A novelty in Ceratozamia (Zamiaceae, Cycadales) from the Sierra Madre del Sur, Mexico: biogeographic and morphological patterns, DNA barcoding and phenology

Abstract Ceratozamia is a genus of cycads occurring in eastern Mexico and Central America. In this study, we describe a new species from the Pacific region of Mexico in Guerrero state. This locality represents the most northwestern Mexico distribution for the genus. We focus the comparison of this species with the most geographically proximate and phenotypically relevant lineages for this taxon. We followed an integrative taxonomy approach to evaluate the classification of these species, including geographic location, morphology, DNA barcoding and phenology as primary sources of systematic data. Within the morphological dataset, reproductive structures are described in detail and new characters are proposed for microsporophylls. The comparative morphology of these structures facilitated the elucidation of differences in forms and species for identification. The two chosen DNA barcoding markers – namely, the chloroplast genome coding region matK and the nuclear ribosomal internal transcribed spacer (ITS) region – had low divergence, allowing only 61% of species identification, suggesting slow molecular evolutionary rates. Besides employing these three basic sources of evidence, we introduced phenology as additional information for species circumscription. In addition, this work includes a brief review of the genus at the species-level. This is therefore the most recent review for Ceratozamia across its full geographic range (latitudinal and elevational). Overall, this work further contributes to a comprehensive framework for systematic studies in Mexican cycads.


Introduction
The Sierra Madre del Sur (SMS) is a biogeographic province assigned to the Mexican Transition Zone, which includes neartic and neotropical ecosystems (Marshall and Liebherr 2000). This Mexican province runs northwest-southwest parallel to the Pacific Ocean coast throughout 1,200 km. It has great diversity of species, due to its high climatic and topographical heterogeneity (Morrone 2005;Santiago-Alvarado et al. 2016). Its rugged orography has been influenced by ancient geological events. In parallel, regional climatic cycles are relevant to explain its ecological heterogeneity, including overlaps between the Neotropical and Nearctic biotas, both of which house high species richness and endemics (Morrone 2010). The SMS has been divided into three subprovinces, in which the eastern Sierra Madre del Sur subprovince is the most heterogeneous in structural composition. This subprovince includes two districts: the Guerrero and Oaxaca highlands (Morrone 2017).
One of the least represented habitats in the SMS, and perhaps the least studied, is the SMS cloud forest. This type of vegetation is distributed unequally between 600 and 1,800 meters of elevation, and has a very disjointed and fragmented range caused by different climatic cycles (Luna-Vega et al. 1999). Moreover, the cloud forest is the most diverse habitat in Mexico per unit area and is found in small fragmented zones mainly in the Guerrero and Oaxaca highland districts (Rzedowski 1996;Challenger 1998).
The distribution of Ceratozamia Brongn. (Cycadales) is restricted to areas with high humidity in the main mountain systems of Mexico and Central America. The genus occurs in a spectrum of habitats: evergreen tropical forest, oak-pine forest, and cloud forest in the Sierra Madre Oriental (SMO), Sierra Madre de Chiapas (SMCh), SMS proper, and mountains of eastern Central America. The greatest diversity of this genus is found in the SMO. In the SMS, Ceratozamia has only been reported in Oaxaca (Vovides et al. 2004;Contreras-Medina 2016), but the north central portion of the SMS also may have suitable habitats for the genus. This is particularly so in Guerrero, which has small patches of cloud forest and is part of this biogeographic province (Luna-Vega et al. 1999).
Ceratozamia is easily diagnosed by the presence of two horns at the distal end of the sporophylls (Norstog and Nicholls 1997). This genus is one of the most complex morphological assemblages within the family Zamiaceae, because some of its species show considerable intrapopulation variation and most characters commonly used to diagnose species in other cycad genera are polymorphic (Martínez-Domínguez et al. 2017). Taxonomic and molecular systematic studies have contributed to clarify species identification and to provide taxonomic revisions with keys at the regional level (e.g. Pérez-Farrera et al. 2009;Martínez-Domínguez et al. 2016, 2017, 2018a. Gener-ally, quantitative morphological characters provide limited information among closely related species (Martínez-Domínguez et al. 2017). However, reproductive characters have been poorly studied at the species level, both morphologically and in terms of phenological cycles. Detailed studies and descriptions of organs in the reproductive structures -mainly micro-and megasporophylls -could yield potentially useful diagnostic characters in the genus (Martínez-Domínguez et al. 2018a,b).
Recently, during a review of collections of Ceratozamia deposited at the FCME herbarium, our research group found a specimen collected in 1984 from the state of Guerrero. However, this material lacked reproductive structures indispensable for unequivocal identification. Because the previously known biogeographic pattern of this genus was restricted to southeastern and central Mexico, this discovery was a novelty, as the corresponding coordinates would represent the northernmost locality of Ceratozamia for the Mexican Pacific. Given that similar specimens from the same geographic point had not been collected again, we explored the corresponding area in search of this underdescribed population. After conducting extensive fieldwork in Guerrero to collect fresh material and monitor the attendant reproductive process, we compared the new specimens with all known species in the genus with an initial focus on similar species using geographic and morphological criteria. Formally, we adopted an integrative taxonomic approach (sensu De Salle et al. 2005;Goldstein and DeSalle 2010) that involves a set of inferential rules for corroboration or refutation of species hypotheses based on multiple sources of evidence ('taxonomic circle' sensu DeSalle et al. 2005). Elsewhere, we have successfully used this approach for species-level identification in Ceratozamia, explicitly considering intraspecific morphological variation (Martínez-Domínguez et al. 2017). To identify this unidentified herbarium sheet and additional specimens from two Guerrero populations, which ultimately turned out to match the new taxon herein described, we use geographic location information, both qualitative and quantitative morphological characters, character-based DNA barcoding and phenological data. Finally, we have explored how this taxonomic discovery might alter our understanding of biogeographic and evolutionary patterns in Ceratozamia.

