Taxonomic revision of the Mesoamerican genus Spathacanthus (Justicieae, Acanthoideae, Acanthaceae)

Abstract Spathacanthus is a Mesoamerican genus that occurs in tropical and temperate regions from southern Mexico to Costa Rica; its taxonomy has not been updated for two decades. In view of the fact that a new species has been discovered and that the interspecific affinities in this genus have not been addressed to date, the present study aims to revise the genus Spathacanthus. Specimens of plants of this genus collected from across the distribution range and deposited in herbaria and digital databases were reviewed. In parallel, a cladistic analysis was carried out, based on morphological characters in order to examine relationships between species. Four species of Spathacanthus were recognised: one endemic to Costa Rica, another micro-endemic to Veracruz in Mexico, one more restricted to the forests of Mexico and Guatemala and the last one more widely distributed. Reflecting the previously limited knowledge of the group, many of the specimens that we studied had been misidentified. A key to differentiate these species is provided, supplemented with photographs, drawings and other illustrations, morphological descriptions, synonymy and ecological data. Results, presented here, extend the distribution range of some taxa and a distribution map is presented. The cladistic analysis recovered the genus as monophyletic, showing that S. hoffmannii and S. hahnianus are sister taxa and S. magdalenae was found to be more closely related to S. parviflorus. These plants are vulnerable to degradation and habitat loss.


Introduction
The family Acanthaceae comprises more than 4,000 species and some 230 genera widely distributed throughout the world. They mostly thrive in tropical and subtropical areas, with the Indo-Malay, African (including Madagascar), South American and Mexican-Central American regions as primary diversity centres. The species of this family thrive in virtually all intertropical habitats, except for high-mountain areas (above 3,000 metres a.s.l.).
Mesoamerica is one of the seven American biodiversity hotpots where species face a high risk of extinction, associated with accelerated deforestation and global warming (Myers et al. 2000;Malcolm et al. 2006). Consequently, an enormous portion of the Mesoamerican flora may well become extinct before species can be described, much less characterised taxonomically and in terms of evolutionary relationships.
Mexico is a major centre of species richness, diversity and endemism for the family Acanthaceae (Daniel 2007), which ranks fifteenth in terms of floristic diversity in the country. Acanthaceae represent 1.6% of the total floristic richness of Mexico (Villaseñor 2016), with around 400 species (47 genera) of herbs, shrubs and trees; of these, 58.8% are endemic to Mexico (Villaseñor and Ortiz 2014), while seven of the 38 native genera documented in Mexico are also endemic to this country. Due to high species richness, no study has covered the entire family and new species are still being discovered (Daniel 2007).
Spathacanthus Baill. is a small genus of shrubs and trees; plants are distinctive amongst Mesoamerican Acanthaceae by having very large capsules and seeds, both of which are amongst the largest known in the family. The genus comprises four species distributed in moist to wet forests of Mexico and Central America. In Mexico, the genus is represented by three native species. Two of them were treated by Daniel (1995) -Spathacanthus hahnianus Baill. and S. parviflorus Leonard -and also inhabit Guatemala, with S. hahnianus reaching Honduras. Spathacanthus magdalenae Cast.-Campos was recently discovered in a riparian forest in Veracruz (Castillo-Campos et al. 2013). Meanwhile, Spathacanthus hoffmannii Lindau is restricted to Costa Rica (Lindau 1895a). Although currently it is accepted that the genus belongs to subfamily Acanthoideae, tribe Justicieae, Pseuderanthemum lineage (McDade et al. 2000), its generic relationships have not been resolved yet and the correct placement of the genus Spathacanthus has not been thoroughly assessed.
In addition to issues at the generic level, there is only one previous taxonomic treatment of the entire genus (Daniel 1999). The discovery of a fourth species (Castillo-Campos et al. 2013) clearly calls for re-assessment of the genus; also, interspecific affinities have yet to be addressed. Therefore, the present study aims to provide an updated taxonomic revision of the genus Spathacanthus, particularly in Mexico, where three of our species occur.

