A new Eocene anagalid (Mammalia: Euarchontoglires) from Mongolia and its implications for the group’s phylogeny and dispersal

Anagalidae are extinct primitive Euarchontoglires from Asia, regarded as relatively closely related to basal Glires. So far, the group has been reported only from China and stratigraphically spans from the early Paleocene to the latest Eocene/earliest Oligocene. Anagalids are characterized by a relatively full dental formula featuring slightly enlarged semi-procumbent incisors, prominent canines, and tall cheek teeth with usually heavily worn crowns, indicative of an abrasive diet. Here we report a new genus and species from the late Eocene Ergilin Dzo Formation in southern Mongolia. The first non-Chinese anagalid is also the northernmost record of the family. Zofiagale ergilinensis gen. and sp.nov. is remarkable for its relatively small size (comparable only to the Paleocene genera Huaiyangale and Stenanagale), lack of P1, and molariform teeth showing almost no wear, suggesting a different diet than most Anagalidae. Furthermore, its molars display a strong buccal cingulum, a character in anagalids shared only with Wanogale. Our phylogenetic analysis of representatives of all anagalid genera based on 82 dental characters places Anagale and Anaptogale as the most basal lineages and clusters Zofiagale gen. nov. together with Qipania and Hsiuannania. These results suggest three independent northward dispersal events within the family in the late Eocene.

SCIEntIfIC REPORTS | (2018) 8:13955 | DOI: 10.1038/s41598-018-32086-x hard to determine the more precise date of extinction of this group of early Euarchontoglires. However, anagalids were certainly present in Asia for nearly 30 million years.
Here we describe a new anagalid from the late Eocene of the Ergilin Dzo Formation in Mongolia (Fig. 1), the first member of Anagalidae found outside of China, and the northernmost representative of the group. It is characterized by a unique dental morphology and relatively small size. We assign this specimen to a new genus and species. Furthermore, we aim at reconstructing the phylogenetic relationships within Anagalidae in only the second cladistic analysis after Hu's 4 , using an extended character-data matrix.
Distribution: As for the type and only species.
Diagnosis: As for the type and only species.   Diagnosis: Smaller than Anagale, Anagalopsis, Hsiuannania, and Qipania; comparable size to Linnania, Huaiyangale, Eosigale, and Interogale; and larger than Stenanagale; based on area of M 1 . Further differs from Anagalopsis, Qipania, and Interogale in having a smaller canine alveolus. Further differs from all anagalids, except for Interogale, in lacking a P 1 . Differs from Anagale, Anagalopsis, and Linnania in lacking a diastema between P 2 and P 3 . Further differs from Linnania and Eosigale in lacking a diastema between P 3 and P 4 . Moreover, it differs from all anagalids, except for Eosigale, in lacking a metaconid on P 4 and further differs from Anagale and Linnania in lacking a paraconid on P 4 . From Qipania and Interogale it differs in lacking a cristid obliqua on P 4 . Further differs from Linnania lofoensis in lacking a paraconid on M 1 , and differs from L. lofoensis, Interogale, and Wanogale in lacking a paraconid on M 2 . Further differs from all anagalids, except for L. lofoensis, L. qinglingensis, Interogale, and Wanogale in having a hypoconulid on M 1 -M 2 . Further differs from L. lofoensis and Interogale in lacking a paraconid on M 3 , and differs from all anagalids, except for Wanogale, in having a buccal cingulum on the lower molars. Additionally it differs from Anagalopsis, Linnania, Hsuiannania, and Qipania in having the enamel not extending into the alveolus.
Description and comparisons: The right body of the mandible is broken at the level of the canine alveolus in the rostral end, and at the level of the M 3 talonid in the caudal end, exposing the posterior root of the M 3 . Two mental foramina are visible. The anterior mental foramen is located below the posterior root of the P 2 , and the posterior mental foramen is smaller and located below the posterior root of P 3 . The total length of the P 4 -M 3 tooth row is 14.5 mm (the total molar row length is 11.5 mm; see Table 1 for more detailed dental measurements). The buccal height of the body of the mandible at the P 4 /M 1 inter-alveolar septum is 5.7 mm, while at the lingual side it is 6.1 mm.
The lower dental formula for Zofiagale ergilinensis is inferred to be? 0.1.3.3. Zofiagale is characterized by the loss of P 1 , which is rare in anagalids. Indeed, only Interogale, which has been tentatively classified as an anagalid in Li 5 is known to lack a P 1 as well 15 . Another rare characteristic for an anagalid seen in Zofiagale is the presence of a buccal cingulid on M 1 -M 3 , whereas most anagalids have a completely smooth buccal surface. There is only one other instance of an anagalid with a buccal cingulid: Wanogale 12 . Wanogale is known from only an M 2 , and also has another feature more typical of euarchontans, which is the presence of a precingulid.
