Scanning Electron Microscopy

Volume 1986 Number 4 Article 35

10-9-1986

The Enamel Ultrastructure of Multituberculate Mammals: A Review

D. W. Krause State University of New York

S. J. Carlson University of California, Davis

Follow this and additional works at: https://digitalcommons.usu.edu/electron

Part of the Life Sciences Commons

Recommended Citation Krause, D. W. and Carlson, S. J. (1986) "The Enamel Ultrastructure of Multituberculate Mammals: A Review," Scanning Electron Microscopy: Vol. 1986 : No. 4 , Article 35. Available at: https://digitalcommons.usu.edu/electron/vol1986/iss4/35

This Article is brought to you for free and open access by the Western Dairy Center at DigitalCommons@USU. It has been accepted for inclusion in Scanning Electron Microscopy by an authorized administrator of DigitalCommons@USU. For more information, please contact [email protected]. SCANNING ELECTRON MICROSCOPY /1986/IV (Pages 1591-1607) 0586-5581/86$1.00+05 SEM Inc., AMF O'Hare (Chicago), IL 60666-0507 USA

THE ENAMELULTRASTRUCTURE OF MULTITUBERCULATEMAMMALS: A REVIEW

1* 2 D. W, Krause and S. J, Carlson

1Department of Anatomical Sciences, State UnivP.rsity of New York, Stony Brook, New York 11794 2Museum of Paleontology and Department of Geological Sciences, The University of Michigan, Ann Arbor, Michigan 48109 2Present address: Department of Geology, University of California Davis, CA 95616 (Received for publication May 09, 1986, and in revised form October 09, 1986) Abstract Introduction

The enamel ultrastructure of multituber­ The enamel ultrastructure of multituber­ culate mammals has been sampled extensively and culate mammals has been sampled extensively and studied intensively and is better known than for studied intensively; it is better known than for any other group of early mammals. The enamel of any other group of early mammals, and perhaps for the earliest multituberculates, those of the Late any other group of fossil mammals save hominoids. Triassic-Early Jurassic suborder Haramiyoidea and Only recently, with the advent of technological the Late Jurassic-early Early Cretaceous suborder advances in scanning electron microscopy, have Plagiaulacoidea, is "preprismatic." With only concerted efforts been made to investigate two exceptions, all Late Cretaceous and early systematically the enamel ultrastructure of Tertiary genera of multituberculates examined mammals for the purpose of providing a new and have prismatic enamel. Prisms are either small independent data set with which to test phylo­ with circular (complete) boundaries or large with genetic hypotheses based on gross morphological arc-shaped (incomplete) boundaries. There is a characters alone (e.g., Gantt et al. 1977; Boyde remarkably consistent relationship between enamel 1978; Vrba and Grine 1978; Gantt 1980, 1983; von ultrastructural type and subordinal taxa in that Koenigswald 1982; Boyde and Martin 1984a, b; small, circular prisms are usually found within Grine et al., 1986a). Historically, multituber­ the suborder Ptilodontoidea and large, arc-shaped culate phylogeny has been determined almost prisms are usually found in the suborder Taenio­ exclusively on the basis of a few gross dental labidoidea and in six Late Cretaceous-Early characters that have proven inadequate to dis­ Tertiary genera of indeterminate subordinal criminate consistently between higher taxa. status. Cranial and postcranial characters are imprac­ Research currently in progress suggests that tical to employ in phylogenetic analyses of both small, circular prisms and large, arc-shaped multituberculates because adequate material is prisms are homologous in all multituberculates in rare. Thus, multituberculates have served as the which they occur, with one exception. Neolio­ focus of considerable research on enamel ultra­ tomus, a taeniolabidoid, appears to have evolved structure because, even though they were among small, circular prisms independently. In addi­ the most evolutionarily successful and taxonom­ tion, it appears that large, arc-shaped prisms ically long-lived of early mammals, their taxon­ represent the primitive condition in multituber­ omy and systematics have been, and still are, in culates with prismatic enamel, not small, circ­ disarray. ular prisms as has been proposed previously. The objectives of this paper are to review and discuss what has been learned from past studies about the enamel ultrastructure of the Multituberculata, to re-evaluate some of that work in the light of previously unpublished data, and to present some preliminary conclusions concerning the homology and polarity of multi­ tuberculate enamel ultrastructural characters. In addition, we wish to suggest some areas for KEY WORDS: Enamel ultrastructure, multituber­ future research on these subjects. But first it culates, mammals, phylogeny, cladistics, homo­ is necessary to provide, as background infor­ logy, polarity, Mesozoic, Paleogene, variability mation, a brief account of the evolutionary history and paleobiological attributes of multi­ tuberculates. *Address for correspondence: David W. Krause, Department of Anatomical What Are Multituberculates? Sciences, State University of New York, Stony Brook, New York 11794 The Multituberculata is the longest-lived Phone No. (516) 444-3117 order of mammals. Their known geologic record is

1591 D, W. Krause and S. J. Carlson from the Late Triassic to the early Oligocene, an Table 1. Classification of multituberculates interval of over 150 million years. Multituber­ (Hahn and Hahn 1983). culates, unlike their dinosaurian contemporaries, survived the Cretaceous-Tertiary boundary with Order MULTITUBERCULATA little apparent ill effect (Archibald 1983). Suborder HARAMIYOIDEA Representatives of the order have been found only Family HARAMIYIDAE on northern continents: in the Late Jurassic to Haramiya, Thomasia, ?Hypsiprymnopsis early Oligocene of North America, in the Late Suborder PLAGIAULACOIDEA Triassic to early Eocene of Europe, and in the Family ARGINBAATARIDAE Early Cretaceous to early Eocene of Asia (Clemens Arginbaatar and Kielan-Jaworowska 1979; Hahn and Hahn 1983). Family Paulchoffatiidae The Order Multituberculata is both a taxonomical­ Subfamily Kuehneodontinae ly diverse and numerically abundant group that Bolodon, Guimarotodon, Henkelodon, includes 57 named genera and over 140 named Kuehneodon, Plioprion species (Hahn and Hahn 1983). During the Late Subfamily Paulchoffatiinae Cretaceous, they comprised as much as 75% of the Paulchoffatia, Pseudobolodon individuals in mammalian local faunas (Van Valen Subfamily indeterminate and Sloan 1966). They appear to have attained Parendotherium peak species diversity in the middle Paleocene, Family Plagiaulacidae approximately 60 million years before present Ctenacodon, Loxaulax, Plagiaulax, (Van Valen and Sloan 1966; Krause 1980). Psalodon The relationships of multituberculates to Suborder PTILODONTOIDEA other higher taxa of mammals are obscure. They Family BOFFIIDAE have long been grouped with docodonts, tricono­ Boffius donts, and monotremes as nontherian mammals, a Family CIMOLODONTIDAE taxonomic arrangement that has recently been Anconodon, Cimolodon, Liotomus challenged (e.g., Presley 1981; Kemp 1982, 1983; Family NEOPLAGIAULACIDAE Archer et al. 1985). Ectypodus, Mesodma, Mimetodon, The Order Multituberculata is generally Neoplagiaulax, Parectypodus, divided into three suborders: Plagiaulacoidea, Xanclomys Ptilodontoidea, and Taeniolabidoidea. Most Family PTILODONTIDAE workers (e.g., Hahn 1973; McKenna 1975; Sloan Kimbetohia, Prochetodon, Ptilodus 1979) also include an enigmatic and poorly known Suborder TAENIOLABIDOIDEA Late Triassic and Early Jurassic (possibly middle Family EUCOSMODONTIDAE Jurassic) group from Europe, the suborder Hara­ Subfamily BUGINBAATARINAE miyoidea, in the Multituberculata (a possible Buginbaatar haramiyid has been described recently from the Subfamily EUCOSMODONTINAE Late Triassic/Early Jurassic of North America Bulganbaatar, Chulsanbaatar, (Jenkins et al. 1983)). The Plagiaulacoidea are Eucosmodon, Kryptobaatar, a Late Jurassic to Early Cretaceous group con­ Nemegtbaatar, Neoliotomus, Stygimys, taining 13 genera, the Ptilodontoidea a Late Tugrigbaatar, ?Xyronomys Cretaceous to early Oligocene group also con­ Subfamily MICROCOSMODONTINAE taining 13 genera, and the Taeniolabidoidea a Acheronodon, Microcosmodon, Late Cretaceous to early Eocene group containing Pentacosmodon 20 genera. Eight Late Cretaceous and early Subfamily SLOANBATAARIDAE Tertiary genera (Allacodon, Cimexomys, Cimolomys, Sloanbaatar Essonodon, Hainina, Meniscoessus, Paracimexomys, Family TAENIOLABIDIDAE and Viridomys), most of which, until recently, Catopsalis, Kamptobaatar, were included in either the Ptilodontoidea or Lambdopsalis, Prionessus, Taeniolabidoidea, are currently placed in Sub­ Sphenopsalis, Taeniolabis order incertae sedis. Table 1, based upon a Suborder indeterminate compilation by Hahn and Hahn (1983), presents the Family CIMOLOMYIDAE most recent comprehensive classification of all Cimolomys, Meniscoessus, ?Essonodon genera currently allocated to the Multitubercu­ Family indeterminate lata. Allacodon, Cimexomys, Hainina, Multituberculates are so-named because of Paracimexomys, Viridomys the possession of multiple cusps, arranged in longitudinal rows, on the molars. Plagiaulacoids are characterized by a greater number of incisors fragmentary cranial material of ptilodontoids and premolars than later forms, ptilodontoids by from North America suggests that these forms were an enlarged, blade-like lower fourth premolar, nocturnal and that their dominant sense was and taeniolabidoids by a restricted band of olfaction rather than vision (Simpson 1937; enamel on the lower incisor (Fig. 1). Hara­ Krause 1986). Although long considered to be miyoids are known only from isolated molariform folivorous mammals, it is clear that not all teeth that closely resemble those of other multituberculates were able to subsist on a diet multituberculates. of leaves (Krause 1982). Consideration of the Multituberculates were typically small physiological constraints of body size suggests mammals, most of them falling within the body that at least the smaller members of the order size range of modern shrews and mice. Some required more protein than is afforded through

