Materials and methods

Specimen revision. Specimens identified as Aechmea bracteata, A. bracteata var. bracteata, and A. bracteata var. pacifica, and housed at the herbaria CHIP, CICY, HGOM, HUAP, IBUG, INIF, MICH, UADY, UAMIZ, UASLP UJAT, XAL, and ZEA (acronyms follow ^{Thiers 2023} and count. updated) were studied and digitized with a scale for further analysis. Also, previously digitized specimens were included from the following herbaria: MEXU (IBdata: www.ibdata.abaco3.org), NY (sweetgum.nybg.org/science/vh/), MO (Tropicos: www.tropicos.org), and the ones included in Red de Herbarios del Noreste de México (herbanwmex.net). We obtained a total of 373 herbarium specimens (331 for A. bracteata var. bracteata and 42 for A. bracteata var. pacifica), including the type specimens for both varieties and that of A. laxiflora. Type specimens were retrieved from JSTOR Global Plants (plants.jstor.org).

Linear morphometric analysis. 82 herbarium specimens (Supplementary material 1) selected from across the distribution of Aechmea bracteata were studied for the morphometric analysis. The specimens used for this analysis were chosen because they were complete (leaves and inflorescence) with well-preserved vegetative and floral structures. In addition, the holotypes for Aechmea bracteata (W. Houston, s.n., BM), Aechmea laxiflora (Sinclair, s.n., K), and the isotype of A. bracteata var. pacifica (C. R. Beutelspacher 51, US) were included, obtaining a total of 85 specimens. Each specimen was measured using the ImageJ version 153q software (^{Schneider et al. 2012}).

A total of 28 quantitative characters were measured considering inflorescence and leaf structures. Characters, used abbreviations, detailed information on how the measures were taken and the number of structures averaged is shown on Table 1. For the characteristics related to the inflorescence, this structure was sectioned into thirds, each referred to as basal, middle and an upper portion.

To test the separation of both taxa into different groups, an exploratory cluster analysis with an UPGMA algorithm was performed using a Gower distance matrix considering the coefficient of cophenetic correlation as an indicator of the goodness of fit for the obtained classification. Additionally, to the quantitative traits, the Gower matrix allows the inclusion of qualitative data, which were codified as follows:

Subsequently, for each of the quantitative traits, the normality of the data sets was evaluated through a Shapiro-Wilk test. After determining the distribution of the data for each character, Student’s T tests were performed for the characters with a normal distribution, and Mann-Whitney tests for characters that lack normality. To detect which characters were statistically different between taxa, an alpha value of P < 0.05 was established and adjusted with the Bonferroni correction. Since primary bracts are absent on the upper portion of A. bracteata var. bracteata, the characters regarding these structures (LPRBU, WPRBU, LWPRBU) were excluded from the normality tests.

Additionally, to identify the statistical significance of each taxon being different, using only the statistically significant characters, a one-way permutated multivariate analysis (PERMANOVA) was performed with 9,999 permutations using a Euclidean similarity index, with an alpha value of P < 0.05 and a subsequent Bonferroni correction. All the analyses were performed according to the classification algorithm UPGMA in the software PAST 4 version 4.10 (^{Hammer et al. 2001}).

For each of the quantitative morphological characters that turned out to be statistically significant, boxplots were made with R (^{R core Team 2022}) using the package ggplot2 (^{Wickham 2016}), to visualize their ranges and distribution of the data obtained.

Living specimens’ revision. To study potential diagnostic traits that are usually not available in herbarium specimens, living specimens cultivated at the Regional Botanical Garden “Roger Orellana” (Centro de Investigación Científica de Yucatán, A. C. (CICY)) and private collections, as well as field specimens and photographic records were analyzed, reviewing 21 individuals of A. bracteata var. bracteata and 15 of A. bracteata var. pacifica (specimens were deposited at CICY and UADY herbaria). This information was complemented with the information available in the protologues and the descriptions available in Flora Neotropica (^{Smith & Downs 1979}, ^{Beutelspacher 1971}). The suggested terminology by ^{Scharf & Gouda (2008)} was used for the morphological descriptions.

Taxonomic sampling for the phylogenetic analysis. All the species belonging to the Aechmea bracteata complex (sensu^{Ramírez-Díaz 2019}) were analyzed in this study (A. bracteata var. bracteata, A. bracteata var. pacifica, A. pubescens, A. pittieri, and A. dactylina).

Fresh foliar tissue from A. bracteata var. bracteata, A. bracteata var. pacifica, A. pittieri and A. pubescens was obtained from wild and cultivated specimens; for both varieties of A. bracteata, tissues were obtained from specimens of different localities of their distribution in order to obtain representation of different populations. In addition, Aechmea aquilega (Salisb.) Griseb. was included as an outgroup rooting due to its phylogenetical position on previous analyses (^{Sass & Specht 2010}, ^{Ramírez-Díaz 2019}), additionally, sequences of A. aquilega, A. dactylina, A. pittieri, and A. pubescens were retrieved from GenBank, obtaining a total of 23 terminals, having nine for A. bracteata var. bracteata, eight for A. bracteata var. pacifica, two for A. pittieri, two for A. pubescens, one for A. dactylina and one for A. aquilega (Table S3).

DNA extraction. Collected leaf tissue was frozen in liquid nitrogen and stored between -20 and -80 °C for later use. The DNA was extracted with the silica gel DNA extraction protocol described by ^{Echevarría-Machado et al. (2005)}, with the following modifications: adding 1 g of leaf tissue, 100 mM of Tris-HCI (ph8) to the lysis buffer, and 5 μl β-mercaptoethanol to the previously mixed lysis buffer, additionally, centrifuge repetitions per minute (RPM) were adjusted, utilizing 14,000 RPM. Finally, DNA quality was verified through electrophoresis with agarose gel with TBE 0.5x at a 1 % concentration.

