On the invalidity of Halitherium schinzii Kaup , 1838 ( Mammalia , Sirenia ) , with comments on systematic consequences

The holotype of the sirenian species Halitherium schinzii Kaup, 1838 from the Alzey Formation (early Oligocene) of the Mainz Basin, western Germany, is reviewed in detail and revised. It is concluded that the type specimen, an isolated premolar, is non-diagnostic, because it reveals no characters of taxonomic value. Therefore, the taxon name H. schinzii is regarded as a nomen dubium, thus cannot be applied to any currently proposed sirenian species. The name of the genus “Halitherium”, which is based on the type species “H. schinzii”, cannot be applied to other species previously assigned to congeneric taxa. Consequently, taxonomic and systematic re-assessment is required. Due to the fact that “Halitherium” is the taxonomical basis of the Halitheriinae, conclusions are drawn on the inappropriateness of this subfamily.


Introduction
The genus Halitherium includes a number of fossil sirenian species, or sea cows, ranging from the late Eocene to early Miocene.Its type species H. schinzii Kaup, 1838 is considered as a widespread Central European taxon that inhabited the early Oligocene coastal waters, especially from Germany (e.g., Voss 2008).The morphological basis for the establishment of this species is a single premolar from the Mainz Basin in western Germany (Kaup 1838;Domning 1996).This premolar forms the focus of the present paper as a first step into a revision of the Halitherium-species complex.

Historical Review
In 1834, Kaup and Scholl mentioned a not otherwise specified tooth of an extinct sea cow from the early Oligocene of Germany near Flonheim (Mainz Basin; Fig. 1) as Pugmeodon schinzii.As Domning (1996) already recognised, both, the genus and species name, have to be considered as nomina nuda, because Kaup and Scholl (1834) neither provided a description nor a figure of the respective tooth.However, a description and figure of another premolar (HLMD-WT Az 48) under the same species name was given by Kaup (1838), who subsequently transferred it to the genus Halitherium, which seemingly is the correct generic name for the species as indicated by later studies (e.g., Kaup 1855;Lepsius 1882;Sickenberg 1934).
Therefore, Domning (1987) opened the nomenclatural case no.2569 to apply for a proper designation of the holotype of Pugmeodon schinzii.By Opinion 1535 of the Bulletin of Zoological Nomenclature Volume 46 (ICZN 1999), all previous designations of type species for the nominal genus Halitherium Kaup, 1838 are set aside and Pugmeodon schinzii Kaup, 1838 is designated as type species under the plenary powers.Furthermore, the names "Halitherium" and "schinzii" are ruled to be placed on the official lists and indexes of names and works in Zoology by the International Commission on Zoological Nomenclature (2001).
In the subsequent decades, early Oligocene sirenian finds from the Mainz Basin comprising cranial and postcranial elements, and occasionally partial skeletons with nearly complete skulls and mandibles, were generally assigned to H. schinzii.Thus, this taxon has become firmly established in the literature as the only sea cow species that had evolved in this specific region to that time (e.g., Krauss 1858;Barthel 1962;Bahlo and Tobien 1982).Apart from its type locality, H. schinzii is also known from the Lower Rhine area in western Germany (Voss 2008) and from the Bay of Leipzig in eastern Germany (Fischer and Krumbiegel 1982).Sirenian finds from the Paris Basin (Bizzarini 1995;Bizzarini and Reggiani 2010), the Antwerp and East Flanders Provinces in North Belgium (Sickenberg 1934), Switzerland (Pilleri 1987) and Hungary (pers.obs.) that are referred to H. schinzii indicate the widespread distribution of this taxon in Central Europe.
However, considerable intraspecific morphological variations have been postulated for H. schinzii since the middle of the 19th century (e.g., Kaup 1855;Lepsius 1882;Sickenberg 1934;Fischer and Krumbiegel 1982).Attempted revisions of H. schinzii consequently resulted in the description of numerous specimens forming the basis of new species, as are, respectively, the skull roofs of H. kaupi and H. bronni (for both taxa see Krauss 1858).However, those are considered to be not valid and synonymous with H. schinzii by Domning (1996).Hartlaub (1886) described another species, Manatherium delheidi, based on cranial remains from the lower Oligocene of Antwerp (North Belgium) in comparison with H. schinzii from the Mainz Basin (Germany).However, Sickenberg (1934) revised Hartlaub's (1886) species, focussing on aspects of juvenile stages and intraspecific variability.Sickenberg (1934), who already mentioned that a revision of Halitherium and H. schinzii was necessary, provided a comprehensive and critical review of the Belgian material, but without considering a separation on species level.According to his investigations, all known Belgian records belong exclusively to smaller animals.Therefore, Sickenberg (1934: 271) defined separate morphological groups, "forma typica" for the German and "forma delheidi" for the Belgian one, and assigned both to H. schinzii (see also Domning 1996).The most recent new combination for H. schinzii refers to skeletal material from the lower Oligocene of France known as H. schinzii lareolensis (Pilleri 1987;Domning 1996).