7592 Enamel Ultrastructure of Multituberculates

described as prismatic (discontinuous), nonpris­ Suborder PLAGIAULACOIDEA matic (continuous, prismless, aprismatic), pseudoprismatic, and/or preprismatic. Although i~~~--?.~.:c,,~~i..::::::;, o._;_J~ .. : .;.::• ~ prismatic and nonprismatic enamels can be easily '--"../ __,_____, <--~~ distinguished, the definitions of pseudoprismatic and preprismatic enamels are less clear (Osborn and Hillman 1979; Frank et al. 1984; Grine et al., in prep.). In any case, there appears to be a discrete type of enamel ultrastructure found in fr,>== certain mammals and reptiles in which prism boundaries are not clearly delineated (as they Suborder PTILODONTOIDEA are in prismatic enamels) but in which the orientation of the £·axes of the hydroxyapatite . ~ :-: -:; .i·.. ;6, crystallites are not all parallel (as they are in ·.> ~ nonprismatic enamels). Until the terminological difficulties concerning this type of ultrastruc­ ' ------ture are resolved (Grine et al., in prep.), we refer to this type as "preprismatic" enamel but recognize that the term may have inappropriate developmental and phylogenetic connotations. ~(~ Prismatic enamel in multituberculates was initially discovered by Carter (1922:605), who Suborder T AENIOLABIDOIDEA noted, and clearly illustrated (his Plate VII: Figs. 2-4) , "a series of horseshoe-shaped bodies" I~ (i.e., prisms) in the enamel of Polymastodon, now ~.-.-_./~~·~ ~--r~tJ••..,j,,.,1-... considered a junior synonym of the taeniolabidoid Taeniolabis. Carter's important early findings, although his paper was cited by Moss (1969) and Sahni (1979), have been ignored by these and all subsequent students of multituberculate enamel. Moss (1969) utilized transmitted, phase contrast, and polarized light microscopy to examine dental material of the following multi­ tuberculates: a Late Jurassic plagiaulacoid, a Figure 1. Representative cranial and dental presumed plagiaulacoid from the Early Cretaceous, morphology of the three major suborders of Meniscoessus (Late Cretaceous - Suborder incertae Multituberculata, exclusive of the suborder sedis), Cimolodon (Late Cretaceous - Suborder Haramiyoidea. For each suborder the upper Ptilodontoidea), Taeniolabis (Paleocene - Sub­ dentition is depicted in occlusal (top) and order Taeniolabidoidea), Ectypodus (Paleocene-Eo­ buccal (bottom) views on the left; the lower cene - Suborder Ptilodontoidea), and two generi­ dentition is depicted in occlusal (top) and cally unidentified forms from the Late Cretaceous buccal (bottom) views on the right. Represen­ and Paleocene. Moss (1969:6) confirmed prelimi­ tative plagiaulacoid is based largely on Ctena­ nary observations by Poole (1967) and by Moss and codon (see Simpson 1929 - Figs. 4-7 and Plate III Kermack (1967) that the enamel of multituber­ Fig. 1), but the upper anterior dentition (to th~ culates, indeed that of all of the earliest left of the dashed line) is based on Kuehneodon mammals, "is non-prismatic or continuous in (see Hahn 1969, Fig. 20); ptilodontoid based on structure, and thus it resembles, but is not Ptilodus and redrawn from Simpson (1937 - Figs. 4 identical with, the enamel structure of advanced and 6) and Krause (1982 - Fig. 2); taeniolabidoid mammal-like reptiles." Moss (1969) reiterated based on Taeniolabis and redrawn from Granger and this conclusion stating: "In the non-therian line Simpson (1929 - Figs. 4, 6, and 8). of mammalian evolution I found no evidence of prismatic enamel in any fossil tooth" (p. 6) and folivory. Furthermore, analysis of various that "non-therians did not evolve prismatic aspects of dental morphology and microwear structure at any time" (p. 35). Moss (1969: 16 indicates that at least those forms that had and figure 12), in fact, noted "a series of large, blade-like posterior premolars in the alternating hemispheres" in the enamel of Menis­ lower dentition (primarily members of the sub­ coessus but did not recognize that they we~ order Ptilodontoidea) ingested large, hard food indeed prisms. items, possibly seeds and nuts. What little is Fosse et al. (1973), the first workers to known of their postcranial anatomy suggests that utilize scanning electron microscopy in the multituberculates, at least those known from analysis of multituberculate enamel ultrastruc­ North America, were arboreal (Jenkins and Krause ture, rediscovered prismatic enamel in the order 1983; Krause and Jenkins 1983). during a survey of six unidentified Late Creta­ ceous multituberculate teeth, thereby confirming Multituberculate Enamel Ultrastructure Carter's (1922) earlier observations and contra­ dicting those of Poole (1967), Moss and Kermack Prismatic. Pseudoprismatic, Preprismatic, and/or (1967), and Moss (1969). Prismatic enamel has Nonprismatic? since been found in every Late Cretaceous and Multituberculate enamel has been variously early Tertiary multituberculate genus examined

7593 D. W. Krause and S. J. Carlson

(Fosse et al. 1978, 1985; Osborn and Hillman early Early Cretaceous Plagiaulacoidea appear to 1979; Sahni 1979; Carlson and Krause 1982, 1985), have "preprismatic" enamel. with two exceptions: Viridomys, a Late Cretaceous Prism size, shape, and density form of indeterminate subordinal status, and an A number of studies have defined and docu­ unidentified ?taeniolabidoid from the early Late mented quantitative parameters of mammalian Cretaceous of North America (see Krause and Baird enamel ultrastructure (e.g., Marcus 1931; Eisen­ 1979). The enamel of Viridomys was described as berg 1938; Fosse 1968a, 1968b; Boyde 1969a; Boyde apparently nonprismatic by Carlson and Krause and Martin 1982; Grine et al. 1986b), but such (1985) but with the qualification that only one parameters have seldom been employed in phyloge­ small area on a single fragmentary tooth had been netic analyses. In contrast, quantitative data examined (because of the rarity of available on enamel ultrastructure have been utilized to a material) and that more specimens needed to be considerable extent in multituberculate taxonomy studied. Although not yet described, prisms are and systematics (Fosse et al. 1973, 1978, 1985; also apparently lacking in the fragmentary Sahni 1979; Carlson and Krause 1982, 1985). Two incisor of the early Late Cretaceous unidentified reasons account for this disproportionate use: ?taeniolabidoid (Carlson and Krause,in prep.). 1) qualitative parameters such as prism packing In all cases where prismatic enamel has been patterns and the degree of prism decussation found in multituberculates, there is also an cannot be consistently employed in multituber­ unusually large proportion of interprismatic culate enamel because of the irregular distri­ material, relative to that found in modern bution of prisms and because of the lack of therians (Fosse et al. 1973, 1978, 1985; Osborn decussation, respectively (see below), and 2) and Hillman 1979; Sahni 1979; Carlson and Krause there is a greater range of prism and inferred 1985). ameloblast sizes within multituberculates than is Frank and Sigogneau-Russell (1984) and Frank known for any other higher taxon of mammals. et al. (1984) recently surveyed the enamel Carter (1922: 605) noted that "The enamel ultrastructure of haramiyoid teeth and found it pattern of the Multituberculates is quite dis­ to be "preprismatic." Fosse et al. (1985) tinctive, and differs fundamentally from all discovered a similar type of enamel ultrastruc­ others which I have examined." He also noted ture lacking discrete prism boundaries in two that Ptilodus, a ptilodontoid, possesses "an genera of plagiaulacoid multituberculates from enamel pattern similar to, but by no means the Late Jurassic of Europe. Preliminary exami­ identical with, that of Polymastodon." Fosse et nation of two specimens of Late Jurassic multi­ al. (1973) quantified several parameters of tuberculates from North America (one of Psalodon enamel ultrastructure in a small sample of and the other unidentified) also reveals that generically unidentified multituberculates from a distinct prism boundaries are absent (Carlson and Late Cretaceous locality in North America. They Krause,in prep.). demonstrated that prism diameter and the distance It is pertinent to point out that there is a between the centers of adjacent prisms, which is large temporal gap, of over 50 million years, equivalent to the diameter of an ameloblast, was between the last undoubted plagiaulacoid (early larger than known for any extant mammal (Boyde Early Cretaceous) and the earliest undoubted and Martin (1984b) have since identified very taeniolabidoid (late Late Cretaceous). Fortu­ large prisms in the extinct hominoid primate nately, multituberculate specimens of intermedi­ Proconsul major). All of the specimens examined ate age (late Early Cretaceous and early Late by Fosse et al. (1973) have center-to-center Cretaceous) are known but these are still largely distances between prisms that are almost three unstudied. As mentioned above, the early Late times larger (x = 16.15 j-llll) than the same mea­ Cretaceous unidentified ?taeniolabidoid does not surement in marsupial enamel (x = 5.61 )-lffi). appear to have prismatic enamel. Specimens of Correspon1ingly, prism density (x = 4,500 late Early Cretaceous age are known from the prisms/mm) is approximately one-n~nth that seen Khovboor fauna of Mongolia and the Trinity fauna in marsupial enamel (x - 37,825/mm ). of Texas (see Clemens et al. 1979). Fosse et al. In a subsequent study, Fosse et al. (1978) (1985) determined that the Khovboor multitubercu­ examined four additional (but this time specifi­ late Arginbaatar does indeed have prismatic cally identified) multituberculate teeth from the enamel, as does a generically unidentified form same Late Cretaceous locality. Represented were from the same locality. Similarly, a preliminary specimens of the taeniolabidoids Catopsalis and survey of the Trinity multituberculates shows Stygimys and the ptilodontoid Mesodma. Fosse et that they have prismatic enamel (Krause et al., al. (1978) determined that the distance between in prep.). The presence of prismatic enamel in centers of prisms in Catopsalis and Stygimys was late Early Cretaceous multituberculates predates, comparable to that observed in the unidentified by approximately 25-30 million years, the previ­ teeth in their earlier study but that, in Mesod­ ously reported earliest occurrence of prismatic ma, the center-to-center distance was much less enamel in multituberculates. (that is, similar to that seen in extant mam­ In sum, therefore, it appears that, with mals). Thus, Fosse et al. concluded that "mutual only two possible exceptions, all known Late central distances are much greater in the repre­ Cretaceous and early Tertiary multituberculates sentatives of Taeniolabidoidea than in the possess prismatic enamel (contra Moss 1969). In representatives of Ptilodontoidea." addition, the enamel of late Early Cretaceous Sahni (1979) examined the enamel ultrastruc­ multituberculates from both North America and ture of six multituberculate genera from the same Asia is prismatic. The Late Triassic and Early Late Cretaceous locality using scanning electron Jurassic Haramiyoidea and the Late Jurassic and microscopy. In addition to the three genera

