40 years (Anon. 2018).
The description above is based mainly on Engler (1926) and Cusset (1987).
Ghogue rediscovered Saxicolella nana in rapids in the Nyong near Mbalmayo, the presumed and likely type locality (details on the type label are indicative of this but not precise) which later in 2007 provided the source of material sequenced in phylogenies such as Koi et al. (2012). Kuetegue et al. (2019) studying the rheophytes of Cameroon did not refind this species but concluded that it should be assessed as Endangered EN B2ab (ii, iii).
When finalising this paper, recent records were found from Gabon on GBIF which are attributed to the species. However, from the images available, they seem to show several important differences from Saxicollela nana and are therefore provisionally identified in this paper as Saxicolella sp. A (see below under unplaced species). If our hypothesis is incorrect and these specimens are shown to be Saxicollela nana, the extinction risk assessment for this species is likely to be reduced from the proposed CR to EN. Butumia marginalis G. Taylor (1953: 55;1954: 124 -125).
Annual herb. Root rosette-like, 5 -12 cm diam., radiating from a central point, the distal ends ribbon-like, 1.5 -5 (-7.5) mm wide, rarely bifurcating. Stems of sessile leaf rosettes, numerous, spaced along the margins of the roots, (0.9 -) 1.  et al. 2000: 69). This publication also formally transferred the species from Butumia to Saxicolella. Ouedrago (2010) further states that "the species may also be present in Ghana and Niger but this needs to be confirmed". No evidence has been found to support this belief and it seems unlikely. Several other extremely rare, range-restricted and threatened species also occur at high altitude at Mt Oku (Cheek et al. 1997(Cheek et al. , 2000 e.g. Scleria cheekii Bauters (Bauters et al. 2018), Deinbollia onanae Cheek (Cheek et al. 2021b). Saxicolella marginalis is genuinely rare in Cameroon, since the first author has targeted searches for it at other falls and rapids in the Mt Oku area and not found it, although two other species of Podostemaceae have been found (Cheek et al. 2000). Nearcomprehensive botanical surveys in other locations S, W and E of Kilum-Ijim have failed to find additional locations although they brought to light several other species of Podostemaceae (see references above under Saxicolella ijim). The status of the subpopulation at the type locality in Nigeria is unknown. The Red Data Book of Cameroon Plants (Onana & Cheek 2011: 253) reassessed the species as EN B2ab(iii) since two locations (above) are recorded. Kuetegue et al. (2019) also assess the species as EN B2ab(iii) citing no new data. PHENOLOGY. Flowering and fruiting in late November to late December. ETYMOLOGY. The specific epithet derives from the rows of sessile shoots that line the margins of the roots. VERNACULAR NAMES. None known. NOTES. In describing this species as a new genus, Butumia, Taylor (1953: 57) stated: "Amongst African genera the plant is most closely related to Saxicolella and Pohliella, in each of which the flowers is unistaminate and erect within the spathella, but it differs from these genera in having entire rosulate leaves, much more shortly pedicellate flowers and complanate stigmas". At that time both genera were only known from their type species and their circumscription was incompletely known.
It is remarkable that the only known Cameroonian site for this species is only a few kilometres distant and at similar altitude (c. 1400 m), to the only known site for Saxicolella ijim (see under that species). However, this submontane altitudinal band of the Cameroon Highlands is immensely rich in endemic rangerestricted species. Other examples include Coffea montekupensis Stoff. (Stoffelen et al. 1997) and Impatiens etindensis Cheek & Eb. Fischer (Cheek & Fischer 1999).
