The enigma of the Yepes ’ armadillo : Dasypus mazzai , D . novemcinctus or D . yepesi ?

The long-nosed armadillos of the Dasypus genus are the richest and more widespread extant xenarthra. In 1933, the prominent Argentinean mammalogist José Yepes studied specimens collected by Salvador Mazza and named a new species: D. mazzai. Several authors interpreted that the holotype of the species was a juvenile of D. novemcinctus. In 1995, Vizcaíno support the synonymy with D. novemcinctus but accepted the existence of an endemic species of northwestern Argentina that dedicated to Yepes (D. yepesi). Morphological studies have questioned the synonymy of the species with D. novemcinctus. In this contribution we reviewed the identity of the type material by sequencing a DNA fragment of 212 bp, product of the 16S ribosomal unit of Dasypodidae. Both samples of the holotype (bone and muscle) and of a specimen from the north of Santa Fe province have a 100% similarity with the D. yepesi / D. sabanicola sequences available in the GenBank. These results indicate that the type specimen of D. mazzai is neither a D. novemcinctus nor another species of the Dasypus genus previously named. Dasypus mazzai is revalidated and the eastern boundary in the Chaco region is confirmed. Field work is required to confirm the conservation status and current distribution of this endemic mulita.


INTRODUCTION
The long-nosed armadillos or "mulitas" belong to the Dasypodidae family, the earliest diverging clade inside the Cingulata by the middle Eocene (Gibb et al., 2016;Delsuc et al., 2016).
With seven species (see below), Dasypus is the extant xenarthra genera most diverse and more widespread (Wetzel et al., 2007;Hautier et al., 2017).The earliest record for the genus comes from the Late Pleistocene (Lujanian) of South America (Vizcaíno et al., 1995;Castro, 2015), but based on molecular evidence, Gibb et al. (2016) suggest that it must have radiated by the middle Miocene.
Morphologically, the genus is characterized by a long and slender rostrum, long ears, long tail and a smooth and a relatively bare carapace (except in D. pilosus which has a dense coat of hair that hides the carapace).Nowadays, Dasypus species occur from the southern United States to center of Argentina (Wetzel et al., 2007;Abba et al., 2012;Loughry & McDonough, 2013).Six species are clearly recognized (but see Feijó et al., 2018): the seven-banded armadillo (D. septemcinctus Linnaeus, 1758), the nine-banded armadillo (D. novemcinctus Linnaeus, 1758), the southern long-nosed armadillo (D. hybridus Desmarest, 1804), the northern long-nosed armadillo (D. sabanicola Mondolfi, 1968), all belonging to the subgenus Dasypus (Wetzel & Mondolfi, 1979); the greater long-nosed armadillo (D. kappleri Krauss, 1862), assigned to its own subgenus Hyperoambon (Wetzel & Mondolfi, 1979); and the hairy long-nosed armadillo (D. pilosus Fitzinger, 1856), assigned to the subgenus Cryptophractus, which has been recently proposed to raise to genus level (Castro et al., 2015).The identity of a seventh taxon has been the subject of discussion over more than seventy years.
In 1933, the prominent Argentinean mammalogist José Yepes (1897Yepes ( -1976) ) named and described Dasypus mazzai Yepes, 1933.The specimens that Yepes used for the original description of this armadillo were donated by the physician and epidemiologist Salvador Mazza (1886Mazza ( -1946)), well-known for his studies in several endemic diseases of South American.During the Regional pathologies study mission, carried on in 1926 in the north of Argentina, Mazza, who had installed his laboratory in a railway car, focused on different potential vectors of trypanosomiasis (Chagas-Mazza disease).In doing so, he gathered a notable collection of mammals currently housed in the Mammalogy Section of the Museo Argentino de Ciencias Naturales "Bernardino Rivadavia" of Buenos Aires (MACN-Ma).Among them the armadillos are particularly conspicuous (see Abba & Vizcaíno, 2008).Hamlett (1939) interpreted that part of the material referred to D. mazzai by Yepes actually belonged to a new species that had to be named and based on the scarce specimens available, added a few more distinctive features.According to Hamlett (1939), Yepes was aware of the issue and reexamined the material.This work was never published but, D. mazzai continued to be accept-ed in general systematic compilations (Cabrera, 1957;Olrog, 1976Olrog, , 1979) ) In their exhaustive revision of the long nosed armadillos, Wetzel & Mondolfi (1979) tried to reconstruct the history of the collections numbers of the specimens assigned to D. mazzai by Yepes.They concluded that "D.mazzai is a composite, a junior synonym of D. novemcinctus and D. hybridus and that use of D. mazzai may also infringe upon D. septemcinctus" Wetzel & Mondolfi (1979: 55).It should be clarified here, that the specimen MACN-Ma 31.213 that Yepes (1933) quote in the page 229 is not the Holotype of D. mazzai, but the holotype of Chlamyphorus truncatus patquiensis described by Yepes (1939), so this was a simple mistake in Yepes (1933).
In 1995 Vizcaíno revised the holotype and paratype of D. mazzai and compared with new specimens collected by him using numerical techniques to analyze the variation observed (Vizcaíno, 1995).In each analysis performed the specimens studied appeared well differentiated from other species of the genus.As a consequence, Vizcaíno named a new species, Dasypus yepesi, dedicating the species to Dr. J. Yepes, to whom he considered corresponded the merit of having noticed the morphological separation of this taxon with respect to those previously recognized.The type specimen was housed in the Museo de La Plata (MLP).
Recently, based on the evidence available in the literature Feijó & Cordeiro-Estrela (2014) claimed that the holotype D. mazzai is not a juvenile of D. novemcinctus and, hence, that D. mazzai is a valid species, with D. yepesi as a junior synonym.Gibb et al. (2016) described a DNA sequence of a specimen of the type series D. yepesi as possessing very similar mitogenomes to D. sabanicola.A recent morphometric analysis of long-nosed armadillos by Hautier et al. (2017) did not include D. mazzai-D.yepesi.Finally, the most recent morphometric analyses of Dasypus genus by Feijó et al. (2018) support the validity of D. mazzai.
In the 90 th anniversary of the incorporation of José Yepes to the Mammalogy Section of the Museo Nacional (today Museo Argentino de Ciencias Naturales "Bernardino Rivadavia") (see Garcia, 2018), and like other colleagues (Galliari, 2018;Ojeda et al., 2018;Gonzalez et al., 2018), we aimed to honored to J. Yepes.The goal of our work is to test the identity of the holotype of D. mazzai (MACN-Ma 31.273) using molecular tools.