The dentition anterior to P 3 is lost, and only two roots and a large alveolus remain. The large alveolus is here considered to have hosted a moderately large, semiprocumbent canine. All the other anagalids for which information about the canines is known (Anagale, Anagalopsis, Eosigale, Qipania, and Interogale) show that they had relatively larger and stronger canines than those of, e.g., zalambdalestids, a group of Cretaceous Eutheria proposed as primitive members of Anagalida sensu lato 6 . The semiprocumbent canines are known for Anagale and Qipania. No anagalid has a P 1 that is significantly larger than the rest of premolars, therefore it is more likely that the large alveolus hosted a canine rather than a P 1 . The two roots between the canine and the P 3 could be interpreted either as a double-rooted P 2 with loss of P 1 , or as a single-rooted P 1 and a single-rooted P 2 . We favor the first interpretation, even taking into account that the loss of P 1 is rare in anagalids, and only Interogale seems to have lost this tooth position, as some morphological traits strongly imply a two-rooted condition on P 2 . The first of the two roots is shifted buccally with respect to the second root, a pattern that coincides with the root positions in P 3 (indeed, the anterior root of P 3 is displaced more buccally than the posterior root, see Fig. 2E,F). This pattern is also seen in other mammals in which there is crowding of the premolars and no lower diastema, such as in the scandentian Ptilocercus lowii (e.g., USNM 488054). Therefore, we interpret the buccal torsion of the P 3 and the lack of diastemata as a clue that the two roots belong to one double-rooted P 2 , the P 1 has been lost, and that there is a fairly high degree of crowding of the premolars in Zofiagale.
The P 3 is broken, with most of the trigonid missing, but the two roots are preserved. Because of the crowding of the anterior dentition, there is no diastema distal to P 3 in Zofiagale, whereas it is observed in Linnania, Eosigale, and Stenanagale. The remains of P 3 indicate presence of a large, tall and presumably single-cusped (protoconid?) trigonid and a small saddle-like talonid ( Fig. 2A,B,E,F). The talonid is not basined and its buccal and lingual margins are much lower than the distal one, indicating the presence of a minuscule hypoconulid. Also, the lingual margin of the talonid bears a well pronounced cingulid-like edge.
The P 4 has no metaconid, usually present in other anagalids. The only other anagalid known to lack a metaconid is Eosigale. The paraconid is also absent, although a paracristid is clearly visible. The lack of paraconid is more common in anagalids, and only Anagale and Linnania retain one. The protocristid directed linguodistally is strong and slopes lingually to the lingual side of the tooth, in fact somehow replacing functionally the missing metaconid (see Fig. 2F). The talonid is not basined and is similarly saddle-like as in P 3 , but the distal margin bears a more pronounced eminence, which is most probably a nascent hypoconulid, developed fully in molars.
The mesiodistal length of P 4 is smaller compared to that of M 1 . This feature is shared by most anagalids, except for Eosigale and Qipania, which have similar mesiodistal lengths of the two tooth positions. The P 4 does not have a cristid obliqua, which is rarely present in angalids (only found in Qipania and Interogale). The P 4 has a buccal cingulid, not seen in any other anagalid.
The M 1 does not have a paraconid, although the paracristid is present, and the trigonid basin is deep (Fig. 2F). Many characters of the dental crown, such as small cusps are sometimes difficult to observe in anagalids, as the crowns in most species are worn down very quickly in ontogeny. In any case, the presence of paraconid on M 1 has only been found in Linnania. The trigonid and the talonid of M 1 have approximately the same area, and the trigonid is less than double of the height of the talonid. The metaconid is higher than the protoconid, and the hypoconid and the entoconid have approximately the same height. The talonid is deeply basined, but the basin area is not very extended. There is a distinct cristid obliqua and small mesoconid at M 1 . The entoconid is higher than hypoconid but its area is slightly smaller than the latter; it is also placed slightly more mesially than the hypoconid. A cusp-like, rounded hypoconulid is placed centrally at the distal margin of the tooth. The tooth bears a strong buccal cingulid, which causes some extension of the buccal side of the tooth.
The M 2 is larger than M 1 (Table 1) and more extended buccally. It also does not have a paraconid, but the paracristid is strong and the trigonid basin larger. This contrasts with the expression of the paraconid in Linnania, Interogale, and Wanogale (although there are generic differences in that respect). The metaconid is higher than the protoconid, and the hypoconid and the entoconid are approximately the same height. Both M 1 and M 2 have similarly developed hypoconulids, only present in a few other anagalids (Linnania, Interogale, and Wanogale). The talonid basin has a greater area than in M 1 . A weak metaconid and cristid obliqua are present.