RECOGNITION. Saxicolella deniseae is most similar morphologically to S. marginalis (G.Taylor) Cheek because both species share numerous rosulate sessile stems inserted at the margin of the ribbon-like roots. Saxicolella deniseae differs in that the three rosette leaves are shorter than the ovary; and in that they are comprised of a concave orbicular sheathing base with only a rudimentary linear apical blade, the apex rounded (in S. marginalis there are 5 -7 rosette leaves, each narrowly triangular-subulate, the sheathing base inconspicuous, the blade apex acute); in S. deniseae the stigmas are narrowly botuliform (in S. marginalis complanate). DISTRIBUTION. Guinée, Kindia-Télimelé, Konkouré River. SPECIMENS EXAMINED. GUINEA. Only known from the type specimen: Kindia-Télimelé Rd, 5 km S of crossing over the Konkouré R., on the Mayankouré near Lamba Sosso village, 10°24'15.4"N, 12°58'30.8"W, 180 m alt. fl. fr. 27 Jan. 2018, Molmou 1683, with Gbamon Konomou (holotype HNG; isotype K [K000593321]). HABITAT. Rapids in river. Growing with several other Podostemaceae species, for which specimens for identification are not yet available. 174 m alt. CONSERVATION STATUS. Currently Saxicolella deniseae is known from a single location which will be impacted by a hydro-electric project in the near future (see notes below). The area of occupation is estimated as 4 km 2 using the IUCN standard. Therefore, we here assess S. deniseae as Critically Endangered (CR B2ab(iiv)). It is to be hoped that further surveys will discover this species at other locations which are secure from threats such as hydroelectric projects otherwise this species is at high risk of extinction. PHENOLOGY. Flowering and fruiting in January as the water level drops with the advance of the dry season. When collected in late January some plants were still emerging from the water, and live and flowering, while others had already been exposed and had concluded fruiting and died. ETYMOLOGY. Named for Denise Molmou of Herbier National de Guinée, Université Gamal Abdel Nasser-Conakry, a leading botanist of Guinea who led the botanical survey team in which this species was discovered and who also collected the type specimen. NOTES. Together with Saxicolella futa, this is the most westerly species of the genus. The two species are geographical outliers of the rest of the genus which is centred in the Cross-Sanaga interval of SE Nigeria and W Cameroon. Saxicolella deniseae cannot be confused with S. futa since in the first the ribbon-like roots are 0.4 cm wide, unbranched, and bear numerous sessile shoots along the root margin, while in S. futa the roots are 0.03 -0.05 (-0.08) cm wide and bifurcate at regular intervals, bearing single sessile shoots only at the sinus of the bifurcations.
No associated endemic plant species are recorded near the location of Saxicolella deniseae because this part of Guinea is otherwise botanically unsurveyed.
RECOGNITION. Differing from other Saxicolella species with sessile spathellae (S. marginalis, S. deniseae) in the shoots not marginal, numerous, closely spaced, but single, at root bifurcations; leaves lacking stipules (not stipulate); root slender (<1 mm wide, repeatedly and regularly bifurcating along its length (not >1 mm wide, not or rarely bifurcating along its length). DISTRIBUTION HABITAT. Waterfalls in the former cloud forest zone. The species grows at three sites each with several other Podostemaceae-Podostemoideae species (see case study above) but flowers, fruits and dies in advance of those, growing higher up in the riverbed than all other species of that group. 710 -877 m alt. CONSERVATION STATUS. Known so far from three locations: 1) Chutes de Salaa near Dalaba (type locality); 2) above the Chutes de Kinkon; 3) Chutes de Kambadga, downstream of Kinkon. The AOO is estimated as 12 km 2 using the IUCN (2012)-preferred 2 km × 2 km grid cells and the extent of occurrence as 81.8 km 2 . Here Saxicolella futa is assessed as Endangered, EN B2ab(iii) since there are severe and imminent threats at all three locations. At locations 1) and 3) there are plans to build hydroelectric dams that are very likely to result in local extinction of the species (pers. obs. Cheek 2018). At site 2) the species is threatened by contamination of the water source by silt from run-off and by eutrophication due to contamination as the Kinkon river traverses the major town of Pita: fewer than 50 plants were seen at this location, and occupied a total area of < 3 m × 3 m. In the next 10 years this species is expected to be reassessed as Critically Endangered (CR), or even "possibly extinct". However, it is possible that further surveys may find additional sites for the species, which would be welcome. Seed banking and public awareness actions will be put in place as soon as possible. PHENOLOGY. Presumably germinating no earlier than May, with the beginning of the wet season, before which its habitat is dry. Flowering may begin as early as November after the conclusion of the wet season in October. By January fruit has formed and the plants are dead. ETYMOLOGY. Named for the Futa (or Fouta) Djalon highlands of the Republic of Guinea to which the species appears to be unique. NOTES. Saxicolella futa has a shorter growing season, (possibly completing its life cycle in only six months or less) than all other Podostemaceae species present at each of the three locations at which it is known, apart from Tristicha trifaria. Both these two species were dead, dried and in fruit when encountered in Jan. 2018, while all other species of Podostemaceae present at these sites were still alive and only, for the most part, just becoming exposed by falling water and were in the process of beginning their flowering. The fact that, in these three locations, S. futa was found only on rock surfaces 30 -100 cm above the waterlevel at which most other Podostemaceae (and all other Podostemoideae) occurred suggests that the species may have evolved into a niche to escape competition from those species. The extremely minute and inconspicuous stems, the diminutive ovaries, diminutive leaves, and slender roots in comparison with other species of the genus, may well all be reductions that have enabled it to occupy a niche where the growing season is short, being the last to be submerged, and the first exposed, of all the Podostemoideae niches present within its range.
