The first living cervid species described in the 21st century and revalidation of Pudella (Artiodactyla)

Abstract

Several issues regarding the systematics and taxonomy of the Neotropical deer tribe Odocoileini, an assemblage of 18 recognized living species currently allocated into 7 genera, remain unclear. The few available phylogenetic analyses indicate that some genera are not monophyletic and that species richness in the group is underestimated. One genus that presents both problems are the stocky and short-legged dwarf deer, Pudu. As currently understood, it includes 2 species, the Northern pudu, Pudu mephistophiles from Peru, Ecuador, and Colombia; and the Southern pudu, P. puda, from southern Chile and nearby Argentina. Here, by means of qualitative and quantitative morphologic analysis and the assessment of genetic variation, we showed that 2 distinct species are encompassed by the current concept of P. mephistophiles. The typical form distributes north of the Huancabamba Depression from northernmost Peru to the north (Ecuador and Colombia), while the other distributes south of the Huancabamba Depression and is a Peruvian endemic. As no name is available for the last one, we describe and name it here. This is the first living cervid species described in the 21st century and the first from the New World in over 60 years. Additionally, as the Southern Pudu (the type species of Pudu) is not sister to the 2 northern pudu species, we revalidate the genus Pudella to allocate the latter 2 species.

Resumen

Distintos aspectos de la sistemática y taxonomía de la tribu de ciervos neotropicales Odocoileini, que incluye un conjunto de 18 especies vivas reconocidas actualmente que se engloban en siete géneros, siguen sin estar resueltos. Los pocos análisis filogenéticos disponibles indican que algunos géneros no son monofiléticos y que la riqueza del grupo esta subestimada. Uno de los géneros que presenta ambos problemas es el género de ciervos pequeños y patas cortas Pudu. Con base en la clasificación actual, éste incluye dos especies, Pudu mephistophiles distribuido en Perú, Ecuador y Colombia y P. puda distribuido en el sur de Chile y áreas cercanas de Argentina. Este estudio mediante análisis morfológicos cualitativo y cuantitativo y la evaluación de la variación genética, muestra que el concepto actual de P. mephistophiles engloba a dos especies distintas. La forma típica se distribuye al norte de la Depresión de Huancabamba desde el extremo norte de Perú hacia el norte (Ecuador y Colombia), mientras que la segunda se distribuye al sur de la Depresión de Huancabamba y es endémica de Perú. Como no hay nombre disponible para esta última, aquí la nominamos y describimos formalmente. Esta es la primera especie viviente de cérvido descrita en el siglo XXI y la primera del Nuevo Mundo en más de 60 años. Además, como el pudú del sur (la especie tipo de Pudu) no es hermano de las dos especies de pudú del norte, revalidamos el género Pudella para colocar a las dos últimas especies.

Cervidae, deer, Odocoileini, Peru, Pudu, South America, Taxonomy

América del Sur, Cervidae, ciervos, Odocoileini, Perú, Pudu, Taxonomía

Available counts of animal species keep increasing globally for most groups. These increments are based both on new field discoveries (e.g. Jayat et al. 2016; Cid-Arcos and Campodonico 2019) as well as the taxonomic revision of already known species (e.g. Krabbe et al. 2020; Oskars and Malaquias 2020; Teta and D’Elía 2020); both scenarios are possible because of the existence of zoological collections, which are fundamental for taxonomy (Rocha et al. 2014). In addition to a shortage in the number of taxonomists that causes the so-called taxonomic impediment (Engel et al. 2021), we are in an age of species discovery (e.g. D’Elía et al. 2019). Naturally, discoveries are not uniformly even across groups, with most new animal species being insects (in particular coleopterans), while for some groups recent increments in species numbers are low or nonexistent. One of the latter groups is the deer family Cervidae, for which no new species has been described in the present century. After a stasis of several decades, 5 Old World muntjac species were described in the 1990s (Ma et al. 1990; Tuoc et al. 1994; Schaller and Vrba 1996; Binh Chau 1997; Giao et al. 1998; Amato et al. 1999). Similarly, the last new cervid from the New World, the brocket deer Mazama “bororo,” was proposed in the same decade (Duarte 1996), although its description does not satisfy the code requirements (see Duarte and Jorge 2003), and its distinction from M. americana jucunda—a form described more than a century ago by Thomas (1913)—remains unclear. A recent study has shown that M. “bororo” and M. jucunda apply to a single species-level lineage (i.e. “bororo” is a synonym of jucunda; Mantellatto et al. 2022). As such, different lines of evidence indicate that the current taxonomic scheme does not reflect species diversity within Cervidae and, in particular of the New World tribe Odocoileini.

