Discovering Karima (Euphorbiaceae), a New Crotonoid Genus from West Tropical Africa Long Hidden within Croton

Croton scarciesii (Euphorbiaceae-Crotonoideae), a rheophytic shrub from West Africa, is shown to have been misplaced in Croton for 120 years, having none of the diagnostic characters of that genus, but rather a set of characters present in no known genus of the family. Pollen analysis shows that the new genus Karima belongs to the inaperturate crotonoid group. Analysis of a concatenated molecular dataset combining trnL-F and rbcL sequences positioned Karima as sister to Neoholstia from south eastern tropical Africa in a well-supported clade comprised of genera of subtribes Grosserineae and Neoboutonieae of the inaperturate crotonoid genera. Several morphological characters support the relationship of Karima with Neoholstia, yet separation is merited by numerous characters usually associated with generic rank in Euphorbiaceae. Quantitative ecological data and a conservation assessment supplement illustrations and descriptions of the taxon.


Introduction
The environmental impact assessment of the Bumbuna-Yiben Hydroelectric Dam project in Sierra Leone which was led by Xander van der Burgt of the Royal Botanic Gardens, Kew, with colleagues from the National Herbarium of Sierra Leone provided numerous plant collections [1]. Among the resultant herbarium specimens, one (Momoh 94) proved of great interest. Momoh 94 clearly represents a member of the Euphorbiaceae since its fruits bear three, bifurcate, persistent styles and are tricoccal, dehiscing septicidally, leaving a central vascular column. The presence of a single ovule in each carpel excludes the material from Phyllanthaceae [2,3], as the presence of petals and absence of white exudate excludes the material from Euphorbiaceae-Euphorbioideae [2,3]. However, although it was eventually matched at K with material named as Croton scarciesii Scott-Elliot, including both syntypes, these authors noted that the main morphological characters of Croton scarciesii disagree with most typical Croton L. features and matched no known genus. Independently at MA, in the context of a taxonomic revision of African Croton, two of the coauthors arrived at the same conclusion. Using evidence from morphology, pollen, molecular phylogenetics and geography, we describe and illustrate a new genus, erecting the name Karima and place it within Euphorbiaceae-Crotonoideae. We compare the new monotypic genus Karima to Croton and to its closest relative in a molecular phylogeny including a subset of other inaperturate crotonoid taxa. A complementary field study of the ecology of Karima was also conducted at the Taia River in Sierra Leone. Despite recent phylogenetic and taxonomic studies, our finding highlights the still problematic classification of Euphorbiaceae, with many poorly known species, issues with generic delimitation, and new genera being described.

Ethics statement
Momoh 94, which triggered this paper, was collected during field studies for an Environmental Impact assessment of the Bumbuna hydroelectric dam, as part of a survey co-ordinated and managed by the international environmental consultancy company ERM for the engineering company Joule Africa. The National Herbarium of Sierra Leone, Njala University, issued a permit to collect herbarium specimens in this context on 9 th May 2014 to the leader of the botanical studies, van der Burgt. A further permit to export herbarium specimens was issued on 3 rd June 2014.
Field studies of Karima at the Taia River reported in S2 Appendix were conducted by the National Herbarium of Sierra Leone, which has the national statutory responsibility for study of the vegetation and plant species of Sierra Leone and requires no individual permits in order to conduct field surveys.
The area of the Bumbuna hydroelectric project is controlled by the state of Sierra Leone. The study area of the Taia river at Njala University is not privately owned, nor protected. Karima scarciesii (Synonym Croton scarciesii) is not a protected species.

Pollen study
Pollen morphology has been traditionally used in Euphorbiaceae taxonomy for delimitation of infrafamilial taxa [3]. Pollen samples were collected from Momoh 94 (FBC, K). Whole, unacetolysed, dehisced anthers from the herbarium specimen were placed on a stub and sputtercoated with platinum in a Quorom Q150T coater and examined in a Hitatchi 54700 scanning electron microscope.