Saxicolella futa is unique in the genus for its very slender (< 1 mm wide) ribbon-like roots, for the small size of its flowers, and for the position of the flowering shoots being only at the point of root bifurcations, so that the roots appear to terminate in a shoot before bifurcating. It is also unique in lacking evidence of stipules, but this is possibly concomitant with the reduction of leaf size. At maturity, more than 90 -95% of the bulk of the plant consists of root, more so than any of the other species of the genus where the shoots are much more numerous and also larger and longer. The flowers and fruits are difficult to discern.
Recent surveys in Guinea connected with the Guinea Tropical Important Plant Area programme (TIPAs) have discovered several other new species to science, all of which are threatened. Several of these are, like Saxicolella futa, also rheophytes, restricted to fast flowing water habitats, such as Prostrate herb with long, ribbon-like roots (black in life), infrequently bifurcating. Shoots sessile, forming 1-flowered leafy rosettes along margins of the root in rows, stem not visible. Leaves 3 -6 per rosette, linear, flattened, as long or longer than the flower, entire, very rarely bifid at apex, apex acute to obtuse, stipules not seen. Spathellum caducous, sometimes visible as a broken ellipsoid remnant. Pedicel cylindrical, stout and white or pink in life, as long as ovary and nearly as broad. Stamen with filament 1.5 -2 times as long as ovary, white to pink. Anthers with cells white, hemiglobose, isodiametric, semi-latrorse, each about as wide as filament. Gynophore absent. Tepals not seen. Ovary sessile narrowly ellipsoid, dull red, longitudinal lines not visible apart from commissure, orbicular in plan view (not laterally compressed). RECOGNITION. Differing from Saxicolella nana in that the root is long and ribbon-like with the shoots lacking visible stems, and arising along the margins of the root in rows (vs disc-like, crustose, the shoots with visible stems, arising from the centre of the root in a cluster); the leaves entire, linear, and not trifid from a point c. 1.5 mm from the base; the ovary sessile (the staminal filament inserted at its base), not with a distinct gynophore; fruit 8-ribbed (not 6-ribbed). DISTRIBUTION. Gabon, Ogooué River and its affluent the Ivindo. Once this species is formally published it will be possible for it to be Red Listed, most likely as Endangered EN B1ab(iii) since three locations are known with threats and extent of occurrence is calculated as 881 km 2 . Although the three points are widely separated, they plot in a straight line forming a nearly linear polygon. Texier 2321 is stated to be "frequent on rocks on Ivindo River" but it is close to the Kongou falls. In 2007 the President of Gabon announced that a hydro-electric dam will be built at the Kongou falls inside the Ivindou National Park to support the Belinga Iron Ore project (Wikipedia, Kongou Falls, 2020). When implemented it can be expected to result in extinction of this species at this location as such projects have caused extinction of Podostemaceae elsewhere in Africa (Cheek et al. 2017b). While threats of this nature have not been found for the other two locations indicated, there is no scope for complacency. Six new large hydroelectric projects are planned in Gabon (Makoni 2020) and the total rises to 39 if potential and smaller projects are included (Cutler 2019). Apart from these, falls and rapids are the ideal sites for placing smaller hydro-power projects needed for future development projects. The location of Nguimbit 26 is close to an undeveloped iron ore project in Congo-Brazzaville, while Boupoya 1952 is about 10 km E of Lastoursville, one of Gabon's major towns. PHENOLOGY. Collected in flower in Feb. and July and August (dry seasons). VERNACULAR NAMES. None are recorded. NOTES. The first author came across this taxon when doing final checks for additional records of the genus on GBIF.org before finishing the manuscript for this paper. The three specimens were identified as Saxicolella nana respectively by Rutishauser, Bidault & Rutishauser, Bidault & Mesterhazy. An excellent image of Texier et al. 2321 and ten for Boupoya 1952 are available online through GBIF and Tropicos, (http://www.tropicos.org/ ImageDownload.aspx?imageid=100597656 and h t t p : / / w w w . t r o p i c o s . o r g / I m a g e D o w n l o a d . aspx?imageid=100891363 respectively) but none for Nguimbit 26 which is included on the assumption that it matches the two specimens which had been named identically.
