We included samples from most of the geographical distribution of Phylloderma stenops in South and Central America (Fig. 1). Morphological analysis included samples of all the named subspecies; in the case of the genetic analysis, DNA sequences could not be retrieved for P. s. boliviensis.

For the morphometric analyses, we examined 58 specimens of Phylloderma from the following collections: AMNH, American Museum of Natural History, New York, NY, USA; FML, Colección de Mamíferos, Fundación Miguel Lillo, Tucumán, Argentina; IAvH, Colección de Mamíferos del Instituto de Investigación de Recursos Biológicos Alexander von Humboldt, Villa de Leyva, Colombia; MEPN, Museo de Historia Natural Gustavo Orcés, Escuela Politécnica Nacional, Quito, Ecuador; QCAZ, Museo de Zoología, División de Mastozoología, Pontificia Universidad Católica del Ecuador, Quito, Ecuador; USNM, National Museum of Natural History, Smithsonian Institution, Washington, DC, USA; UV, Colección de Mamíferos de la Universidad del Valle, Cali, Colombia (Table SS1).

We obtained tissue samples from 18 Phylloderma specimens that are housed at QCAZ or were loaned from the AMNH; FMNH, Field Museum of Natural History, Chicago, IL, USA; MSB, Museum of Southwestern Biology, Albuquerque, NM, USA; and ROM, Royal Ontario Museum, Toronto, Canada (Table SS1).

Lastly, we included published information of locality data and measurements, listed in Medellín et al. (1986), Sánchez-Palomino et al. (1993), Trujillo and Albuja (2005), Esbérard and Faria (2006), Santos-Moreno and Gallardo (2014), Medina-Fitoria et al. (2015), Bomfim et al. (2017), Martínez-Cerón et al. (2019), and Verde et al. (2021) (Table SS1).

We gathered a set of 18 tissue samples from Phylloderma (Table 1). Some of these samples were old, dating from 1935. The tissues included heart or liver (c. 40 mg), claws (c. 3 mm), and biopsy punch wing snippets (c. 3 x 3 mm diameter).

Complete mitochondrial genome sequencing and assembly was performed using a genome skimming procedure, as recently performed for other Phyllostomidae groups (Camacho et al. 2022, 2024). Laboratory procedures were carried out at the NGS Core Facility of the Bernhard Nocht Institute for Tropical Medicine in Hamburg. The extraction and amplification of DNA were performed in a dedicated clean room facility. Contamination prevention protocols and negative controls were also implemented. The DNA extraction process for various sample types (dried skin, ethanol, or formaldehyde-preserved tissues) involved proteinase K digestion at 55°C using 20 µl of proteinase K and 220 µl of ATL lysis buffer (MinElute Reaction Cleanup kit, Qiagen). Prior to digestion, the samples were thoroughly washed with nuclease-free water (Qiagen). The incubation time for proteinase K digestion varied depending on the tissue type and sample preservation, ranging from 5 to 24 hours. Following digestion, DNA was extracted and purified using the Qiagen MinElute kit, with each sample eluted to a final volume of 60 µl. DNA concentration was measured using Qubit and Bioanalyzer instruments. For library preparation, the QIAseq FX DNA Library Kit (Qiagen) was used, with double index barcode labeling according to the manufacturer’s instructions. DNA fragmentation was often avoided due to the high degradation of nucleic acid (DNA ≤ 500 bp) material and low concentration (≤ 3 ng/µl). The HiFi PCR Master Mix from the QIAseq FX kit was utilized to amplify DNA regions with varying GC contents, minimizing sequencing bias caused by PCR, such as nucleotide misincorporations from cytosine deamination. The libraries underwent quality control to determine fragment size by using the Agilent 2100 Bioanalyzer, and the concentration was assessed using a Qubit 2.0 Fluorometer. After normalization, the samples and negative controls were pooled and subjected to sequencing on the NextSeq 2000 platform (2 × 100 cycles) (Illumina, San Diego, CA, USA). A detailed description of the whole protocol, including extraction and library preparation, is available at protocols.io (https://doi.org/10.17504/protocols.io.5jyl8q3r7l2w/v1)