This review documents that the attempts of different authors to identify morphological distinctions on species level remain ambiguous until today.The debate on splitting and lumping of species currently referred to H. schinzii was recently revived by Voss (2012).This author stated that morphological variations within German and Belgian early Oligocene Sirenia might be related to the species level and postulated the presence of an additional species.
The hypothesis of two morphospecies necessitates the examination of the holotype of H. schinzii as a target for a revision of this type species.However, the preservation and nature of the holotype material pose major problems.Voss (2010) already postulated that no significant taxonomic definition for H. schinzii can be deduced from the premolar HLMD-WT Az 48 and, consequently, this species and the thereon based genus lack a diagnosis.
The present paper aims to clarify this issue by elucidating the morphology and taxonomic value of the premolar and provides perspectives for new taxonomic and systematic approaches.
Type horizon and locality.Alzey Formation of the Selztal Group (lower Oligocene) from Flonheim in the Mainz Basin, western Germany.
Description.The tooth (Fig. 2) represents a premolar.However, its exact position within the tooth arcade cannot be certainly determined due to incomplete preservation and wear.A single root was present, but is no more preserved.The crown measures 11 mm in maximum diameters with the lingual side being broader mesiodistally than the labial one.Though worn, the central cusp shows a smooth and convex side facing labially.Lingually, five accessory cuspules are present.A further, medium sized cusp being also in wear is present mesially or distally depending on the side and locus the premolar inhabits within the jaw.The overall shape of the crown resembles a clenched fist, which tempted Kaup (1838: 319) to introduce for this specimen the new genus Pugmeodon ("Faustzahn"), which he transferred into Halitherium (Kaup 1838: 536) in the same volume.
Remarks.As it can be inferred from the description above, the taxonomic value of the holotype of H. schinzii is doubtful and its assignment to a certain species remains ambiguous.This is additionally corroborated by comparative morphological investigations with species lumped under the genus Halitherium.
Compared to Halitherium taulannense (Sagne 2001), known by upper and lower premolars, and the H. schinzii specimen BSPG 1956 I 540 (see also Barthel 1962), comprising the right P2 and left P3-P4, HLMD-WT Az 48 most likely represents an upper posterior premolar.Using the preserved teeth of both species as reference points, HLMD-WT Az 48 could have occupied either the third or the fourth locus based on its complex cusp pattern.However, a clear assignment of HLMD-WT Az 48 to P3 or P4 of BSPG 1956 I 540 or any upper premolars observable in H. taulannense is not possible.On the one hand, this is related to the high degree of wear that precludes comparisons of this tooth with a number of nearly unworn premolars of H. taulannense (Sagne 2001), for example.On the other hand, the special crown morphology of HLMD-WT Az 48 described above is considered to hamper a proper identification.The specific arrangement of cusps is found neither in isolated nor associated premolars even of a comparable degree of wear of any known sirenian species.
Apart from BSPG 1956 I 540, complete series of upper premolars are not known in specimens hitherto assigned to H. schinzii making further comparative studies difficult.Additionally, other fossil sirenian taxa like H. cristolii (Spillmann 1959) have no premolars preserved and only reveal the root alveoli in the bone.Thus, it remains unresolved whether HLMD-WT Az 48 belongs to the left or the right quadrant of the jaw and the identification of its exact position within the tooth arcade is still an open question.
In conclusion, the relation of premolar HLMD-WT Az 48 to the nominal species H. schinzii is doubtful or at least cannot be established, because this tooth does not yield any significant characters to diagnose both, the genus and the species.Consequently, the name "H. schinzii" is considered a nomen dubium as is the genus "Halitherium" since "H.schinzii" is the type species of that genus.Both terms can be applied to the premolar specimen only and are therefore rejected for further nomenclatural purposes.
Following the principle of priority, article 23.3.5 (ICZN 1999) states that an invalid taxon name must be replaced by the oldest available junior synonym.In this case, three nominal synonyms exist, which however are each considered as inappropriate replacement for "Halitherium" due to their poor material basis, uncertain taxonomic assignment and juvenile status or a combination of these factors.
The oldest genus is Trachytherium (Gervais 1849; Domning 1996) and is based on a single, isolated lower M3.According to Domning (1988) and Domning and Pervesler (2001) sirenian molars are often characterised by a high degree of intraspecific variation.Beside the fact that this tooth represents the only known specimen, it is therefore emphasised here that the establishment of a new sirenian genus on a molar alone is not sufficient.