Sequence sampling and PCR amplification. The regions selected were two nuclear genes: the nuclear ribosomal external transcribed spacer (ETS) and the region between the 8^th and the 10^th exons of glyceraldehydes-3-phosphate-dehydrogenase (g3pdh), and one chloroplast region: the trnL intron and the trnL-trnF intergenic spacer (trnL-F). The previously mentioned regions were selected due to their variability, the resolution and statistical support values obtained in previous studies (i.e., ^{Sass & Specht 2010}, ^{Ramírez-Díaz 2019}).

PCR amplification was performed from genomic DNA with Invitrogen (Carlsband, California) or Qiagen (Germantown, Maryland) Taq polymerase. The PCR conditions and their primers are shown on Table 2. After performing the PCR reaction, the quality and the approximate size of the PCR products was verified in an electrophoresis with an agarose gel with TBE 0.5x at a 1 % concentration. After checking their quality, the products were sent to Macrogen (South Korea) for sequencing with the Sanger method. All the obtained sequences were deposited in GenBank (Table S3).

Sequence alignment and phylogenetic reconstruction. Sequence assembling and editing was performed in Sequencher version 4.1 (Gene Codes Corporation www.genecodes.com). The matrix assembling and manual edition was performed utilizing BioEdit version 7.0.5.3 (^{Hall 1999}), while the alignment was performed using the algorithm MUSCLE (^{Edgar 2004}) on the software MEGA version 11.0.13 (^{Tamura et al. 2021}). The evaluation of the most adequate nucleotide substitution model for each region, according to the Akaike Information Criteria, was analyzed with the software JModelTest version 2.1.7 (^{Guindon & Gascuel 2003}, ^{Darriba et al. 2012}), where the selected nucleotide substitution model for the ETS and trnL-F regions were GTR + G, while for the g3pdh region was HKY + I.

For the phylogenetic analyses, Bayesian Inference (BI) and Maximum Likelihood (ML) analyses were performed. For the BI analysis the phylogenetic reconstruction was carried out with MrBayes version 3.2.5 (^{Ronquist & Huelsenbeck 2003}) in the CIPRES portal 2.0 (^{Miller et al. 2010}) with four independent Markov Chain analysis consisting of 1,000,000 generations with a sampling frequency of 1,000 generations, discarding the 25 % as burn-in. While for the ML analysis, 1,000 Bootstrap replicates were performed in the MEGA 11 (^{Tamura et al. 2021}) software. Individual analyses were performed for each marker, as well as a concatenated analyses for both nDNA markers, and a concatenated analysis of all nDNA and cpDNA evidence.

A topological comparison was carried out between the phylograms obtained from the concatenated and individual analyses to detect topological inconsistencies, considering as inconsistencies those nodes whose values are less than 70 BS and 0.70 PP (^{Mason-Gamer & Kellog 1996}, ^{Barber et al. 2007}). The phylograms were edited using the platform Interactive Tree of Life (https://itol.embl.de/) (^{Letunic & Bork 2021}).

Geographic distribution map. Distribution records were obtained from the collection sites from the 373 herbarium specimens. For specimens without coordinates, these were georeferenced with Google Earth (^{Google Earth 2023}) considering the information provided in the herbarium labels. In addition, the data was complemented with accessions from the Global Biodiversity Information Facility (^{GBIF 2022}). After reviewing and determinate which taxa, the herbarium specimens belonged to, data bases of the coordinates were made. The distribution map was generated with QGIS version 3.30 (^{QGIS Development Team 2023}), projecting the coordinates on a layer of the biogeographic provinces for the Neotropics (^{Morrone et al. 2022}) from Mexico to northern South America.

Results

Linear morphometric analysis. The dendrogram obtained by the conglomerate analysis resulted into two groups (Figure 1) with a cophenetic correlation coefficient of 0.764. According to the position of the included nomenclatural types of each taxon, group 1 represents a cluster that can be characterized as A. bracteata var. pacifica, in which the type specimen of A. laxiflora is included; as for group 2, is composed merely of the typical variety of A. bracteata where the holotype of the species belongs.

Figure 1 Dendrogram resulting from the cluster analysis of the Gower distance matrix. Group 1 A. bracteata var. pacifica, black circles; purple circle: isotype; green circle: A. laxiflora type specimen; Group 2 A. bracteata var. bracteata, red circles; dark purple circle: A. bracteata holotype. Herbarium specimens of the photographs: A. bracteata var. pacifica J. P. Pinzón & J. J. Ancona 339 (UADY); A. bracteata var. bracteata M. Soto David et al. 6. (UADY). 

As for the qualitative morphological traits included in the data sets, the most variable character between both taxa was the number of secondary branches on the basal branches of the inflorescence. Out of the 45 evaluated herbarium specimens of A. bracteata var. bracteata, 23 had four or more secondary branches, whilst the rest had two or no secondary branches, as for the 37 specimens of A. bracteata var. pacifica, 19 of the evaluated specimens did not have secondary branches, 14 had two secondary branches, and only three had 3 or more secondary branches in the basal branches of the inflorescence. For the peduncle and primary bracts margin there was no variation, except for three specimens of A. bracteata var. bracteata that possessed a serrate margin, the rest of the specimens for both taxa showed entire margins.

Out of the 25 quantitative characters analyzed, according to the Shapiro-Wilk’s tests, 14 of them turned out to be normally distributed, whereas 11 did not (Table S1). For the parametric character analyses (Student’s T test), the alpha value with the Bonferroni correction was of P < 0.038 whilst for the non-parametric character analyses (Mann-Whitney test) the obtained value was P < 0.004. Seven characters resulted as statistically significant between the two varieties of A. bracteata, such as IFL, IBLU, LPB, LWPB, LPRBB, WPRBM, and the LFB (Figure 2). P values of the Student’s T and Mann-Whitney tests are provided in Table S2.

Figure 2 Boxplots diagrams for the statistical significative morphological characters according to the univariate analyses. ABB = Aechmea bracteata var. bracteata; ABP = Aechmea bracteata var. pacifica. 