7594 Enamel Ultrastructure of Multituberculates examined by Fosse et al. (1978), Sahni (1979) studied specimens of Cimexomys, Cimolodon, and Meniscoessus. Sahni noted, as had Fosse et al. (1973), that the prisms of multituberculates were large, approximately twice as large as those of fossil and Recent placentals. The values ob­ tained by Sahni (1979) are often at variance with those obtained by Fosse et al. (1978, 1985) and by Carlson and Krause (1985); Fosse et al. (1985) have demonstrated that measurements of Sahni's published figures do not yield the results that he obtained. Carlson and Krause (1985) obtained quanti­ tative data on the enamel ultrastructure of all 13 recognized ptilodontoid genera, 12 of the 20 taeniolabidoid genera, and six of the eight genera placed in Suborder incertae sedis (the taxonomic validity of one of the genera, Allaco­ don, is questionable and, as mentioned previous­ ly, Viridomys does not appear to have prismatic enamel). Fosse et al. (1985), in addition to examining some of the same genera studied by Carlson and Krause (1985), added information on three more taeniolabidoid genera (Chulsanbaatar, Neme~tbaatar, and Kamptobaatar). As a conse­ quence, data on the enamel ultrastructure of 34 of the 41 recognized genera of Late Cretaceous and early Tertiary multituberculates are avail­ able for analysis. Here we present a selection of those quanti­ tative parameters that are regarded as the most Figure 2. Model of congruent, contiguous circles valuable in interpreting differences in the in symmetrical hexagonal distribution represen­ ultrastructure of prismatic enamel and for ting an idealized prism packing pattern (Pattern inferring details of development of the enamel 1 or 3). The centers of each circle represent (these parameters are explained in greater detail the centers of prisms. Sides d, y, and x of by Fosse (1968a, b) and Grine et al. (1986b)). The triangle ADE represent distances between prism parameters used for comparative analysis in the centers and h represents the height of triangle present study, with citation of those workers who ADE. B, C, E, and F designate the angles of a employed them in the past (if any), include the parallelogram formed by two triangles with one following: side in common. Each such parallelogram is a) Prism diameter (PDi) - Sahni (1979), equivalent in area to the secretory territory of Carlson and Krause (1982, 1985). This value is one ameloblast and the shaded area is equivalent directly measured on scanning electron micro­ to the area of a single prism. The number of graphs of sections tangential to the enamel such parallelograms equals the number of circles surface as the maximum diameter between prism (in this case, prisms) per unit area. See text boundaries perpendicular to the apicocervical for additional explanation. (Adapted from Fosse axis of the tooth crown. 2 1968a, Fig. 1). b) Numerical prism density per mm (PDe) Fosse et al. (1973, 1978, 1985), Sahni (1979), Carlson and Krause (1985). The basic unit In this study, the distances between the describing prism density is a triangle with sides centers of any three adjacent prisms were mea­ plotted between the centers of three closely sured on transparent acetate sheets on which the adjacent prisms (Fig. 2). Two such triangles centers of prisms had been plotted directly from with one side in common constitute a parallelo­ scanning electron micrographs of tangential gram. As a result, there are as many parallelo­ sections of enamel. They were measured by means grams as there are prisms in a given area. of a Fowler digital caliper connected to a Fowler Therefore, measuring the areas o~ a number of EDP microprinter. As pointed out by Fosse et al. paralle ograms and dividing 1 mm (transformed 2 2 (1973), it would be preferable to take sections into µm's) by the mean parallelogram 2rea in µm along the incremental lines that demarcate the will yield the number of prisms per mm. Since developing surface of the enamel. However, since prisms, particularly in multituberculates, are incremental lines are difficult to consistently never arranged in a precisely regular geometrical detect in multituberculate enamel and since pattern, tetragons, not parallelograms, are destructive longitudinal sections were generally employed to describe the relationship between the not taken, all measurements were taken on sec­ centers of four adjacent prisms. Therefore, in tions that were tangential to the enamel surface. order to "idealize" prism spacing, the mean area The enamel of multituberculates is thin and Fosse of a parallelogram is determined by measuring the et al. (1973) have calculated that the errors of sides of several triangles, calculating the mean measurement in tangential sections of such thin area of the triangles, and doubling it. enamel are probably less than 1%.

7595 D, W. Krause and S. J. Carlson

•-Circular ■- Suborder Ptilodontoidea

0- Arcade-shaped □- Suborder Taeniolabidoideo

~-Circular and arcade-shaped @-suborder indeterminate 0 5 6 ~4

<.'.)"' 3 02 z,0

18 2.0 22 24 Ln Prism Diameter (µm)

1.8 20 2.2 2.4 2.6 2.8 3.0 3.2 34 ■- Suborder Pt1lodonto1dea Ln Computed Central Distance ( JJ m)

0- Suborder Taeniolabidoideo Figure 4. Frequency distribution of computed central distance between contiguous prisms among 34 Late Cretaceous and Early Tertiary multituber­ culate genera. The computed central distance between contiguous prisms is generally small in genera of the Suborder Ptilodontoidea and gener­ ally large in genera of the Suborder Taeniolabi­ doidea. The six indeterminate genera group with those of the Suborder Taeniolabidoidea. Notable 0.6 0.8 1.0 I 2 14 1.6 18 20 2 2 Ln Prism Diameter (µm) exceptions to this pattern include Boffius (~). Cimolodon (Q), Microcosmodon (tl), Neoliotomus (N), and Xyronomys (X) (see text for additional Figure 3. a) Frequency distribution of prism explanation). Data from Table 2. size (diameter) and shape among 31 Late Cretace­ ous and Early Tertiary multituberculate genera. b) The same distribution of prism size indicating large. Therefore, the central distance of previous subordinal assignment of individual equilateral triangles is computed by first genera (Hahn and Hahn 1983). Note that the determining the height of the triangle with sides bimodal distribution consists of ptilodontoid d, y, and x. This is calculated from Fosse's genera with small, circular prisms on the left equation III: and taeniolabidoid and Suborder indeterminate genera with large, arc-shaped prisms on the right. Notable exceptions to this pattern h = ------(2) include Boffius (~). Cimolodon (Q), Microcosmodon 2d (tl), Neoliotomus (N), and Xyronomys (X) (see text for additional explanation). Data from Table 2. The new value, h, which represents the height of an idealized equilateral triangle,can be used to The three distances measured on adjacent calculate the central distance (CD) between the prisms were d, y, and x, each of which represents corners of an idealized equilateral triangle with the side of a triangle (Fig. 2). In order to be based as follows (Fosse's equation VI): consistent, d was always taken as near to hori­ zontal as possible and y and x were the consecu­ tive counterclockwise directions. The mean (3) values of d, y, and x were then used to compute 2 numerical prism density (PDe) per mm using d) Cross-sectional Prism Area (PA) - This Fosse's (1968a) equation II: value was derived by directly tracing, on acetate 6 sheets, the outlines of prisms with a Graf/Pen 2 X 10 Sonic Digitizer (Science Accessories Corpora­ PDe - ;,-=-;;-:-,;~~'""'"""~-~=-,;...,~ ( 1) tion), which was programmed to automatically v~~2~2-~-(~2-~-~2-~-~2;2---- calculate the areas of irregular geometrical figures. This procedure is preferable to that c) Computed central distance between prisms used by Sahni (1979) and Carlson and Krause iQD_ - Fosse et al. (1973, 1978, 1985). Even in (1985) in which prism diameter was simply squared the most geometrically regular enamels, a precise to obtain prism area. hexagonal closest packing arrangement (as illus­ e) Cross-sectional Ameloblast Area (AA) - trated in Fig. 2) is never attained and thus the Fosse et al. (1985). The area of the secretory central distances between three adjacent prisms territory of an ameloblast (represented by the will never consistently form equilateral trian­ area of a parallelogram) theoretically represents gles. The three central distances (d, y, and x) an area equivalent to one prism (shaded area in cannot be averaged to obtain an arithmetic mean Fig. 2) with its surrounding interprismatic central distance since this computation will material. Arneloblast area was calculated simply result in an equilateral triangle that is too by multiplying the average base (d) by the

