Introduction

In 1998 at Park Hill, Croydon, Greater London, a section in the Woolwich Formation of earliest Eocene age, originally exposed in 1882 during excavation of a railway cutting, was reexposed when the cutting was widened for the Croydon Tramlink (Hooker et al. 2009). In 1882, the segment of the cutting known as Sandilands showed what is now identified as the upper shelly clay unit of the Woolwich Formation, overlying the Reading Formation and underlying the Blackheath Formation (Klaassen 1883; Hooker et al. 2009; Hooker 2010). Within apparently channelling blue clays (base not seen in 1998), Klaassen (1883) recorded three small channel fills. In the central of these three channels he found bones of the flightless bird Gastornis (Newton 1886) and an ulna of the pantodont mammal Coryphodon (Newton 1883). In 1998, thanks to Klaassen's accurate measurements, the central channel was relocated and bulk sampled. Among the small assemblage of mammals recovered (Hooker et al. 2009) were a lower jaw and three isolated lower molars of the omomyid Melaneremia. These differ from the only previously described species of the genus, Melaneremia bryanti Hooker, 2007, and here form the basis of a new species.

The annotation of the synonymy lists follows Matthews (1973). Scanning electron micrographs of casts of specimens were taken on a Philips XL30.

Institutional abbreviations.—CM, Carnegie Museum of Natural History, Pittsburgh, USA; NHMUK, Palaeontology Department, Natural History Museum, London, UK.

Other abbreviations.—PETM, Paleocene—Eocene Thermal Maximum; RHTS, Red Hot Truck Stop.

Systematic palaeontology

Fig. 1.

Scanning electron micrographs of gold—palladium coated epoxy casts of the holotype left dentary of the omomyid primate Melaneremia schrevei sp. nov., NHMUK.M85501, central channel, upper shelly clays, Woolwich Formation, Sandilands cutting, Park Hill, Croydon, in occlusal (A), buccal/lateral (B), and lingual/medial (C) views.

Fig. 2.

Scanning electron micrographs of gold—palladium coated epoxy casts of lower molars of the omomyid primate Melaneremia schrevei sp. nov., central channel, upper shelly clays, Woolwich Formation, Sandilands cutting, Park Hill, Croydon. A. Right ml (reversed), NHMUK.M85504. B. Right m2 (reversed), NHMUK.M85502. C. Left m2, NHMUK.M85503. Views are occlusal (A₁, B₁, C₁), buccal (A₃, B₂, C₂), and lingual (A₂, B₃, C₃).

Fig. 3.

Scatter diagram of length vs width (in mm) of lower cheek teeth of Melaneremia schrevei and M. bryanti. Lines join teeth of single individuals. The length dimension is measured lingually from the bulge of the distal crown base to the tip of the paraconid.