Regarding the one-way PERMANOVA analysis with the statistical significative data sets, A. bracteata var. bracteata and A. bracteata var. pacifica represent two different groups with statistical significance, since a P value of 0.0001 and a F value of 31.19 were obtained.

Living specimens’ revision. In addition to the differences on the inflorescence size and the distribution of the primary bracts (Figures 3A, 4A), there are some other morphological traits that provide elements to diagnose both taxa that are usually lost or hard to identify in herbarium specimens.

Figure 3 Aechmea bracteata (Sw.) Griseb. var. bracteata A. Inflorescence. B. Inflorescence branches. C. Fruits in different maturation stages. D. Detail of the leaf blade spines. E. Floral bract. F. Habitat. Photographs: Mauricio Soto David (A-D) and Claudia Ramírez-Díaz (E-F). 

Figure 4 Aechmea bracteata var. pacifica Beutelsp. (Aechmea laxiflora Benth.). A. Inflorescence. B. Secondary inflorescence branches. C. Fruits in different maturation stages. D. Detail of the leaf blade spines. E. Floral bract. F. Habitat. Photographs: Mauricio Soto-David (A-D) and Claudia Ramírez-Díaz (E-F). 

The morphology of the corolla and the floral bracts shows differences between both taxa. The corolla, for A. bracteata var. pacifica is usually wider (2-3 mm), compared to the typical variety where it is narrower (1.5-1.8 mm), regarding the petals, in A. bracteata var. bracteata they are pale yellow while the petals of A. bracteata var. pacifica are bright yellow colored (Figures 3B, 4B). As for the floral bracts, in A. bracteata var. bracteata they are ovate to elliptic and acuminate, while in A. bracteata var. pacifica, these tend to be ovate-triangular and apiculate (Figures 3E, 4E).

Additionally, differences can be appreciated during the ripening process of the fruits. In A. bracteata var. bracteata the exocarp is provided with a white punctuated indument, shortly after fertilization, during the widening and maturation of the fruit the exocarp presents a green coloration, and the persistent sepals a rufous red coloration; after the fruit ripens the sepals acquire a purple coloration and the exocarp becomes bluish black (Figure 3C). In the other hand, the fruits of A. bracteata var. pacifica possess a stellate white indument, after fertilization, the exocarp presents a yellowish green coloration as well as the persistent sepals; when the fruit matures, both structures acquire a black coloration (Figure 4C).

Furthermore, in the vegetative structures other differences can be noticed. The leaf blades in A. bracteata var. bracteata possess are coriaceous, parallelodromous trichomes abaxially, a ligulate to triangular shape, with spines from the leaf margin being uncinate to antrorse (Figure 3D). In contrast, leaf blades of A. bracteata var. pacifica are more flexible, with less and irregularly distributed trichomes abaxially, and has ligulate to broadly triangular leaf blades with antrorse spines (Figure 4D).

Phylogenetic analysis. For the 23 terminals included in this analysis, size, variability, and informative levels of the data sets for each region and combined are shown in Table 4. From this data we can identify that ETS possessed the higher values of informativity whilst trnL-F resulted in the least informative region.

While analyzing the markers individually and combined in the BI and ML, topologies with different resolution levels and statistical support values were obtained. Regarding the observed topological incongruences, these were distinguished as soft incongruences, mostly due to the phylogenetic position of the individuals within the clades for both varieties of A. bracteata.

In the analyses performed for the nDNA markers (Figure 5) both varieties of A. bracteata shown reciprocal monophyly with high statistical support (PP: 1, BS: 98), their clade is divided into two subclades, one conformed solely of the typical variety of the species (PP: 1, BS: 99), and the other composed by A. bracteata var. pacifica (PP: 0.99, BS: 94). For the rest of the species of the complex, a clade containing A. pittieri, A. pubescens and A. dactylina was retrieved with a high statistical support (PP: 1, BS: 100). The monophyly of A. pittieri (PP: 1, BS: 100) and A. pubescens (PP: 1, BS: 100) was retrieved, however, the phylogenetic position of A. dactylina could not be determined due to topological incongruences on its position on the BI and ML analyses as well as low statistical support. Individual phylogenetic analyses of ETS and g3pdh are shown in Figures S1 and S2, respectively.

Figure 5 Majority Rule Consensus tree from a Bayesian Inference analysis of the Aechmea bracteata complex with evidence of nuclear DNA markers (ETS and g3pdh). Above the branches: Posterior Probability; below the branches: Bootstrap from a Maximum Likelihood analysis. PP values above 0.50 and BS values above 50 are shown. 

Contrary to the results obtained with nuclear DNA regions, the BI and ML analyses for chloroplast region trnL-F (Figure 6), showed a low phylogenetic resolution, resulting in a polytomy with two main clades: a polytomy including A. bracteata var. pacifica variety as well as A. pittieri and A. pubescens, with a higher statistical support (PP: 0.99, BS: 84), and a polytomy composed of the majority of the A. bracteata var. bracteata sampled specimens (PP: 0.76, BS: 64). However, three of the sampled specimens of A. bracteata var. bracteata (663, 664 and 665) appeared collapsed at the base of the tree.

Figure 6 Majority Rule Consensus tree from a Bayesian Inference analysis of the Aechmea bracteata complex with evidence of chloroplast DNA marker trnL-F. Above branches: Posterior Probability; below the branches: Bootstrap from a Maximum Likelihood analysis. PP values above 0.50 and BS values above 50 are shown. 

In the resulting phylogram obtained from the concatenated matrix of the two nDNA and the cpDNA markers (Figure 7), the topology is identical to that obtained from the nDNA markers analysis, but with slightly lower statistical values. In this analysis with the three markers, both varieties of A. bracteata showed reciprocal monophyly (PP: 1, BS: 87), the subclade conformed by A. bracteata var. bracteata showed high statistical support (PP: 1, BS: 99), whereas the subclade conformed by A. bracteata var. pacifica showed a high PP value but a low BS support (PP: 0.99, BS: 72). The remaining species of the complex formed their own clade with a high statistical support (PP: 1, BS: 99). The monophyly of A. pittieri (PP: 1, BS: 86) and A. pubescens (PP: 1, BS: 92) was retrieved, however, the phylogenetic position of A. dactylina could not be determined either in this concatenated analysis due to low statistical support and topological incongruences on its position on the BI and ML analyses.