The tribe Odocoileini constitutes an ecologically diverse assemblage of 18 recognized living species currently allocated into 6 genera (Duarte and González 2010). Two main morphotypes exist: small forms (i.e. Mazama, Subulo, and Pudu), mostly adapted to forested areas; and large forms (Odocoileus, Hippocamelus, Ozotoceros, and Blastocerus), mostly inhabitants of open forest, grasslands, and wetlands (Merino and Rossi 2010). Besides being charismatic animals of cultural and social relevance (e.g. source of food) and being the focus of numerous studies (e.g. Duarte and González 2010), several issues regarding the systematics and taxonomy of the group remain unclear. For instance, the few available phylogenetic analyses, based basically on mitochondrial DNA sequences, indicate that Hippocamelus, Mazama, and Pudu are not monophyletic (Duarte et al. 2008; Hassanin et al. 2012; Heckeberg et al. 2016; Gutiérrez et al. 2017; Heckeberg 2020). The extent to which those gene trees reflect the species tree is so far not completely known, as nuclear loci have not generally been included in the assessment of phylogenetic relationships among Neotropical cervids. Similarly, several lines of evidence indicate that the species richness of the group is underestimated. For instance, distinct cytogenetic and genetic studies have shown that the binomen Mazama americana encompasses a complex of distinct species (e.g. Duarte and Jorge 1996; Duarte et al. 2008; Abril et al. 2010; Gutiérrez et al. 2017; Cifuentes-Rincón et al. 2020).

One taxon that presents both enunciated deficiencies is the genus PuduGray, 1852, comprised of stocky and short-legged dwarf deer (Gutiérrez et al. 2017). As currently understood, it includes 2 species: the Northern Pudu, Pudu mephistophilesde Winton, 1896, from the paramo and Andean forests of Colombia, Ecuador, and Peru; and the Southern Pudu, Pudu puda (Molina, 1782), from the Valdivian forests of southern Chile and neighboring Argentina (Czernay 1987; Geist 1998). These dwarf deer are distributed in 2 disjunct areas along the Andes. Overall, species of Pudu differ from other cervids primarily by their size—both species are shorter than 43 cm in height at the shoulder due to the coalition in a single element of the cuboid-navicular bone and the medial and external cuneiform tarsal bones (Hershkovitz 1982; Escamilo et al. 2010). This latter trait is also present in the Asian cervids Muntiacus and Elaphodus (Hershkovitz 1982). Additionally, pudus have shorter cervical vertebrae and 8 caudal vertebrae, not 10 or 11 as in other deer species (Hershkovitz 1982; Geist 1998). Thomas (1913) considered that the cranial differences between P. puda and P. mephistophiles were large enough to recognize 2 genera; he named Pudella, with P. mephistophiles as its type species. The most notorious differences between both taxa are the lachrymal fossa, which is deep in Pudu but shallow in Pudella, and the presence of upper canines in the latter. Only 2 years after the proposal of Thomas (1913), both pudu species were considered cogeneric, with Pudella regarded as a subgenus of Pudu (Lydekker 1915). This latter scheme has predominated since then and is the one currently used (e.g. Mochi and Carter 1953; Whitehead 1972; Weber and González 2003; Grubb 2005; Mammal Diversity Database 2022). However, some authors (e.g. Cabrera and Yepes 1940) recognized both genera, while Haltenorth (1963) regarded Pudu as a synonym of Mazama. However, even when phylogenetic analyses have shown that both pudu species are not sister to each other (Heckeberg et al. 2016; Gutiérrez et al. 2017; Heckeberg 2020), Pudella remains in the synonymy of Pudu.

The northern pudu, P. mephistophiles, as currently delimited—even when its distribution has been depicted as continuous by some authors (e.g. Eisenberg and Redford 1993)—distributes in 2 geographically disjunct areas (Escamilo et al. 2010; Barrio and Tirira 2019; Fig. 1). The northern population distributes from the central Andes of Colombia to the northern Peruvian Andes (ca. 5°30ʹN to 5°14ʹS; Hershkovitz 1982; Tirira 2007; Barrio and Tirira 2019). The second population ranges along the northern and central parts of the eastern Peruvian Andes, east of the Marañon River (ca. 5°42ʹS to 11°15ʹS). The nearest distance between the 2 populations, about 50 km, lays at the Huancabamba Depression, a well-characterized barrier of low elevation and drier environments along both the Huancabamba and Marañon rivers that splits forested humid areas (Vuilleumier 1984). Interestingly, Rivera-Portilla et al. (2022) recently showed that a distribution model based on records of the northern population of P. mephistophiles is a poor predictor of the known distribution of the southern population; i.e., the northern population of P. mephistophiles has different climatic preferences than the southern population.

Fig. 1.

Distributional ranges of the 2 species of Pudella (pink: P. mephistophila s.s.; blue: P. carlae n. sp.). Numbers (1–9) signals of locality of the specimens of Odocoileini sequenced in this study (see Table 1 for more details; the placement of locality 9 is approximated). Map developed by Oscar Chipana.

Additionally, the northern population of P. mephistophiles presents color variation, which was the basis for a subspecific proposal. Lehmann (1945) described the subspecies Pudu (Pudella) mephistophiles wetmorei for populations from southern Colombia, retaining the typical form, P. m. mephistophiles, for populations from the Páramos de Papallacta in northern Ecuador and further south. However, Frädrich (1975) and Hershkovitz (1982), analyzing specimens from several localities in Ecuador and Colombia, showed that variation in coat coloration within each of the proposed subspecies was greater than between them. Consequently, currently, P. mephistophiles is considered monotypic (Escamilo et al. 2010).

Here, we present a taxonomic study focused on both northern and southern populations of P. mephistophiles to further evaluate their distinction. The study is based on analysis of morphological characters and mitochondrial DNA sequences. Results allow us to hypothesize that Peruvian populations south of the Huancabamba Depression represent an undescribed species of pudu that we describe and name here.

Materials and methods

This research did not involve live animals as the study specimens were already housed in collections.