7596 Enamel Ultrastructure of Multituberculates

■- Suborder Ptilodonto,deo ■ -Suborder Ptilodontoidea

□ -Suborder Toeniolobidoideo □ -Suborder Toeniolabidoidea []-Suborder 1ndeterrn1note Wj-Suborder indeterminate c' 4 "'C 5 ~3 N 02 24 (I) 0 C z, ~3

26 3.0 3:4 3.8 4.2 4.6 50 oz 0 Ln Prism Area (;,m 2 ) z, Figure 5. Frequency distribution of prism area among 31 Late Cretaceous and Early Tertiary 22 28 34 40 46 52 multituberculate genera. Prism area is generally Prism Area I Ameloblast Area x 100 small in genera of the Suborder Ptilodontoidea and generally large in genera of the Suborder Figure 6. Frequency distribution of prism Taeniolabidoidea. The six indeterminate genera area/ameloblast area among 31 Late Cretaceous and group with those of the Suborder Taeniolabidoid­ Early Tertiary multituberculate genera. Multitu­ ea. Notable exceptions to this pattern include berculates, in general, display relatively small Boffius (~), Cimolodon (~), Microcosmodon (tl), prism area relative to ameloblast area, that is, Neoliotomus (~), and Xyronomys (X) (see text for they have relatively large amounts of interpris­ additional explanation). Data from Table 2. matic material. Clear-cut subordinal distinc­ tions on the basis of this ratio, however, are absent. Boffius (~). Cimolodon (~). Microcosmo­ don (tl), Neoliotomus (~). and Xyronomys (X), computed height (h). Data from Table 2. f) Ratio of Prism Area to Ameloblast Area (PA/AA) - Carlson and Krause (1985). This value was calculated in order to quantify the relative amount of interprismatic material (1 minus the the large, arc-shaped distribution, as do the above ratio), which is generally high in multi­ genera of indeterminate subordinal status (Cimex­ tuberculates. omys. Cimolomys, Essonodon, Hainina, Meniscoessus A seventh value, K, which indicates the - Fig. 7c, Paracimexomys). amount of vertical compression or distension There are, however, several notable excep­ between adjacent prisms, has not been calculated. tions to the correlation of ultrastructural type Although this value appears to have some utility and subordinal assignment on the basis of gross in differentiating between enamels of therian dental characters: 1) Cimolodon, a ptilodontoid, taxa (e.g., Grine et al., 1986a) and was initially possesses large, arc-shaped prisms rather than used by Fosse et al. (1973, 1978) for multituber­ small, circular prisms (Fig. 7d); 2) Neoliotomus, culates, multituberculate enamel is, in general, a taeniolabidoid, possesses small, circular less regularly organized than that of therian prisms rather than large, arc-shaped prisms (Fig. mammals. K is thus ineffective as a diagnostic 7e); and 3) Microcosmodon, a taeniolabidoid, has parameter. a combination of both small, circular and small, The data generated from the calculation of arc-shaped prisms (Fig. 7f). The significance of prism diameter, computed central distance, amelo­ the ultrastructural pattern seen in Microcosmodon blast area, prism area, prism area/ameloblast cannot yet be explained and requires further area, and prism density are presented in Table 2. analysis (Carlson and Krause, 1985) but the Generic averages for prism diameter, computed presence of more than one pattern in the same central distance, prism area, and prism taxon is not unusual among therian mammals (e.g., area/ameloblast area are graphically depicted by Boyde and Martin 1982, 1984a, b; Fortelius 1985; histograms in Figs. 3-6 (histograms for amelo­ Ishiyama 1984). In addition, Boffius, a purpor­ blast area and prism density are unnecessary ted ptilodontoid, has large, arc-shaped prisms; since they are almost identical to that for its assignment to the Ptilodontoidea, however, computed central distance). As determined by was not based on diagnostic, associated material Carlson and Krause (1985), there appears to be a and is questionable (Vianey-Liaud 1979). Simi­ remarkably consistent relationship between prism larly, Xyronomys, a purported taeniolabidoid, has size and shape and subordinal distinctions among small, circular prisms but its assignment to the Late Cretaceous and early Tertiary multitubercu­ Taeniolabidoidea was also not based on diagnos­ lates that have prismatic enamel. As depicted in tic, associated material and is questionable Fig. 3, large, arc-shaped prisms are generally (Rigby 1980). The latter two taxa therefore found in taeniolabidoids (e.g., Taeniolabis - cannot be regarded as legitimate exceptions to Fig. 7a) and small, circular prisms are generally the distribution of prism types among subordinal found in ptilodontoids (e.g., Prochetodon - Fig. taxa. 7b). The three additional taeniolabidoid genera The distributions for computed central studied by Fosse et al. (1985) also fall within distance and prism area exhibit the same general

7597 D. W. Krause and S. J. Carlson

Table 2. Measurements and calculations of various parameters of prismatic enamel in multituberculate mammals. PDi - pris2 diameter in µrn; CD - c~mputed central distance in >-112; AA - ameloblast cross­ sectional area in µrn PA - prism area in µm; PDe prism density per mm; I upper incisor; i - lower incisor; P - upper premolar; p - lower premolar; M - upper molar; m - lower molar. Data for PDi are taken from Carlson and Krause (1985). All of the other data derive from measurements or calcula­ tions taken during the course of this study, except for those genera marked with*, which refers to data from Fosse et al. (1985).

Taxon Tooth PDi CD AA PA PA/AA PDe

Suborder PLAGIAULACOIDEA Family PAULCHOFFATIIDAE Subfamily KUEHNEODONTINAE *Kuehneodon sp. PS 4.75 19.6 50,983 Subfamily PAULCHOFFATIINAE *Paulchoffatia sp. Pl 4.78 19.9 51,534 Family ARGINBATAARIDAE *Arginbaatar dimitrievae p4 13 .66 162.3 6,274 Suborder PTILODONTOIDEA Family BOFFIIDAE Boffius splendidus M 11.8 24.88 536.0 118 .6 0.22 1,866 Family CIMOLODONTIDAE 6_11.Q.9nodonrusselli p4 3.4 7.15 44.3 14.1 0.32 22,557 Cimolodon nitidus p4 8.0 16.48 235.1 71.1 0.30 4,253 ml 7.0 12. 77 141.2 35.2 0.25 7,083 Liotomus marshi p4 3.5 8.30 59.6 15.2 0.26 16,773 Family NEOPLAGIAULACIDAE Ectypodus powelli il 3.8 9.75 82.2 24.6 0. 30 12,159 p4 2.9 7. 72 51. 6 19,387 Mesodma sp. p4 3.6 8.55 63.3 22.3 0.35 15, 791 *Mesodma sp. p4 6.49 36.4 27,474 Ml 6.95 42.0 24,126 Mimetodon silberlingi p4 3.1 7.43 47.8 13. 7 0.29 20,908 ml 3.1 6.79 39.9 18.8 0.47 25,070 Neoplagiaulax hunteri p4 3.2 6.99 42.3 10.6 0.25 23,632 Parectypodus lunatus il 2.7 7.86 53.5 10.2 0.19 18,703 p4 2.7 7.34 46.7 21,414 Xanclomys mcgrewi p4 1. 9 6 .11 32.3 9.4 0.29 30,927 Family PTILODONTIDAE Kimbetohia campi p4 3.9 7.48 48.4 16.4 0.34 20,644 Prochetodon cavus p4 2.6 6.66 38.4 26,071 Ml 3.7 7.52 49.0 17.2 0.35 20,395 Ptilodus new species B il 4.8 10.57 96.8 34.8 0.36 10,331 Ptilodus wyomingensis p4 3.5 8.22 58.6 18.4 0.31 17,073 ml 3.2 7.81 52.9 18,910 Suborder TAENIOLABIDOIDEA Family EUCOSMODONTIDAE Subfamily EUCOSMODONTINAE *Chulsanbaatar vulgaris il 13. 77 165.0 6,154 p4 12.60 139. 7 7,627 ml 10.35 93.0 10,812 Eucosmodon primus il 9.9 17.94 278.7 78.1 0.28 3,589 p4 6.9 13. 69 162.3 43 .4 0.27 6,163 Kryptobaatar dashzevegi p4 6.0 13 .45 156.6 45.3 0.29 6,386 ml 7.3 13. 81 165.2 53.5 0.32 6,052 m2 8.3 14.62 185.2 68.4 0.37 5,401 *Kryptobaatar dashzevegi il 17.40 263.0 3,851 p4 14.24 176.2 5,731 ml 16 .13 227.7 4,591 *Nemegtbaatar gobiensis il 14.60 184. 5 5,417 p4 16.49 235.7 4,241 ml 10.78 101.7 10,202 Neoliotomus ultimus il 5.3 10.03 87.2 26.2 0.30 11,468 p4 4.0 10.45 94.6 28.4 0.30 10,567 Stygimys kuszmauli ml 9.4 17.19 255.9 79.7 0.38 3,908 12 7.9 15.56 209.6 118.4 0.46 4,770 *Stygimys kuszmauli il 17.29 259.0 3,860 Subfamily EUCOSMODONTINAE? Xyronomys sp. p4 4.2 9.26 74.3 28.5 0.38 13,454