Figure 7 Majority Rule Consensus tree from a Bayesian Inference analysis of the Aechmea bracteata complex with the concatenated evidence of two markers of the nDNA (ETS and g3pdh) and the cpDNA marker (trnL-F). Above branches: Posterior Probability; below the branches: Bootstrap from a Maximum Likelihood analysis. PP values above 0.50 and BS values above 50 are shown. 

Geographic distribution. Based on the georeferenced collection sites reported on herbarium records, two population groups are distinguished, each one corresponding to a variety of A. bracteata (Figure 8). According to the obtained data, these varieties are not sympatric. Our results show that the distribution of Aechmea bracteata var. pacifica is restricted to Mexico, from the south of Sinaloa to western Oaxaca, occurring in the biogeographic province of the Pacific Lowlands with some records in transition areas with the Sierra Madre del Sur province.

Figure 8 Geographic distribution of the varieties of Aechmea bracteata projected over the Biogeographic Provinces of the Neotropics. 

Meanwhile, A. bracteata var. bracteata possesses a wider distribution from Mexico to northern Colombia and Venezuela, being found throughout various Neotropical biogeographic provinces. In Mexico this taxon occurs in the Veracruzan Province, the Yucatan Peninsula, some localities in transitional areas with the Sierra Madre Oriental, a small portion within the Chiapas Highlands, and Pacific Lowlands in the region of the Isthmus of Tehuantepec. In Central America it is found in the biogeographic provinces of Mosquito, Pacific Lowlands, Guatuso-Talamanca and Punta Arenas-Chiriquí, and in South America, records were found in the provinces of Guajira, Magdalena, and the Venezuelan Province. Records were not found in the Chocó-Darién and Cauca provinces.

Discussion

Morphometry and morphology of A. bracteata. Different kinds of linear and geometrical morphometric analyses have resulted useful to study the Bromeliaceae family, allowing to elucidate, validate, recognize new taxa, or change the taxonomic status of varieties and synonyms (e.g., ^{Wendt et al. 2000}, ^{Hornung-Leoni & Sosa 2008}, ^{Ferreira et al. 2009}, ^{de Faria et al. 2010}, ^{Pinzón et al. 2011}, ^{Castello & Galetto 2013}, ^{Guarçoni et al. 2017}, ^{Neves et al. 2018}, ^{González-Rocha et al. 2018}, ^{Martínez-García et al. 2022}). In this study, the morphometric cluster analysis grouped the individuals into two clusters, each corresponding to a variety of A. bracteata, according to the characters previously recognized by ^{Beutelspacher (1971)} and ^{Smith & Downs (1979)}. In addition, the PERMANOVA analysis showed that both groups are statistically different.

Among the morphological characters that showed differences between both groups in the univariate analyses, are the IFL, IBLU, LPB, LWPB, LPRBB, WPRBM and LFB.

According to the original description for A. bracteata var. pacifica by ^{Beutelspacher (1971)}, the inflorescence in this variety is shorter than the typical variety of the species. Although the description mentions that such structure size is up to 90 cm (probably including the peduncle), despite not including the peduncle in the herein study, the obtained data for the IFL resulted in A. bracteata var. pacifica, having a tendency to possess a smaller size range ([12.3-] 20.1-33.6 [-43.9] cm) in comparison to A. bracteata var. bracteata ([16.5-] 33.5-50.5 [-70.7] cm).

As to the LPRB, the values obtained for A. bracteata var. pacifica were close to the description by ^{Beutelspacher (1971)}, in which it is said that these structures are larger, being up to 10 cm in length, smaller than the ones on the typical variety, in the present study, for the LPRB of the basal portion of the inflorescence the obtained range was (4.8-) 6.3-9 (-11.9) cm, contrasting with the range obtained for A. bracteata var. bracteata (5.5-) 7.5-12.9 (-18.4) cm, indicating a higher variation and size of this structure.

Moreover, according to the same description, it is mentioned that the branches of the inflorescence have a regular longitude, being almost of the same size across the whole inflorescence which is up to 6 cm (^{Beutelspacher 1971}). The specimens analyzed in the present study showed some concordance with the description of A. bracteata var. pacifica, as the obtained range of the IBL was (0.5-) 1. 6-6.9 (-10.6) cm long, contrasting with A. bracteata var. bracteata in which IBL range was (0.7-) 1.2-12.4 (-23.6) cm long, being larger than those observed for A. bracteata var. pacifica.

According to the cluster analysis, the type specimen of A. laxiflora is placed within the A. bracteata var. pacifica group, supporting the fact that both taxa belong to the same entity as proposed by the synonymy of ^{Smith & Downs (1979)}.

In addition to morphometric analyses, some of the qualitative characters observed in living specimens are useful for the differentiation between taxa. According to the observed traits, A. bracteata var. bracteata has green fruits with persistent red sepals when immature, and after ripening, the berry turns bluish black and the persistent sepals acquire a purple coloration, while A. bracteata var. pacifica has yellowish green fruits and when ripe they turn black as well as the persistent sepals. In previous studies, fruit and floral characteristics, among others, have been used to provide further support for the characterization and delimitation between closely related taxa within the subfamily Bromelioideae (e.g., ^{Leme 1997}, ¹⁹⁹⁸, ²⁰⁰⁰, ²⁰⁰⁷, ^{Leme et al. 2017}). These traits were not revised until the herein study for species of the A. bracteata complex.