The next older genus Crassitherium (Van Beneden 1871; Sickenberg 1934;Domning 1996) was established on the basis of a skull fragment and eight vertebrae.While the vertebrae are clearly identified as belonging to a sirenian, there are serious doubts on the taxonomic assignment of the skull fragment, which might be referable to a reptile (Sickenberg 1934: 207).This is also supported by personal investigations of the material in question hence Crassitherium is not considered as a useful or valid replacement name.
The third genus Manatherium (Hartlaub 1886; Domning 1996) is based on skull fragments of a juvenile.On the basis of personal investigations these fragments are identified as belonging to the frontal, the sphenoid and the right maxilla.Beside the juvenile status of this specimen that already tempted Sickenberg (1934) to declare this genus to be invalid, the preserved material is not considered to reveal diagnostic features for a reliable taxonomic assignment on the species level.
Article 23.3.5 (ICZN 1999) also states that if the rejected name has no valid synonym a new substitute name must be established in its place.Therefore, all taxa originally assigned to "Halitherium" require a new designation.A taxonomic and systematic revision of the "Ha- litherium"-species complex is in progress to be published and beyond the scope of the present paper.
It is refrained here to designate a neotype, because the conditions specified in article 75.3 (ICZN 1999) do not apply.Most importantly, the name-bearing type specimen of "H.schinzii" is neither lost nor destroyed, but still extant and represented by the premolar HLMD-WT Az 48.Additionally, there has never been designated a paratype or paralectotype, which could be chosen as neotype according to recommendation 75A (ICZN 1999).The choice of a neotype originating from the type locality or even the same geological horizon is difficult for several reasons.An obvious potential neotype could be an incomplete skull cap representing the holotype and only specimen of "Halitherium" kaupi (Krauss 1858), which was assigned as junior synonym of "H.schinzii" by Domning (1996).Presenting mainly indistinctive features, which are typical for a juvenile, this specimen is just fragmentarily preserved including only the supraoccipital and parietal in addition to remnants of the frontal.This incompleteness and poor preservation, but mainly the juvenile status prevents a designation of this specimen as a neotype with sufficient diagnostic characters distinguishing this taxon from other sirenian genera.Another reason for the difficulty finding a neotype is the hypothesis by Voss (2012) that a second sirenian species, viz."H." bronni, might have been present in the early Oligocene of Germany, and therefore the respective type specimen is also not an available choice.Due to the presumed presence of two sympatric species, suitable material deriving from a host species for the potential neotype is not determinable unambiguously.Therefore, the replacement of an unidentifiable name-bearing type by a neotype (article 75.5;ICZN 1999), which would be possible considering "H.schinzii" as a nomen dubium, is not performed.
The Prorastomidae and Protosirenidae, both representing semiaquatic quadrupeds, are confined to the Eocene and considered paraphyletic (Domning 1994;2001a, b).The Dugongidae and Trichechidae represent fully aquatic animals and also include the few extant members belonging to the genera Dugong and Trichechus (Domning 2001a).According to Domning (1994), trichechids are monophyletic and nested within the Dugongidae.Consequently, the Dugongidae are paraphyletic.Within dugongids only two monophyletic groups are distinguished, the Dugonginae and Hydrodamalinae (Fig. 3).The most diverse dugongid subfamily and largest group within sirenians at all is represented by the Halitheriinae comprising the genera Caribosiren Reinhart, 1959, Eosiren Andrews, 1902, Eotheroides Palmer, 1899, Halitherium Kaup, 1838, Metaxytherium De Christol, 1840, and Prototherium De Zigno, 1887.However, this exclusively extinct group is paraphyletic indicating that higher-level phylogeny and systematics of the whole order Sirenia is paraphyly-based.Moreover, Domning's phylogenetic analysis (1994) points out that "halitheriines" itself mainly comprise genera that are invariably paraphyletic and in addition the type genus of this subfamily, "Halitherium", is also not monophyletic (Fig. 3).Domning (1994: 187) stated that his phylogenetic analysis is a preliminary work and in need of systematic revisions in many parts.Nevertheless, the traditional classification concept of four sirenian families is unrevised subject of systematic studies up to now (e.g., Velez-Juarbe et al. 2012).This causes many systematic problems due to missing diagnoses and hampers a reliable taxonomic assignment of new sirenian finds.According to Domning (1994), the most controversial groups refer to the Eocene dugongids and Eocene sirenian taxa in general, the incompletely resolved clade of the dugongines and the inter-and intrageneric relationships of Metaxytherium.The need of revision of these groups is beyond doubt.Additionally, the present study emphasises that the "Halitheriinae", comprising amongst others the Eocene dugongids and the genus Metaxytherium, constitute one of the most questionable sirenian groups, which is the focus of the following discussion.