1598 Enamel Ultrastructure of Multituberculates

Table 2 (continued) ------Taxon Tooth PDi CD AA PA PA/AA PDe ------Subfamily MICROCOSMODONTINAE 28.9 0.32 10,959 p4 5.4 10.26 91. 3 Microcosmodon £.Q.ill!.§_ 32.0 0.54 14,337 ml 4.7 8.97 69.8 Microcosmodon rosei 31. 3 0.45 16,789 12 4.3 8.29 59.6 40.2 0.36 8,943 p4 8.3 11. 36 111. 8 Pentacosmodon pronus 58.2 0.50 8,512 ml 6.8 11.65 117. 5 Family TAENIOIABIDIDAE 120.3 0.50 4,132 p4 8.1 16.71 242.0 Catopsalis joyneri 0.34 4,528 7.5 15.97 220.9 74.2 ml 3,208 9.8 18.97 311. 7 112.1 0.36 12 4,470 *Catopsalis catopsaloides il 15.60 210. 9 13 .48 157.6 6,357 ml 6,776 p4 13 .OS 147.5 *Kamptobaatar kuczynskii 0. 21 3,828 M2 7.7 17.37 261. 2 54. 3 Lambdopsalis bulla 61. 7 0.30 4,794 Prionessus lucifer m2 8.4 15.52 208.6 13.38 155.2 6,440 *Prionessus lucifer il 5,775 ml 14.13 173 .1 6,138 m2 13.71 162.9 77 .0 0.44 5,744 Sphenopsalis nobilis m2 7.1 14.18 174.1 122.3 0.41 3,330 il 10.1 18.62 300.3 Taeniolabis taoensis 0.49 4,080 p4 8.3 16.82 245.1 119. 5 3,146 M2 10.2 19.16 317.9 98 .1 0.31 Suborder indeterminate Family CIMOLOMYIDAE 296 p4 8.5 14. 77 188.8 82.0 0.43 s. Cimolomys clarki 5,538 ml 8.5 14.44 180.6 85.9 0.48 0.46 3,832 p4 9.8 17.36 261. 0 121. l Meniscoessus robustus 3,343 m2 9.5 18.58 299.1 82.4 0.28 4,516 12 8.6 15.99 221. 5 77 .0 0.35 4,088 *Meniscoessus sp. m2 16.80 244.5 Family CIMOLOMYIDAE? 70.3 0.51 5,950 Essonodon browni ml 8.1 13. 94 168.l Family indeterminate 0.40 8,835 p4 6.9 11.43 113. 2 45.3 Cimexomys minor 10,131 Ml 6.3 10.68 98.7 45.3 0.46 62.9 0.52 8,310 Hainina belgica Ml 6.9 11. 79 120.3 43.8 0.38 8,603 Paracimexomys magister Ml 6.1 11.59 116. 2 15.36 204.4 4,891 *Khovboor spec. GI PST 10/29 12 11,365 *Khovboor spec. GI PST 10/23 p4 10.07 87.9

pattern (Figs. 4 and 5). Ptilodontoids, in the proportionate amount of interprismatic general, have prism center-to-center distances material in deep, intermediate, and superficial and prism areas that are small relative to those levels in the enamel of Ovis aries (sheep) and of taeniolabidoids and genera of indeterminate Capra hircus (goat). On average, however, the subordinal status. The exceptions, in terms of proportionate amount of interprismatic matrix is shape of prisms, are the same as those listed only approximately 30% in Ovis aries and 35% in above for prism diameter. Capra hircus. The pattern for prism area/ameloblast area Finally, it should be added that Fosse et al. (1985) provided quantitative information for does not serve to distinguish suborders as well the "preprismatic" enamel of two genera of as the parameters already discussed (Fig. 6). plagiaulacoids and for the prismatic enamel of The six genera of indeterminate subordinal status late Early Cretaceous multituberculates from have less interprismatic material (x = 0.56, sd - Asia. The center-to-center distances between 0.07) than do members of the Ptilodontoidea "preprisms" of plagiaulacoids are equivalent to (excluding Boffius) (n = 12, x - 0.70, sd = 0.05) those found between the prisms of most ptilodon­ but taeniolabidoid genera (excluding Xyronomys) toids. The late Early Cretaceous Asian multi­ (n - 11, x - 0.64, sd - 0.08) overlap both ptilo­ tuberculates, however, have large, arc-shaped dontoids and the indeterminate genera. These prisms, as in most taeniolabidoids. Similarly, data serve to document that multituberculates do the Trinity multituberculates from the late Early indeed have a large amount of interprismatic Cretaceous of North America possess large, material (n - 31, x - 0.65, sd - 0.08); the arc-shaped prisms (Krause et al., in prep.). enamel of almost all genera is comprised of more Prism Packing Patterns than 50% interprismatic material, and in some Boyde (1964, 1965, 1976) described three cases as much as 80%. Comparable data are primary types of mammalian enamel ultrastructure largely unavailable for therian mammals. Grine designated Patterns 1, 2, and 3 (Fig. 8). These• et al. (1986b) found considerable differences in

1599 D. W. Krause and S. J. Carlson

Figure 7. Scanning electron rnicrographs of enamel ultrastructure in selected Late Cretaceous and Early Tertiary rnultituberculate genera. a) Taeniolabis - Suborder Taeniolabidoidea; b) Prochetodon - Suborder Ptilodontoidea; c) Meniscoessus - Suborder indeterminate; d) Cirnolodon - Suborder Ptilodontoidea; e) Neoliotornus - Suborder Taeniolabidoidea; and f) Microcosrnodon (Suborder Taeniolabidoidea). Bar - 10 µrn.

patterns were hypothesized to reflect the three­ more abundant than Pattern 3, which was rare or dimensional arrangement and shape of the Tornes' perhaps even absent. Sahni's conclusions are processes of the arneloblasts that secrete the puzzling in that five of the six taxa examined by enamel. When viewed in sections tangential to him have since been found to exhibit arc-shaped the enamel surface, Pattern 1 prisms are circular prisms (Fosse et al. 1978; Carlson and Krause (complete prism boundaries), separated by a 1985) and thus could not possibly conform to a discrete and continuous interprisrnatic region, Pattern 1 arrangement. Mesodrna is the only genus and are diagrammatically arranged in off-set of the six illustrated by Sahni that could be horizontal rows. Pattern 2 and Pattern 3 prisms described as having Pattern 1 enamel and even are arc-shaped (incomplete prism boundaries) but this genus has regions in which arc-shaped prisms Pattern 2 prisms are aligned in apicocervical predominate (Carlson and Krause 1985; Fosse et columns whereas Pattern 3 prisms are arranged in al. 1985). Later, Sahni (1985) suggested that off-set horizontal rows. The open sides of Pattern 1 prisms are characteristic of ptilodon­ Pattern 2 and 3 prisms are directed cervically; tid rnultituberculates (presumably ptilodontoids those of Pattern 2 therefore open onto or towards since Sahni did not examine specimens of ptilo­ the apical boundary of the cervically adjacent dontids) and that arc-shaped prisms, separated by prism whereas those of Pattern 3 are directed large interprisrnatic regions, were present in between the lateral boundaries of prisms in the Stygirnys, a taeniolabidoid. Sahni (1985:140) cervically adjacent row. The number of arnelo­ asserted that the latter arrangement was not blasts contributing to the formation of a single "strictly homologous to Pattern Three prisms as prism is one for Pattern 1, two for Pattern 2, defined by Boyde (1964), as the latter have a and four for Pattern 3. Boyde (1969a) also poorly developed interprisrnatic phase" and determined that arneloblasts associated with therefore considered the pattern represented in Pattern 2 are consistently the smallest, those of Stygirnys "to be a modified version of Pattern Pattern 3 the largest, and those of Pattern 1 Three" (no mention was made of Pattern 2 prisms). intermediate in size. The amount of interprisrnatic material, however, Sahni (1979) reported that, in the rnultitu­ does not affect the spatial arrangement of berculate taxa he examined, the Pattern 1 ar­ Pattern 3 prisms and hence there appears to be rangement predominated, and that Pattern 2 was little justification for referring to the pattern