Phylogenetic relationships. According to the phylogenetic analysis by ^{Sass & Specht (2010)}, which is based on the concatenation of three nDNA markers (g3pdh, ETS and rpb2) and one from cpDNA (trnL-F); the Aechmea bracteata complex represents a monophyletic group with high statistical support (PP: 1, BS: 99) and its composed by both varieties of A. bracteata, A. pubescens, A. dactylina, and A. pittieri, as well as three unidentified samples. Even though our analysis and ^{Sass & Specht (2010)} analysis are congruent with A. bracteata not being a monophyletic species, the topology of the clade is different than the herein obtained, positioning A. bracteata var. pacifica in a clade with an undetermined specimen (Aechmea sp.) as the sister clade of the rest of the complex. Regarding A. bracteata var. bracteata, in ^{Sass & Specht (2010)} its phylogenetic position is as the sister species of A. pubescens. However, it is possible that the differences among the topology of ^{Sass & Specht’s (2010)} analysis and ours, rely on the inclusion of three undetermined specimens (which were excluded in our analysis), and the inclusion of the rpb2 sequences in their analysis.

On the other hand, the obtained results by ^{Ramírez-Díaz (2019)} in a consensus analysis for two nDNA markers (agt1 and g3pdh), are congruent with the results in the present publication, since A. bracteata var. bracteata and A. bracteata var. pacifica also showed to be reciprocally monophyletic (PP: 0.85, BS: 87), with high statistical values for A. bracteata var. bracteata (PP: 0.90, BS: 86.2) but lower statistical support values for A. bracteata var. pacifica (not shown). In our analysis with ETS and g3pdh markers (nDNA), the node support for both taxa being reciprocally monophyletic was higher (PP: 1, BS: 98), as well as the statistical support for the clades conformed by individuals of each taxa (for A. bracteata var. bracteata PP: 1, BS: 99; for A. bracteata var. pacifica PP: 0.99, BS: 94).

Regarding the cpDNA evidence obtained by ^{Ramírez-Díaz (2019)} with the markers rpl32-trnL and ycf1, the Aechmea bracteata complex presented a polytomy with high PP value (0.99) but with low BS value (not shown), A. pittieri and A. pubescens appeared collapsed towards the base of the polytomy, while A. bracteata var. pacifica presented a polytomy by its own (PP: 099, BS: 77.3). Concerning A. bracteata var. bracteata, the individuals sampled by Ramírez-Díaz formed a clade with a high BS value (98.6) but a low PP value (not shown). In the herein study, the phylogenetic tree obtained with the cpDNA marker trnL-F, showed some congruence with the results obtained by Ramírez-Díaz, however, the specimens of A. bracteata var. pacifica grouped in a politomy with A. pittieri and A. pubescens (PP: 0.99, BS: 84), whereas many of the A. bracteata var. bracteata specimens conformed an independent polytomy but with low statistical support (PP: 0.76, BS: 64) and others appear collapsed at the base of the tree. The difference among the obtained topology and the one obtained by ^{Ramírez-Díaz (2019)} might be due to the low resolution of the trnL-F marker and the low differentiation in the sequences among the sampled species and individuals. However, in both cases, cpDNA evidence suggests that the monophyly of A. bracteata as a single species is not supported.

Due to the topological inconsistencies regarding the phylogenetic position of A. bracteata var. pacifica, and considering that in angiosperms the inheritance of nDNA is biparental, whereas the inheritance of cpDNA is uniparental, mostly maternal, and highly conserved due to its slow molecular evolution and low mutation rates (^{Small et al. 2004}, ^{Robbins & Kelly 2023}), it can be hypothesized that the origin of this taxon might be due to potential events of ancestral hybridization, introgression and/or incomplete lineage sorting (^{Barber et al. 2007}, ^{Kim & Donoghue 2008}, ^{Jabaily & Systma 2010}). Incongruences between nDNA and cpDNA evidence has been constantly observed in Bromeliaceae, suggesting that phenomena such as hybridization, introgression and incomplete lineage sorting have been of importance to the diversification of bromeliads and are present in several groups within the family (e.g., ^{Jabaily & Sytsma 2010}, ^{Schulte et al. 2010}, ^{Palma-Silva et al. 2011}, ^{Versieux et al. 2012}, ^{Krapp et al. 2014}, ^{Goetze et al. 2017}, ^{Mota et al. 2019}, ^{Ancona et al. 2022}), being observed mostly in taxa resulting from a recent adaptative radiation in which the lineages experiment a process of diversification into various ecological niches in a short time period (^{Goetze et al. 2017}). However, to determine whether these events influenced the evolutionary history of A. bracteata var. pacifica, studies involving population genetics, phylogenetic networks, phylogeographic analyses and ancestral area estimation could address this more explicitly.

Geographic distribution. Aechmea bracteata var. pacifica is distributed mainly over the biogeographic province of the Pacific Lowlands and some transitional areas with Sierra Madre del Sur. This variety, according to the reviewed herbarium specimens and ^{Flores-Argüelles et al. (2023)}, occurs in riparian forests, low and medium deciduous to sub-deciduous forests, oak forests, and pine-oak forests, at elevations lower than 1,250 m asl, with a distribution that forms a strip from southern Sinaloa to western Oaxaca.

As previously mentioned, A. bracteata var. pacifica has been collected in the southern slope of the Sierra Madre del Sur at elevations below 1,250 m asl. This may be due to the presence of transition areas, divisions and ravines across the Sierra Madre del Sur caused by the mouths of rivers towards the Pacific Ocean (^{Santiago-Alvarado et al. 2016}). In the rest of the region the elevation can be higher than 3,500 m asl with temperate climates, or on the slope facing the Balsas Basin, there are dry climates with xeric scrubs (^{Espinosa et al. 2008}, ^{Santiago-Alvarado et al. 2016}), environmental conditions that represent a geographical barrier for this taxon.

Other geographic limitations and barriers for the distribution of A. bracteata var. pacifica are the Sierra Madre Occidental to the north, and the Trans Mexican Volcanic Belt. These provinces represent mountainous chains with elevations above 1,000 m asl and reaching elevations up to 3,000 m, with warm to temperate and semi cold climates (^{Espinosa et al. 2008}, ^{González-Elizondo et al. 2012}, ^{Gámez et al. 2012}, ^{Morrone 2014}).