Specimens
Dasypus mazzai: MACN-Ma 31.273(Holotype, Fig. 1.A-B); Supp.Matterial 1) a cephalic shield, carapace and tail, body (without head) conserved in fluid, skull figured in Yepes (1933) Yepes (1933), the skull number JVY 32 actually lost (Figure 2B).Locality: According to Yepes (1933) Argentina, Salta, Dto.Orán, Tabacal (Fig. 2).Collected by Salvador Mazza.Vizcaíno (1995) considered it unknown, because that was the town where Mazza labeled the specimens collected by the inhabitants of the area, so that it cannot accurately determine where the specimen came from.Sample for DNA analyses includes dry tissue in the tail from the mounted skin.

DNA sequence
We choose a 16s rRNA of 212 bp length in which D. novemcinctus, D. hybridus and D. yepesi have three different haplotypes determined by 15 variable sites.The two sequences of D. novemcinctus available in the GenBank (KT818542.1 from French Guiana and Y11832 from USA) present a percentage of differentiation of 2% for the Dasypodidae 16S ribosomal RNA (16S rRNA).This differentiation is less than that presented by most of the sequences belonging to the different species studied.However, this high differentiation inside D. novemcinctus could be reflecting the differentiation of two geographical lineages, of perhaps two different species (see Gibb et al., 2016).In order to evaluate the variability in the sequenced fragment we included four specimens of D. novemcinctus from Argentina in the analysis (see above).In addition, the sequences of D. hybridus (GenBank KT818540) and D. septemcinctus (GenBank KT818546) show a similarity of 100% for the chosen fragment.Gibb et al. (2016) found that this two mitogenome sequences have a 99.3% identity and referred as probably misidentification.Indeed, the D. septemcinctus sequence belongs to an animal collected in north Argentina and up to date there as a wide amount of evidence that this species is absent from this country (Abba & Vizcaíno, 2008;Abba et al., 2012).Based on morphologic and morphometric analysis, Feijó et al. (2018) proposed that D. hybridus and D. septemcinctus are synonyms but this assignation is uncertain until have been a support by a robust molecular analysis.