1600 Enamel Ultrastructure of Multituberculates

the enamel (e.g., Kawai 1955; Boyde 1964, 1969b). nn The prisms in a single band are parallel to one another and bend from side to side (horizontal n~nn n@n 0@0 nnnn rth decussation) or up and down (vertical decussa­ ooom0000 nnnn tion) relative to the root-crown axis as they 000 n~n~ nfin extend outwards towards the enamel surface. The 0000 nnnn nn 000 °"I orientation of prisms within each band is at an nnn angle to prisms in adjacent bands. Bands may be Pattern 1 Pattern 2 Pattern 3 a single layer thick (uniserial enamel) or several layers thick (multiserial and pauciserial enamels). The transition in orientation of prisms between bands is gradual in multiserial Figure 8. Schematic representation of Boyde's enamel and abrupt in uniserial and pauciserial (1965, 1969a) three patterns of enamel ultra­ types. structure. Forty-two prisms of each type are It appears that true prism decussation is shown illustrating their relative sizes and not present in the Multituberculata, despite the spatial arrangement in tangential sections identification by Moss (1969:16) of "a possible through the enamel. Hexagonal outlines represent decussating arrangement of the enamel bands" in the secretory territories of individual amelo­ the enamel of an unidentified Late Cretaceous blasts. Circular or arc-shaped lines depict multituberculate or the observation by Osborn and prism boundaries. Stippled regions illustrate Hillman (1979:58) that prisms "crossed each other the area of an individual prism, in each of the rarely" in Cretaceous mammals (including the three patterns (from Carlson and Krause, 1985, multituberculates Stygimys and Catopsalis). Fig. 3). Sahni (1979:42) identified bending of prisms in Mesodma incisors but stated that "there is no in Stygimys as "a modified version of Pattern evidence of prism decussation" and, later (1984: Three." 459), that "there is no clearcut zonation caused No worker other than Sahni (1979) has by the crossing over of prisms." Carlson and systematically utilized Boyde's system of prism Krause (1985) also observed distinct changes in packing patterns to describe the spatial arrange­ orientation of prisms in lower incisors of ment of prisms in multituberculate enamel. The Taeniolabis and Ptilodus but in neither case was ultrastructure of multituberculate enamel is there evidence of bands of prisms crossing over conspicuously less regular in its geometrical one another. arrangement than is that of most other mammals. Enamel Tubules Consistent patterns of spatial arrangement are Enamel tubules are radially-directed cylin­ observable over very small areas only (Fosse et ders present in the enamel of some, but not all, al. 1973; Carlson and Krause 1985). As detailed mammalian taxa, and in some mammal-like reptiles above, prismatic enamel in multituberculates can (Poole 1956; Moss 1969; Cooper and Poole 1973; be divided into two types: 1) small and circular Osborn and Hillman 1979). Their developmental prisms, and 2) large and arc-shaped prisms. If origin is controversial (e.g., Boyde and Lester the arrangement of prisms relative to one another 1967, 1984; Lester, 1970; Risnes and Fosse 1974; are ignored and only aspects of size and shape Osborn 1974). The presence of tubules in multi­ are considered, then it would appear that Pat­ tuberculate enamel was first discovered by Carter terns 1 (small and circular prisms) and 3 (large (1922), who observed them in the genera Taenio­ and arc-shaped prisms) are represented (Carlson labis and Ptilodus and referred to them as and Krause 1985). Even considering spatial "tubes." Carter observed that the tubules pass distributions, it appears that a Pattern 3 across the enamel-dentine junction, from the arrangement, in which the large, arc-shaped dentine into the enamel. Moss (1969), Fosse et prisms are distributed in offset horizontal rows, al. (1973), Osborn and Hillman (1979), and others is more frequently observed than a Pattern 2 have confirmed the presence of tubules in multi­ arrangement (also see Fosse et al. 1973). tuberculate enamel, that the tubules are direct Furthermore, even if it were accepted that the extensions of dentinal tubules, and that they small, circular prisms should be designated cross the enamel-dentine junction perpendicular­ Pattern 1 and the large, arc-shaped prisms should ly, as in marsupials (Boyde and Lester 1967). be designated Pattern 3, the increased sampling The tubules are seen to have a "zig-zag" or of multituberculate genera (Carlson and Krause spiral centrifugal course through the enamel and 1985; Fosse et al. 1985) reveals that "Pattern l" are found in both prism and interprism areas, is not the predominate pattern (contra Sahni). also as seen in marsupials (Boyde and Lester Of a total of 31 genera with prismatic enamel 1984) (Fig. 7). Moss (1969) and Sahni (1979) sampled, 13 exhibit "Pattern l" prisms and 17 found the tubules confined to the inner layers of exhibit "Pattern 3" prisms (Carlson and Krause enamel, at least in some genera (e.g., Meniscoes­ 1985). One genus (Microcosmodon) has small, fil!2.). Sahni (1979) further noted that tubules circular and small, arc-shaped prisms. are largely restricted to interprismatic regions Prism Decussation in the inner enamel, to prismatic regions at more Prism decussation refers to the phenomenon superficial levels, and that, at intermediate in which bands of prisms, expressed optically as enamel depths, tubules can be found in both prism alternating light and dark bands called Hunter­ and interprism areas. Schreger bands, lie at distinct angles to one Enamel tubules have been recognized in another as they pass towards the outer surface of plagiaulacoids (Moss 1969; Fosse et al. 1985),

1601 D. W. Krause and S. J. Carlson

although in the two genera examined by Fosse et accordingly. al. (1985: 444) they were described as "very Tubules were not observed by Frank et al. scarce." Tubules are also present in the enamel (1984) in the "preprismatic" enamel of harami­ of Late Cretaceous and Early Tertiary multituber­ yoids. Preprismatic enamel was defined, in part, culates (Fosse et al. 1978, 1985; Osborn and as being devoid of enamel tubules (Frank et al. Hillman 1979; Sahni 1979; Carlson and Krause 1984; Sigogneau-Russell et al. 1984). Poole 1985), as well as in the Early Cretaceous multi­ (1957:364) stated that "tubular enamel possesses tuberculates from Texas (Moss 1969). Based on both prisms and tubules" and that "tubular enamel measurements of tubule diameter, tubule density, ... must be considered to be a later special­ and tubule number/prism number, Sahni (1979:47) ization of the simpler prismatic type." Moss estimated that "tubules are more common and of (1969), however, correctly pointed out that larger diameter in taeniolabidoid multitubercu­ tubules are found in both nonprismatic and lates than in ptilodontoid and are usually prismatic enamel. Their presence in the enamel restricted to the inner half of the enamel." of plagiaulacoids demonstrates that they exist in While Sahni's data appear to support his conclu­ "preprismatic" enamel as well (Moss 1969; Fosse sions concerning tubule diameter (although only et al. 1985). one ptilodontoid genus is represented in his Variability in Enamel Ultrastructural Parameters sample), his data on tubule density are equivo­ Although qualitative and quantitative cal. Mesodma, a ptilodontoid, is l~sted as parameters of mammalian enamel have been employed having a tubule density of 6,000/mm; Cimexomys, increasingly in phylogenetic analyses, the then considered a ptilodontoid but now of inde­ variability of these features has generally not terminat2 subordinal status, with a density of received adequate consideration. A study of 5,000/mm; Meniscoessus, then considered a absolute and relative variability between two taeniolabidoid but now of indeterminate su~­ closely-related extant therian species, Ovis ordinal status, with a density of 8,000/mm; and aries (sheep) and Capra hircus (goat) by Grine et Catopsalis and Stygimys, both taeniol2bidoids, al. (1986b), has revealed that measurements of with densities of 28,500 and 2,500/mm, respec­ enamel ultrastructure at undefined enamel depths tively. These data are insufficient to establish in single specimens that are regarded as specifi­ clear-cut tendencies. cally or generically representative should be Carlson and Krause (1985:10) in a survey of treated with circumspection. 31 Late Cretaceous and early Tertiary genera with Carlson and Krause (1985) examined sources prismatic enamel concluded that the presence or of variability in the enamel ultrastructure of location of enamel tubules "does not seem to vary multituberculates at a number of hierarchical in any consistent manner." They provided quali­ levels: 1) different positions on a single tooth; tative assessments of the relative abundance of 2) different depths and orientations of a pre­ tubules observed in scanning electron micrographs pared enamel surface; 3) different teeth from a of tangential sections of the enamel. A summary single individual; 4) isolated teeth assigned to of these assessments is recorded in Table 3. a single species; 5) between congeneric species; There appears to be little or no concordance 6) between genera; and 7) within supraspecific between higher taxa and relative abundance of taxa. The technical difficulties of examining tubules. The vast majority of the genera sampled enamel as thin as in multituberculate teeth at have no to few enamel tubules visible in SEM precisely specified depths, and the impossibility tangential sections. The relative abundance of (as with almost all fossil material) of obtaining enamel tubules, however, is not adequately large samples of conspecific teeth, precluded an assessed by means of tangential sections alone, analysis of the type performed by Grine et al. especially if tubules are largely confined to the (1986b). Nonetheless, the study by Carlson and inner layers of enamel. Until a comparative Krause (1985) demonstrated that, at intra-tooth, survey at precisely specified depths in the intra-individual, intra-specific, and intra­ enamel can be completed, conclusions concerning generic levels, the ranges of variability in the relative density of tubules must be tempered ultrastructural parameters examined overlapped

Table 3. Qualitative assessment of abundance of enamel tubules in Late Cretaceous and early Tertiary multituberculates according to Carlson and Krause (1985). Arranged alphabetically within each category - taxonomic assignments are provided in Table 1.