Genetic analyses

Genetic analyses are based on a fragment (801 bp) of the mitochondrial gene Cytochrome b (Cytb). We sequenced 5 specimens of southern P. mephistophiles, representing the new species described herein, another specimen of northern P. mephistophiles (i.e. P. mephistophiles sensu stricto, see below), one of M. americana, one of M. chunyi, and another of Hippocamelus antisensis. Locality information for the sequenced specimens is given in Table 1 and Fig. 1. These new sequences were integrated into a large matrix of sequences retrieved from other 46 specimens of Odocoileini, with sampling guided by the results of Duarte et al. (2008), Gutiérrez et al. (2017), Heckeberg et al. (2016), Cifuentes-Rincón et al. (2020), and Sandoval et al. (2022) in order to have a broad sampling of species-level lineages whether they are recognized in current classification as distinct species or not (e.g. main lineages of M. americana and H. antisensis). Regarding pudus, the matrix includes a total of 12 sequences of P. puda and 11 specimens of P. mephistophiles sensu lato (6 of P. mephistophiles s.s. and 5 of a new species described below). The matrix also includes 1 sequence of the Caribou (Rangifer tarandus); however, as in some phylogenetics analysis, Rangiferini falls nested within Odocoileini (see Fig. 3 in Heckeberg et al. 2016); the outgroup was formed with sequences of representatives of the tribes Alceini (Alces alces), Capreolini (Capreolus capreolus), Muntiacini (Muntiacus muntjak), and Cervini (Cervus elaphus). Sequence details, including species name, specimen catalog number, and GenBank accession numbers (including those corresponding to the sequences acquired here), are presented in Fig. 2.

Table 1.

List of the specimens newly sequenced in this study (locality numbers are those of Fig. 1) including catalog number, locality data, and GenBank accession numbers of the newly gathered sequences.

Species	Catalog number	Country	Locality	Locality number	Type of sample	Accession number
Hippocamelus antisensis	CORBIDI 38	Peru	Huancavelica	1	Dry skin	ON721323
Mazama americana	CORBIDI 183	Peru	Madre de Dios	3	Dry skin	ON721322
Mazama chunyi	CORBIDI 26	Peru	Cusco	2	Dry muscle	ON721321
Pudella carlae n. sp.	MUSM 23144	Peru	San Martín	4	Dry muscle	ON721318
Pudella carlae n. sp.	MUSM 25713	Peru	Amazonas	5	Dry muscle	ON721319
Pudella carlae n. sp.	MUSM 55639	Peru	Amazonas	6	Fresh muscle	ON721317
Pudella carlae n. sp.	CORBIDI 39	Peru	Junin	7	Dry muscle	ON721315
Pudella carlae n. sp.	CORBIDI 40	Peru	Junin	8	Fresh muscle	ON721316
Pudella mephistophila	MEPN 10944	Ecuador	No locality data	9	Dry skin	ON721320

Fig. 2.

Cytb gene tree of haplotypes of specimens of the Odocoileini. Majority-rule consensus tree obtained from the Bayesian analysis of 53 Cytb gene sequences of odocoileines. Numbers indicate PP (left of the diagonal) and ultrafast bootstrap of the MaxL analysis (right of the diagonal) values of adjacent nodes; a < indicates that the given node has less than 50% of ultrafast bootstrap support; a dash indicates that the given node was not recovered in the MaxL analysis. Terminal designations are species names and GenBank accession numbers, respectively. The haplotype of the type specimen of Pudella carlae n. sp. is marked by #.

DNA was purified from skin or muscle samples (Table 1) using a commercial DNA purification kit (Wizard SV Genomic DNA Purification System, Promega) following the manufacturer’s instructions. A fragment of the Cytb gene was amplified using primers MVZ05 5ʹ CGAAGCTTGATATGAAAAACCATCGTTG 3ʹ and MVZ16 5ʹ AAATAGGAARTATCAYTCTGGTTTRAT 3ʹ (Smith and Patton 1993) for samples CORBIDI26, CORBIDI39, CORBIDI40, and MUSM55639. Meanwhile, samples CORBIDI38, and CORBIDI183, MEPN10944, MUSM23144, and MUSM25713 were amplified in 2 fragments using primers MVZ05-OCT439R and OCT406F-MVZ16 (see Cadenillas and D'Elía 2021). All PCR was conducted with the following conditions: an initial denaturation period of 5 min at 94 °C, followed by 37 cycles of 45 s of denaturation at 94 °C, 30 s of annealing at 51 °C, and 90 s of extension at 72 °C, and then a final extension period of 10 min at 72 °C. Amplicons were sequenced at an external service (Macrogen, Korea). New sequences were submitted to GenBank (ON721315 to ON721323).