Raw reads were first subjected to a qualitative assessment, followed by the removal of adaptor sequences and the filtration of polyclonal and low-quality reads (<55 bases long) using CLC Genomics Workbench 24 (Qiagen; https://digitalinsights.qiagen.com). Overlapping paired-end (PE) reads were merged to improve quality, while non-overlapping pairs and orphan reads were left unchanged. Deduplication was performed with an assumed 100% identity using BBTools v.39.27 (Bushnell 2014), expanding the length of contigs produced during de novo assembly. Custom assembly was conducted using MEGAHIT v.1.2.9 (Li et al. 2016) and SPAdes v.4.1.0 (Prjibelski et al. 2020; Wang et al. 2021) applications. All the assembled mitochondrial genomes were annotated using the MITOS2 metazoan pipeline (Al Arab et al. 2017), followed by visual inspection and manual adjustment in GENEIOUS v.9.0.5 (https://www.geneious.com).

Ribosomal RNA (rRNA) and transfer RNA (tRNA) loci were aligned using MUSCLE (Edgar 2004), whereas protein-coding gene sequences were aligned using TranslatorX (Abascal et al. 2010). We used the optimal partitioning schemes previously estimated by Camacho et al. (2022). The phylogenetic analysis was conducted using 38 partitions: one partition for rRNA genes, one partition for tRNA genes, and 36 partitions corresponding to the three codon positions of each of the 12 protein-coding genes. Phylogenetic relationships were inferred under maximum likelihood (ML) in RAxML-NG v.1.2.0 (Kozlov et al. 2019), using a partitioned GTR+I+G model selected under the Akaike Information Criterion (AIC). The analysis was started from 10 random and 10 parsimony trees. Nodal support was estimated using both Felsenstein bootstrap proportions (FBP) and transfer bootstrap expectation (TBE) (Lemoine et al. 2018), with the autoMRE bootstopping criterion. Bootstrapping converged after 200 replicates. Support values are reported in Figure 2 as FBP/TBE, both expressed as percentages. Trees were visualized and edited in FigTree v.1.4.4 (http://tree.bio.ed.ac.uk/software/figtree). Bootstrap values of 75% and higher indicated robust support, 50–75% indicated moderate support, and 50% or lower indicated minimal support (Moratelli et al. 2017). Finally, we estimated evolutionary divergence between groups in MEGA 11 (Tamura et al. 2021) as the mean number of base differences per site averaged over all sequence pairs between groups, with among-site rate heterogeneity modeled using a gamma distribution.

To root the phylogenetic analysis and ensure broad taxonomic representation, we selected outgroups that included at least one representative of each genus within the subfamily Phyllostominae (except Mimon, for which no complete mitochondrial genome was available), as well as one representative from the subfamilies Lonchophyllinae and Glossophaginae (Table 1), whose phylogenetic relationships to Phyllostominae are unclear (Camacho et al. 2022). We newly sequenced complete mitochondrial genomes from Gardnerycteris crenulata (QCAZ18512), Tonatia bakeri (QCAZ18699), Vampyrum spectrum (QCAZ18135), Macrophyllum macrophyllum (QCAZ15869), and Chrotopterus auritus (QCAZ17594). In addition, we retrieved six complete mitogenomes from GenBank: Trachops cirrhosus (PP410143), Lonchorhina aurita (KU743908), Glossophaga soricina (ON321893), Lophostoma nicaraguae (ON310506), Lonchophylla robusta (ON357721), and Phyllostomus discolor (ON357733). This outgroup sampling was intended to minimize the effect of any single outgroup on tree rooting; the taxonomic conclusions of this study are based on the reciprocal monophyly and genetic divergence of the Phylloderma lineages, rather than on the deeper relationships among outgroups.

We examined 58 Phylloderma specimens to obtain external and craniodental measurements. Only fully mature individuals were included, ensuring consistency by excluding subadult stages (Kunz et al. 1996; Novaes et al. 2021).