Specimen collection and morphological character coding
Twenty-one specimens were collected for the newly described taxon from two localities in Guerrero, Mexico. Leaf tissue was collected from all individuals for DNA sequencing and preserved in silica-gel. In total, we sampled 8 to 10 specimens approximately per population for Ceratozamia robusta Miq., and C. subroseophylla Mart.-Domínguez & Nic.-Mor., three and two populations, respectively (Appendix I). These specimens were collected during 2014 and 2017 by our research group, which have been deposited at the CIB and MEXU. In these collections, we have included the type locality and associated populations that correspond to the distribution range. Species selection was further based upon morphological similarities with the initial Ceratozamia specimen from Guerrero. For this species, a set of 40 continuous, quantitative characters were coded along with another series of 39 discrete, qualitative characters. Additionally, we examined all herbarium specimens for the 30 species recognized in the genus, which are deposited at CIB, ENCB, FCME, MEXU, US, NY, and XAL. From the herbarium specimens for all species, we evaluated a set of 25 continuous and 23 discrete characters.

Biogeographic information for specimens
Herbarium specimens of Ceratozamia were reviewed and their geographic coordinates were used to a compile a database. This information was verified in the geographic information system ArcMap GIS v.10.2. Ambiguous and/or doubtful geographic data were omitted; in cases where precise locality data were available, we georeferenced each locality with Google Earth Pro (2020) (http://earth.google.es/). To determine elevation for all registered samples, points of occurrence were superimposed on the 'digital elevation model' available from CONABIO (Guevara and Arroyo-Cruz 2016). The vegetation type for the populations of Guerrero taxon was characterized and classified following the biogeographical provinces of cloud forest according to CONABIO (2008). For amplification of the nuclear ribosomal internal transcribed spacer (ITS) region, primers 5a fwd (5'CCTTATCATTTAGAGGAAGGAG3') and 4 rev (5'TCCTCCGCTTATT-GATATGC3') were used, whereas primers fwd (5'ATACCCCATTTTATTCATCC3') and rev (5'GTACTTTTATGTTTACGAGC3') were used for the maturase K (matK) chloroplast genome locus. The latter region was amplified in Ceratozamia subroseophylla (an ITS region sequence was previously available) and included in the molecular matrix. Choice of markers followed previous publications where they performed best in terms of number of diagnostic sites (Martínez-Domínguez et al. 2016, 2017. Amplification products were visualized in 1% ethidium bromide-stained, agarose gels. Bands of the expected molecular weight were purified with a QIAquick PCR Purification Kit (QIAGEN) and sent to Laboratorio de Secuenciación Genómica de la Biodiversidad y la Salud (LANABIO; Instituto de Biología, UNAM) for automated DNA sequencing. All DNA sequence data were deposited in GenBank.