Phylogenetic relationships

Choice of taxa.—M. schrevei is here included in a character-taxon matrix (Appendix 2) similar to that analysed when M. bryanti was described (Hooker 2007). As the aim of this paper is not to analyse the relationships of the Omomyidae as a whole, but to try to establish the relationships of the Microchoerinae to the rest, emphasis is placed on analysing the better known of the more primitive taxa. As the anaptomorphine tribe Trogolemurini has previously been implicated in microchoerine phylogeny (Ni et al. 2004), the primitive trogolemurins Anemorhysis savagei and Teilhardina demissa (see Tornow 2008) are also included here. Since the paper describing M. bryanti was accepted for publication, Teilhardina magnoliana Beard, 2008a has been named and T. brandti Gingerich, 1993 has become better known (Smith et al. 2006). In addition, Late Eocene Paraloris bavaricus Fahlbusch, 1995 was found to branch at the base of the Microchoerinae (Hooker and Harrison 2008). Being at the stem of omomyid evolution, these species are also included in the analysis here, with some modifications to and augmentation of the 2007 character list (see Appendix 1 for changes to the characters). Apart from the well-represented Teilhardina asiatica Ni, Wang, Hu, and Li, 2004, several other genera of primitive omomyids are now known from Asia, including: Kohatius Russell and Gingerich, 1980; Vastanomys Bajpai, Kapur, Thewissen, Das, Tiwari, Sharma, and Saravanan, 2005 (see also Rose et al. 2009); Baataromomys Ni, Beard, Meng, Wang, and Gebo, 2007; and Indusius Gunnell, Gingerich, U1-Haq, Bloch, Khan, and Clyde, 2008b. However, these are currently too poorly known to be usefully included in this analysis. Other microchoerines included are Nannopithex zuccolae and Vectipithex raabi (Godinot et al. 1992; Hooker 1996b; Hooker and Harrison 2008). Once again Purgatorius Van Valen and Sloan, 1965 is chosen as outgroup, and in the PAUP 4.0b10 (Swofford 2002) analysis, this is treated as paraphyletic and the ingroup as monophyletic, thus characters found only in and throughout the ingroup have been excluded as having no influence on the result. As the morphological gulf between Purgatorius and the most primitive omomyids is large, two primitive adapiforms, Cantius eppsi and Donrussellia, are added to the ingroup to aid character polarity. They are not, however, treated as outgroup taxa, as they are potentially as derived in their own ways as omomyids.

Notes on characters.—Height of the p4 metaconid has previously been used in phylogenetic analyses (Hooker 2007: character 10; Smith et al. 2006: character 9). With the inclusion here of taxa with different overall relative heights of p4 (the protoconid being the tallest cusp), and given the substantial intraspecific variation in height of the metaconid in the well-represented species Teilhardina americana Bown, 1976 and Tetonius matthewi Bown and Rose, 1987 (Rose and Bown 1986), this character is abandoned.

Combining the relative sizes of the first incisor and canine (Hooker 2007: character 1) is also abandoned in favour of separate characters (1 and 2 herein). This accords with evidence for decoupling in several cases.

Taxa previously scored as having no p4 postmetacristid (Hooker 2007: character 11) are found to have faint basal development of this crest. The character is emended thus and the scoring slightly revised (character 16 herein).

Strength of the lower molar ectocingulid has also been used in previous analyses (e.g., Smith et al. 2006). However, it was considered too variable to be used here for the taxa involved.

Results.—Using the heuristic search and the stepwise addition option, random, with 1000 addition sequence replicates, PAUP found four maximum parsimony trees (MPTs) of 186 steps, a consistency index (CI) of 0.3602 (all characters parsimony informative) and a retention index (RI) of 0.6007. The strict consensus is poorly resolved. Thus, essentially an omomyid polytomy, only resolved into the following clades, (Altanius, Anemorhysis savagei), (Nannopithex zuccolae, Vectipithex raabi) and (Steinius vespertinus (Loveina, Omomys carteri)), is sister group to Teilhardina asiatica. This omomyid clade is sister group to the adapiform clade (Cantius eppsi + Donrussellia).

Fig. 4.

Maximum parsimony tree resembling the majority rule consensus (Fig. 5A), generated by PAUP 4.0b10 from the character-taxon matrix in Appendix 2, showing character state changes; see Appendix 1 for description of numbered characters. Broad bar, synapomorphy; narrow bar, normal polarity homoplasy; X, reversal. Characters that vary with different optimisations are enclosed between [ ] for Acctran and between ( ) for Deltran.

In fact, there is little difference between the four MPTs. The largest difference is in the position of the Nannopithex zuccolae + Vectipithex raabi clade. In one tree it is nested successively with the North American Uintanius rutherfurdi, Jemezius, and Teilhardina demissa. This is an unlikely position on biogeographical grounds, given that the opening of the North Atlantic had cut off land routes between Europe and North America several million years earlier (Hooker 2007, 2010, and references therein). Jemezius is from Wasatchian Zone Wa-6 (52.9–53.5 Ma), whilst the earliest Uintanius rutherfurdi (Robinson, 1966) is from Bridgerian Zone Br-1a (52–52.4 Ma) (Beard 1987; Beard et al. 1992; Clyde et al. 2001; Luterbacher et al. 2004; Gunnell et al. 2008a, 2009).

