Le présent volume est l'aboutissement d'un projet né il y a presque cinq ans. En décembre 1991, l'un d'entre nous (MG) prenait des contacts en vue de proposer un symposium sur les mammifères fossiles, dédié à D.E. Russell, dans le programme du 4e Congrès de la European Society for Evolutionary Biology. Ce congrès, baptisé "Evolution 93", devait se tenir à Montpellier en août 1993. Son Comité d'Organisation, animé par F. Catzeflis, recherchait des organisateurs de symposiums. L'idée fut acceptée avec enthousiasme par le second d'entre nous (PDG), et le titre de notre Symposium fut précisé: " Palaeobiology and Evolution of Early Cenozoic Mammals - A Symposium in Honor of D.E. Russell". Le projet fut formellement accepté par le Comité d'Organisation en avril 1992. 

The scientific career of D.E. Russell began with a Pliocene fauna from Oregon, and then turned in the direction of European Paleogene mammals. Field work followed by study of the mammals that were collected, firstly in the Paleocene and later in the early Eocene, greatly rejuvenated learning in this field. Syntheses on the Northwest European Tertiary basin and on European marnmals and stratigraphy came next. Research on the Eocene of Asia was carried out jointly with Gingerich on Pakistan and with Dashzeveg on the faunas of Mongolia. An important synthesis on the entire Paleogene of Asia, joint with Zhai, followed. Field work in Africa with Sigogneau-Russell led to the discovery of Mesozoic mammals there. A synthesis of mammalian paleofaunas of the world was written with Savage, and a similar synthesis of Cenozoic vertebrate faunas is currently being prepared. These achievements reflect the perennial importance of field work, numerous collaborations with both amateurs and professionals, and the human qualities of this author.
  

A brief historical account of fossil vertebrate discoveries in the Eastern part of the Paris Basin between the beginning of the nineteenth century and 1950 is given. Other localities discovered since then are presented. A reconstruction of past landscapes is briefly elaborated. A biozonation based on mammals is proposed, from the Late Thanetian to the Middle Bartonian. Paleogeographical considerations are added. Suggestions regarding the search for new marnmal localities are made. 

Amongst a collection of selachian teeth made at Hoegaarden in a marine level of Bruxellian (Lutetian) age, containing a reworked Landenian (Sparnacian) fauna mixed with a contemporaneous one, a few teeth of  terrestrial mammals have been discovered. They comprise two rare European taxa: ? Hallensia sp. and Palaeonictis gigantea, both known from the Landenian. Even though the ?Hallensia has not been definitely identified, il differs from the only perissodactyl of this age previously recorded from Belgium (Cymbalophus cuniculus). 

Two new finds of pantodont materials from southern China, assigned to Archaeolambda, are described in this paper. One, a new species from the Nanxiong Basin, Guangdong Province, is similar to Alcidedorbignya inopinata from the early Paleocene of Tiupampa, Bolivia in size. It provides reliable evidence of the occurrence of Archaeolambda in the early-middle Paleocene of southern China. The second find includes specimens of Archaeolambda sp. cf. A. planicanina from the ?late Paleocene of Hengyang Basin, Hunan Province, which are the first record of a fossil mammal from the area near Hengyang city. The only vertebrate fossils previously found here were two genera of crocodiles discovered in 1938. This find sheds new light on the local biostratigraphy. 

Re-evaluation of the small collection of mammal fossils from the Changxindian Formation near Beijing, China indicates the following taxa are present: Eutheria, Hypsimilus beifingensis, cf. Miacis sp., Anthracotheriidae and Forstercooperia grandis. The presence of Forstercooperia grandis indicates an Irdinmanhan age and does not support previous assignment of a Sharamurunian age to the Changxindian Formation. 