Taxon sampling and molecular data
Guided by the results from pollen morphology, we limited our taxon sampling for phylogenetic analyses to the two clades of inaperturate crotonoids ("C1" and "C2" sensu Wurdack et al. ([2]: figure 4). Based on the results of a preliminary Maximum Parsimony phylogenetic analysis (data not shown) of all sequences of inaperturate crotonoids included in [2] using PAUP Ã v.4.0b.10 [4], our final molecular data matrix consisted of all (26) species of clade C2 of the combined analysis of Wurdack et al (2005), three representatives from clade C1, and newly generated sequences of Neoholstia tenuifolia and two accessions of Karima scarciesii (Momoh 94, Sierra Leone and Jongkind 4208, Cote D'Ivoire). A non-crotonoid species, Nealchornea yapurensis Huber (Euphorbioideae), was also included as the most distant outgroup. A complete list of accessions, including newly generated sequences and their respective Genbank numbers, are listed in S1 Appendix.
For Karima scarciesii and Neoholstia tenuifolia samples, DNA extraction, polymerase chain reaction (PCR) amplification, and sequencing followed the laboratory procedures described in [5]. For the region comprising the trnL intron and trnL-F intergenic spacer (hereafter referred to as "trnL-F") we used the primers indicated in [2]. The rbcL region was partially sequenced (first half) using primers rbcL1F (ATG-TCA-CCA-CAA-ACA-GAR-AC) and rbcL724R  (TCG-CAT-GTA-CCC-TGC-AGT-TGC). The new sequences were assembled and edited using Staden Package v.2003.0b1 [6] and were deposited in GenBank (S1 Appendix). Previously published sequences were obtained from NCBI GenBank (www.ncbi.nlm.nih.gov). Sequences were aligned manually with MacClade v.4.08a [7], following the similarity criterion as suggested by [8]. Alignment of the rbcL data was straightforward and gap-free, whereas the trnL-F matrix required the inclusion of several gaps. The final alignment of each matrix (rbcL, trnL-F) was end-trimmed to remove most of the characters with missing data. The two matrices were combined into a single data set with two partitions corresponding to each molecular marker (rbcL, trnL-F). The combined matrix is available at TreeBase (http://purl.org/phylo/ treebase/phylows/study/TB2:S18891).

Phylogenetic analyses
The concatenated matrix with two data partitions was analyzed using a Bayesian approach. MrModeltest v.3.7 [9] was used to estimate the most appropriate model of sequence evolution for each data partition under the Akaike Information Criterion (AIC) [10]. The HKY+G and GTR+I+G models were selected for the trnL-F and rbcL datasets, respectively.
Bayesian inference, based on a Markov Chain Monte Carlo (MCMC) approach [11], was conducted in MrBayes v.3.2.1 [12]. Base equilibrium frequencies, instantaneous substitution rates, and among-site rate variation values were estimated independently for each partition on shared topologies. Two runs of 10 million generations were conducted, and trees were sampled every 1,000 generations. Each run consisted of four independent Markov chains initiated from a random starting tree and using the default temperature (0.2). The resultant Ln likelihood and model parameters from the MCMC runs were inspected using Tracer v.1.5 [13] to determine run convergence and stationarity as indicated by the effective sample size (ESS) values, which should be higher than 100. One-fourth of the MCMC samples from each run was discarded as "burn-in." The remaining trees were pooled into a 50% majority rule consensus tree with clade credibility values. The consensus tree was visualized and edited in FigTree v.1.3.1. Five hundred maximum likelihood (ML) [14] bootstrap replicates, using the model GTR + G and the same data partitions as in the Bayesian analysis above, were implemented in RAxML v.7.0.3 [15] to generate an additional measure of clade support.

Taxonomic treatment
A comparative morphological study of the new taxon was conducted using collections from the following herbaria: B, BM, BR, E, FBC, FHO, GC, IFAN, K, LISC, MO, P, SL, UCJ, US, WAG, Z. Codes for cited herbaria follow Index Herbariorum [16]. Cited specimens which have been seen by one or more author are annotated "!", prefixed by a seven digit number for those with barcodes. The overall morphology was documented, described and illustrated following botanical standard procedures as documented in [17]. Information about habit, habitat, and distribution was taken from specimen labels and field observations. Specimens and protologues of all inaperturate Euphorbiaceae genera, and especially genera indicated as being closely related to Karima according to the phylogenetic results, were studied and compared with the new taxon. The conservation status was evaluated using IUCN criteria [18]. fieldworkers being swept away by the currents. The plots were placed around patches of the study species to obtain quantitative data on plant densities, spacings and heights, which were measured by metre rules and tape measures. Identifications of Karima were confirmed at the National Herbarium of Sierra Leone at the University of Njala, and further confirmed at K using photographs of the study subjects. All photographs associated with this paper were taken during this study.

Nomenclature
The electronic version of this article in Portable Document Format (PDF) in a work with an ISSN or ISBN will represent a published work according to the International Code of Nomenclature for algae, fungi, and plants [19], and hence the new names contained in the electronic publication of a PLOS article are effectively published under that Code from the electronic edition alone, so there is no need to provide printed copies.
In addition, the new names contained in this work have been submitted to IPNI, from where they will be made available to the Global Names Index. The IPNI LSIDs can be resolved and the associated information viewed through any standard web browser by appending the LSID contained in this publication to the prefix http://ipni.org/. The online version of this work is archived and available from the following digital repositories: PubMed Central, LOCKSS.