For each specimen, 26 metrics were recorded, including 16 craniomandibular and 10 external measurements. Total length (TL), tail length (T), hind foot (HF), ear length (E), and mass (W) were obtained from specimen tags when available. The measurements were taken by a single observer (M.A.C.) across all museum visits with a digital caliper (0.01 mm precision) and following a standardized protocol (measurement landmarks and orientation kept constant) to minimize observer-related variance. The measurements of P. stenops boliviensis specimens from the FML (n = 2) were provided by the curator of this museum. Because the Bolivian sample size was very small, these specimens were not included in multivariate analyses and were considered only for descriptive comparisons.

We adopted the cranial and dental terminology proposed by Velazco (2005), Tavares et al. (2014), and Garbino et al. (2020). For lower premolars, we followed the nomenclature of Miller (1907) and Cirranello et al. (2016), recognizing three teeth (p2, p3, and p4), with absent p1. We used standard abbreviations for tooth position, with upper-case letters for upper teeth and lower-case letters for lower teeth: I/i, upper/lower incisors; C/c, upper/lower canines; P/p, upper/lower premolars; M/m, upper/lower molars; and tooth positions indicated by numbers.

The external and craniomandibular measurements, and their abbreviations and descriptions are as follows: calcar length (CL), distance from the joint with the ankle to the tip of the calcar; ear length (E), distance from the inter-tragic notch of the ear to its tip; forearm length (FA), distance from the elbow (external edge of the olecranon process) to the wrist (including the carpals); hindfoot length (HF), distance from the ankle to the tip of the claws; metacarpal III (MET-III), distance from the joint to the wrist (carpal bones) with the 3rd metacarpal to the metacarpophalangeal joint of the 3rd digit; metacarpal IV (MET-IV), distance from the joint of the wrist (carpal bones) and the 4th metacarpal to the metacarpophalangeal joint of the 4th digit; metacarpal V (MET-V), distance from the joint of the wrist (carpal bones) and the 5th metacarpal to the metacarpophalangeal joint of the 5th digit; tail length (T), distance from the base of the tail to its tip; tibia length (TiL), distance from the proximal end of the tibia to the distal base of the calcar; total length (TL), head and body length excluding tail; weight (W), body mass in grams; braincase height (BCH), height of the braincase, posteriorly to the auditory bullae, from the basioccipital bone to the sagittal crest; breadth across upper molars (M2-M2), greatest breadth of the palate across the labial margins of the alveoli of M2s; breadth of brain case (BB), greatest breadth of the globular part of the braincase, excluding the mastoid and paraoccipital processes; condylocanine length (CCL), distance from the occipital condyles to the anterior border of the upper canines; condyloincisive length (CIL), distance between a line connecting the posteriormost margins of the occipital condyles and the anteriormost edge of the upper incisors; coronoid height (COH), perpendicular height from the ventral margin of the mandible to the tip of the coronoid process; dentary length (DENL), distance from the midpoint of the condyle to the anteriormost point of the dentary; greatest length of skull (GLS), distance from the occiput to the anteriormost edge of the premaxilla (including the incisors); mandibular toothrow length (MANDL), distance from the anteriormost edge of the lower canine to the posteriormost edge of m3; mastoid (process) breadth (MPW), greatest breadth across the skull, including the mastoid processes; maxillary toothrow (MTRL), distance from the anteriormost edge of the upper canine to the posteriormost edge of the crown of M3; molariform toothrow (MLTRL), distance from the anteriormost edge of P3 to the posteriormost edge of the crown of M3; palatal width at canines (C-C), distance between the outermost edges of the cinguli of upper canines; palatal length (PL), distance from the posterior palatal notch to the anteriormost edge of the incisive alveoli; postorbital constriction breadth (PB), least breadth at the postorbital constriction; zygomatic breadth (ZB), greatest breadth across the zygomatic arches.