On the other hand, Aechmea bracteata var. bracteata is found throughout the Veracruzan Province, the Yucatan Peninsula, in low elevation points in the eastern slope of the Sierra Madre Oriental, Mosquito, Guatuso-Talamanca, Puntarenas-Chiriquí, some projections of the Pacific Lowlands, the provinces of Guajira, Magdalena and the Venezuelan Province. This taxon is found in sites from sea level to 1,000 m of elevation, occurring in low to medium deciduous and sub-deciduous forests, semi-evergreen forests, medium and high evergreen forests, and savannas. Likewise, it is found in areas with tropical sub-humid and even semi-arid climates such as the north of the Yucatan Peninsula (^{Espinosa et al. 2008}, ^{Muchoney et al. 2000}).

Within the Sierra Madre Oriental, this taxon occurs in areas below 1,000 m asl, in transitional areas with the Veracruz Province, growing in the eastern slope where different types of jungle and forest ecosystems are favored by humidity (^{Espinosa et al. 2008}, ^{Suárez-Mota et al. 2017}). On the other hand, most of this province acts as one of the geographical barriers that limit the distribution of this taxon due to high altitudes and the dry slope in the west portion (^{Espinosa et al. 2008}, ^{Suárez-Mota et al. 2017}).

Other barriers to the distribution of A. bracteata var. bracteata include the Trans Mexican Volcanic Belt, the northern and eastern portions of the Sierra Madre del Sur, and the Chiapas Highlands, all of them composed of mountains with elevations greater than the requirements that the taxon seems to have. Regarding the Chiapas Highlands, it is possible that it is due to this mountain chain that A. bracteata var. bracteata does not occur in the southern portions of Guatemala and Honduras, and El Salvador, being found only in the Caribbean slope up to Nicaragua.

The Chocó-Darién and Cauca provinces interrupt the distribution of A. bracteata var. bracteata. This is probably because the Chocó-Darién is an area of high humidity and precipitation (8,000 to 13,000 mm) (^{Fagua et al. 2019}). In the case of the Cauca province, this province consists of the western and central mountain ranges of Colombia (^{Kattan et al. 2004}, ^{Morrone 2014}). This coincides with the reports published by ^{Aguirre-Santoro & Betancur (2008}), where A. bracteata is only reported for areas that belong to the biogeographical provinces of Magdalena and Guajira. Likewise, A. bracteata has not been reported for the Chocó-Darien region (e.g., ^{Cárdenas-López 2003}).

In Mexico both taxa are found in the state of Oaxaca, however, there are no sympatric areas in their distribution. This could be due to the presence of the Sierra Madre del Sur and the Balsas Basin in central Oaxaca, where the elevational ranges vary between 1,500 to 2,000 m, having cooler temperatures (^{Ortiz et al. 2004}, ^{Trejo 2004}). Aechmea bracteata var. bracteata is found in the Tehuantepec isthmus region, which is a low elevation area with warm humid and sub-humid climates (^{Ortiz et al. 2004}, ^{Trejo 2004}). Whereas A. bracteata var. pacifica is reported for the west of the state, near the locality of Villa de Putla de Guerrero, at 730 m (A. Espejo, A. R. López-Ferrari, J. Ceja & A. Mendoza R. 6794 CICY; UAMIZ), a region which possesses a warm sub-humid climate. However, considering the elevational distribution of this taxon, it could be found at minor elevations in areas with the same type of climate.

Re-establishment of Aechmea laxiflora Benth. According to the gathered evidence, it is suggested that both varieties of Aechmea bracteata can be treated under the unified species concept proposed by ^{De Queiroz (2005}, ²⁰⁰⁷⁾ which primary property establishes that a species is the segment of a population that have evolved separately conforming its own lineage. It can be considered that both varieties represent their own lineage, since nDNA evidence separates them with reciprocal monophyly, although cpDNA evidence groups A. bracteata var. pacifica in a polytomy with the Central American species of the complex, possibly due the low variability of the analyzed region.

^{De Queiroz (2007)} also proposes that the traits considered by other species concepts serve as secondary or contingent properties that could be or not be acquired through the evolution of the populations. The secondary criteria include intrinsic reproductive isolation, monophyly, exclusive coalescence, diagnosability, deficits of genetic intermediates, and ecological divergence (^{De Queiroz 2007}). For the studied taxa, some of the contingent properties than can be considered are the capability of being identified through their morphology (by qualitative and quantitative methods), their reciprocal monophyly (according to nDNA), and their disjunct geographical distribution, that also serve as an extrinsic reproductive barrier. However, it is not possible to determine if they somehow differ ecologically or possess intrinsic reproductive barriers because those approaches were not considered in the present study.

Due to the gathered evidence, it is possible to elevate to a specific level both taxonomic entities. According to the grouping of the specimens and nomenclatural types included in the morphometric cluster analysis, and the obtained results of the phylogenetic analyses, it is suggested that the circumscription of Aechmea bracteata is restricted to the populations previously considered as the typical variety of the species. The populations restricted to the Pacific slopes of Mexico (from Sinaloa to western Oaxaca), includes the collections and nomenclatural types of A. bracteata var. pacifica and A. laxiflora. Considering that and the fact that they share morphological traits, we propose to reestablish the name Aechmea laxiflora and turn A. bracteata var. pacifica into its synonym.

Since Aechmea laxiflora has a restricted geographic distribution alongside the Pacific Slopes of Mexico, it is possible to recognize it as an endemic species to Mexico. This elevates the number of species from the genus Aechmea reported for Mexico from ten, and the endemic species number from three (^{Espejo-Serna & López-Ferrari 2018}, ^{Flores-Argüelles et al. 2020}) to eleven and four, respectively (Table 5).

Aechmea bracteata (Sw.) Griseb. Fl. Brit. W. I.: 592. 1864.

Bromelia bracteata Sw., Prodr.: 56. 1788. Type: Mexico, Veracruz, 1731, W. Houston, s.n. (holotype: BM).