None Very Few Few Few-Sev. Several Several-Many Many

Anconodon Boffius Neoplagiaulax Xanclomys Essonodon Meniscoessus Cimexomys Cimolodon Catopsalis Parectypodus Kryptobaatar Paracimexomys Ec~us Cimolomys Pentacosmodon Microcosmodon Eucosmodon Lambdopsalis Sphenopsalis Hainina Liotomus Taeniolabis Kimbetohia Mimeto don Mesodma Ptilodus Neoliotomus Stygimys Prionessus Prochetodon Xyronomys

1602 Enamel Ultrastructure of Multituberculates significantly. It is only above the species and our preliminary cladogram, it appears that level that consistent differences in multituber­ large, arc-shaped, not small, circular, prisms culate enamel ultrastructure were detected. represent the primitive condition. Within Late These differences are, in general, of such a Cretaceous and early Tertiary multituberculates, magnitude that "noise" introduced by relatively Paracimexomys, Cimexomys. cimolomyids (Meniscoes­ imprecise sampling of enamel depth is insuffi­ sus, Cimolomys, and possibly Essonodon), the rel­ cient to mask the differences at the generic atively primitive ptilodontoid Cimolodon, and all level and above. taeniolabidoids except Neoliotomus have large, Homology and Polarity of Enamel Ultrastructural arc-shaped prisms. This result allows us to ~ reject the hypothesis that circular prisms Fosse et al. (1978, p. 60) proposed that the represent the primitive condition within later presence of "gigantoprismatic" enamel in taenio­ multituberculates. Such a conclusion is corrobo­ labidoids might be of "diagnostic value in rated by the discovery of large, arc-shaped multituberculate taxonomy" and suggested that prisms in late Early Cretaceous multituberculates other multituberculate genera should be examined. from Asia (Fosse et al. 1985) and North America Later, Fosse et al. (1985) used the presence of (Krause et al., in prep.). It remains to be gigantoprismatic enamel in cimolomyids to suggest demonstrated, however, whether Pattern 1 prisms that members of the family should be removed from are primitive for mammals as a whole. Suborder incertae sedis and allied with the It seems an inescapable conclusion that if Taeniolabidoidea, to which the family was origi­ plagiaulacoids are somehow involved in the nally allocated by Sloan and Van Valen (1965). ancestry of later multituberculates, as they Sahni (1984, 1985) concluded that circular prisms almost certainly were, then fully prismatic are the most primitive prismatic type among enamel evolved in multituberculates independent mammals, implying that they are also primitive of its evolution in other mammalian taxa. This within multituberculates. Kozawa (1984:437) result confirms an earlier speculation by Clemens stated that "round-shaped enamel prisms which are (1979:199). Furthermore, the small, circular surrounded by interprismatic enamel, and do not prisms found in later multituberculates are not have the Schreger band ... are considered the homologous with those seen in other mammals early prismatic enamel pattern" and, further, exhibiting Pattern 1 enamel. that "the primitive prismatic enamel structure has irregularly arranged round prisms surrounded Future Research by broad interprismatic enamel." Similarly, Boyde and Martin (1984b:419) stated that "by We offer several suggestions for developing outgroup comparison with the other mammalian a research program to further study the enamel orders, it appears probable that pattern 1 would ultrastructure of multituberculates and its be the primitive pattern for Mammalia." Martin utility in phylogenetic analysis. One of the (1981) came to a similar conclusion based on the greatest difficulties in studying the properties purported presence of Pattern 1 enamel in the of multituberculate enamel, as with most fossil Late Triassic triconodont Eozostrodon (Grine et material, has been the rarity of specimens and al. 1979). However, no tests, within the frame­ their potential damage during preparation for work of explicit phylogenetic hypotheses, have scanning electron microscope analysis. The been performed to determine whether individual invention and modification of a new microscope, ultrastructural characters are homologous in all the Tandem-Scanning Reflected-Light Microscope forms in which they occur and whether arc-shaped (TSRLM) has provided a means to circumvent many prisms are derived, at least within the Multitu­ of these problems (Petran et al. 1985). The berculata. TSRLM permits examination of subsurface structure A test of the homology of enamel ultrastruc­ of enamel without the physical removal of super­ tural types in later multituberculates, within ficial enamel by grinding or sectioning. It has the context of Archibald's (1982) cladogram of already been profitably employed in the examina­ relationships among some Late Cretaceous genera tion of primate enamel ultrastructure (Boyde and and a preliminary cladogram of relationships Martin, 1984b). Taxa that are known from only among all Late Cretaceous and Early Tertiary one or a few specimens (e.g., Acheronodon, genera (Krause and Carlson, in prep.), indicates Viridomys) can now be examined without fear of that both large, arc-shaped prisms and small, substantially reducing the hypodigm. Further­ circular prisms appear to be homologous, with one more, some of the problems involved in the exception. Neoliotomus exhibits prismatic enamel preparation of specimens for SEM analysis and the that is virtually indistinguishable from ptilo­ interpretation of structure revealed after dontoid enamel yet, based on gross dental morph­ preparation, particularly acid etching, are ology, groups with the taeniolabidoid clade. avoided. Although the resolution of detail is Neoliotomus appears to have evolved small, less on TSRLM micrographs than on SEM micro­ circular prisms independently; its condition graphs, all of the measurements that have been represents a homoplasious similarity rather than employed in the above analysis can be taken on similarity due to a common origin with ptilodon­ TSRLM micrographs. And finally, since the TSRLM toids. The presence of large, arc-shaped prisms can optically section materials, it will be in the ptilodontoid Cimolodon does not represent possible to reconstruct, in three dimensions, a homoplasious condition since Cimolodon is many of the details of enamel ultrastructure that phylogenetically the most primitive ptilodontoid. have previously proven elusive (e.g., the course Based on the distribution of enamel ultra­ of prisms and tubules from the enamel-dentine structural types on Archibald's (1982) cladogram junction to the outer enamel surface).

1603 D. W, Krause and S. J, Carlson

It would appear that the major gap in our on turnover rates and trophic structure. Acta current knowledge of multituberculate enamel Palaeont. Polonica 28, 7-17. ultrastructure concerns earlier taxa, particu­ Boyde A. (1964). The structure and devlopment larly plagiaulacoids. Only a few plagiaulacoid of mammalian enamel. Ph.D. thesis, University of genera have been examined to date (Fosse et al. London. 1985) and therefore the breadth of sampling is not comparable to that for later multitubercu­ Boyde A. (1965). The structure of developing lates. The major transition in ultrastructural mammalian dental enamel. In: Tooth Enamel -- its types, between "preprismatic" and prismatic Composition, Properties, and Fundamental Struc­ types, appears to have occurred sometime during ture, Stack MV, Fearnhead RW (eds), pp. 163-167. the Early Cretaceous. Further study of the known John Wright and Sons Ltd., Bristol. early Early Cretaceous and late Early Cretaceous Boyde A. (1969a). Correlation of ameloblast forms is warranted. Hopefully, multituberculates size with enamel prism pattern: use of scanning of intermediate age will be discovered and their electron microscope to make surface area measure­ presumably transitional enamel ultrastructure ments. Z. Zellforsch. 93, 583-593. documented. Finally, a review of the history of past Boyde A. (1969b). Electron microscope obser­ work on enamel ultrastructure in multitubercu­ vations relating to the nature and development of lates is instructive with regard to the caution prism decussation in mammalian dental enamel. that should be exercised concerning generaliza­ Bull. Group Int. Rech. Sc. Stomat 12, 151-207. tions derived from sampling only a few specimens Boyde A. (1976). Amelogenesis and the structure of a few taxa. Conclusions regarding the homolo­ of enamel. In: Scientific Foundations of Den­ gy and polarity of enamel ultrastructural types, tistry, Cohen B, Kramer IRH (eds), pp. 335-352. when not drawn within the context of robust William Heineman Medical Books Ltd., London. phylogenetic hypotheses, must also be viewed with circumspection. With the continued emphasis on Boyde A. (1978). Development and structure of extensive sampling of enamel ultrastructural data the enamel of the incisor in the three classical from all known multituberculate taxa and with subordinal groups of the Rodentia. In: Develop­ continuing efforts to develop explicit, testable ment, Function and Evolution of Teeth, Butler PM, phylogenies, it will be possible to learn a great Joysey KA (eds), pp. 43-58. Academic Press, New deal more than is currently known about the York. evolution of enamel ultrastructure within the Boyde A, Lester KA. (1967). The structure and Multituberculata. development of marsupial enamel tubules. Z. Zellforsch. ~. 558-576. Boyde A, Lester KA. (1984). Further SEM studies Acknowledgements of marsupial enamel. In: Tooth Enamel IV, Fearnhead RW, Suga S (eds), pp. 442-446. We are extremely grateful to the following Elsevier Science Publishers, Amsterdam. individuals for the various ways in which they Boyde A, Martin L. (1982). Enamel microstruc­ assisted in this study: A. Boyde, G. Fosse, F. ture determination in hominoid and cercopithecoid E. Grine, Z. Kielan-Jaworowska, M. C. Maas, and primates. Anat. Embryo. 165, 193-212. L. B. Martin for informative discussion; N. Creel, G. Fosse, S. Hartman, and J. T. Boyde A, Martin L. (1984a). The microstructure Stern for computer assistance; J. Cohen, G. of primate dental enamel. In: Food Acquisition Resnick, and particularly M. C. Maas for help and Processing in Primates, Chivers DJ, Wood BA, with some 11th hour measuring; L. Betti for Bilsborough A (eds), pp. 341-367. Plenum Press, rendering the graphs and drawings; J. Muennig for New York. photographs; and N. Glover for typing portions of Boyde A, Martin L. (1984b). A non-destructive the manuscript. This work was supported by survey of prism packing patterns in primate National Science Foundation grant BSR-84-06707 enamel. In: Tooth Enamel IV, Fearnhead RW, Suga and a Biomedical Research Support Grant from the S (eds), pp. 417-421. Elsevier Science State University of New York Research Foundation Publishers, Amsterdam. to DWK, and a Geological Society of America Research Grant to SJC. Carlson SJ, Krause DW. (1982). Multituberculate phylogeny: evidence from tooth enamel ultrastruc­ References ture. Ann. Mtg. 1982 Geol. Soc. Amer. Abstr., 14, 460. Archer M, Flannery TF, Ritchie A, Molnar RE. Carlson SJ, Krause DW. (1985). Enamel ultra­ (1985). First Mesozoic mammal from Australia structure. of multituberculate mammals: an inves­ an early Cretaceous monotreme. Nature 318, tigation of variability. Contrib. Mus. Paleont. 363-366. Univ. Mich., 27, 1-50. Archibald JD. (1982). A study of the Mammalia Carter JT. (1922). On the structure of the and geology across the Cretaceous-Tertiary enamel in the primates and some other mammals. boundary in Garfield County, Montana. Univ. Proc. Zool. Soc. London, 599-608. Calif. Publ. Geol. Sci., 122, 1-286. Clemens WA. (1979). Marsupialia. In: Mesozoic Archibald JD. (1983). Structure of the K-T Mammals -- The First Two-Thirds of Mammalian mammal radiation in North America: speculations