Sequence alignment was done with Clustal as implemented in MEGA 6 (Tamura et al. 2013) using the default parameters. The presence of internal stop codons and reading frame shifts was checked by a visual inspection of the alignment; no correction was needed. Relationships among Cytb sequences were inferred using 2 methods: Bayesian inference (BI) as implemented in MrBayes 3.1 (Ronquist and Huelsenbeck 2003); and maximum likelihood (MaxL) using IQ-TREE (Nguyen et al. 2015) implemented on the IQ-TREE web server (Trifinopoulos et al. 2016). The GTR+G+I model, selected with jModelTest (Darriba et al. 2012), was implemented in both analyses. The BI analysis was conducted with 2 independent runs with 5 heated and 1 cold Markov chains each. Model parameters were estimated in MrBayes; base composition and GTR parameters assumed a Dirichlet process prior; all other parameters were set as uniform interval priors. Runs were run for 10⁶ generations, with trees sampled every 1,000 generations. To check if runs converged on a stable log-likelihood value, we plotted log-likelihood values against generation time and checked the average standard deviation in split frequencies. The first 25% of the trees sampled were discarded as burn-in; remaining trees were used to compute a 50% majority-rule consensus tree and to obtain posterior probability (PP) values for each clade. The MaxL analysis was performed in the online implementation (Trifinopoulos et al. 2016 at https://www.hiv.lanl.gov/content/sequence/IQTREE/iqtree.html) of IQ-TREE specifying the selected models of molecular evolution, perturbation strength set to 0.5, and number of unsuccessful iterations set to 100. Branch support was estimated via ultrafast bootstrap (UFB; Hoang et al. 2018). Observed percentage of sequence divergence (p-distances) between pairs of haplotypes, local samples, and species was calculated with MEGA 6 ignoring sites with missing data.

Morphological analyses

The morphological assessment included analysis of 15 skulls of P. puda from Chile and Argentina (data for 12 taken from Hershkovitz 1982), and 25 skulls of specimens of P. mephistophiles s.l. (22 of P. mephistophiles s.s. and 3 of a new species described below). Our sampling did not include the type specimen of P. mephistophiles. Analyzed specimens (Appendix  I) are housed at the collections of the Museo de la Universidad del Valle (UV, Cali, Colombia), Museo de Historia Natural de la Universidad del Cauca (MHNUC, Popayán, Colombia), Museo de Historia Natural Gustavo Orcés de la Escuela Politécnica Nacional (MEPN, Quito, Ecuador), Museo de Historia Natural Javier Prado de la Universidad Nacional Mayor de San Marcos (MUSM, Lima, Perú), Museo Argentino de Ciencias Naturales (MACN, Buenos Aires, Argentina), and Museo de La Plata (MLP, La Plata, Argentina).

The following 12 skull measurements were taken using a caliper to the nearest 0.1 mm: maximum length (ML), condylo basal length (CBL), condylo premolar length (CPL), zygomatic breadth (ZB), greatest cranium breadth (GCB), nasal length (NL), postorbital width on zygomatic arch (POO), preorbital width on frontal (PRO), deep or shallow lachrymal fossa (LF), distance between pedicels (DBP), length of left antler (LAL), and length of right antler (LAR). A principal component analysis (PCA) was performed using IBM SPSS Statistics 21 (IBM 2012) based on the skull measurements recorded in all specimens (males and females): GCB, ZB, NL, and ML. The length of 2 skulls with broken premaxilla of the southern P. mephistophiles considers an estimated addition of 1 mm.

Results

Phylogenetic analysis

Both BI and MaxL analyses recovered mostly resolved and congruent topologies, although some relevant differences are found. In the Bayesian analysis (Fig. 2) Odocoileini is recovered as monophyletic, although lacking significant support (PP = 0.52) and appears sister to Rangiferini in a highly supported relationship (PP = 1). Meanwhile, in the MaxL analysis Rangifer is nested between the odocoileine clade (UB = 93). Relationships within the odocoiline clade are mostly resolved in both analyses; polytomies mostly involve intraspecific relationships. Differences between the Bayesian and MaxL topologies pertain to relationships toward the base of the odocoileine–rangiferine clade involving the 4 main lineages of this clade (Fig. 2). Haplotypes of specimens currently assigned to Mazama fall into distinct lineages that together do not form a monophyletic group. Similarly, haplotypes recovered from specimens of Hippocamelus do not form a clade, but 2 clades that are not sister to other; one of these clades (PP = 1; UB = 100) is formed by 3 haplotypes of H. antisensis and the other clade (PP = 1; UB = 99) included 1 haplotype of H. bisulcus and another of H. antisensis. Therefore, not only is Hippocamelus polyphyletic, but also the species H. antisensis. Finally, the genus Pudu as currently delimited is not monophyletic. Clades that included exclusively haplotypes of P. puda (PP = 1; UB = 100) and P. mephistophiles s.l. (PP = 0.95; UB = 98), respectively, are not sister to each other. Within the latter variants recovered from specimens of P. mephistophiles collected south of the Huancabamaba Depression form a clade (PP = 0.99; UB = 99) that in the Bayesian tree is sister to the clade form by haplotypes of P. mephistophiles from the north of the Huancabamaba Depression (PP = 0.56). In the MaxL tree haplotypes of P. mephistophiles from north of the Huancabamaba Depression form a paraphyletic group to the clade of southern haplotypes. On average haplotypes from the south and the north groups of P. mephistophiles s.l. diverge by 4.6% (Supplementary Data SD1).