Mitochondrial DNA extraction and PCR amplification
The first steps of DNA extraction protocols differed depending on the sample source.Dried tissue samples were incubated overnight at 37° C in extraction buffer containing 10 μl of proteinase K, 10 mg/ml; 5 μl of RNase, 20 mg/ml and 10% SDS.For bone samples, approximately 50 mg of bone powder was incubated with stirring for 24 hours in 1 ml of EDTA 0.5 M, pH = 8.The mix was then centrifuged for 10 min at 14,000 rpm and the precipitate was stored for the posterior extraction and purification of DNA.The sample stored in formaldehyde was first washed in 70% ethanol for 30 min and then incubated with stirring for 24 hs in TE buffer at 37° C, replacing the buffer every 8 hours.After these first steps, DNA was isolated from all samples by phenol-chloroform extraction and alcohol precipitation.
Aliquots of total DNA were used as templates in Polymerase Chain Reaction (PCR) in order to amplify a 212 bp double-stranded DNA product from the Dasypodidae 16S rRNA mitochondrial gene.Each PCR had a reaction volume of 20 μl and contained 1X Easy Taq buffer (20 mM Tris-HCl, pH = 8.3, 20 mM KCl, 10 mM NH4 2 S0 4 , 2 mM MgS0 4 ), 0.4 mM of each deoxynucleotide triphosphate, 0.5 μM of forward and reverse primers, 0.5U of Easy Taq DNA polymerase (TransGen Biotech), 3.5 mM MgCl 2 , 2 μg/μl BSA (New England Biolabs), 4 μl of DNA extract and water to reach the final volume reaction.The primer pair used was, arm16SF (5'-TAAG-TAGAGGTGAAAG ACCAAC -3') and arm16SR (5'-TGCTTTAGGCCAACTATGT-3').It was developed using Primer3 online software using the complete sequences of the mitochondrial genome of D. hybridus, D. novemcinctus and D. yepesi as reference (GenBank: KT818540, KT818542, and KT818547, respectively).We chose this short DNA fragment due to the difficulty to amplify longer fragments from old bone and tissue sample and because it contains several variable sites that allow the distinction of the species to which the studied samples could belong (see above).Amplification protocol consisted in a single denaturation step at 94° C for 4 min, followed by 20 cycles of denaturation at 94° C for 30 s, annealing at 43° C (annealing temperature -5° C) for 30 s, and extension at 72° C for 30 s; 20 cycles of denaturation at 94° C for 30 s, annealing at 48° C for 30 s, and extension at 72° C for 30 s and a final extension at 72° C for 5 min.PCR products were resolved in 1% agarose gel electrophoresis and visualized under UV light.A 10 μl aliquot of the amplification products were purified using the EXO/SAP enzymatic method and sent to an external laboratory (Biotechnology Institute of INTA Castelar, Argentina) where sequencing was performed with the same primers used in PCR reactions in an ABI 377 Automated DNA PrismTM Sequencer (Applied Biosystems Inc).

RESULTS
Ten of the eleven samples analyzed could be amplified and sequenced successfully (Table 1; Fig. 3; Suppl.Material 3).The amplified 16s rRNA fragments obtained had a length of 212 bp and consisted in three haplotypes determined by 15 variable sites (Table 1).These haplotypes were compared with those 16S rRNA sequences previously published for the Dasypodidae family (Figure 3).The four D. novemcinctus samples analyzed showed a 100% similarity with D. novemcinctus GenBank sequence KT818542.1 (Table 1), a 98% similarity with D. novemcinctus GenBank sequence Y11832.1 and 93-97% similarity with 16S rRNA sequences from other Dasypodidae species.Samples MACN-Ma 31.273(b= bone), MACN-Ma 31.273(t= tisuue) and MACN-Ma 33.176 showed a 100% similarity with D. yepesi and D. sabanicola sequences (Genbank: KT818547 and KT818545, respectively; Table 1) and a 95-97% similarity with 16S rRNA sequences from other Dasypodidae species, suggesting that these individuals belong to the species D. yepesi/sabanicola.Instead, samples MACN-Ma 30.17, 30.18 and MACN-Ma 36.980showed a 100% similarity with D. hybridus sequences (Genbank: KT818540 and KT818546, respectively; Table 1) and a 95-97% similarity with 16S rRNA sequences from other Dasypodidae species, suggesting that these three individuals belong to D. hybridus.Finally, the sample MACN-Ma 13222 presented sequences of very low quality that prevented to identify the species to which they belong.