In the other three trees, the Nannopithex zuccolae + Vectipithex raabi clade forms part of a monophyletic Microchoerinae. The only differences between these three trees involve the combination of two alternatives. The first is whether Teilhardina belgica is sister taxon to a trichotomous clade comprising T. magnoliana, a monophyletic Microchoerinae and North American and Asian omomyids (Omomyinae + Anaptomorphinae), or is part of that clade. The second is whether Microchoerinae plus North American and Asian omomyids form a clade independent of T. belgica and T. magnoliana, or separately form a polytomy with them.

One of these three MPTs resembles the majority rule consensus and is shown with its character transformations (Figs. 4, 5A). It shows Microchoerinae, Omomyinae + Anaptomorphinae, Teilhardina belgica and T. magnoliana forming a quadritomy, which is sister group to T. asiatica. The collapsing of T. belgica into this polytomy in this MPT by PAUP has created two extra steps and changed the CI to 0.3564 and the RI to 0.5940. Teilhardina is clearly paraphyletic as has been pointed out before (e.g., Hooker 2007; Tornow 2008), but three species in particular, T. asiatica, T. belgica and T. magnoliana, cannot be accommodated in any of the three currently named subfamilies.

Three key characters unequivocally support the M. bryanti + N. zuccolae + V. raabi clade. They are: p4 distal talonid wall sloping, with extensive contact with ml (character 18); m3 metaconid with accessory cuspule in some individuals (character 38: unique, but insufficiently tested as only one M. schrevei m3 is known); and procumbent lower molar paraconid (character 40). A long M1–2 trigon basin (character 22) may also characterise this clade (Deltran). However, it could also have evolved at the base of the Microchoerinae clade (Acctran), but upper teeth are currently unknown for M. schrevei and Paraloris. The positioning of other characters here or at lower microchoerine nodes also varies according to the optimisation method, reflecting missing data in one or another taxon.

Five key characters support the M. schrevei + M. bryanti + N. zuccolae + V. raabi clade. They are: an open m1 trigonid (character 32); a stepped ml protocristid (character 33); an enlarged lower molar paraconid (character 39); a wide m3 hypoconulid lobe (character 51); and secondary development of a tall p4 paraconid (character 15).

A concave postmetacristid is shown as an autapomorphy of M. bryanti (character 48), shared convergently only with T. magnoliana. Strict autapomorphies were not included in the analysis (see differential diagnosis of M. schrevei above). However, intermediate grade m2 paraconid separation from the protocristid (character 35), low-crowned p3–4 (character 13), both under Deltran, and angle of buccal m3 crown base shift to below protoconid (character 46), under Acctran, are shown as autapomorphies of M. schrevei. However, character state 35.1 may be the result of enlargement of the paraconid in the microchoerine clade and therefore in this case a transient feature, reversed by subsequent procumbent orientation of the paraconid (character 40). Character state 13.2 may also not really be an autapomorphy, since p3 is unknown in M. bryanti and Paraloris. The primitive state of character 46 is also an unlikely reversal for M. schrevei. However, the very small size and relatively buccal position of the P3 protocone (Hooker 2010) may be an autapomorphy of M. bryanti or of the genus Melaneremia, although the tooth type is unknown in M. schrevei. Therefore, with no unequivocal autapomorphies and in view of its stratigraphic occurrence, there is the possibility that M. schrevei and M. bryanti are stages in the evolution of a single biological species, as has been found in some North American omomyid species and genera (Rose and Bown 1986). However, in the absence of demonstrable character overlap, M. schrevei and M. bryanti are conservatively regarded as specifically distinct.