Stratigraphic studies in the Aktau Mountains bordering the Dzhungarian Alatau Range in southeastern Kazakhstan have included mapping of Tertiary lithostratigraphic units, documentation of fossiliferous deposits, correlation of sections, etc. These investigations have led in turn to revised interpretation of the Tertiary geology of the area. The Tertiary sequence in the Aktau Mountains is represented by three lithostratigraphic units (in ascending order): (1) the middle Eocene Akbulak Formation; (2) the Oligocene Aktau Formation with a lower member including white quartz sands that contain fossil mammals, and an upper member including red-colored clays and sandstones, brick red clays, an anhydrite and gypsum clayey horizon, and bright brown-red clays; and (3) the upper Oligocene-Miocene Chul'adyr Formation with a lower member of greenish and yellowish conglomerates and gritstones, a middle member including grayish and yellowish sands and gritstones, and an upper member including brown and red clays and carbonate- and anhydrite-rich clays. The Aktau and Chul”adyr Formations represent separate cycles of sedimentation. Mammalian biostratigraphy and biochronology of the three vertebrate faunas in Aktau Mountains are reviewed. The mammalian fauna from white sands of the lower Aktau Formation is small but includes Ardynia and is thought to be early Oligocene in age. The mammalian fauna from conglomerates and gritstones of the lower member of the Chul”adyr Formation is also small but includes Paraceratherium and is thought to be late Oligocene in age. The mammalian fauna from sands of the middle member of the Chul'adyr Formation is extensive, with micro- and macrofauna attributed to Neogene mammal zones MN4 to MN 6, indicating a latest early Miocene to earliest middle Miocene age (Orleanian-Astaracian). Most genera of middle Chul”adyr mammals are known from the middle Miocene Shanwang faunas of China and from the Castelnau-d”Arbieu faunal assemblage (MN4-MN6) of southwestern France. 

New records of terrestrial mammals from the Qasr el Sagha Formation, Fayum Depression, Egypt are reported, and the stratigraphic occurrences of these fossils noted. These include additional specimens of Moeritheríum, Barytherium, and anthracotheres, as well as the oldest record of a hyracoid in the Fayum.These Eocene mammals occur almost exclusively in the alluvial deposits of the Dir Abu Lifa Member of the Qasr el Sagha Formation and show close affinities to the faunas from the lower sequence of the Jebel Qatrani Formation. There is no evidence of a more marked faunal discontinuity between the Qasr el Sagha and Jebel Qatrani Formations than there is across any of the three major breaks in sedimentation that exist within the Jebel Qatrani Formation. The faunal similarities between fossils of the lower sequence of the Jebel Qatrani Formation and of the upper part of the Qasr el Sagha Formation is consistent with recent paleomagnetic dating that suggests that these rocks differ in age by only one to two million years. 

Two species lineages of North American late Paleocene Plesiadapis exhibit a pattern of size divergence from a common ancestral lineage. Time series of fossils in each of these lineages are analyzed to test the idea that size divergence represents competitive character displacement. The critical factor in a test of character divergence is showing that divergent lineages evolved directionally rather than randomly (multifactorially). Analysis of evolutionary rates and their temporal scaling in Plesiadapis shows that both divergent species lineages have the scaling slope expected for lineages evolving randomly rather than directionally, and size divergence in Plesíadapis does not represent character displacement. Rates of evolution commonly observed on a per-generation time scale are high enough to produce character displacement within a few generations. Thus character displacement is not likely to be visible on scales of time that can be studied in the fossil record. 

In dilambdodont molars the primitive crest between paracone and metacone (centrocrista) is represented by a pair of crests that join the mesostyle (postparacrista, premetacrista). The cutting action of these crests against the crests of the hypoconid is described. Dilambdodonty is a derived adaptation for greater cutting efficiency. It has evolved several times and in more than one way. 

Three erinaceomorph species are known from the early Middle Eocene of Grube Messel near Darmstadt, Germany, which are referred to the family Amphilemuridae. Pholidocercus hassiacus, Macrocranion tupaiodon, and Macrocranion tenerum showed extraordinary adaptations to their different life strategies, and several of their specializations are unknown among living insectivores. Pholídocercus was a well-defended robust animal with an opportunistic feeding strategy. Macrocraníon zupaiodon was a slender forest floor-dweller with saltatorial specializations to escape from predators; fishes were the preferred component of its omnivorous diet. Macrocranion tenerum exhibited a combination of both survival strategies, extremely elongated hind limbs for rapid and even ricochetal flight and a spiny exterior as an effective protective device; it was probably specialized for feeding on ants. Thus, closely related, omnivorous-insectivorous forest floor-dwellers could exploit the Messel ecosystem. 