Character-based DNA barcoding
Electropherograms were edited and assembled using the program Sequencher v.4.8 (Gene Codes Corp., Ann Arbor, MI, USA). Sequences were aligned in BioEdit v.7.0.9 using the 'multiple alignment' option in Clustal X (Thompson et al. 1997;Hall 1999). We manually checked and edited these alignments with MacClade v.4.03 (Maddison and Maddison 2001). A matrix of these sequences, plus a subset of sequences for Ceratozamia including previously published ITS and matK sequences currently available in GenBank, was then saved in Nexus format for further use (Suppl. material 1: File S1). All markers were assembled with SequenceMatrix v.1.7.8 (Vaidya et al. 2011). The resulting dataset corresponding to all species currently and correctly ascribed to the genus in Mexico (Nicolalde-Morejón et al. 2011;Martínez-Domínguez et al. 2017) was subject to character-based DNA barcoding in CAOS (Character Attribute Organization System; Sarkar et al. 2008). Following the recommendations in the CAOS manual, a guide tree was prepared using the molecular matrix of Ceratozamia DNA sequences. The total length of the concatenated ITS and matK data matrix was 1813 nucleotide sites, including gaps. Cladistic analyses on this matrix in TNT v.1.5 (Goloboff et al. 2008) generated nine shortest, equally parsimonious trees. Because relationships in the corresponding strict consensus tree were not completely resolved (Suppl. material 2: Fig. S1), this tree was manually edited to avoid polytomies, preserving the arrangement of resolved clades in all trees. The topology of the resulting guide tree was saved in the Nexus format for further CAOS analyses. Identification of DNA diagnostics was carried out with the P-gnome program. Only simple pure ('sPu' sensu Sarkar et al. 2008) characters (i.e. attributes) with confidence value of 1.00 were selected.

Reproductive phenology data
The populations registered for the Guerrero taxon, as well as two populations for both Ceratozamia robusta and C. subroseophylla, were monitored for phenological observations of ovulate strobili (Appendix I). In addition, herbarium specimens of known wild collections in CIB, MEXU, and XAL were examined for complementary information in this regard (Suppl. material 3: File S2). All observations were categorized into one of the following three states, which were determined according to Martínez-Domínguez et al. (2018b): (i) receptivity (R); (ii) late ovulate (LO); and (iii) degraded (D). Determining phenological patterns for other species in the genus was not possible due to the lack of reproductive structures in herbaria.

Biogeographic patterns
The genus Ceratozamia has a continuous but restricted distribution along two of the major mountain ranges in Mexico -namely, the Sierra Madre Oriental (SMO) and the Sierra Madre del Sur (SMS), Belize, and some lowlands in Los Tuxtlas (Veracruz, Mexico) and Honduras (Fig. 1). Elevational ranges for most species are broad, usually occurring between 800 to 1,400 meters. However, some species are distributed at the extremes of this general pattern. Ceratozamia miqueliana H.Wendl., can occur at elevations as low as 19 meters, and C. zaragozae Medellín-Leal at elevations up to 2,030 meters (Fig. 2). The species with the greatest elevation range are C. fuscoviridis W. Bull, C. miqueliana, C. robusta, and C. zoquorum Pérez-Farr., Vovides & Iglesias. In turn, the two species with the narrowest range of elevational variation are C. hildae G.P.Landry & M.C.Wilson and C. euryphyllidia Vázq.Torres, Sabato & D.W.Stev. The new populations registered from Guerrero are the northernmost for the genus on the western ranges of the coastal Pacific Ocean region; this location corresponds to the northern end of the SMS province, from 1,100 to 1,400 meters of elevation (Fig. 1).
The Ceratozamia taxon herein described from Guerrero is found in relictual cloud forest and the transition zone between cloud forest and oak forest on rocky limestone slopes. In contrast, C. robusta is found in evergreen tropical forest and oak forest, whereas C. subroseophylla only inhabits evergreen tropical forest. These two related Ceratozamia species occur on clay soils with isolated rocks of volcanic origin, whereas the Guerrero taxon occurs on karstic rocks. In the context of classifications of biogeographical provinces of cloud forest, the Guerrero taxon occurs in southern coastal mountain range and the Putla subregion.