Three key characters support the Paraloris + M. schrevei M. bryanti + N. zuccolae + V. raabi clade (i.e., Microchoerinae). They are: a steep m2–3 trigonid back wall (character 37); m2–3 paraconid sublingually positioned (character 43); and reduction in molar crown height (character 49: reversed in the N. zuccolae + V. raabi clade).

The Omomyinae/Anaptomorphinae part of the tree differs from the results of previous recent analyses (e.g., Ni et al. 2004; Tornow 2008), although it is uncertain whether this is because of the addition of Paraloris and Melaneremia or because of the limited number of omomyine and anaptomorphine taxa used. However, the microchoerines included in the analyses of these authors were also limited (partly by what was available) to only derived members, whilst Asian Altanius was excluded from the analysis of Tornow (2008). Ni et al. (2004) included Altanius, but found that it grouped with plesiadapiforms. However, Altanius lacks the enlarged incisors, reduced canines and semimolariform last premolar of most plesiadapiforms. It was compared favourably with the carpolestid Elphidotarsius by Rose and Krause (1984) before the anterior dentition was known (Gingerich et al. 1991). However, unlike this genus, it has a well developed p4 trigonid, not an incipient slicing blade. Also, the paraconid of m2–3 is distally positioned with respect to the paracristid, subequal in height to the metaconid and has a strong buccal rib, which instead resembles the condition in Teilhardina demissa, Tetonoides, Anemorhysis, and Arapahovius. The molar similarities to Elphidotarsius can be largely accounted for by simple mesiodistal shortening.

As in the analysis originally including M. bryanti (Hooker 2007), neither Omomyinae nor Anaptomorphinae are monophyletic in the new analysis. However, by contrast, here it is the uintaniin omomyines Jemezius and Uintanius that nest within the Anaptomorphinae, with other omomyines (primitive members of the tribes Omomyini and Washakiini) remaining distinct, but with Teilhardina brandti being the sister taxon to the rest. The new analysis also shows Altanius as sister taxon to Anemorhysis. What are considered primitive characters of Altanius, such as large canines and the presence of a first premolar, are shown here as reversals to the primitive state. The absence of close relatives of Altanius suggests that there may be more homoplasy than currently recognised involving modification of the anterior dentition within Omomyidae. Such homoplasy is already increased (Hooker 2007, herein) by the addition of the primitive microchoerine Melaneremia to analyses, distancing derived microchoerines from a relationship with derived Trogolemurini (cf. Ni et al. 2004). Additional parallelism in this trend (if found) could result in retention of, rather than reversal to, various primitive states in Altanius.

Transformations of the anterior dentition.—Large canines and small incisors in mammals are universally regarded as the primitive states for size of these tooth types. They are also usually treated as primitive when encountered in primates, e.g., the primitive omomyids Teilhardina asiatica and T. belgica, and the primitive adapiforms Cantius and Donrussellia (e.g., Rose and Bown 1991 and references therein). However, the Plesiadapiformes, the generally accepted sister group to primates (e.g., Bloch et al. 2007; Silcox 2003) universally have enlarged first incisors and usually reduced canines (except Purgatorius, where both teeth are large: Clemens 2004). Moreover, a similar pattern exists in the other two (and more remote) modern euarchontan orders, Scandentia and Dermoptera, which have enlarged incisors (in Dermoptera only in the lower jaw) and lack large canines (Butler 1980; Rose 2006). The extinct euarchontan families Mixodectidae, Plagiomenidae (Rose 2008) and Nyctitheriidae (Sigé 1976; Hooker 2001) likewise have enlarged incisors and reduced canines. Therefore, use of any non-primate euarchontan as outgroup will produce a derived state for small incisors in primates. Herein, Purgatorius is used as outgroup, so large canine is nevertheless taken as primitive. However, small incisors in Teilhardina asiatica and T. belgica become derived.