Tarsals that can be confidently attributed to Paschatheríum dolloi from Dormaal (Belgium) are described. The astragalus is short; its broad neck is set at an angle of 30 degrees to the trochlea. The trochlea is pulley-shaped and proximally extended. There is no astragalar foramen. The medial tibial facet extends distally in a cup deeply excavated in the neck and buttressed. The sustentacular facet extends toward the navicular facet but is not fully confluent with it. The calcaneum bears a proximo-distally extended proximal facet for the astragalus. A relatively large peroneal tubercle projects from the body, and is situated distally. Functionally, the trochlea indicates extensive flexion-extension movements of the foot. Calcaneo-astragalar facets indicate sliding and rotation between the two tarsals. The inclination of the navicular facet suggests frequent foot inversion. The peroneal tubercle reflects good muscular capacities for foot rotation. Overall morphology is interpreted as a scansorial adaptation similar to that of sciurids.Comparisons are made with the tarsals of Macrocranion vandebroeki from Dormaal and Hyopsodus paulus from the Bridgerian of Wyoming. Some similarities between the astragali of Paschatherium and Macrocraníon (trochlea) are interpreted as convergences for rapid locomotion. However absence of mobility at the lower ankle and midtarsal joints in M. vandebroekzi suggests that this species was cursorial, as is known for M. tenerum from Messel. Similarities in the calcanea of Paschatherium and Hyopsodus are probably the result of close phylogenetic relations, and confirm the placement of Paschatherium in the hyopsodontids. The differences in the calcanea of Paschatherium and Macrocranion underline that Paschatherium is distinct from erinaceomorph insectivores. The differences in astragalar morphology between Paschatherium and Hyopsodus show that a marked adaptive divergence separates the two genera. We speculate about the common occurrence of a deep tibial cup (“cotylar fossa”) and a pulley-shaped trochlea in Paschatherium and hyracoids, suggesting that an adaptive scenario similar to that having led to Paschatherium (scansoriality) might explain the acquisition of the peculiar hyracoid tarsal characters; such a scenario contradicts the concept of Pantomesaxonia. Other peculiar characters of Hyopsodus suggest that hyopsodontids might be given more consideration in the search for hyracoid (and tethythere) origins. 

The first good skeletal remains of Homogalax protapirinus from the Wasatchian of the Bighorn Basin, Wyoming, indicate that this primitive tapiromorph was more plesiomorphic in many features than primitive equoids including Hyracotherium. Compared to Hyracotherium, Homogalax more closely resembles Phenacodonta (the closest outgroup of Perissodactyla for which postcrania are known) in various details of articular surfaces, muscle attachments, and proportions of the humerus, manus, and pes.Among known taxa, Homogalax most nearly approximates the plesiomorphic postcranial skeletal anatomy of Perissodactyla. 

Paleosols and the taphonomy of vertebrate fossils in the Eocene Kuldana Formation of northern Pakistan provide important information on the preservation and time-averaging of fossil assemblages. Morphologic, mineralogic, and chemical data as well as comparisons with Quaternary soils suggest that Kuldana paleosols formed under generally dry and oxidizing conditions over time intervals of less than 100 000 years and perhaps as short as 1000 years. The distribution of carbonate in Kuldana paleosols further indicates that the upper half of the profiles were acidic whereas the lower halves were alkaline. Vertebrate fossils are rare in Kuldana paleosols and occur primarily in well-cemented sandstones and conglomerates with abundant micritic and iron-stained nodules that were reworked from floodplain soils. The scarcity of vertebrate remains in Kuldana paleosols probably reflects a combination of acidic, dry, and oxidizing conditions in the upper half of the profiles and rapid floodplain sedimentation. Comparisons between the taphonomic characteristics of Kuldana channel fossil assemblages and bone accumulations in modem rivers provide a basis for estimating the length of time represented by Kuldana fossils from several important localities. Vertebrate fossil assemblages from Barbora Banda are characterized by a low-diversity paleofauna, partially articulated skeletons, and bones that are sorted by size and shape. Comparison with bone accumulations in modern rivers suggests that the fossils from Barbora Banda accumulated in 1 to 10 years. Vertebrate fossils from the Lower Kuldana in the Kala Chitta Hills region, typified by locality H-GSP 62, are characterized by a high-diversity paleofauna and generally random and unsorted fossil bone distributions, which suggest that the fossils from these localities represent longer time intervals than the Barbora Banda fossils. Based on the time estimates for Kuldana paleosol development, fossil assemblages in Kuldana channel deposits in the Kala Chitta Hills region probably represent time intervals of about 1000 years. 