Methods
This study was based primarily on herbarium specimens deposited in the following herbaria: BIGU, CHAPA, CR, ENCB, IEB, IBUG, MEXU, TEFH, USJ and XAL, as well as field observations of living plants. In total, 97 Spathacanthus specimens were examined. Digital specimens from the herbaria at CAS, DS, DUKE, F, K, MO, NY, TEX, US and UT were also consulted, all of them available via the TROPICOS (http://www.tropicos.org) and JSTOR Global Plants (https://plants.jstor.org/collection/TYPSPE) websites. Herbarium acronyms follow Thiers (2019). Protologues and type material of all species and synonyms involved were consulted. This allowed verifying or, where appropriate, re-identifying specimens from morphological features. We used a stereomicroscope (Stemi 2000-C, Carl Zeiss, Cd. Mx., Mexico) to study morphological variations amongst plants of all four species. New descriptions prepared for each species reflect the careful comparative work undertaken.
The descriptions, newly presented here, include taxonomic and nomenclatural synonyms of each taxon, along with phenological information, habitat, distribution, elevation and vegetation types. The names of the species were verified on TROPICOS, The International Plant Name Index (https://www.ipni.org/) and Catalogue of Life 2020 Annual Checklist (http://www.catalogueoflife.org/col). Endemism was determined according to the descriptions in specialised studies, floristic listings, distribution data as reported in international databases (IPNI, TROPICOS, Global Biodiversity Information Facility (https://www.gbif.org/)) and information from herbarium specimens. A species was considered endemic if its distribution range is restricted to a given territory and as micro-endemic if exclusive to a limited site within a country. Distribution maps were prepared using geographic data via ARCGIS v.10.2.2 (ESRI 2014), based on the collection data compiled from herbarium specimens.
We also conducted a cladistic analysis, based on morphological characters as a first approach to explore relationships amongst species. We examined all four species of Spathacanthus as the in-group, plus Odontonema callistachyum Kuntze and Odontonema cuspidatum (Nees) Kuntze, as out-groups. Both genera -Odontonema and Spathacanthus -have been recognised as being the closest American relatives within the Pseuderanthemum lineage, in the tribe Justicieae (McDade et al. 2000). Based on the work of Daniel (1999) and after a detailed review of herbarium specimens, morphological characters derived from traditional external morphology were selected to build a matrix of morphological traits for cladistic analysis. A total of 15 binary and multistate characters were scored; the character by taxon matrix was built in MESQUITE v.3.2 (Maddison and Maddison 2017).
Character definitions and states are given in Appendices 1 and 2. A maximum-parsimony analysis (MP) was performed using PAUP* v.4.0a (build 165) (Swofford 2002); to this end, an exhaustive search was conducted involving 15 unordered characters given equal weight; multistate taxa were interpreted as polymorphisms. A total of 1000 replicates were carried out, auto-increased by 100, collapsing zero-length branches and with MulTrees in effect. The shortest trees were saved and a majority-rule consensus tree was produced. Statistical branch support was determined with 1000 non-parametric boot-strap (BS) iterations each with ten replicates, saving five trees per replicate. In addition, a Jackknife analysis (JK) was carried out with a 50% deletion and a full heuristic search; groups with a frequency > 50% were retained, with 1000 replicates and using the TBR branch-swapping algorithm. Finally, the tree was visualised with FIGTREE v.1.4.

Key to
Distribution, habitat and phenology. Spathacanthus magdalenae is endemic to southern Mexico where it is restricted to the riparian vegetation of central Veracruz (Fig. 1B). It is frequent near rivers or humid canyons in cloud forests and oak forests in tropical to temperate zones, at 1300 to 1700 m a.s.l. Flowering takes place from November to March; mature fruits can be found from January to March. Description. Small trees or shrubs, up to 8 m tall, branched, internodes glabrous or nearly glabrous. Stems subquadrate to somewhat compressed when young, nodes sparsely pubescent with rigid to flexible eglandular trichomes. Leaves petiolate, petioles 5-45 mm long, blade elliptic to obovate-elliptic, 20-155 mm × 11-100 mm, apically acute to acuminate, basally acute, marginally entire, both surfaces glabrous or with eglandular trichomes along main veins, these conspicuous on abaxial surface. Inflorescence terminal, rarely axillary, rachis glabrous to pubescent with eglandular trichomas; bracts triangular to subulate, 1.3-6 mm × 0.8-1.4 mm, abaxial surface glabrous or pubescent like rachis; bracteoles triangular to subulate, 1-4 mm × 0.6-1 mm, abaxial surface glabrous or pubescent like rachis. Flowers pedicellate, pedicels 1-13 mm long,

Phylogenetic relationships
A total of 15 characters were analysed, 14 (93.3%) of which were parsimony-informative. The MP analysis resulted in five most parsimonious trees of 24 steps, a consistency index (CI) of 0.88 (excluding uninformative characters) and a retention index (RI) of 0.80. The majority rule consensus tree is shown in Fig. 10.
The monophyly of the genus Spathacanthus was strongly supported (BS and JK = 100%). Within the genus, two clades were recovered, the first one includes the species S. hahnianus and S. hoffmannii as more closely related, with moderate support (BS = 60% and JK = 50%) (Fig.10). In the second clade, two of the white flower species, S. parviflorus, and S. magdalenae were recovered as sister taxa, with support values of BS = 60% and JK = 52%.