Fig. 5.

A. 50% majority rule consensus of four maximum parsimony trees generated by PAUP 4.0b10, from the character-taxon matrix in Appendix 2. Numbers in square boxes delimiting nodes are decay indices (Bremer 1994), those in circles are consensus percentages. B. The same tree constrained by stratigraphic ranges of the taxa. Thickened sections of branches indicate stratigraphic ranges. Those shown as solid reflect the level of accuracy provided by the cited references and in some cases are known from only one horizon somewhere in that range. That shown as vertically hatched indicates uncertain age (see text). The timescale and correlation follow Luterbacher et al. (2004). Ranges are from: Rose and Bown (1986), Beard (1987, 2008b), Beard et al. (1992), Thalmann (1994), Hooker (1996b, 2010), Smith et al. (2006), Gingerich et al. (2008), Gunnell et al. (2008a, 2009), Beard and Dawson (2009), and herein. Abbreviations: Ma, millions of years before present; ME, Middle Eocene; P, Paleocene.

As no anterior dentitions (incisors and canine) or even alveoli are known for either species of Melaneremia, characters 1 and 2 are coded as “?” (Appendix 2). Therefore, the position at which canine size reduced to small in N. zuccolae and V. raabi varies between the basal microchoerine node in one step (Acctran) and the node subtending N. zuccolae and V. raabi in two steps (Deltran). The same situation exists for the enlargement of the incisors. The enlarged state of i1 is unlikely in either species of Melaneremia for two reasons. The dentary in M. schrevei is shallower than in N. zuccolae or V. raabi. However, it is still deepening slightly, anteriorly below p4, unlike in T. belgica and Paraloris, where anterior shallowing begins just in front of m1 (Teilhard 1927; Fahlbusch 1995). This suggests that M. schrevei had an i1 that was larger than in T. belgica or Paraloris, but smaller than in N. zuccolae or V. raabi (i.e., state 1 of character 1), as well as probably a slightly reduced canine. The dentary of M. bryanti is also shallower than that of N. zuccolae, and its p2 is procumbent with the mesial interstitial facet underneath the mesial overhang. This contrasts with e.g., T. americana, where the anterior premolars are well spaced and there is little or no contact between the canine/p1 and p2, and suggests crowding in M. bryanti, consistent with incipient enlargement of i1 and incipient reduction of the canine as suggested for M. schrevei. Therefore, the implication is that incisor enlargement and canine reduction in Microchoerinae took place in two steps, to states 1.1 and 2.1 at the node above Paraloris, and to states 1.0 and 2.2 at the node above M. bryanti.

Stratigraphie issues.—Figure 5B shows the majority rule consensus from Fig. 5A plotted against stratigraphy. It reflects the absence of an early record of Omomyinae (nothing in the first nearly 2 million years of the Eocene: Gunnell et al. 2008a) and a gap of 4 million years in the early record of Microchoerinae between the youngest known Melaneremia and the oldest known Nannopithex. The most incongruent stratigraphic range, however, is that of Paraloris bavaricus Fahlbusch, 1995, which is known only from the holotype dentary from Priabonian strata in southern Germany. The cladistic analysis shows it to be the most primitive microchoerine, with the implication of a ghost record for its clade of about 20 million years. Only the discovery of more material of this enigmatic taxon and its earlier relatives will resolve the questions surrounding its evolution. Melaneremia remains the key genus at the root of mainstream microchoerine evolution as recorded in most of Europe.