Inspection of these images shows several characters that differ from those of S. nana which are given under "Recognition" above. No scale was available, so the description is made without giving dimensions, entirely on the basis of the online photos. Therefore, a full formal description of this taxon is still advisable, including observation to confirm the expected unilocularity of the ovary. Since there are limitations to studying specimens only from images which can be misleading, Saxicolella sp. A is treated as unplaced here. While it appears to belong to subgenus Butumia, it is indeed similar to S. nana. It is possible that when the specimens themselves are studied, the dissimilarities with S. nana will be reduced and that Saxicolella sp. A will be reduced to synonymy. Yet this seems unlikely given the number of qualitative morphological differences enumerated above, sufficient to result in separation at subgenus level. The specimens themselves could not be reviewed on loan because of technical difficulties due to their transfer from BRLU to ZT (Bidault pers. comm. to Cheek June 2021).
Nguimbit 26 was collected at a point so close (< 4 km) to the border with Congo-Brazzaville that it maps just inside that country on Tropicos. Since the river concerned extends far into that country it is very likely that the species occurs there and that it is not endemic to Gabon.
Gabon is seeing an upsurge in description of new species with 37 published in 2019, second only after Cameroon in tropical Africa, many of these being rare, range-restricted and endemic (Cheek et al. 2020b), joining those already documented narrow and threatened endemics as such Diospyros rabiensis Breteler (Breteler 1994), Whitfieldia purpurata Heine and Whitfieldia rutilans Heine (Grall & Darbyshire 2021), some of which are already extinct such as Pseudohydrosme buettneri Engl. (Cheek et al. 2021a).

Discussion
The massive range extension of Saxicolella to the Guinea Highlands (due to recent collections representing new species to science) was unexpected, but recently has also been seen in other genera which were also considered previously to be confined to Lower Guinea (the region around the Bight of Biafra, far to the east). These genera are Ternstroemia Mutis ex L.f. and Talbotiella  In contrast, the range extension to Angola was not a surprise given that the first collection of the new taxon was made nearly a century ago, and its identification as a new species of Saxicolella more than 40 years ago. The delay in its formal publication speaks of both the shortage of taxonomists to do the work needed and the barriers to taking scientific discoveries through to formal publication. Despite the dramatic rise in numbers of the genus from three to eight, we still know very little of this genus in comparison with other African podostemoid genera. Additional species unknown to science of this and other Podostemaceae genera almost certainly remain to be discovered for science in Africa.
Until species are known to science, they cannot be assessed for their conservation status and the possibility of protecting them is minimal. About 2000 species of vascular plant have been described as new to science each year for the last decade or more (Cheek et al. 2020b). To maximise the survival prospects of range-restricted species such as Podostemaceae, there is an urgent need not only to document them formally but to assess them for their extinction risk, using the widely accepted IUCN Red List of Threatened Species (Bachman et al. 2019). Despite rapid increases over recent years in numbers of plant species represented by assessments on the Red List, the vast majority of plant species still lack such assessments (Nic Lughadha et al. 2020) and Podostemaceae are no exception.