Morphological analysis

The PCA revealed that all variables were positively correlated with the first principal component (Table 2). PC1 explains 76.1% of the observed variation and PC2 13.9% (Table 3). The plot of individual scores against the first 2 PCs shows that specimens of P. mephistophiles s.l. separate into 2 well-differentiated, nonoverlapping clusters. These groups include individuals north and south of the Huancabamba Depression, respectively, and segregate both along PC1 and PC2. Specimens of P. puda form a third cluster that differs along PC2 from P. mephistophiles south of the Huancabamba Depression and separates along PC1 from individuals of P. mephistophiles north of the Huancabamaba Depression (Fig. 3A) When plotting scores along PC1 and PC3 the 3 clusters (i.e. P. puda, P. mephistophiles south, and P. mephistophiles north of the Huancabamaba Depression) segregate completely in multivariate space; in particular, P. mephistophiles south and P. mephistophiles north of the Huancabamaba Depression are well segregated (Fig. 3B). Similarly, these clusters show significant differences in skull measurements (Supplementary Data SD1). Finally, we note that nasal length differs between sexes of P. puda (females 34.1 mm, males 41.7 mm), while no measurements seem to differ between sexes of the sample of P. mephistophiles north of the Huancabamaba Depression (Supplementary Data SD1). As only females compose the sample of P. mephistophiles south of the Huancabamaba Depression, differences between sexes could not be evaluated.

Table 2.

Component matrix and correlations of a PCA performed on skull measurements of specimens of Pudu and Pudella.

	Component and correlation
	1	2	3	4
ML	0.933	−0.192	−0.059	−0.300
GCB	0.904	0.144	−0.374	0.148
NL	0.847	−0.453	0.218	0.173
ZB	0.799	0.541	0.261	−0.001

Extraction method: principal component analysis.

Table 3.

Total variance explained per the first 4 principal components of a PCA performed on skull measurements of specimens of Pudu and Pudella.

Component	Total	% variance	% accumulated
1	3.044	76.092	76.092
2	0.556	13.894	89.985
3	0.259	6.478	96.463
4	0.141	3.537	100.000

Fig. 3.

Individual scores of a PCA performed on cranial measurements of Pudu puda (squares), Pudella mephistophila (crosses), and Pudella carlae n. sp. (circles) on (A) the first 2 principal components (PC1 = 76.092%; PC2 = 13.894%), and (B) the first and third principal components (PC1 = 76.092%; PC3 = 6.478%).

Discussion

The present study corroborates early claims stating that species richness of the cervid tribe Odocoileini is underestimated, and as such, current classification does not mirror their evolutionary history.

Genetic and morphologic evidence allowed identification of an unnamed lineage of pudu distributed in the Peruvian Andes south of the Huancabamba Depression. This lineage is strongly supported in the phylogenetic analyses. The observed divergence value between the south and north groups of P. mephistophiles s.l. falls in the range observed for comparisons between odocoileine species pairs, with several species pairs showing values below that of both lineages of P. mephistophiles s.l.; moreover, the majority of values above that observed between both main lineages of P. mephistophiles s.l. involve comparisons between species of distinct genera (Supplementary Data SD1). Remembering that no value is indicative of species-level distinction and the fact that species limits as well as genus delimitation within Odocoileini are still poorly understood, along with several known cases of historical hybridization (e.g Cathey et al. 1998; Latch et al. 2011), factors that would bias comparison values, we suggest that these values should be interpreted which caution. Importantly, cranial morphological comparisons (Fig. 3; Supplementary Data SD1) of members of this lineage with those of its sister lineage corroborates its distinction. However, we acknowledge that the sample size of the Peruvian lineage is small; a situation that is commonly seen in taxonomic studies addressing large mammal species (Moratelli 2014). Interestingly, the lineages of pudu to the south and north of the Huancabamba Depression show distinct climatic preferences (Rivera-Portilla et al. 2022). Available lines of evidence allow us to hypothesize that the newly uncovered Peruvian lineage of pudu represents a species distinct from mephistophiles s.s. We note that both nominal forms associated with mephistophiles—wetmoreiLehmann, 1945 and fusca Spillmann, 1931—apply to specimens north of the Huancabamba Depression. The form wetmorei originally proposed as a subspecies of mephistophiles was based on putative color coat differences of specimens from southern Colombia from those of typical Ecuadorian mephistophiles. However, the distinction between both nominal forms was earlier disregarded by a detailed assessment of color variation conducted by Hershkovitz (1982; see also Frädrich 1975). Similarly, the form fusca Spillmann, 1931, from the surroundings of Quito, Ecuador, is a full synonym of mephistophiles s.s., which also comes from the same general area from nearby Quito (see Hershkovitz 1982). As such, no name is available for the Peruvian pudu; therefore, we describe it below as a new species.

In current classifications mephistophiles s.l. is allocated to the genus Pudu, whose type species is puda. However, all phylogenetic analyses (including ours; Fig. 2) show that puda and mephistophiles s.l. are not sister to each other. As such, the genus Pudu is here restricted to the southern pudu P. puda. The generic name Pudella is available for mephistophiles, given that this form is its type species (Thomas 1913). As such, we remove Pudella from the synonymy of Pudu and allocate to it mephistophiles s.s. and the new species described below. The usage of Pudella, a name of feminine gender, conveys that to accomplish gender agreement the species name mephistophiles needs to be adjusted to mephistophila (see article 31.2 of the Code ICZN 1999).

Family Cervidae Goldfuss, 1820

Genus PudellaThomas, 1913

Pudella carlae new species

Peruvian Yungas Pudu

(Figs. 4 and 5)

• Pudu mephistophilesde Winton, 1896 (part)

Holotype

An adult female (MUSM 55639), collected by Fanny Cornejo and Carlos Tello on 10 December 2009 (field number CT1053) and preserved as skin and skull (Fig. 4).