Descriptive statistics (mean, range and standard deviation) were calculated for all measurements. Only craniomandibular metrics were used in statistical analyses. Missing measurements represented 3% of the morphometric dataset and were estimated using multiple imputation in the SPSS Statistics v. 20 software (IBM Corp. 2011). Data normality was assessed using logarithmic transformation and the Shapiro-Wilk test, which is effective for small to moderate sample sizes. Molecular findings guided an exploratory ANOVA to assess size differences across sexes within phylogroups. The Panamanian-South American form was treated as distinct from the Honduran form (P. s. septentrionalis). Phylloderma stenops boliviensis was excluded due to the limited number of specimens. The analysis focused on the greatest length of the skull (GLS). Because no marked size differences between sexes were detected, males and females were pooled for subsequent morphometric analyses. Using the same phylogroups, morphometric variation in 16 craniomandibular dimensions was assessed using independent-samples t-tests on log10-transformed measurements. Homogeneity of variances was evaluated with Levene’s test; however, given the strongly unbalanced sample sizes between groups, we report Welch’s t-test results (equal variances not assumed) for all comparisons as a conservative and robust approach.

Finally, to evaluate whether morphometric differentiation reflected proportional cranial differences rather than only overall size, we calculated Mosimann size-adjusted variables as the natural logarithm of each measurement minus the individual mean of all log-transformed measurements. A principal component analysis was then performed on these Mosimann variables in R (R Core Team 2026).

We successfully sequenced 15 Phylloderma mitogenomes (Table SS1). The phylogenetic analysis recovered two highly supported clades: one containing individuals distributed from Panama to South America, corresponding to P. stenops (Clade 1; Fig. 2), and another from Honduras, corresponding to the northern Central American lineage historically treated as P. s. septentrionalis (Clade 2; Fig. 2). Mitogenomic sequence divergence estimates revealed substantial genetic differentiation between these taxa. A complete mitogenome genetic distance of 8.60% was observed between P. stenops sensu stricto and the northern Central American lineage here revalidated as P. septentrionalis (Table 2).

Phylloderma septentrionalis is larger in craniomandibular and external measurements than P. s. stenops and P. s. boliviensis (Table 3). Log10-transformed variables met normality assumptions (Shapiro–Wilk, P > 0.05). An exploratory ANOVA using GLS as the dependent variable detected no significant size differences between sexes within phylogroups (P > 0.05), therefore, males and females were pooled. Because sample sizes were strongly unbalanced between phylogroups, we report Welch’s independent-samples t-tests (equal variances not assumed) for all craniomandibular comparisons. Welch’s t-tests indicated significant differences between P. septentrionalis and P. stenops across the examined cranial dimensions (Table 4), supporting the usefulness of craniometric traits for distinguishing these taxa.

The PCA based on Mosimann size-adjusted variables explained 25.14% of the total variance along PC1 and 21.53% along PC2 (Fig. 3). The plot showed partial displacement of P. septentrionalis from most specimens of P. stenops along PC1, although overlap among taxa indicates that proportional cranial variation is not fully discrete. The two specimens of P. s. boliviensis were positioned toward negative PC1 values, but this pattern should be interpreted cautiously because of the limited sample size. PC1 showed the strongest loadings for BCH and MLTRL, with additional contributions from M2M2, PL, and CC. PC2 showed the strongest loadings for MLTRL, BB, MPW, MTRL, PB, and MANDL (Table SS2).

Our results support the distinctiveness of northern Central American populations historically treated as Phylloderma stenops septentrionalis. This study confirms that this northern lineage is generally larger than P. stenops in both craniomandibular and external measurements, corroborating previous findings (Peters 1865; Goodwin 1940; Bárquez and Ojeda 1979). Welch’s t-tests recovered significant differences in craniomandibular measurements, indicating a marked size component in the morphometric differentiation between the two taxa. After size adjustment using Mosimann variables, the PCA showed partial differentiation of P. septentrionalis from most P. stenops specimens, indicating that proportional cranial differences are present but not fully discrete. Thus, the morphometric evidence is best interpreted as reflecting pronounced overall size differences, accompanied by subtler proportional variation and diagnostic qualitative dental traits.