Aechmea schiedeana Schltdl., Linnaea 18: 437-439. 1845 "1844". Type: Mexico, Veracruz, Hacienda de la Laguna, without collection date, Schiede & Deppe, s.n. (holotype: B, destroyed).

Hoplophytum bracteatum (Sw.) K. Koch, Wochenschr. Vereines Beförd. Gartenbaues Königl. Preuss. Staaten 3: 306. 1860.

Hohenbergia bracteata (Sw.) Baker in: Saunders, Refug. bot. 4: sub t. 284, no. 3. 1871.

Aechmea barleei Baker, Gard. Chron. n.s., 20: 102. 1883. Type: Belize, British Honduras, 1877, Barlee s.n. (holotype: K).

Aechmea isabellina Baker, J. Bot. 28: 305. 1890. Type: Guatemala, Izabal, boca del Polochic, April 1889, J. Donell Smith 1824 (holotype: US).

Description. Plants epiphytic, rarely saxicolous herbs, from 50 cm to over 2 m when flowering, polycarpic, funnelform rosette. Leaves erect to recurved, coriaceous; leaf sheaths 13-26 × 10-16 cm, elliptic to obovate, adaxially pardus lepidote, abaxially green, lepidote, coriaceous; leaf blades 26-123 × 3-9 cm, ligulate to broadly triangular, green on both sides but sometimes red when growing exposed, slightly lepidote adaxially, densely lepidote abaxially, parallelodromous trichomes, conspicuous veining, serrate margins; spines 2-7 × 1-3 mm long, decreasing its size gradually towards the leaf blade apex, straight to uncinate, antrorse, green with yellowish tips. Inflorescence terminal, usually the basal spikes twice-branched and the rest once-branched, occasionally completely once-branched, erect or curved, paniculated; peduncle , 46-67 (-95) cm long, 2-9 mm in diameter, cylindrical, erect to curved, green or red colored axis, covered with white floccose indument; peduncle bracts 9-22 × 1-2 cm, broadly elliptic to lanceolate, acute, imbricated, entire margin, bright red colored, alternate phyllotaxis; fertile portion of the inflorescence 16-70 cm long, erect to curved, green to red colored axis, covered with white indument; spikes 0.4-7 cm long, ca. 30 to over 100, decreasing in size towards the apex, alternate phyllotaxis, green or red colored axis covered with white floccose indument; secondary spikes, 1-4 cm, alternate phyllotaxis, green or red colored axis covered with white floccose indument; primary bracts along the inflorescence up to the middle portion of the inflorescence and decreasing in size, 3-21 × (0.30-) 0.40-2 cm in the basal portion, 1-7 (-10) × 0.1-0.6 (-1) cm in the middle portion, generally absent in de upper portion towards the apex, lanceolate to linear, acute, margin entire to minutely serrated, bright red, glabrous; floral bracts 4-7 × 1-5 mm, ovate to elliptic, acuminate, with a small spine in the apex, white indument abaxially, glabrous adaxially, green to red, apical spine green or red to pale brown. Flowers 10-13 × 2-2.5 mm, sessile, appressed to the spike rachis; sepals 3-4 × 2-2.5 mm, triangular, slightly acuminated, free, glabrous, smooth, coriaceous, green with hyaline margins; corolla tubular, actinomorphic, 5-8 × 1.5-1.8 mm; petals 5-8 × 2-2.5 mm, oblanceolate, acute, slightly incurved, hyaline margins, free, glabrous on both surfaces, pale yellow, scaled crenated appendages; stamens 6, 3 attached to each one of the petals and the other 3 attached to the ovary, free, shorter than the petals, white filaments, dorsoventrally flattened, 3-5 mm long; anthers 1-1.1 mm long, oblong, monothecal, sub basifixed, yellow; pollen yellow; style 6-8 mm long, erect, inserted, white to pale yellow, cupulate stigma; ovary 3-4 × 3-3.5 mm, oblong, green, floccose white lepidote. Berry 7-8 × 6-7 mm, globose, green when immature, bluish-black when mature, floccose to dotted white lepidote; persistent sepals, red when the fruit is immature, purple when mature; seeds 3.2-3.8 × 1-1.6 mm long, cuneiform, light brown to reddish-brown, covered with mucilage.

Distribution and ecology. Aechmea bracteata occurs from Mexico (alongside the slopes of the Gulf of Mexico, from southern Tamaulipas to the Yucatan Peninsula; and eastern Oaxaca and Chiapas) to northern Colombia and Venezuela. This species occurs in oak forests, riparian forests, tropical deciduous forests, tropical sub-deciduous forests, tropical sub-evergreen forests, mangroves, seasonally flooded savannahs, and secondary vegetation from sea level to ca. 1,000 m asl (according to herbarium records, ^{Ramírez et al. 2004}, ^{Espejo-Serna et al. 2005}).

Phenology. This species has been collected flowering and fructifying almost all year long, however, flowering occurs most commonly from October to May, while fructification mostly occurs from January to August.

Etymology. Not specified in the protologue, however, “bracteata” might be a reference to the inflorescence bracts this species possesses.

Taxonomic notes. Initially being described as Bromelia bracteata by ^{Swartz (1788)} from a specimen collected in Veracruz, Mexico (Houston, W. s.n., BM), and later transferred to the genus Aechmea Ruiz & Pav. by ^{Grisebach (1864)}, Aechmea bracteata possess a considerable taxonomic history due to various taxa described from various populations during the XIX century. Among the taxa considered as synonyms of A. bracteata, Aechmea schiedeana Schltdl. was described based on a specimen collected from Veracruz, Mexico (Schiede & Deppe, s.n., B, destroyed). Although its type specimen is declared as destroyed, it is due to its geographic origin and characteristics mentioned on its protologue, such as the leaf size and margin, the parallelodromous trichomes, the size of the inflorescence, branches and primary bracts, and the floral bracts having a “piercing point” (^{Schlechtendal 1845}); that is possible to infer its identity as A. bracteata. Other synonyms known for A. bracteata are A. barleei Baker and A. isabellina Baker, both described in Central America, in the countries of Belize (Barlee, s.n., K (photo, US)), and Guatemala (J. D. Smith 1824, US), respectively (^{Baker 1883}, ¹⁸⁹⁰). However, the descriptions for each of the previously mentioned taxa, match with the morphology of A. bracteata, regarding their type specimen’s collection sites, these belong to the known geographical distribution of A. bracteata, supporting the synonymy initially proposed by ^{Smith (1958)} and ^{Smith & Downs (1979)}.