Fig. 4.

Holotype (MUSM 55639; female) of Pudella carlae n. sp.: (A) dorsal view of the skull; (B) ventral view of the skull; (C) lateral view of the skull; and (D) lateral view of the mandible.

Type locality

Peru, Amazonas, Rodríguez de Mendoza, Valle de la Colpa (−6.3884, −77.2325, 2,360 m).

Diagnosis

Pelage intense orange brown, with coarse, long hair, darker on the back, with facial mask dark brown not reaching the forehead. Yellowish cream-colored collar mostly absent or much reduced. Ears oval-shaped, with brownish external surface and inner surface with a grayish to pure white coloration. Abdomen and ventral side of legs vary between ochre and light reddish; flanks with tonalities between dark orange and reddish; legs and feet dark brown. First incisor (I1) moderately spatulated.

Skull measurements (mm) of the holotype

ML 148.2; CBL 141.7; CPL 81.8; ZB 74.7; GCB 50.9; NL 37.5; POO 67.3; PRO 32.5.

Comparisons

Pudella carlae n. sp. is larger (ca. 7 to 9 kg, possibly heavier) than P. mephistophila (ca. 5 to 6 kg; Czernay 1987). Besides size, the main differences between both species are associated with the pelage (Fig. 5). The coat of P. carlae n. sp. is less coarse than that of P. mephistophila and the general coloration of the body is rich reddish brown or orange-red, while in P. mephistophila ranges from dull dark brown to dull reddish brown. The flanks of the new species are brown to dark reddish and the whole legs and feet are very dark brown, close to black, while in P. mephistophila the lower abdomen and the ventral side of the hind legs vary between ochre and dark reddish. Pudella carlae n. sp. mostly lacks the cream-colored collar that varies in tones and extension shown by P. mephistophila. Pudella mephistophila presents a dark brown head and face; the muzzle is also brown and the rhinarium is black and bulbous. Pudella carlae n. sp. presents the external surface of the ears brownish but the inner surface is mostly covered by white hair, giving a grayish to white coloration, while in P. mephistophila the amount of white is reduced.

Fig. 5.

Illustrations of (1) Pudella mephistophila, (2) Pudella carlae n. sp., and (3) Pudu puda (Drawings by Omar Custodio).

The new species is smaller (ca. 7–9 kg) than P. puda (ca. 7–13 kg; Czernay 1987) and also is morphologically different. The most remarkable difference in its external morphology is coloration (Fig. 5). Pudu puda is mostly bicolored, where the general body is reddish to olive-brown, the whole underside including the inner front legs, and the distal part of the legs, as well as the inner side of the ears, a small patch over the eyes—sometimes as an ocular ring, are ochre-colored. The front part of the neck is variegated ochre-colored mixed with the general coloration of the body.

Both species of Pudella do not differentiate in their tegumentary glands, which appear rudimentary. The preorbital gland is contained in an inconspicuous shallow opening. Foot glands are located within shallow, skin folds at the front of the feet just above the hooves.

As in P. mephistophila the upper canine persists in most adults of P. carlae n. sp. The first incisor (I1) of P. carlae n. sp. is characterized by being more spatulated than P. mephistophila (Fig. 6), almost as spatulated as in P. puda. The 3 pudu species differentiate in skull morphology (Fig. 7; Supplementary Data SD2). A marked difference is that P. puda presents a deep lacrimal fossa, absent in both species of Pudella. In P. carlae n. sp. the skull is proportionally more elongated when considering the premaxilla and the nasal bone lengths. This skull characteristic is referable when comparing the new species to both P. puda and P. mephistophila. The premaxilla and nasals are larger in the P. carlae n. sp. than in P. mephistophila. The lacrimal fossa is small and shallow as in P. mephistophila. Similarly, P. carlae n. sp. presents wider brain case and zygomatic breadth than that of P. mephistophila.

Fig. 6.

Incisors of (left) the holotype (MUSM 55639) of Pudella carlae n. sp. and (right) an Ecuadorian specimen of Pudella mephistophila (MEPN 113-2-CB).

Fig. 7.

Skulls of Pudu puda and Pudella spp. (A) Pudu puda from Argentina (MACN 47.219); (B) holotype of Pudella carlae n. sp. (MUSM 55639); and (C) Pudella mephistophila from Ecuador (MEPN 09212)—note that the premaxillary bone of this specimen is broken (see a photo of an unbroken subadult specimen at Supplementary Data SD2).

Distribution

Pudella carlae n. sp. distributes from the southeast of the Huancabamba Depression in northern Peru, along the Peruvian Yungas (elfin and cloud forests) through the northern and central part of the eastern side of the Peruvian Andes (5°42ʹ30″S to 11°15ʹ00″S), at elevations between 1,800 and 3,300 m (Fig. 1). As such, the distribution of P. carlae n. sp. is confined between 2 dry areas, the Marañon River dry valley in the north and the Mantaro River valley in the south.

Etymology

The specific name carlae is given after the biologist Carla Gazzolo as a form of showing a high esteem to her. Her actions helped to save the life of the first author after a life-threatening vascular problem.