The invalidity of Halitherium and its effects on sirenian systematics
According to article 11.7.1.1 (ICZN 1999) a family or subfamily, respectively, should be based on a valid genus.However, as outlined above this does not apply here, because the genus "Halitherium" is considered a nomen dubium and therefore is rejected.Since the "Halitheriinae" comprise a number of genera this problem could be solved by the designation of a new type genus.This taxonomic practise would be reasonable if the group in question represents a clade.However, the subfamily "Halitheriinae" has never been shown to form a monophyletic group in phylogenetic analyses considering Sirenia as a whole (Domning 1994;Fig.3) as well as on subsets of the order (e.g., Bajpai and Domning 1997;Domning and Pervesler 2001;Domning and Aguilera 2008;Velez-Juarbe et al. 2012).Doubts on the "halitheriine" monophyly are reinforced by the most up to date phylogenetic analysis on Sirenia recently performed by Voss (2013) in her doctoral thesis, which will be published at a later date.
In fact, Simpson (1945) and Domning (1996) demonstrate that the "Halitheriinae" rather represent a historically grown assemblage of taxa characterised by morphological features that neither substantiate a monophyletic grouping nor distinguish it from other subfamilies.The latter is amongst others corroborated by the overlap of characters such as "cheek teeth enamelled, closed roots" that were provided by Simpson (1932: 423-424) to define the "Halitheriinae", Miosireninae and Rytiodontinae.Consequently, the designation of a new type genus is not possible.
Likewise, the type genus "Halitherium" is traditionally considered to be monophyletic, because this has never been substantiated with sufficient evidence from any phylogenetic analysis up to now.The "Halitherium"-species complex regarded as valid within the European Palaeogene includes "H.schinzii" Kaup, 1838 (early Oligocene of Europe), "H."cristolii Fitzinger, 1842 (late Oligocene of Upper Austria) and "H." taulannense Sagne, 2001 (late Eocene of France).As indicated above, none of these taxa form a clade.Instead, each species represents individual branches in phylogenetic studies (e.g., Domning 1994;Domning and Aguilera 2008).This paraphyly of "Halitherium" is the main reason for the conclusion drawn herein to question the current use of the subfamily "Halitheriinae".Based on the fact that "Halitherium" is the type genus of the "Halitheriinae", this subfamily is to be reduced to a single species, the type species "H.schinzii", when all hitherto known phylogenies (e.g., Domning 1994;Domning and Aguilera 2008) are considered.Even if a new substitute name would be established for the genus "Halitherium" and the respective taxa of the former "Halitherium"-species complex receive a new designation, the taxonomic content of this subfamily will still comprise a single species.Therefore, the use of the "Halitheriinae" is not considered essential, because it does not contribute to a better understanding of sirenian diversity, but would rather lead to confusion.Instead, the rejection of the "Halitheriinae" is considered as a target-oriented solution into a more rigid taxonomic framework.

Conclusions
In the course of the revision of the genus "Halitherium", it is concluded that the premolar holotype of the type species "H.schinzii" neither can be clearly identified nor is assignable to a certain species or genus, and therefore it is non-diagnostic.Accordingly, these taxonomic terms are considered as nomina dubia.A neotype for "H.schinzii" is not designated here due, among other things, to the fact that the presence of two sympatric species in the German early Oligocene is hypothesised (Voss 2012), which makes it difficult to unambiguously define a suitable neotype.As a further consequence, the extinct subfamily "Halitheriinae" is refuted here.On the one hand, this is substantiated by the status of the type genus "Halitherium" as nomen dubium and, on the other hand, by the paraphyly of "halitheriines" and "Halitherium".Consi dering all previous phylogenies, especially the order-based ana lysis by Domning (1994), this subfamily consequently includes a single species only, viz."H.schinzii".This situation would remain unchanged even though a substitute name would be erected for "Halitherium" and "H.schinzii", respectively.Therefore, the use of the "Halitheriinae" is not considered practicable in this study.
As there have been made very few changes to the traditional paraphyly-based classification (see Domning 1994), a first step towards a revised taxonomic concept for the entire sirenian order is suggested here by rejecting the subfamily "Halitheriinae".Following this recommendation, not only one of the main sources of sirenian paraphyla is resolved, but also new approaches are provided to interpret the taxa originally lumped in this category.

Figure 1 .
Figure 1.Geographic setting and palaeogeography of the Mainz Basin (black star) in western Germany during the early Oligocene (~ 32 Ma).White areas indicate land, grey areas represent marine conditions (modified after Probst 1986).