Figure 1. Transformation of jaw bones in nonmammalian taxa into ear ossicles in mammals. In the nonmammalian synapsid, Thrinaxodon (a ‘mammal-like’ reptile) (bottom), the jaw joint consisted of an articulation between two bones: the quadrate in the cranium, and the articular in the mandible. Transitional stages, or ‘missing links’, in evolution between reptiles and mammals are represented here by the late cynodont therapsid, Probainognathus (second from bottom) and Morganucodon (a mammaliform taxon) (third from bottom; also see Table 1). In these forms two or more jaw joint articulations existed simultaneously. In Probainognathus, the jaw joint consisted of a typical reptilian quadrate–articular joint, with an additional squamosal–surangular joint (also reptilian character). In Morganucodon, the jaw joint was even more complex, with five bones involved: the quadratojugal and quadrate (not visible in this side view) articulated with the articular bone in the lower jaw and, in addition, the squamosal articulated with the dentary (the latter joint is found in modern mammals). In Mammalia (top right), only the squamosal–dentary articulation is present. During millions of years of evolution, jaw joint bones were transformed into ear ossicles: the quadrate became the incus (‘anvil’) and the articular became the malleus (‘hammer’). The stapes (‘stirrup’) was present in the middle ear from the early stages of vertebrate evolution. In addition, a portion of the angular (the reflected lamina) in the lower jawbone was transformed into the ectotympanic – the rim of the external acoustic meatus – which forms the outer opening of the bony ear. This is shown in the top left illustration, depicting the ear ossicles (and associated structures), of the marsupial mammal, Didelphis. The fenestra vestibuli is the oval window leading to the inner ear; the ‘footplate’ of the stapes fits into this window. Modified from Carroll (1988); Pough et al. (1996); Kardong (1995) and Vaughan et al. (2000).

Figure 2. A Mammalia cladogram for living taxa based on combined morphological evidence for living and extinct taxa (after Shoshani and McKenna (1998, p. 576); compare the branching in this figure to Table 2 and Figure 3). Outgroups include Chelydra, Sphenodon, Varanus and Cynognathus. Heavy solid lines represent known fossil occurrences (gaps in the records are ignored) (updated after Archibald (1996) and Gheerbrant et al. (1996). Fine solid lines indicate suggested relationships. Dashed lines and question marks indicate uncertain placement when a relationship is based on data from living taxa alone, or on combined data from extinct and extant taxa, based on branch-swapping tests, low bootstrap values (BV) and Bremer support values (BSV), and incongruities between morphological and molecular results. A question mark implies not so ‘good’ a character as others. An asterisk indicates congruence, whilst a bracketed asterisk indicates partial congruence, with molecular data. A hash sign indicates that no BV or BSV values given. Tree length for this figure is 664 steps (consistency index = 0.45). Numbers at nodes are BV for 1000 replications (bottom) and BSV (top). Selected synapomorphies for mammalian clades are provided. Some of these synapomorphies do not appear for certain taxa in previous literature.