As a global standard, the IUCN Red List supports the safeguarding and sustainability frameworks used by businesses and their major lenders (Bennun et al. 2018;Juffe-Bignoli et al. 2016). For example, clients of the International Finance Corporation (World Bank Group) are required to use the Red List to inform project risks and to refrain from activities leading to a net reduction in populations of species assessed on the Red List as Endangered (EN) or Critically Endangered (CR), over a reasonable timescale (IFC 2019). This is relevant because the International Finance Corporation have been linked with funding the hydro-electric projects that are major extinction risks for African Podostemaceae species.
Documented extinctions of plant species are increasing (Humphreys et al. 2019) and recent estimates suggest that as many as two fifths of the world's plant species are now threatened with extinction (Nic Lughadha et al. 2020). That 100% of the species of a plant group should be threatened as documented in this paper for Saxicolella is unusual and may be unprecedented.
Extinctions of plant species are becoming documented throughout the range of Saxicolella. In Cameroon, the centre of diversity of Saxicolella, the narrowly endemic species Oxygyne triandra Schltr. and Afrothismia pachyantha Schltr. are now known to be globally extinct (Cheek & Williams 1999;Cheek et al. 2018c;Cheek et al. 2019c) and Vepris bali Cheek was extinct even before it was known to science (Cheek et al. 2018d). Focussing on Podostemaceae, in Angola Ledermanniella lunda Cheek (Cheek et al. 2015) is thought to be globally extinct and in Guinea Inversodicraea pygmaea G. Taylor (Cheek & Magassouba 2018;Couch et al. 2019), both due to hydro-electric projects.
Efforts are now being made to delimit the highest priority areas in Cameroon for plant conservation as Tropical Important Plant Areas (TIPAs) using the revised IPA criteria set out in Darbyshire et al. (2017), as has already been completed for Guinea (Couch et al. 2019). These two countries have five of the eight species of Saxicolella between them and species such as S. marginalis and S. ijim (Cameroon) and S. futa (Guinea) are already included or set to be included in TIPAs. However, there is no such TIPA programme yet for Angola, Gabon and Nigeria, the other countries with species of Saxicolella. Even inclusion in an officially protected area is not a guarantee that a species will be safe from extinction. As in the case of Saxicolella sp. A of Gabon (see above), its most threatened location is inside a National Park, because a major Hydro-electric project is planned there. If new hydro-electric projects continue to be constructed there seems no doubt that global extinctions of Podostemaceae species will continue. Alternatives to protecting species other than in their natural habitat are not yet available. While Podostemaceae have orthodox seed that can be banked, there are as yet, no documented cases of new populations being established artificially either experimentally or in the wild.
collections above cited were made. Jean Michel Onana, Florence Ngo Ngwe, Eric Nana, and Jean Betti Lagarde, their successors, are thanked for support in maintaining this partnership.
Janis Shillito typed the manuscript. Eimear Nic Lughadha and two anonymous reviewers gave painstaking advice on earlier versions of the manuscript. Aurelie Grall is thanked for attempting to make contact with Colette Cusset. Julia Buckley of the Department of Library, Art and Archives, Royal Botanic Gardens, Kew is thanked for arranging with the estate of Margaret Stones for permission to re-use Figures 1 and 5, which were originally labelled as Pohliella flabellata and Butumia marginalis and first appeared in Taylor (1953 and1954) respectively. Andrea Hart of the National History Museum also supported our re-use of Figure 1. Robert Vogt, Collections manager (B) is thanked for expediting the loan to K of the type of Saxicolella nana. We thank Joaquim Santos, Collections Manager, Herbarium of the University of Coimbra, Portugal (COI) for help accessing the electronic collections databases of COI and LISC.
The authors declare that they have no conflict of interest.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/ licenses/by/4.0/ .