Comparisons of quantitative and qualitative morphological characters
In terms of vegetative morphology, three general groups of species within Ceratozamia can be distinguished. These groups include plants with (i) very wide leaflets, between (2.5) 4.5-17.6 cm, oblong to oblanceolate and obovate, (ii) wide leaflets of 2.3-4.6 cm, lanceolate to linear, and (iii) narrow leaflets, 0.8-2.2 cm wide, linear to lanceolate. The first group includes the species C. miqueliana, C. zoquorum, C. latifolia Miq.,   In preparation for the integrative taxonomy analysis, morphological comparisons revealed that the taxon from Guerrero has epigeous stems, median leaflets of 1.9-2.8 cm width, and long ovulate strobili. These data strongly suggest affinity of the Guerrero taxon to the second group.
In this context, the Guerrero taxon has a close morphological similarity to Ceratozamia subroseophylla and C. robusta; however, it has differences when compared to all species within the second group (see the taxonomic key below, for more details). Further morphological comparisons between these three entities reveal that only a few vegetative morphological characters provide support for species delimitation. The detailed morphological differences between C. subroseophylla, C. robusta and the Guerrero taxon are listed in Tables 1, 2. In this regard, the leaves of the Guerrero taxon show great morphometric affinity with C. robusta and C. subroseophylla with a leaf length of ranging from 1.20 to 2.80 meters. In contrast, the leaf position of the new taxon is descending, whereas in the other two species it is erect (Table 1). Additionally, the Guerrero taxon bears linear and membranaceous leaflets in contrast to C. robusta and C. subroseophylla, which have lanceolate and papyraceous to subcoriaeous leaflets ( Fig. 3). Furthermore, the shape of prickles on the petiole are robust in C. subroseophylla and C. robusta, whereas in the Guerrero taxon the prickles are thin. The primary diagnostic morphological characters lie in the reproductive structures. However, quantitative characters traditionally used for species distinction overlap in this case, both in length and diameter of ovulate strobili and peduncle length of ovulate strobili ( Table 2). The most relevant and diagnostic characters are, instead, the number of megasporophylls per orthostichy and the length of horns in the megasporophylls (Table 2). Moreover, the pollen strobilus in the Guerrero taxon has several differences when compared to species in the second group, especially in terms of microsporophyll morphology: in the Guerrero taxon, the horns of microsporophylls are straight, thin, the infertile portion is linear, and the fertile portion is deeply lobate (Fig. 4).

Character-based DNA barcoding
Sequences of matK and ITS allowed the molecular identification of 19 out of 31 Ceratozamia species (Table 3). The ITS region alone granted 45% identification success of recognized species, which increased to 61% after the inclusion of matK. Surprisingly, molecular diagnosability determined through character-based DNA barcoding in CAOS indicated that the taxon from Guerrero has two autapomorphies within matK [i.e. 'simple pure' (sPu) characteristic attributes sensu Sarkar et al. 2008] in all   For more detail of differences in character states see Table 1. five sequenced individuals (G: 1592; C: 1807). No interindividual or intrapopulation polymorphic sites were detected. Among the species with morphological affinity to the Guerrero taxon, Ceratozamia robusta and C. subroseophylla were diagnosable with ITS and ITS+matK, respectively. Notably, C. subroseophylla has a 15 nucleotide deletion in region 569-583 of the Table 3. Species identification using the candidate combination of loci for character-based DNA barcoding in Ceratozamia. The dash indicates absence of diagnostic sites.

ITS region matK
Total DNA diagnostic sites C. alvarezii Gray squares indicate the presence of at least one DNA diagnostic site for the corresponding locus. For species with an asterisk, only ITS sequences are available. aligned matrix; this type of deletion is not present in any other Ceratozamia species (Table 3; Suppl. material 4: Table S1) and, thus, represents an autapomorphy. Our samples from Guerrero are more distant genetically from other phenotypically similar species such as C. mexicana, C. mixeorum, C. delucana, C. whitelockiana and C. brevifrons. The latter three species did not have any diagnostic sites; however, a closer visual inspection of groups generated by CAOS showed divergences with the taxon from Guerrero and greater similarities with C. miqueliana and C. morettii. Ceratozamia kuesteriana had the greatest number of diagnostic sites (nine in total). This species was followed by C. matudae and C. zoquorum, with six and five diagnostic characters, respectively. The remaining species of the genus had low values of DNA diagnostics; the number of diagnostic sites by species ranges from one to three DNA diagnostics, and for nine species diagnostic sites were consistently absent (Table 3). In the case of C. chamberlainii only a gap in matK was recovered as a potential diagnostic site and it was included as a diagnostic character due to its location in a coding region.

Ovuliferous reproductive phenology
Overall, the Guerrero taxon has a phenological reproductive pattern that differs from its most morphologically related species (Fig. 5). The phenological cycle of Ceratozamia robusta and C. subroseophylla is mutually more similar; however, in the first species, the R phase occurs from May to June, whereas in C. subroseophylla this phase occurs from June to July. OT and D phases for these species have different offsets (Fig. 5). In contrast, the R phase in the Guerrero taxon occurs from March to April, while the D phase runs from September to October.