Mammalia: character no. 2 – quadrate–articular contact is not involved in the jaw-joint articulation (manifested in the presence of three middle ear ossicles); character no. 44 – dentary is the largest or the only bone in the hemimandible; character no. 59 – cribriform plate present. Theria: character no.4 – prefrontal and/or post frontal absent; character no. 33 – cochlea auris with at least one full turn; character no. 219 – ovipary absent. Placentalia: character no. 67 – optic foramen is separate i.e. not confluent with the orbital fissure; character no. 75 – foramen ovale located within alisphenoid; character no. 228 – chorioallantoic placenta; character no. 254 – os caruncula or ‘egg tooth’ absent. Epitheria: character no. 130 – fibula–femur contact absent; character no. 204 – stylohyoid muscle is a derivative of posterior digastric; character no. 221 – vagina's longitudinal divisions, or traces of them, absent; character no. 234 – penis developed. Preptotheria: character no. 109 – pars chondralis of the interclavicle absent; character no. 215 – cloaca absent; character no. 237 – encephalization quotient more than 0.21. Unguiculata: character no. 22 (?) – tegmen tympani partly fused with tubal cartilage; character no. 103 (?) – scapular spine only –¾ of scapular length. Anagalida: character no. 72 (?) – masticatory and buccinator foramen(ina) present; character no. 226 – embryonic disc orientated toward mesometrial pole of uterus at time of implantation. Glires: character no. 11 – premaxilla–frontal narrow contact present before or at anterior border of orbits, nasals long (not retracted); character no. 32 – glenoid fossa elongate high with no postglenoid process; character no. 223 – uterus duplex. Archonta: character no. 198 (?) – flexor digitorum brevis manus absent; character no. 235 – penis free and pendulous. Primatomorpha: character no. 121 – shape of acetabulum is elliptical in outline; character no. 251 – neurovisual character complex. Volitantia: character no. 24 – fenestra rotundum of cochlea faces directly posteriorly; character no. 96 – ribs flattened; character no. 156 – patagium (including associated skeletal and muscular characters) present. Fereuungulata: character no. 159 – monoceps brachii i.e. short head is reduced or absent; character no. 242 – flocculus of cerebellum vestigal or not visible in dorsal view; character no. 255 (?) – enamel prism non-cylindrical and not surrounded by interprismatic matrix (no data for Pholidota and Tubulidentata). Ferae: character no. 261 – tentorium osseum well-developed, with possible parallelism in Tubulidentata; certain molecular data (see text in Shoshani and McKenna, 1998). Ungulata: character no. 83 – opening of stylomastoid foramen is dorsal to alveoli of cheek teeth; character no. 129 (?) – rounded or approximately ball-shaped patella (flattish and/or elongate in Hyracoidea); character no. 137 (?) – distal phalanges spatulate (non-spatulate in earliest Artiodactyla); character no. 158 – accessorius pedes (quadratus plantae) extremely reduced or absent. Cetungulata: character no. 154 – calcaneal peroneal tubercle indistinguishable or absent (no data for Cetacea for this and other characters for this node); character no. 162 – common calcaneal tendon = ‘hamstring’ for tendo Achilles absent; character no. 168 – semi- and presemimembranosus muscles fused almost to insertion; character no. 205 – iliocostalis and longissimus muscles are fused. Eparctocyona: character no. 134 – pes paraxonic i.e. digits III and IV are subequal; character no. 185 – lumbricales muscles reduced or absent; character no. 253 – incus crus breve is longer than crus longum. Altungulata: character no. 3 – petrosal and basiocciptital contact absent; character no. 108 – clavicle (even as vestigal or cartilage) absent; character no. 151 – astragalar head with short neck, or neck absent with flat head; character no. 236 – penial glandular fossa present. Uranotheria: character no. 12 – anterior border of orbit shifted anteriorly relative to cheek teeth, opening of infraorbital canal forward of orbit; character no. 140 – taxeopody; character no. 173 – ceratohyoideus absent; character no. 225 – reduced, free yolk sac (in later stages of development), associated with zonary placenta. Tethytheria: character no. 20 – zygoma thick and laterally expanded; character no. 182 – digastricus originates on stylohyal; character no. 207 – heart is bifid (with two apexes). Abbreviations: K–T, a standard geological term for the Cretaceous–Tertiary boundary (‘K’ is from the German ‘kreide’, for chalk); Ma, Mega annum.

Figure 3. Phylogram depicting relationships among major mammalian taxa based on molecular data (from the cover of Systematic Biology (1999) 48(1)). Considering the uncertainties expressed in this figure (see arrows) and in Figure 2 (see dashed lines and question marks), there are more agreements than disagreements between results obtained from morphological and molecular methods. In Figures 2 and 3, these are the clades with support from morphological and molecular methods. Placentalia; Tethytheria (Sirenia, Proboscidea; shown but not labelled in Figure 3); Uranotheria (Paenungulata in Figure 3, including Tethytheria and Hyracodiea); Eparctocyona (Cetacea and Artiodactyla; Certartiodactyla in Figure 3); Ferae (Carnivora and Pholidota; Zooamata shown but not labelled in Figure 3); Primatomorpha (in Figure 2, Dermoptera and Chiroptera are grouped under Volitantia, then Volitantia and Primates make up the Primatomorpha; in Figure 3, Dermoptera and Primates make up the Primatomorpha, and Chiroptera may join them, see arrow); Glires (Rodentia and Lagomorpha in both figures). It is noted that of the 18 placental mammalian order compared here, 12 are incorporated in the congruity assessment. Each method has its advantages and disadvantages. Incongruent results between morphology and molecules should be interpreted as challenges for future investigators since there is only one phylogenetic history for Mammalia. Abbreviations: K–T, a standard geological term for the Cretaceous–Tertiary boundary (‘K’ is from the German ‘kreide’, for chalk); Ma, Mega annum.