In this article is described a new species of Nyctitheriidae with primitive characters: Leptacodon nascimentoi n. sp. from the early Eocene of Silveirinha (Portugal). 

A new genus, Eppsillycteris, is erected for Adapisorex? allglicus COOPER, 1932, from the earliest Eocene Blackheath Beds of Abbey Wood, London, England. Various derived character states indicate that it belongs to the order Chiroptera (bats) rather than to the extinct "insectivore" family Adapisoricidae. Other derived character states are shared with fossil and modern members of the family Emballonuridae. Placement of the new genus in this family extends the record of the Emballonuridae back in time by about 10 million years. It is the earliest record of a modern bat family and one of the earliest bats. This implies that the differentiation of at least some modern bat families took place in the Palaeocene, where no authenticated records of bats yet exist. The primitive characters of the earliest bats make the family Nyctitheriidae an unlikely stem group for the order Chiroptera. A tentative plausible alternative exists in some unnamed upper molars from the Palaeocene of Walbeck, Germany. Wyollycteris chalix, described as a bat from the Late Palaeocene of Wyoming, U,S.A., fits better in the family Nyctitheriidae. 

The earliest North American rodents occur in basal Clarkforkian beds of the Fort Union Formation at Big Multi Quarry near Bitter Creek, northern Washakie Basin, Sweetwater County, Wyoming, and in closely correlative Fort Union beds formerly accessible in the Eagle Coal Mine near Bear Creek, northern Clark's Fork Basin, Carbon County, Montana. Two new species of early Clarkforkian rodents, Paramys adamus and Alagomys russelli, are described from Big Multi Quarry. Paramys adamus is represented by virtually complete upper and lower dentitions, which demonstrate that this species is one of the most primitive North American paramyids yet discovered. These specimens form the basis for a reevaluation of the content and stratigraphic range of P. atavus, which is known with certainty only from Bear Creek. Alagomys russelli is the first North American record for the enigmatic rodent family Alagomyidae, otherwise known from ?late Paleocene-early Eocene localities in Mongolia and China. Phylogenetic analysis of dental and gnathic traits suggests that Alagomyidae form the sister group of all other undoubted rodents. At least two rodent clades, alagomyids and basal paramyids, seem to have invaded North America from Asia at the beginning of Clarkforkian time, but only the paramyids persisted to undergo a significant evolutionary radiation in North America. 

A new European rodent from the middle Eocene of Spain, Zamoramys extraneus n. gen., n. sp., appears to be closely related to the middle Eocene chapattimyid rodents of Indo-Pakistan. This contradicts the generally accepted paleobiogeographic hypothesis of a Tethyian barrier between Europe and Asia isolating Europe during the middle Eocene. Because of this barrier, some authors have proposed that European and Asian rodents were not closely related, their similarity being the result of morphological convergence. Here monophyly has been tested, using the parsimony criterion, based on an analysis of dental characters (including discussing of homology and the validity of some characteristics). Our results indicate a phylogenetic relationship among the Asiatic Ctenodactyloidea, Zamoramys from Spain, and the European endemic Theridomyoidea. We also conclude from our analysis that theridomyoids and European ischyromyoids are probably not closely related phylogenetically. 

A partial skull of Ardynomys russelli sp. nov. (Rodentia, Cylindrodontidae) is described. This was collected in the late Eocene of Alag Tsab locality in the eastem Gobi Desert, Mongolia. Ardynomys russelli sp. nov. is characterized by small size, brachyodont molars, and retention of P3. It represents the earliest record of the genus Ardynomys MATTHEW & GRANGER, 1925, in Asia. 

A new hypothesis for the phylogenetic relationships of recent anomalurids and graphiurids is proposed, based on information from evolutionary lineages of Paleogene European rodents, particularly Gliridae, and Eocene Algerian Zegdoumyidae. Differences in first occurrences, in paleogeography, and in infraorbital structure in glirids (protrogomorphy and pseudomyomorphy) and graphiurids (hystricomorphy) separate Graphiuridae from Gliridae (Graphiurinae is here raised to family rank). Similar considerations, and dental morphology, suggest that Anomaluridae (appearing in the late Eocene) and Graphiuridae (appearing in the Pliocene) are related to early Eocene Zegdoumyidae. 