Discussion
Spathacanthus is composed of four recognised species, all of which are restricted to Mesoamerica. Spathacanthus hahnianus is the most widely distributed species, spanning from southern Mexico and Guatemala to Honduras. Spathacanthus parviflorus occurs only in southern Mexico and Guatemala; S. magdalenae and S. hoffmannii are geographically restricted, the first as micro-endemic to the centre of Veracruz, Mexico and the second as endemic to Costa Rica, where it is widely distributed throughout that country. Narrowly distributed species like S. magdalenae are of conservation concern as they may be threatened by the effects of environmental deterioration and habitat loss (Castillo-Campos et al. 2005;Mooers and Redding 2009). Despite this, none of these species has been formally assessed using the IUCN Red List standard. This highlights the need for more studies on each of the species in the group that allow us to know if it is necessary to take measures for conservation.
Plants of Spathacanthus occur mainly in tropical rainforests and cloud forests and at elevations between 100 and 2,300 m a.s.l. Of the four species, S. hahnianus and S. hoffmannii thrive in areas below 1,000 m a.s.l. Plants occur less commonly in higher elevation temperate forests. Particularly, S. parviflorus is restricted to cloud forests, S. hoffmannii inhabits humid tropical forests and S. magdalenae is usually found in cloud forests and oak forests. Meanwhile, S. hahnianus occurs in all of these environments. This information is relevant because, globally, montane cloud forests and tropical rainforests are threatened ecosystems. Particularly, the first is considered rare due to its restricted extent of coverage. A mere 2.5% of the total area of the tropical forests worldwide is cloud forest (Bruijnzeel et al. 2010;Sánchez-Ramos and Dirzo 2014). Three of the four species occur naturally in Mexico where tropical forests occupy only 1% of the territory; nevertheless, these ecosystems harbour a large number of species, representing 27% of the floristic richness of the country (Gual-Díaz and Rendón-Correa 2014). Worldwide, Central America is one of the regions most affected by deforestation. An important issue is that no collections of Spathacanthus have been made from Nicaragua or El Salvador. It is not clear whether these countries remain inadequately known botanically or whether plants of Spathacanthus are genuinely absent from these countries. Based on proximity and shared climate and vegetation types, further botanical exploration is warranted.
The detailed taxonomic review of the four species revealed a number of specimens that were misidentified. For example, Daniel (1999) noted that a white corolla had been reported for S. hahnianus by Ventura A. 19361 (XAL, IEB). However, this specimen corresponds to S. magdalenae (revised and corrected in this study). (XAL)) have been misidentified as S. parviflorus. As species are circumscribed here, intraspecific variation in corolla colour can be ruled out (see Daniel 1999 for discussion). This study demonstrates corolla colour is a key morphological character for differentiating species (see the identification key above). However, since the work by Daniel (1999), no taxonomic has been carried out on Spathacanthus, such that such errors were perpetuated. Consequently, we also find labels with scientific names that are no longer valid today. In addition, our study found that only S. magdalenae has monothecate anthers, an autapomorphy for the species. Therefore, this and other slight modifications were made in the description of the genus, since it was originally described as having dithecate anthers only. These results highlight the importance of taxonomic research as we report here.
Until recently, S. hahnianus was known only from Mexico and Honduras. Daniel (1999) reported a single specimen of Spathacanthus from Guatemala but the condition of the specimen did not allow it to be identified. A decade later, Daniel (2010) suggested that the species was either rare or extirpated in Guatemala. In 2012, in a report on new distribution records of Acanthaceae in Guatemala, the same author noted the presence of the species in the montane cloud forests of Quiché. Similarly, in Honduras, the species was only known from Yoro, but is here reported also from the provinces of Cortés and Lempira. Thus, this species is now known to range more widely in Guatemala and Honduras than had been previously recognised.

Phylogenetic analysis
The parsimony phylogenetic analysis retrieved Spathacanthus as a monophyletic group, which is consistent with the results of Daniel (1999). In his previous cladistic analysis, S. hoffmannii was more closely related to S. hahnianus because both species have long corollas and the lower lip ends in relatively long lobes. On the other hand, based on similarities and differences according to the morphological key proposed by Castillo-Campos et al. (2013), it is suggested that S. magdalenae was closest to S. parviflorus, although the former has longer flowers, fruits and seeds, as well as wider leaves, relative to S. parviflorus.
Our results are concordant with those suggested from both of these works. However, the basal position of S. parviflorus, resulting from Daniels's analysis, changes when S. magdalenae is included, supporting the hypothesis, proposed by Castillo-Campos et al. (2013), that S. magdalenae and S. parviflorus are closely related species; both share white corollas (as does their sister taxon, S. hoffmannii) and leaves petiolate. The plants of these species are mainly distributed in the cloud forest of the Pacific regions, only above 1000 m a.s.l. For their part, S. hahnianus and S. hoffmannii share long corollas and they are the most widely distributed species in terms of habitats and altitude. Spathacanthus hahnianus is the most widely distributed geographically, meanwhile, S. magdalenae and S. hoffmannii have a more limited distribution on the Mexican Pacific slopes and in Costa Rica, respectively. Although the results are concordant with previous studies, it should not be forgotten that the cladistic analysis, presented here, is only exploratory, so increased character sampling and the acquisition of molecular data integrated into a framework of total evidence will support a test of the phylogenetic hypothesis proposed herein.

Conclusions
Our analysis yielded a more in-depth insight into the distribution, characters and ecological features of plants of the Mesoamerican genus Spathacanthus; however, this genus remains poorly known in Mexico and Central America. Future research should specifically seek to collect plants of Spathacanthus species in suitable habitats across the Mesoamerican region. We expect that extensive fieldwork will reveal new areas of distribution and perhaps new species of Spathacanthus. Molecular data should be gathered to test the phylogenetic hypothesis presented here; notably, strategic sampling within species would shed light on the processes of speciation.