The M 3 is smaller than M 2 and also has no paraconid, although the paracristid is present. This contrasts with Linnania and Interogale, which have a paraconid in this locus. The trigonid and talonid are similar in breadth. This is a rare feature in anagalids, since all other anagalids, except for Anagalopsis, have narrower talonids than trigonids. The talonid is taller than the trigonid, which is due to the upward curvature of the M 3 around the area of the hypoconulid lobe. This feature also occurs in Huaiyangale, Eosigale, and Qipania. The hypoconid and the entoconid are approximately the same height, but they are not quite aligned, the entoconid being marginally more distal. The talonid basin is extended and shallower having a relatively flat bottom; it forms a small extension towards a large and distally elongated hypoconulid and invades its surface to some extent (Fig. 2F). All molars have a distinct hypoflexid. The enamel does not extend into the alveolus, in contrast to a few anagalids (Anagalopsis, Linnania, Hsiuannania, and Qipania; Hu 4 ); furthermore, the molar crowns are relatively wear resistant, a rare trait for anagalids.
Phylogenetic analysis. In order to assess the phylogenetic relationships of Zofiagale with other anagalid genera, we conducted a cladistic analysis. A list of anagalid-specific characters was created based on character diagnoses from Chow et al. 10 Table S1). The primitive eutherian Zalambdalestes lechei was chosen as the outgroup for Anagalidae following some broader analyses on Glires 1,2,20,21 .
To correctly place Zofiagale in a phylogenetic context, a representative for each of the 14 anagalid genera were included. A total of 82 dental characters were scored for 15 taxa (see Supplementary Information 1 (Table S1, and Dataset S1). Color code for character states: yellow (0), red (1), blue (2) yielded a single most parsimonious tree (Fig. 3). Anaptogale and Anagale are the most basal anagalids in this tree, and constitute two separate primitive lineages (Fig. 3). The rest of anagalid taxa are evenly distributed between two distict clades. Zofiagale is a sister taxon to Hsiuannania and Qipania, and, in turn, this clade is the sister group to a clade composed of Chianshania and Eosigale, and Huaiyangale at a more basal position. The second major clade contains a clade grouping most genera presumed tentative anagalids (as per Hu 4 ), such as Stenanagale, Wanogale, and Interogale, with Linnania being the most basally nested, and Anagalopsis clustered with Diacronus.

Discussion
Anagalidae are a poorly known group of placental mammals with no extant representatives. The type and best known genus is Anagale gobiensis. By virtue of its almost complete skull and some postcranial remains 8 , Anagale has been often used in phylogenetic analyses of Eutheria where it has appeared well nested within Euarchontoglires 1,2,22 . Our results challenge the general topology of Hu's 4 hypothesis of relationships among anagalids, relegating Anagale to a basal position on our tree, and separating Linnania and Anagalopsis from Eosigale, Huaiyangale, and Hsiuannania-Qipania cluster. Our tree, however, agrees with Hu's 4 on the close relationship between Hsiunnania and Qipania, and Huaiyangale and Eosigale. Our tree shows a split between two major groups of anagalids. One of these clades includes Hsiuannania, Qipania, Zofiagale, Huaiyangale, Chianshania, and Eosigale. This clade appears well supported by a few synapomorphies: a P 3 smaller than P 4 , presence of a paracingulum on M 1 , absence of a paraconule and a metaconule on M 1-2 , absence of a diastema distal to P 2 , a talonid taller than the trigonid on M 3 , and a smaller area of M 1 compared to M 2 . The clade includes four out of seven taxa of Hu's 4 unambiguous anagalids. The other major group of anagalids is composed of Interogale, Wanogale, Stenanagale, Linnania, Anagalopsis, and Diacronus. This clade is well supported by the following synapomorphies: presence of a minute parastyle on M 2 , a distinctive paraconule and metaconule on M 1-2 , an erect rather than semi-procumbent lower canine, a large root of the lower canine, and a trigonid taller than the talonid on M 3 .
A new genus Zofiagale is well nested within the anagalid tree; in fact, it represents one of the most derived taxa, closely related to Hsiunnania and Qipania. It shows some unique dental characters which stress its independent phylogenetic heritage and imply a long evolutionary history. Interestingly, Zofiagale shares its two most characteristic features with species which do not show the most typical anagalid dental morphology and whose anagalid status has been questioned. With Wanogale it shares a strong buccal cingulum on lower molars and crowns that do not heavily wear down, and with Interogale, a lack of P 1 . Although Hu 4 did not consider Wanogale to be an anagalid, Li 5 placed it as a tentative member of the family. On the other hand, Interogale, also of questionable status, was considered as belonging to Tillodontia by Wang and Jin 23 .