Pollen description
The pollen grains are spheroidal, inaperturate and have a crotonoid exine sculpture pattern with supratectal subunits attached to the upper edge of low, circular muri each enclosing a central lacuna. The diameter of the grains is (35-)38-40 μm as measured by SEM from a sample of 20 pollen grains. The subunit rings are (2.8-)3.5-4(-5.2) μm diam., with (5-)7(-9) subunits per ring. Subunits are ovoid-spheroidal, 1-1.3 x 1-1.5 μm, strongly longitudinally ridged, with 12-14 ridges arising from the base, several of which unite with each other before converging at the obtuse apex. The floors of the central lacunae each have 7-20 scattered granules, each c. 0.1 μm diam. (Fig 1).

Molecular phylogeny
The concatenated dataset combining trnL-F and part of rbcL contained 33 accessions and 2728 aligned positions, of which 202 were parsimony-informative. In the case of rbcL we were only able to amplify part of this region for the newly sampled taxa, therefore we only used the corresponding fragment of the published sequences downloaded from Genbank. The trnL-F dataset with 1365 aligned positions has more parsimony-informative characters (118) than the rbcL data set with 666 aligned positions and 84 parsimony-informative characters. The rbcL fragments for the two accessions of Karima scarciesii are identical. The two trnL-F sequences are nearly identical and differ only in one position, however, one of the accessions (Jongkind 4208) has 55 characters missing at the beginning of the sequence. Preliminary Bayesian analyses of the individual datasets did not show evidence of topological incongruence between the two genetic markers (results not shown). Given the lack of obvious conflict between the topologies resulting from individual matrices, and following the same strategy as [2], we analyzed the combined dataset (rbcL + trnL-F). The Bayesian consensus tree (Fig 2) resulting from the analysis of the concatenated matrix shows a phylogenetic structure congruent with the phylogenetic analyses of [2] (their figures 1, 2, 3). Representatives of clades C1 and C2 are sister groups with 100% Bayesian posterior probability (Fig 2). As in [2], clade C2 is highly supported but resolution at the backbone is quite low.

Morphological differences
The morphological differences of Karima from Croton sensu stricto are extensive (Table 1). It is surprising that no other researchers have remarked upon the absence in Croton scarciesii (that is, the new genus, Karima) of the key diagnostic characters of Croton, such as lepidote scales or stellate hairs, the presence of nectary glands at the junction of the petiole and leaf blade, a crotonoid inflorescence (thyrsoid inflorescence with female and male flowers at base, male towards apex), and inflexed staminal filaments in flower buds.
The placement of Karima near Neoholstia was reached by molecular analysis but is strongly supported by morphology ( Table 2). The two genera share stalked, glandular, red-apexed glands on the bracts (unusual in Crotonoideae), and leaf texture and venation are also similar, as is the general indumentum type (simple, erect, robust, translucent hairs). The tuberculate fruits of both genera are also similar, apart from differences in the styles. However there are major qualitative characters that divide the two taxa. These characters are usual in serving to distinguish genera, rather than species, in Euphorbiaceae as can be seen in [21]. These major qualitative characters concern differences in gross architecture, in sexuality, in inflorescence structure and placement, in presence/absence of both bud-scales and of leaf heteromorphy ( Table 2).

Ecology
The results of the field studies are in S2 Appendix.

Former placement in Croton
Karima was erroneously misplaced in Croton as Croton scarciesii by Scott-Elliot [24]. The protologue is part of a paper [24] describing botanical novelties resulting from his participation in the survey of the boundary between Sierra Leone and the then French Guinea (now Republique de Guinée), also known as Guinea-Conakry. In that paper most of the species are described from his specimens by the corresponding family specialists at Kew. Where no family specialist was available, Scott-Elliot himself took on responsibility for placing and describing the taxa that he considered to be new to science. Evidently, at that time, Kew had no Euphorbiaceae specialist. The placement of the taxon in Croton appears never to have been challenged. Opportunities to detect the error were missed when the Croton accounts for the Flora of West Tropical Africa were written by the respective editors Hutchinson [25] and, in the revised edition, by Keay [26]. It is possible that they were under pressure of time to complete these accounts and so could not critically review Scott-Elliot's generic placement of this species. Perhaps they attributed the obviously anomalous morphology within Croton of the species to its rheophytic ecology. It is also possible that they did not dissect the flowers, which would have revealed additional characters that militated against placement in Croton (Table 1).