DISCUSSION
The four specimens of D. novemcinctus from Argentina show no variation in the amplified 16s rRNA fragments and share the same sequence with the Gibb et al. (2016) specimen from French Guiana (GenBank KT818542).In addition, this sequence has a 98% of similarity with the Arnason et al. (1997) from USA (GenBank Y11832).The divergence between the French Guiana and USA lineages was already established since Huchon et al. (1999) study.Further studies are needed to confirm that the specimens from French Guiana and Argentina belong to the same lineage.If this is the case, as the type locality for D. novemcinctus corresponds to "America Meridionali" (i.e., South America) and restricted to Pernambuco, Brazil, by Cabrera (1957: p. 225; see also Feijó et al., 2018), the USA lineage might represent a previously unrecognized taxon.
Our results indicate that the holotype of D. mazzai (MACN-Ma 31.273) is not a specimen of D. novemcinctus (neither the South American nor USA lineages) as proposed by Hamlett (1939), Wetzel & Mondolfi (1979) and Vizcaíno (1995), nor any other of the Dasypus species previously nominated.Consequently, D. mazzai must be considered a valid species as originally described by Yepes (1933).Furthermore, the sequences of the holotype of D. mazzai are identical to that of one the specimens of the hypodigm of D. yepesi, housed in the Museo de La Plata (MLP 30.III.90.2).Thus, nowadays the specimens of D. mazzai are housed in the collections of the MACN and MLP.
The recognition of D. mazzai as a valid species has further taxonomic consequences for the systematics of the genus that remain to be tested.Following Gibb et al. (2016) the complete mitogenomic sequence of the specimen of D. yepesi Vizcaíno, 1995 (MLP 30.III.90.2), show 98.7% identity with that of D. sabanicola Mondolfi, 1968 housed at National Museum of Natural History, Washington, USA (USNM 372834).Therefore if D. sabanicola and D. yepesi were considered the same species, D. sabanicola would be a junior synonym of D. mazzai Yepes, 1933.This opens an interesting distributional issue, because the same species would be represented by populations restricted to very distinct localities of South America, with a separation at least of 3000 km: the llanos of Venezuela and Colombia and in Northern Argentina.Further studies are needed to understand this possibly disjunct distribution.Yepes (1933) stated that the type locality of D. mazzai in the Argentinean Yungas represents the western limit of its distribution, which extended eastwards to the North of Santiago del Estero and Santa Fe provinces, in the Chaco region.The identification of the sequence of the specimen Department) as belonging to D. mazzai confirms the distributional range of D. mazzai in the Chaco region of Argentina (Fig. 2) proposed in the 1930's.As Yungas and the Chaco regions extended in the neighbor countries of Paraguay and Bolivia, the species could be present there too, which makes the distribution of D. mazzai needs to be updated.
Other Dasypus species reported in the distribution area of D. mazzai are D. novemcinctus, D. hybridus and probably D. septemcinctus (Wetzel & Mondolfi, 1979).Yepes (1933) used the MACN-Ma 30.17 and 30.18 specimens as D. septemcintus in his cranial measurement comparison with those of the Holotype and Paratype of D. mazzai (both actually lost).Wetzel & Mondolfi (1979) assigned one of them to D. septemcinctus (MACN-Ma 30.18) and the other to D. hybridus (MACN-Ma 30.17) based on their canonical analyses of cranial measurements.Following Wetzel & Mondolfi (Page 54, 1979), the evidence of sympatry between both species in the Chaco region of Argentina was based in few specimens (including both before mentioned).In this work we follow the criteria of Gibb et al. (2016) and considered the sequence of D. septemcinctus ISEM T-3002 (Genbank: KT818546) from Argentina as misidentification, and considered it as belonging to D. hybridus.According the morphological characters and the DNA sequences of the specimens MACN-Ma 30.17,MACN-Ma 30.18 and MACN-Ma 36.980we assign them to D. hybridus.
The DNA extraction using old samples (from 1930 decade) allowed us to evaluate the taxonomic assignation of Dasypus species from Yungas and Chaco of Argentina.These two forested regions are actually among the more degraded areas in the world (Malizia et al., 2012, Kuemmerlet et al., 2017).In addition, there are very poor data about conservation status, distribution, abundance and other biological parameters of the Dasypus armadillos inhabiting these regions.Their conservation status is quite preoccupant.Dasypus hybridus and D. sabanicola are catalogued as Near Threatened due to severe habitat loss and hunting throughout its range (Abba & Gonzales, 2014, Superina et al., 2014) and "D.yepesi" is catalogued as Data Deficient (Abba & Vizcaíno, 2014).A reassessment of the conservation status of these species is highly needed, which encourages performing new field efforts to acquire new data of distribution and current population status of the Dasypus species in these regions.
As a closing remark, we regret that Yepes no longer has an armadillo nominated after him.However, we understand that our impartial scientific results honor his work and his intention to pay homage to Dr. Salvador Mazza, by recognizing the correctness of his original identification of an endemic armadillo from northern Argentina.

CONCLUSIONS
The S16 rDNA sequences of the Holotype of Dasypus mazzai Yepes, 1933 MACN-Ma 31.273 are not compatible with the specimens of Dasypus novemcinctus from South America and USA and must be considered as corresponding to a different taxa.
The specimen MACN-Ma 33.176 from Villa Guillermina, Santa Fe province (General Obligado Department) belongs to D. mazzai and confirms the distributional range in the Chaco region of Argentina as was supposed by Yepes (1933).

Fig. 3 .
Fig. 3. Alignment of the 212 bp fragment of the 16S ribosomalRNA mitochondrial gene amplified from museum specimens and GenBank sequences.First base of museum specimen sequences corresponds to position 1421 in the Dasypus yepesi mtDNA complete genome (GenBank: KT818547).Color key: red A, yellow G, green C and blue T. Asterisk indicates the sequence here referred as D. hybridus.Holotype of D. mazzai in bold.