The taxon sample for our tree does not allow us to decide on what is and what is not an anagalid. In order to do so, other non-anagalid euarchontogliran representatives (i.e., primates, scandentians, lagomorphs, rodents, etc.) should be included in a broader phylogenetic analysis; of course, this raises the question if anagalids are paraphyletic, especially because many tentative anagalids according to Hu 4 and Li 5 are well nested within our sample. Such phylogenetic analysis including representatives of euarchontan and Glires lineages is necessary also to properly assess the position of anagalids within the broader framework of Euarchontoglires. However, an analysis of this magnitude is beyond the scope of this study.
The time bracket for the anagalid evolution spans from the early Paleocene to late Eocene/earliest Oligocene, with the timing of Anagalopsis ocurrence still a matter of debate 5,9,16 . Nevertheless, it is one of the last occurring anagalids, together with Anagale and Zofiagale (Fig. 4). The fact that all three northern anagalids belong to three   distinct lineages within the family suggests that disperal events between southern and northern China (including the Mongolian Plateau) were probably common sometime between the late Paleocene and late Eocene, and these happened at least three times within Anaglidae.
Most anagalids have significantly worn-down cheek teeth, which has led McKenna 24 to suggest that they may have obtained fairly abrasive food below the surface of the ground. This view endorsed Bohlin's 9 idea that anagalids were potentially fossorial. However, back in 1963, the only anagalids known were Anagale and Anagalopsis, and the variety of dental shape and body masses reported in anagalids have increased greatly (Fig. 5). The present record of anagalids shows taxa with non-worn or lightly worn cheek teeth, such as Zofiagale and Wanogale. This suggests that their diet could not be primarily composed of abrasive foods, and most probably would differ from that suggested for Anagale and Anagalopsis. It is, however, worth noting that this inference derives from a sample size of a single specimen and should therefore be taken with caution. Alternatively, ZPAL MgM-II/100 could belong to a very young adult (the M 3 is erupted) who has not started wearing down the crowns. However, if that was the case and this animal had an abrasive diet, it would be expected to find differential wear between M 1 and M 3 (i.e., M 3 erupts later), and there is not.
Zofiagale is one of the smallest anagalids, with an estimated body mass of 516 g, Kay's 25 threshold sets a minimum of 500 g for primates to acquire their protein mainly from leaves based on observations of modern primates, with species below that mass being mostly insectivorous. Besides Eocene euprimates, this threshold has been used in Paleocene and Eocene plesiadapiforms and non-primate euarchontans as well [26][27][28][29] and could be tentatively informative in other early Paleogene mammals of generalized morphotypes similar to those of plesiadapiforms. The estimated body mass for Zofiagale sits closely to the 500 g threshold, suggesting that an animal this size could easily acquire protein from both insects and leaves, without relying solely on any of the two. Given the blunt cusps, it is likely that the diet of Zofiagale might have been largely enriched with fruits. Whereas the presence of a buccal cingulid (rare in anagalids) helps providing gingival protection from the opposing upper tooth and/or food [30][31][32] , it also effectively broadens the tooth, generating more surface for pressing a bolus of fruit.

Methods
The specimen was photographed with a scanning microscope Hitachi S−3400 N without coating, in a natural mode (low vacuum) at the Museum and Institute of Zoology, Polish Academy of Sciences, Warsaw, Poland. Furthermore, we documented the specimen using a Keyence digital microscope VHX 5000, and a Nikon (SMZ-800) stereo-microscope at the Institute of Paleobiology (IPAL), Polish Academy of Sciences, Warsaw. The measurements (Table 1) were taken using SYLVAC digital caliper and Keyence digital microscope software with an accuracy to the nearest 0.01 mm. Dental terminology generally follows Meng and Wyss 22 .
Size estimation. We estimated the body mass of Zofiagale and comparative anagalid taxa using Conroy's 51 equation for prosimians. Although Zofiagale is not a primate, Conroy's 51 equations have been used to estimate body mass for plesiadapiforms 28,29,51-55 , and anagalids have a more similar generalized morphotype to plesiadapiforms than to modern Glires. Therefore, an equation based on a prosimian subset would be preferable for anagalids, rather than equations based on living rodents 56 or lagomorphs 57 .
Nomenclatural acts. This published work and the nomenclatural acts it contains have been registered in ZooBank, the Official Register of for the International Code of Zoological Nomenclature (IZCN), mandatory for electronic-only publications. The ZooBank LSIDs (Life Science Identifiers) can be resolved by appending the LSID to the prefix "http://zoobank.org". The LSID for this publication is urn:lsid:zoobank.org:pub:65C21FB2-3E88-4A00-8900-6380A4661E81. The identifiers for taxa appear following each new name (see Systematic Paleontology).

Data Availability
All data generated or analyzed during this study are included in this published article (and its Supplementary Information files).