The dating of Teilhardina magnoliana from the Red Hot Truck Stop site (RHTS), Meridian, Mississippi, is controversial (Beard 2008b; Gingerich et al. 2008). The presence of the dinocyst Apectodinium augustum reported by Beard and Dawson (2001) is relevant. This species is restricted to the Paleocene—Eocene Thermal Maximum (PETM), which records an Apectodinium acme (Bujak and Brinkhuis 1998). However, acme percentages for the RHTS have not been given by Beard and Dawson (2001, 2009). Moreover (contra Beard 2008b), it is the onset of the Apectodinium acme, not the acme itself, which at some sites antedates the PETM Carbon Isotope Excursion (Sluijs et al. 2007b). There is a major sequence boundary between the top of the Tuscahoma Formation (slightly above the RHTS mammal horizon, the T4 channel) and the overlying Bashi Formation (Ingram 1991). However, the Bashi Formation belongs to calcareous nannoplankton zone NP10 (Frederiksen et al. 1982), which begins at least 700ky after the end of the PETM (Luterbacher et al. 2004). Thus, if the T4 channel does date from the PETM, it implies a larger hiatus than usually considered present below the Bashi Formation (e.g., Frederiksen et al. 1982). The basal Bashi sequence boundary in any case (contra Beard 2008b) is younger than the Tht-5 sequence boundary, which underlies the Dormaal Sand Member in Belgium (Steurbaut et al. 1999). The former instead probably equates with that of Ypr-2 in Europe (Bujak and Brinkhuis 1998). Accepting that the T4 channel does indeed belong to the PETM, the presence of typical Eocene pollen taxa well below in the upper Tuscahoma Formation (Harrington 2003a, b, 2004; Harrington and Jaramillo 2007) implies that the T4 channel should belong to the later part of the PETM. The range given for T. magnoliana (Fig. 5B) attempts to reflect current knowledge.

Dating of Altanius orlovi Dashzeveg and McKenna, 1977 within the Early Eocene is largely speculative (Fig. 5B). There is no clear evidence that it is from strata as old as the PETM, as there are no species in common with those from the documented PETM Carbon Isotope Excursion at Lingcha, China (Ting et al. 2011).

Relevance of Melaneremia to European bio stratigraphy

The Woolwich Formation is spanned by an acme of the dinoflagellate Apectodinium that proxies the PETM (Collinson et al. 2009). Underlying this formation locally in Kent is the Cobham Lignite Bed, which records a negative carbon isotope shift that marks the onset of the PETM (Collinson et al. 2003; Pancost et al. 2007). The occurrence of Melaneremia schrevei in the upper part of the Woolwich Formation, the upper shelly clay unit (Hooker et al. 2009), therefore links its occurrence to the later part of the PETM.

Two isolated teeth, a p4 and an m2, from the Conglomérat de Meudon, Vaugirard Formation of Meudon, Paris Basin, figured by Godinot et al. (1998: pl. 11.1 : k, 1) bear a striking resemblance to the same teeth of M. schrevei, suggesting that they are conspecific. The m2 in particular shows spacing of the paraconid and protocristid identical to that of M. schrevei. The p4 differs slightly in having a crestiform paraconid. The mammal fauna of Meudon formed the basis of mammal Biozone PE II (Hooker 1996a), and the apparent occurrence of M. schrevei at both Meudon and Croydon suggests attribution of the Croydon fauna to the same zone. The underlying Biozone PE I is based on the mammal fauna of Dormaal, Belgium, which is correlated to the early part of the PETM (Steurbaut et al. 1999). Dormaal and other PE I sites contain a different, more primitive omomyid, Teilhardina belgica. The more derived species of Melaneremia, M. bryanti, occurs in the Black-heath Formation, which overlies the Woolwich Formation in the London Basin and whose mammal fauna has formed the basis of Biozone PE III (Hooker 1996a, 2010). Zone PE II is therefore tentatively anchored to the later part of the PETM, correlating approximately with the lower part of Wasatchian Zone Wa-1 of the Bighorn Basin, Wyoming, USA. As the omomyids at this time in Europe and North America are more different than in the early part of the PETM, this appears to be when a restriction to dispersal and the beginning of differentiation took place. This is rapid given the 170 ky extent (Sluijs et al. 2007a) of the Paleocene—Eocene Thermal Maximum.