Placement and establishment of Karima based on multiple evidence
Despite the strong morphological separation of Karima from Croton (Table 1), examination of the pollen indicates that grains are spheroidal, inaperturate and with crotonoid exine sculpturing. This characterises subfamily Crotonoideae, and together with the gross morphological characters of the indumentum, leaf and flower and utilisation of the keys in Radcliffe-Smith [21] and Webster [27], suggests that Karima is best placed in Crotonoideae in the region of Codiaeae and Aleuritidieae. The molecular phylogeny also confirms that Karima is part of Crotonoideae, and that it belongs to the inaperturate crotonoid clade C2 of Wurdack et al. [2]. This clade, emended in this paper (Fig 2), is comprised of all sampled members (27 genera  The DNA sequence data indicate that Karima scarciesii is most closely related to the southeastern tropical African Neoholstia tenuifolia (Fig 2). We did not include the new taxon within the monotypic genus Neoholstia [21] because morphological differences between the two taxa are considerable (see above and Table 2). Furthermore, the current phylogeny of the C2 clade of Wurdack et al. [2] is only sampled at the level of a single species per genus and so provides little resolution to assess generic delimitation with confidence. Phylogenetic analyses including additional species from each genus and using more informative molecular markers are still needed to better address taxonomic delimitation within clade C2 of the inaperturate crotonoids.

Ecology
Totally submerged in the wet season, exposed in the dry season, with a patchy distribution, in river beds, mainly rocky, in the open, less usually on sand or silt, or in the shade (when fewer stems per plant, leaf-blades larger, petioles longer) of riverbank trees, sometimes locally common, often with other rheophytes, especially Podostemaceae and Rotula aquatica (Boraginaceae), usually in the evergreen forest zone (less usually in predominantly wooded grassland At study sites of patches of Karima at the Taia river (S2 Appendix), plants occur at densities of 24-78 per 100 m², with spacings between clumps of plants between 20-400 cm. See detailed quantitative and qualitative ecological field data in S2 Appendix.

Local names and uses
Taduεme (Mende language, Sierra Leone) and àsá (Temne language, Sierra Leone), both according to N.W. Thompson, recorded in Burkhill [36]. No uses are recorded but we note that this species has potential applications in reducing erosion and in stabilising the banks of fast-flowing water-courses. However, Burkhill [36] also states that in northern Ghana this shrub "offers harbour for the tsetse fly", citing Vigne 3864, (K).

Etymology
From Karim meaning generous (Arabic), one of the 99 names of God, used as a name in West Africa, and especially in Sierra Leone, specifically commemorating Dr Karim, Dean of Science at Fourah Bay College, Freetown, Sierra Leone.

Conservation
Karima scarciesii is known from c. 26 locations (see map , Fig 4), having an extent of occurrence (in the sense of IUCN [18]) of 890,258 km² (calculated with Geocat, Bachman et al. [37]) and an area of occupancy of c. 30 km² (calculated using the IUCN [18] advised cell size on 1 km² for riverine species, based on 30 sites being known).
The major and only known threat to Karmia scarciesii is from hydroelectric projects which permanently submerge locations for this species, as has happened at Kete-Krachi in Ghana (Volta Dam), more recently in Sierra Leone at the Bumbuna Phase 1 dam (Hawthorne pers. comm. 2013) and, imminently, the second phase Yiben Dam (van der Burgt pers. comm. 2014). Total inundation and therefore local extinctions are also expected shortly to happen in Cote D'Ivoire, at Sassandra River, Chutes de Soufré, and at Chutes de Naoua, also at the Black Volta River at Bui, Ghana. It is likely that in future further hydroelectric projects will destroy additional locations for this species. Despite these threats, the taxon is here assessed as Near Threatened since the number of locations thought to survive exceeds the threshold of 10 usually required before an assessment of threatened is made with Criterion B of IUCN [18]. A detailed study has not been made of the survival of the species at its historical sites. Should this be done it is quite possible that Karima scarciesii might be shown to have been lost at >30% of these sites in the last 90 years, in which case, should the generation time be found to be as long as 30 years, the taxon would be eligible for reassessment as Vulnerable under Criterion A of IUCN [18]. There is insufficiently detailed population data to make IUCN assessments under either Criterion C or D.

Morphological variation
Material illustrated from Senegal in the extreme western part of the range by Berhaut [30,31] is atypical in having much larger leaf-blades than is normal elsewhere within the range, and also unusual crotonoid inflorescences. These are bisexual with the female at the base of the raceme, the remaining flowers being male (instead of having separate male and female inflorescences). This material merits further study since it may be a second species of the genus rather an interesting local variant.
Supporting Information S1 Appendix. Species sampled with GenBank accession numbers for rbcL and trnL-F sequences, respectively. Voucher information is given for newly sequenced species (boldfaced).