Two distinct clades were identified, with a complete mitogenome divergence of 8.60% between the Honduran P. septentrionalis and the Panamanian-South American P. stenops. Because this estimate is based on complete mitochondrial genomes, it is not strictly comparable with single-locus distances such as cyt b or COI, but it provides a broader mitochondrial estimate of divergence across multiple genes. The magnitude of this divergence is consistent with values reported among congeneric chiropteran species based on mitochondrial markers, including the 2–11% cyt b range discussed by Bradley and Baker (2001); and is even higher than the divergence reported between Vampyressa villai and its sister species V. thyone (4.66%; Garbino et al. 2024), and comparable to mean mitochondrial divergence found between congeneric species of other phyllostomids: 7% for Sturnira, 9% for Artibeus, 7.5% for Carollia, 5.5% for Chiroderma, 10% for Dermanura, or 7.2% for Platyrrhinus (Bradley and Baker 2001; Velazco and Patterson 2008; Velazco 2013; Molinari et al. 2017).

The Northern Central American P. septentrionalis is generally larger than the Panamanian-South American form, with considerable size differences in GLS and MET III to MET V. Skull size varies, and even though skull shape is similar in both groups, there are notable dental differences (see Taxonomy Section). Phylloderma stenops boliviensis is intermediate in size between the other taxa. Thus, P. septentrionalis is distinguished by its larger size, diagnostic dental traits, and deep complete mitogenome divergence. Although some morphometric traits exhibit overlapping ranges and the size-adjusted PCA does not show complete morphometric separation, consistent trends in trait means, diagnostic cranial characters, and mitogenomic divergence support species-level differentiation. Based on this evidence, we propose revalidating Phylloderma septentrionalis Goodwin, 1940 as a species-level taxon. We acknowledge, however, that this interpretation is constrained by the limited number of specimens available from Central America and by morphological differences that are not equally marked across all characters. Therefore, we treat the revalidation of P. septentrionalis as a well-supported taxonomic hypothesis that should be further tested with additional material, especially from Costa Rica and adjacent regions, and with complementary genetic analyses using nuclear markers.

While multivariate analyses can summarize patterns of morphological variation, we also emphasize classical taxonomic comparison. We acknowledge the disparity in sample sizes among Phylloderma taxa, which can reduce statistical power (Cohen 1988). Nevertheless, the congruence between morphological differentiation and genetic divergence supports species-level distinction under an integrative lineage framework (de Queiroz 2007). This framework is particularly suited for lineages showing limited diagnosability, where morphological diagnosability may be limited but genetic divergence indicates evolutionary independence (Fišer et al. 2018). Consistent with the General Lineage Concept previously outlined (de Queiroz 2007), we interpret the congruent divergence observed across mitogenomic and morphometric datasets as evidence of evolutionary independence in Phylloderma populations.

This taxonomic decision also has direct biogeographic implications. Phylloderma septentrionalis appears to have a distribution restricted to the region between southern Mexico (Oaxaca and Chiapas) and Costa Rica (Carter et al. 1966; Timm et al. 1989; Santos-Moreno and Gallardo 2014). Phylloderma stenops is known to occur from southern Panama to South America. We here amend the distribution proposed by Solari et al. (2019), who suggested the range of P. s. septentrionalis to be from southern Mexico and Central America to northwestern South America (western Colombia, western Ecuador, and northwestern Peru). We also disagree with Martínez-Cerón et al. (2019) who suggested a trans-Andean clade for Phylloderma, distinct from a cis-Andean clade more closely related to specimens from eastern Ecuador and the Guiana Shield. The trans-Andean clade recovered by Martínez-Cerón et al. (2019), composed of samples from Colombia and Panama, appears nested in the larger cis-Andean clade, thus not supporting the existence of a monophyletic trans-Andean clade. Additionally, that study did not include samples from northern Central America, leaving the phylogeographic limits unclear.