Integrative taxonomic inference of a new Ceratozamia species
According to the inferential rules of the 'taxonomic circle' in our integrative taxonomy approach, the specimens collected from Guerrero were marked with a 'red flag' -i.e. as a hypothetical species demanding test. This hypothesis was corroborated after detection of morphological diagnostic characters, DNA diagnostics and phenological differences, and after the establishment of the particular geographic locality of the collected specimens. Formally, these specimens are recognized here as belonging to a new species based on the presence of (i) distinct morphological qualitative characters, particularly in reproductive structures; (ii) exclusive DNA diagnostic sites in matK; (3) distinctions in the phenological pattern in comparison to similar species; and (4) a separate geographical range, which suggests allopatric geographic isolation (i.e. barriers to gene flow). In summary, the new Ceratozamia species from Guerrero is diagnosable according to all tested criteria. Ceratozamia leptoceras is most similar to C. robusta, but can be distinguished by its linear membranaceous leaflets and petioles with thin prickles. In addition, C. leptoceras is easily distinguished from its congeners by having obconic microsporophylls with a long, linear infertile portion (0.83-0.96 cm), and two thin horns; ovulate strobilus with abundant pubescence at base of megasporophylls, 8-9 orthostichies, and 7-9 sporophylls per orthostichy.
Etymology. The specific epithet alludes to microsporophylls horns shape, which are short and thin. This name comes from the Greek "lepto", which means thin or fine, and "ceras" in reference to horns.

Distribution and habitat.
Only known from Guerrero, Mexico, on the karstic rocks within the elevation range of 1,170-1,400 m of the Sierra Madre del Sur subprovince of the Guerreran district (Morrone 2017) (Fig. 1). This species occurs in cloud forest. The climate type is (A) C (w2)-semi-warm temperate subhumid with summer rains, and the annual precipitation is from 2, 000 to 2, 500 mm (García 2004).
Phenology. The leaves are produced in groups of 9 to 15 and mature almost simultaneously. Ovulate strobili mature from June to July; seeds mature from August to September. Pollen strobili mature from January to May.
Common names. The common local name for this species by the "Mixteco" ethnic group is Shalukaá.