Regarding, Aechmea regularis Baker, a taxon which has been previously considered as a synonym of A. bracteata in other publications (i.e., ^{Smith 1958}, ^{Smith & Downs 1979}); it was described from a specimen allegedly collected in South Brazil (Weir, s.n., K) and is described to have white petals (^{Baker 1879}). Due to disagreement with the known distribution and petal color described for A. bracteata, A. regularis will not be treated as a synonym in this publication.

Specimens examined. Supplementary material 2.

Aechmea laxiflora Benth., Bot. Voy. Sulphur [Bentham] 173. 1846. Type: Mexico, Guerrero: Acapulco, without collection date, Sinclair s.n., (holotype K).

Hohenbergia laxiflora (Benth.) Baker, Refug. Bot. 4: sub t. 284, no. 7. 1871.

Aechmea bracteata var. pacifica Beutelsp. Cact. Suc. Mex. 16: 44, f. 26-27. 1971. Type: Mexico, Guerrero, Mun. Manilinaltepec, El Rincón, 45 km de Chilpancingo camino a Acapulco, 26 March 1971, C. R. Beutelspacher s.n. (holotype, MEXU)

Description. Plants epiphytic, rarely saxicolous herbs, from 50 cm to less than 2 m when flowering, polycarpic, funnelform rosette. Leaves, recurved to erect, flexible, coriaceous; leaf sheaths 13-26 × 9-15 cm, elliptic, adaxially red-brown lepidote, abaxially yellowish green, lepidote, coriaceous; leaf blades 39-96 × 2-9 cm, ligulate to broadly triangular, slightly lepidote adaxially, lepidote with irregularly distributed trichomes abaxially, flexible coriaceous, yellowish green both sides, inconspicuous veining, serrate margins; spines 3-7 mm long, decreasing its size gradually towards the leaf blade apex, uncinate, antrorse, yellowish green with pale brown tips. Inflorescence terminal, usually once-branched, rarely twice-branched at the basal spikes, erect or curved, paniculate; peduncle 39-67 cm long, 3-6 mm in diameter, cylindrical, erect to slightly curved, green to red colored axis, with thin white indument; peduncle bracts 9-17 × 1-2 cm, broadly elliptic, attenuate, imbricated, entire margin, pinkish red colored, alternate phyllotaxis; fertile portion of the inflorescence 12-43 (59) cm long, erect to curved, green to red axis with thin white indument; spikes 0.5-8 cm long, ca. 20 to over a 100, decreasing in size towards the apex, alternate phyllotaxis, green colored axis covered with thin white indument; secondary spikes 2-8 cm long, green colored axis covered with thin white indument; primary bracts along the whole inflorescence decreasing in size towards the apex, 4-12 × 0.50-2 cm in the basal portion, 1-5 × 0.30-0.75 cm in the middle portion, 0.98-2 × 0.20-0.55 cm in the superior portion, lanceolate to linear, acute, entire margin, pinkish red colored, glabrous on both sides; floral bracts 3-6 × 2-5 mm, ovate-triangular, apiculate, cucullate, green, glabrous on both surfaces, smooth. Flowers distichous, ca. 10 × 3 mm, sessile, appressed to the spike rachis; sepals 3-5 × 2-3 mm, obovate, rounded apex, free, asymmetric, coriaceous, green with hyaline margins, glabrous on both surfaces, smooth; corolla tubular, actinomorphic, 5-6 × 2-3 mm; petals 8-9 × 2-4 mm, oblanceolate, bright yellow, hyaline margin, free, glabrous on both surfaces, scaled crenated appendages inserted at the base, slightly incurved towards its apex, apiculate to slightly acuminated; stamens 6, 3 attached to each one of the petals and the other 3 attached to the ovary, free, shorter than the petals, inserted, white filaments, dorsoventrally flattened, 6-7 mm long; anthers 1.4-1.7 mm long, oblong, monothecal, sub basifixed, yellow, pollen yellow; style 8-9 mm long, erect, inserted, yellow, cupulate stigma; ovary 3-4 × 3-3.6 mm oblong, green, scarcely white-lepidote. Berry 7.5-8 × 6-7 mm, globose, yellowish green when immature, black when mature; persistent sepals, coloration before and after maturation is the same as the fruit; seeds 3.2-3.3 × 1-1.4 mm, fusiform, reddish brown, covered with mucilage.

Distribution and ecology. Aechmea laxiflora is endemic to western Mexico, from southern Sinaloa to western Oaxaca. This species occurs in oak forests, pine-oak forests, riparian forests, tropical deciduous forests, and tropical sub-deciduous forests from sea level to ca. 1,250 m asl (according to herbarium records, ^{Flores-Argüelles et al. 2023}).

Phenology. According to the examined herbarium specimens, A. laxiflora can be found flowering and fructifying almost all year long, however, flowering occurs most commonly from September to July, while fructification mostly occurs from January to April.

Etymology. Not specified in the protologue, but probably named because of the lax arrangement of the inflorescence branches.

Taxonomic notes. This taxon was considered initially as a synonym for A. bracteata by ^{Smith (1958)}, but after A. bracteata var. pacifica was described (^{Beutelspacher 1971}), A. laxiflora was proposed as a its synonym (^{Smith & Downs 1979}), however this proposal was not explicitly justified in such publication.

Additional specimens examined. Supplementary material 3

Supplementary material

Supplemental material for this article can be accessed here: https://doi.org/10.17129/botsci.3389

Supplemental material.