To interpret these patterns, geology provides a context. The tectonic and geological history of Central America is known to have influenced bat distribution and speciation (Hernández-Canchola and León-Paniagua 2017). Geology has shaped the landscape and restricted gene flow, thus has been crucial to species diversification (Marshall 2007). The genetic distinction between the northern Central American and Panamanian-South American Phylloderma is consistent with major geological and topographical discontinuities in Central America, including the Talamanca region and the Hess Escarpment. The latter, which is a steep slope marking the boundary of the Eastern Pacific Rise, has been shown to be important as a barrier separating related species in several taxonomic groups, including bats (Hernández-Canchola and León-Paniagua 2017), insects (Halffter 1987), amphibians (Smith et al. 2007), reptiles (Jadin et al. 2012; Suárez-Atilano et al. 2014), and birds (Navarro et al. 2001; Rocha-Méndez et al. 2019).

At a finer scale, the Talamanca Range’s diverse ecological zones shape bat diversity and distribution. Species compositions in Costa Rica and Panama differ from those in the northern transition zone, including Nicaragua (Ortega and Arita 1998). Gutiérrez-García and Vázquez-Domínguez’s (2013) meta-analysis identified the “Panamanian clade”, a distinct genetic assemblage of species influenced by the formation of the Isthmus of Panama, which disperses towards Costa Rica along the Chorotega block, an important tectonic feature extending from Costa Rica to the Panamanian highlands (Giunta and Oliveri 2009). This highlights the impact of geological events on the evolutionary trajectories of species, with the Hess Escarpment playing a fundamental role both in defining the boundaries of tectonic activity and influencing species distribution.

Our study shows northern Central American (Mexico to Nicaragua) Phylloderma to represent a distinct clade. A critical limitation of this study is the absence of samples from Costa Rica, the region where the Hess Escarpment and the Talamanca Range are most prominent. This gap prevents precise localization of the phylogeographic break and limits our ability to assess which species these populations belong to under our revised taxonomy or even whether they represent an intermediate lineage. Consequently, we treat the role of specific barriers as an informed hypothesis that requires direct sampling across Costa Rica for confirmation.

While our study spans a broad geographic range, we acknowledge that the number of sequenced specimens is modest relative to the spatial scale. However, the use of complete mitochondrial genomes provides high-resolution insights into lineage divergence, and the geographic breadth of our sampling captures key biogeographic transitions. These data allowed us to formulate our species hypothesis based on mitogenomic divergence and spatial structure, but we emphasize that broader sampling and nuclear genomic validation will be essential to confirm species boundaries and assess population-level variation.

Further genetic analysis is needed to clarify the taxonomic status of Costa Rican and Panamanian specimens of Phylloderma. Despite being rare in collections, they are widespread in the Neotropics, thus are ideal for studying geographically structured diversification and hidden lineage boundaries (Stevens 2023). Our findings reveal that not only the northern Central American samples of this genus are distinct, but also that there is an underlying biogeographic element determining their diversification. This pattern is not unique to our studied species but is evident in other bats such as Lophostoma evote (Davis and Carter 1978), Uroderma davisi (Baker and McDaniel 1972; Mantilla-Meluk 2014), Sturnira parvidens (Goldman 1917), Glossophaga leachii (Gray 1844), G. morenoi (Martínez and Villa-Ramírez 1938), and Vampyressa villai (Garbino et al. 2024).

Despite the recent uncovering of previously unrecognized species endemic to the region, owing to the limited number of specimens in scientific collections (Nassar et al. 2020), and to the challenges in obtaining or accessing fresh or adequately preserved tissue samples for genetic analysis, there is likely still an underestimation of the bat biodiversity of Central America (Castellanos 2019). Our findings are consistent with a geographically structured divergence within Phylloderma, and they support a taxonomic reassessment of the genus. Future sampling across Costa Rica and adjacent regions will be necessary to evaluate whether the Talamanca Cordillera, the Hess Escarpment, or other geographic features correspond to the transition between P. septentrionalis and P. stenops.

To facilitate visualization of diagnostic differences among the Phylloderma taxa treated here, we summarize key qualitative characters and selected external and craniodental measurements in Table 5. Quantitative values are provided as ranges based on examined adult specimens; values for P. s. boliviensis are included for reference (n = 2) but were not used in statistical analyses.

Family Phyllostomidae Gray, 1825

Genus Phylloderma Peters, 1865