Discussion
In this synthesis of Ceratozamia, we constructed a database for the genus based with geographic, morphological, and molecular data through the use of comparative and character-based DNA barcoding methods. In addition, we investigated the potential value of phenological reproductive patterns for species delimitation; this type of ecological information has been scarcely studied in cycads (Stevenson 1981;Clugston et al. 2016) or in integrative taxonomy studies of gymnosperms. Considering the different male plants in a population, the phenology of pollen strobili shows a broader cycle at the population level; the lifespan of a pollen strobilus is shorter than for ovulate strobili, and an open pollen phase is prolonged in different individual plants (Martínez-Domínguez et al. 2018b). In contrast, duration of ovulate strobili exhibits high specificity in the receptivity phase, which suggests that these data are taxonomically informative. Our findings on the length of the lifespan of reproductive structures and phenological phases coincide with our previous observations in C. tenuis (Martínez-Domínguez et al. 2018b). Ceratozamia leptoceras has a different phenological pattern for ovulate strobili compared to its morphologically similar congeners, which perhaps represents a reproductive barrier in the field (Fig. 5). Unfortunately, these data are not available for all Ceratozamia species. Despite the species diversity of Ceratozamia and its restricted geographic distribution, a formal infrageneric classification for the genus has not been proposed. This is mainly because previously reconstructed phylogenetic relationships display remarkable differences in the number of clades and the members of the clades (cf. Vovides 2002, 2012;Medina-Villarreal et al. 2019). This situation has led to an incorrect identification of specimens that unfortunately has limited the inference of evolutionary relationships, with contradictory results inexplicably unrecognized (cf. Medina-Villarreal and González-Astorga 2016;Medina-Villarreal et al. 2019). The groups within the genus are based on similarities in geographic distribution and morphology (Vovides et al. 2004); however, Ceratozamia species complexes defined by these authors are more influenced by geography. In this work, we follow the proposal of Stevenson et al. (1986) to address morphology in this genus. Our morphological description for all Ceratozamia species recognized tree groups based on the affinity of morphological characters, both vegetative and reproductive, which facilitates comparisons between species.
Under integrative taxonomy criteria (sensu De Salle et al. 2005), we have proposed a new Ceratozamia species from the Sierra Madre del Sur. This biogeographic province is a salient area in terms of gymnosperm diversity in the country (Contreras-Medina 2016). This new record for the genus in Guerrero represents an expansion of the distribution pattern for the genus on the Pacific seaboard, which opens new questions on the influence of mountain systems on the diversity of the Mexican flora. Our survey shows that the vast majority of Ceratozamia species grow at relatively high elevations (Fig. 2); the genus has standard elevation gradient patterns and does not exhibit occurrences at higher elevations or lower elevations. Therefore, it appears that Ceratozamia is poorly adapted at lower and very high elevations (Fig. 2). Species with the lowest elevation distributions occur in southern Mexico and Central America, and those at higher elevations occur in the SMO, Highlands of Chiapas, and SMS. This atypical distribution pattern seems to be favored by mountain regions, mainly in areas with cloud forest.
Historically, taxonomic research in the genus Ceratozamia has been characterized by difficulties in species identification (Miquel 1868 In terms of morphological taxonomic evidence, vegetative characters have been widely used in the genus to identify species (Vovides et al. 2004;Whitelock 2004;Pérez-Farrera et al. 2009). However, these characters are polymorphic in some species, which hinders the construction of dichotomous keys (Martínez-Domínguez et al. 2017). The classification of strictly allopatric taxa, as is the case in some Ceratozamia species, will always remain arbitrary to some degree so that the evaluation of new characters is a pressing task. In this work, we introduced previously unreported reproductive characters such as the internal parts of the strobilus, which have not been described before in detail for the genus Ceratozamia ( Fig. 4; Table 1). Comparisons of pollen strobili and the microsporophylls among different species indicated strong morphological divergence limited to certain groups of related species (Fig. 4). For this reason, we only present a key for the species group that includes the newly described taxon. Thus, it is necessary to explore these reproductive characters for the rest of the species in the genus. Keys based on vegetative features still preserve their relevance due to the dioic condition of these gymnosperms, and the lack of reproductive structures in materials deposited at the consulted herbaria (Morettii et al. 1989). However, we consider that, in view of the complexity of vegetative morphological variation, keys based on reproductive characters could function as excellent support tools in the systematics of cycad genera such as Ceratozamia.
Species delimitation in Ceratozamia exhibits a high degree of complexity (Whitelock 2004). The morphological and molecular evidence available for the genus at pre-sent does not allow the diagnosability of all species. For example, C. hildae is a morphologically distinctive species, but lacks molecular diagnostics; in contrast, C. vovidesii and C. mirandae are species that share many morphological affinities, but variation at the molecular level facilitates their taxonomic differentiation (Table 3). This taxonomic-systematic scenario points to the need to explore new molecular markers and the evaluation of variation between populations with morphological, molecular, and phenological evidence for some closely related species. In particular, there are some populations in Oaxaca and Chiapas that still await closer inspection. In the present work, 31 species have been recognized for the genus (Stevenson et al. 1986;Nicolalde-Morejón et al. 2014;Vovides et al. 2016;Pérez-Farrera et al. 2017;Martínez-Domínguez et al. 2017, 2018a, with C. becerrae and C. microstrobila treated as synonyms of C. zoquorum and C. latifolia, respectively (Stevenson et al. 1986;Martínez-Domínguez et al. 2017, 2018a.

Conclusions and future directions
This study adds to recent research that suggests the significant role of topography in SMS as a speciation driver in shaping its high species diversity (Santiago-Alvarado et al. 2016). The geographical barriers and ecological changes in this biogeographic province could have allowed the intermingling of different species; these processes could have favored the development of a number of centers of endemism (Morrone 2010;Santiago-Alvarado et al. 2016).
Our integrative taxonomic assessment provided support for the recognition of a new species, Ceratozamia leptoceras. However, the taxonomic complexity of the genus indicates the need of further systematic revisions using multiple sources of evidence, particularly in some groups of species with problematic boundaries. In addition, we have demonstrated the value of investigating ovulate and pollen strobili -particularly, microsporophylls -for the construction of refined morphological matrices for Ceratozamia; and finally, that the construction of dichotomous keys with vegetative characters should consider variation at the population level. stituto de Biología, UNAM) kindly performed DNA sequencing. Finally, we thank Dr. David Gernandt (IBUNAM) for his valuable comments and editorial suggestions on early versions of this paper. This work was financed by Instituto de Investigaciones Biológicas (Universidad Veracruzana, Mexico) and by NSF Grants BSR-8607049 and EF-0629817, awarded to DWS, and logistically supported by FVS.