Common name

We propose the common name for P. carlae n. sp. to be Peruvian Yungas Pudu highlighting the ecoregion and country where the species distributes. The use of Pudu as part of the common name is in order to keep the tradition of using this epitope to refer to stocky and short-legged dwarf deer along the Andean Cordillera. Agreeing with the common name for the new species, it is suggested to change the common name for P. mephistophila to Ecuadorian Paramo Pudu given the ecosystem and country it primarily occurs.

Natural history

Not much is known about the biology of the Peruvian Yungas Pudu; we note that Grimwood revealed its presence in Peru in 1969 (Grimwood 1969). As referred by local people and observed with camera traps, P. carlae n. sp. feeds on forbs and leaves from woody plants, preferably shrubs and small trees; it sometimes even climbs inclined tree trunks to reach higher leaves and probably fruits (as told by local people from several places). Personal observations (JB) indicate that it also feeds on fruits, especially those found on the ground and growing on shrubs. In general, regarding its dietary items, P. carlae n. sp. is similar to P. mephistophila, and even P. puda (Pavez-Fox et al. 2015); differences relate to the species of plants eaten as there are differences in the vegetational composition of the areas where these species distribute. In areas close to elfin forest and shrubs, the Peruvian Yungas Pudu occasionally ventures into the wet eastern puna grassland, which vegetationally resembles the paramo to the north of the Huancabamba Depression.

Conservation

Pudella carlae n. sp. has not been assessed by the IUCN Red List. While a previous assessment (Barrio and Tirira 2008) regarded P. mephistophiles s.l. as vulnerable (VU), in a 2019 assessment it was considered that there was no sufficient available information to quantitatively measure threats and rates of decline for P. mephistophiles s.l.; as such, the species was categorized as Data Deficient (Barrio and Tirira 2019).

The distribution area estimated for P. carlae n. sp., based on data that were compiled for the currently in process 2022 Red List of Peruvian Species, is below 27,000 km², giving a measurement of Extent of Occurrence (EOO) above the 20,000 km² threshold to be considered as a Vulnerable Species. However, most of this EOO includes large areas that contain cities, large towns, and surrounding anthropogenic areas. The actual area of suitable habitat might be under 20,000 km². Moreover, there is also a continued decline in extent and quality of habitat. If we base the EOO on the available area of distribution, P. carlae n. sp. could be regarded, under the criteria B1b(iii), as a species vulnerable to extinction (VU).

Pudella carlae n. sp. occurs in several Peruvian national protected areas: Río Abiseo National Park, Yanachaga Chemillén National Park, Pampa Hermosa National Sanctuary, Cordillera Colán National Sanctuary, Pui Pui Protection Forest, Alto Mayo Protection Forest, and Chayu Nain Communal Reserve. It also occurs in Vista Alegre–Omia Regional Conservation Area, where the type specimen was collected. While the conservation perspective may seem positive based on the large number of protected areas where the species occurs, the real level of protection in each of these areas varies based on the conservation measures applied.

Final considerations

Our study corroborates early claims regarding that species richness of Neotropical cervids is still underestimated. The same is true related to the need of adjusting the generic classification of the tribe Odocoileini. Both scenarios, even after the description of P. carlae n. sp. with the revalidation of Pudella, still hold for the tribe. For instance, species currently allocated to Mazama do not form a monophyletic group (see also Smith et al. 1986; Gilbert et al. 2006; Duarte et al. 2008; Hassanin et al. 2012; Escobedo-Morales et al. 2016; Heckeberg et al. 2016; Cifuentes-Rincón et al. 2020). Therefore, Mazama, as currently delimited, even after the recent revalidation of Subulo to include gouazoubira (Bernegossi et al. 2023), represents a remarkable case of morphological convergence (Duarte et al. 2008) or more likely of a conservative morphotype that has persisted during diversification of Odocoileini. Moreover, as has been shown by Cifuentes-Rincón et al. (2020) even M. americana is not recovered as monophyletic and represents a cryptic complex of brocket deer species. Similarly, results suggest that H. antisensis may represent a complex of 2 species that are not even sister to each other (Fig. 2; see also Heckeberg 2020). As such, we expect that as new studies are published, which should ideally be based on larger samples and include nuclear DNA sequences, several departures from the current classification will be formally proposed. Furthermore, the inclusion of new cytogenetic data would prove useful to further assess species limits within Odocoileini, including that between the species of Pudella (see an example in Duarte and Jorge 2003).

Noting that several classificatory inconsistencies of Odocoileini have been known for several years now (e.g. Gilbert et al. 2006; Duarte et al. 2008), we visualize our study as a contemporaneous step toward a classification that more accurately reflects diversity and phylogenetic relationships of the group (see also Bernegossi et al. 2023)—providing a more robust foundation from which to address other aspects of the evolutionary history of Odocoileini, including historical biogeography, and the tempo and rhythm of group diversification.

Supplementary data

Supplementary data available at Journal of Mammalogy online.

Supplementary Data SD1.—Supplementary tables.

Supplementary Data SD2.—Supplementary figures.

Version of Record, first published online March 1, 2024, with fixed content and layout in compliance with Art. 8.1.3.2 ICZN.