Isolated upper cheek teeth of the primitive artiodactyl Diacodexis sp., upper molars of Eolantianius russelli gen. et sp. nov. (Diacodexeidae), two lower molars tentatively referred to Eolantianius russelli gen. et sp. nov., and astragali of Diacodexeidae indet. are described from the early Eocene (late Ypresian) of locality Andarak 2 in Kyrgyzstan. 

Dichobunoid artiodactyls are described in this paper from the middle Eocene Geiseltal lignite deposits near Halle (Sachsen-Anhalt, Germany). The genera Eurodexis and Parahexacodus are established based on odontologica1 studies. The type-species are E. ceciliensis (FRANZEN & KRUMBIEGEL, 1981) and P. germanicus n. sp. from the "obere Mittelkohle" (see text), MP 13. The genera are referred to the new subfamily Eurodexeinae (Dichobunidae, Artiodactyla, Mammalia), which also contains the monospecific genus Eygalayodon SUDRE & MARANDAT, 1993. These new genera show many features similar to North American homacodontids and antiacodontids. They exemplify a high degree of diversity in European dichobunids. Moreover, the family Dichobunidae includes the Dichobuninae and the Hyperdichobuninae. The connection of the posthypocristid with the postentocristid, the shift of a medial to a lingual position of the hypoconulid on the lower molars and the development of a large, caniniform P1 are regarded as the principal tendencies of the eurodexeines. The entoconid is larger and more mesially situated compared to that of Diacodexis. The hypoconulid has a lingual position on a broad postcingulid. The elongation of the premolars and the presence of diastemata are considered associated with extension of the muzzle. This and the acute tubercles of the molars could indicate a limited degree of insectivory. Protodichobune cf. oweni and Diacodexis cf. varleli, first reported here from the "untere Unterkohle" (MP 11) of the Geiseltal, demonstrate the existence of forms in the basal part of the middle Eocene, that are direct descendants of lower Eocene ones. Eurodexis russelli n. sp. from MP 10 of Premontre is the most primitive representative of Eurodexis. A further part of this lineage could be Messelobunodon sp. from MP 11 of Messel (Germany). Another form from Premontre, referred to Eurodexeinae indet., represents the origin of the lineage to Parahexacodus germanicus n. sp. from the Geiseltal (MP 13). The identification of several lineages in MP 10 is evidence of an adaptive radiation during this time. Unfortunately the history of the earlier species of Diacodexis is not known well enough to exclude earlier diversifications. Our study shows the independent development of artiodactyls at the end of the lower Eocene in Europe, North America (and maybe in Asia) with much convergence. The question of the geographic origin of the genus Diacodexis is still open. 

The artiodactyls (Mammalia) from the latest Ypresian locality of Prémontré from the Paris Basin (niveau repère MP 10 in the lower Eocene of the Paris Basin) are described in this paper. Three species have been identified: 1) Diacodexis cf. varleti SUDRE et al., 1983; 2) a new species of Eurodexis ERFURT & SUDRE (E. russelli nov. sp.) defined after the revision of the species Messelobunodon? ceciliensis from the Lutetian beds of Geiseltal (Germany); and 3) Eurodexeinae indet., a probable ancestor of another form from the Geiseltal which was previously recorded as Homacodon? sp. (Erfurt 1993) and now named Parahexacodus germanicus. The two later forms are referred to the new subfamily Eurodexeinae (Erfurt & Sudre 1996). The analysis of these forms as weIl as comparative studies have led us to reconsider our previous conclusions regarding the content of the species Protodichobune oweni LEMOINE 1878 and some aspects of Ypresian diacodexid evolution. One can postulate that the divergence of E. russelli nov. sp. occurred during the first radiation of these primitive artiodactyls. Some other stem form with bunodont teeth such as Protodichobune and Aumelasia have also differentiated from Diacodexis. Like Eurodexis, these two genera persist during the middle Eocene. The absence of Protodichobune and Aumelasia at Prémontré is probably due to particular ecological conditions.