Nomenclatural statement.—A Life Science Identifier (LSID) number was obtained for this publication: urn:lsid:zoobank.org:pub: 6D981A05-B780-472E-9F75-125665FFF0E6

Acknowledgments

Carlos Tello and Fanny Cornejo allowed access to the skull and skin of the type specimen housed at the mammal collection at the Javier Prado National Museum of Natural History of the Universidad San Marcos, Lima, Peru. Several curators at museum mammal collections helped to access species samples: Victor Pacheco granted access to several deer specimens housed at the Museo de Historia Natural Javier Prado, Universidad Nacional Mayor de San Marcos, Lima, Peru; María del Pilar Rivas granted access to the mammal collection of the Museo de Historia Natural, Universidad del Cauca, Popayán, Colombia; Alan Giraldo granted access to the mammal collection in Universidad del Valle, Cali, Colombia; Luis Albuja granted access to the mammal collection of the Museo de Historia Natural, Escuela Politécnica Nacional, Quito, Ecuador; David Flores granted access to the mammal collection in the Museo Argentino de Ciencias Naturales, Buenos Aires, Argentina; and Diego Verzi, who granted access to the mammal collection at Museo de la Universidad de La Plata, La Plata, Argentina. Special thanks go to Omar Custodio for illustrations of the species, Oscar Chipana for elaboration of the map, and Pablo Teta for assistance with Figure 3. Antonio García, Carla Gazzolo, and Fabiana Barrio accompanied the senior author during fieldwork. Two anonymous reviewers made valuable suggestions on an earlier version of this study.

Author contributions

JB: conducted the initial conceptualization of the study; performed sample collection and field observations; analyzed morphologic data; wrote an initial draft of the manuscript; got financial support. EEG participated in the initial conceptualization of the study. GD gathered and analyzed genetic data; wrote the final manuscript; got financial support. All authors read the final manuscript and accepted submission of this work for publication.

Funding

The study was mainly funded by the Mohamed Bin Zayed Species Conservation Fund (Project 0925185). Fondo Nacional de Desarrollo Científico y Tecnológico (FONDECYT 1221115) provided financial support for the genetic-based analysis.

Conflict of interest

None declared.

Data availability

New sequences are available in GenBank (ON721315 to ON721323).

Appendix I

List of specimens of Pudu and Pudella included in the morphologic-based analyses. Species allocation follows the taxonomic scheme here proposed (see text). For each specimen we provide available locality information and catalog number. Acronyms are as follows: AMNH, American Museum of Natural History (New York, United States); CORBIDI, Centro de Ornitología y Biodiversidad (Lima, Perú); FMNH, Field Museum of Natural History (Chicago, United States); MACN, Museo Argentino de Ciencias Naturales Bernardino Rivadavia (Buenos Aires, Argentina); MCZ, Museum of Comparative Zoology, Harvard University (Cambridge, United States); MEPN, Museo de la Escuela Politécnica Nacional (Quito, Ecuador); MHNUC, Museo de Historia Natural de la Universidad del Cauca (Popayán, Colombia); MLP, Museo La Plata, Universidad Nacional de La Plata (La Plata, Argentina); MUSM, Museo de Historia Natural Javier Prado de la Universidad Nacional Mayor de San Marcos (Lima, Perú); USNM, National Museum of Natural History, Smithsonian Institution (Washington, District of Columbia, United States); UV, Colección de la Universidad del Valle (Cali, Colombia).

Pudella carlae n. sp. (n = 3): Peru: Amazonas, Quebrada Begazán (−6.1336°, −77.4669°), CORBIDI 558; Amazonas, near Abra Patricia (−5.7078°, −77.7881°), MUSM 23144; Amazonas, Valle Colpa (−6.3883°, −77.2325°), MUSM 55639.

Pudella mephistophila (n = 22): Colombia: Cali, Zoo UV 7410; Cauca, La Vega (2.0°, −76.71°), FMNH (1 specimen without specified number in Hershkovitz 1982); Cauca, Malvasá, FMNH (10 specimens without specified number in Hershkovitz 1982); Cauca, Paramo de Guanacas (2.39°, −76.28°), MHNUC 168; Cauca, Parque Nacional Puracé (2.25°, −76.35°), MHNUC 172; Cauca Puracé, AMNH (2 specimens without specified number in Hershkovitz 1982), FMNH (1 specimen without specified number in Hershkovitz 1982), USNM (1 specimen without specified number in Hershkovitz 1982), MHNUC 171; Quindio, Guayabla, Genova (2.235°, −75.66°), UV 13305. Ecuador: no data, MEPN 113-2-CB; Cayambé, San Marcos (0.023°, −78.03°), MEPN 9212.

Pudu puda (n = 15): Argentina: no data, MACN 22368; zoo, MACN 47.219; Neuquén, Parque Nacional Nahuel Huapi, Lago Espejo, MLP 26.XI.09.1. Chile: Araucania, Angol AMNH (1 specimen without specified number in Hershkovitz 1982); Bio Bio, Isla Mocha, AMNH (1 specimen without specified number in Hershkovitz 1982); Bio Bio, San Pedro, FMNH (1 specimen without specified number in Hershkovitz 1982); Los Lagos, Isla de Chiloé, FMNH (6 specimen without specified number in Hershkovitz 1982), MCZ (2 specimen without specified number in Hershkovitz 1982); USNM (1 specimen without specified number in Hershkovitz 1982); Los Lagos, Petrohué, FMNH (1 specimen without specified number in Hershkovitz 1982).