1604 Enamel Ultrastructure of Multituberculates

History, Lillegraven JA, Kielan-Jaworowska Z, Gantt D. (1983). The enamel of Neogene homi­ Clemens WA (eds). Pp. 192-220. Univ. Calif. noids: structural and phyletic implications. In: Press, Berkeley. New Interpretations of Ape and Human Ancestry, Ciochon RL, Corruccini RS (eds). Pp. 249-298. Clemens WA, Kielan-Jaworowska, Z. (1979). Plenum Press, New York. Multituberculata. In: Mesozoic Mammals -- The First Two-Thirds of Mammalian History, Lille­ Gantt D, Pilbeam DR, Steward GP. (1977). graven JA, Kielan-Jaworowska Z, Clemens WA (eds). Hominoid prism patterns. Science 198, 1155-1157. Pp. 99-149. Univ. Calif. Press, Granger WK, Simpson GG. (1929). A revision of Berkeley. the Tertiary Multituberculata. Bull. Amer. Mus. Clemens WA, Lillegraven JA, Lindsay EH, Simpson Nat. Hist. 56, 601-676. GG. (1979). Where, when, and what -- a survey Grine FE, Fosse G, Krause DW, Jungers WL. of known Mesozoic mammal distribution. In: (1986a). Analysis of enamel ultrastructure in Mesozoic Mammals -- The First Two-Thirds of archeology: the identification of Ovis aries and Mammalian History, Lillegraven JA, Kielan­ Capra hircus dental remains. J. Archaeo. Sci., Jaworowska Z, Clemens WA (eds). Pp. 7-58. Univ. in press. Calif. Press, Berkeley. Grine FE, Krause DW, Fosse G, Jungers WL. Cooper JS, Poole DFG. (1973). The dentition and (1986b). Analysis of individual, intraspecific dental tissues of the agamid lizard Uromastyx. and interspecific variability in quantitative J. Zool., Lond. 169, 85-100. parameters of caprine tooth enamel structure. Eisenberg MJ. (1938). A microscopic study of Acta Odontol. Scand., in press. the surface enamel of human teeth. Anat. Rec. Grine FE, Vrba ES, Cruickshank ARI. (1979), 71, 221-226. Enamel prisms and diphyodonty: linked apomorphies Fortelius M. (1985). Ungulate cheek teeth: of Mammalia. S. Afr. J. Sci. 75, 114-120. developmental, functional, and evolutionary Hahn G. (1969). Beitrage zur Fauna der Grube interrelations. Acta Zool. Fenn. 180, 1-76. Guimarota Nr. 3, die Multituberculata. Palaeon­ Fosse G. (1968a). A quantitative analysis of togr. Abt. A, ill, 1-100. the numerical density and the distributional Hahn G. (1973). Neue Zahne von Haramiyiden aus pattern of prisms and ameloblasts in dental der deutschen Ober-Trias und ihre Beziehungen zu enamel and tooth germs. III. The calculation of den Multituberculaten. Palaeontogr. Abt. A, 142, prism diameters and numbers of prisms per unit 1-15. area in dental enamel. Acta Odontol. Scand. 26, 315-336. Hahn G, Hahn R. (1983). Multituberculata. In: Fossilium Catalogus I: Animalia, Westphal F Fosse G. (1968b). A quantitative analysis of (ed.). Pp. 1-409. Kugler Publications, Amster­ the numerical density and the distributional dam. pattern of prisms and ameloblasts in dental enamel and tooth germs. IV. The vertical compres­ Ishiyama M. (1984). Comparative histology of sion of the prism pattern on the outer enamel tooth enamel in several toothed whales. In: surface of human permanent teeth. Acta Odontol. Tooth Enamel IV, Fearnhead RW, Suga S (eds.). Scand. 26, 545-572. Pp. 432-436. Elsevier Science Publishers, Amsterdam. Fosse G, Eskildsen 0, Risnes S, Sloan RE. (1978). Prism size in tooth enamel of some Late Jenkins FA, Crompton AW, Downs WR. (1983). Cretaceous mammals and its value in multituber­ Mesozoic mammals from Arizona: new evidence on culate taxonomy. Zool. Scripta, l, 57-61. mammalian evolution. Science 222, 1233-1235. Fosse G, Kielan-Jaworowska Z, Skaale SG. (1985). Jenkins FA, Krause DW. (1983). Adaptations for The microstructure of tooth enamel in multituber­ climbing in North American multituberculates culate mammals. Palaeontol. 28, 435-449. (Mammalia). Science 220, 712- 715. Fosse G, Risnes S, Holmbakken N. (1973). Prisms Kawai N. (1955). Comparative anatomy of the and tubules in multituberculate enamel. Cale. bands of Schreger. Okaj. Folia Anat. Jap. 27, Tiss. Res. 11, 133-150. 115-131.

Frank RM, Sigogneau-Russell D. (1984). Ultra­ Kemp TS. (1982). Mammal-like reptiles and the structure of enamel and dentin of Haramiyidae origin of mammals. Academic Press, New York. teeth. J. Dent. Res. 63, 531. 289-293. Frank RM, Sigogneau-Russell D, Voegel JC. Kemp TS. (1983). The relationships of mammals. (1984). Tooth ultrastructure of Late Triassic Zool. J. Linn. Soc. 11, 353-384. Haramiyidae. J. Dent. Res. 63, 661-664. Koenigswald W von. (1982). Enamel structure in Gantt D. (1980). Implications of enamel prism the molars of Arvicolidae (Rodentia, Mammalia), a pattern for the origin of New World monkeys. In: key to functional morphology and phylogeny. In: Evolutionary Biology of the New World Monkeys and Teeth: Form, Function and Evolution, Kurten B Continental Drift, Ciochon RL, Chiarelli AB (ed). Pp. 109-122. Columbia Univ. Press, New (eds). Pp. 201-217. Plenum Press, New York. York.

1605 D. W. Krause and S. J. Carlson

Kozawa Y. (1984). The development and the Poole DFG. (1957). The formation and properties evolution of mammalian enamel structure. In: of the organic matrix of reptilian tooth enamel. Tooth Enamel IV, Fearnhead RW, Suga S (eds.). Quart. J. Microsc. Sci. 98, 349-367. Pp. 437-441. Elsevier Science Publishers, Poole DFG. (1967). Enamel structure in primi­ Amsterdam. tive mammals. J. Dent. Res. 46, 124. Krause DW. (1980). Multituberculates from the Presley R. (1981). Alisphenoid equivalents in Clarkforkian Land-Mammal Age, late Paleocene placentals, marsupials, monotremes and fossils. early Eocene, of western North America. J. Nature 294, 668-670. Paleontol. 54, 1163-1183. Rigby Jr. JK (1980). Swain Quarry of the Fort Krause DW. (1982). Jaw movement, dental func­ Union Formation, middle Paleocene (Torrejonian), tion, and diet in the Paleocene multituberculate Carbon County, Wyoming: geologic setting and Ptilodus. Paleobiology ~. 265-281. mammalian fauna. Evol. Monogr. l, 1-179. Krause DW. (1986). Competitive exclusion and Risnes S, Fosse G. (1974). The origin of taxonomic displacement in the fossil record: the marsupial enamel tubules. Acta Anat. fil, case of rodents and multituberculates in North 275-282. America. In: Vertebrates, Phylogeny, and Philo­ sophy: a Tribute to George Gaylord Simpson, Sahni A. (1979). Enamel ultrastructure of Flanagan K, Lillegraven JA (eds.). Univ. Wyoming certain North American Cretaceous mammals. Contrib. Geol., Spec. Paper No. 3, in press. Palaeontogr. Abt. A, 166, 37-49. Krause DW, Baird D. (1979). Late Cretaceous Sahni A. (1984). The evolution of mammalian mammals east of the North American Western enamels: evidenca from MultituberculatR (Allo­ Interior Seaway. J. Paleontol. 53, 562-565. theria, extinct); primitive whales (archaeocete Cetacea) and early rodents. In: Tooth Enamel IV, Krause DW, Jenkins FA. (1983). The postcranial Fearnhead RW, Suga S (eds.). Pp. 457-461. skeleton of North American multituberculates Elsevier Science Publishers, Amsterdam. (Mammalia). Bull. Mus. Comp. Zool., Harvard Univ. 150, 199-246. Sahni A. (1985). Enamel structure of early mammals and its role in evaluating relationships Lester KS. (1970). On the nature of "fibrils" among rodents. In: Evolutionary Relationships and tubules in developing enamel of the opossum, Among Rodents, A Multidisciplinary Analysis, Didell1bis !!!_ar~ialis. J. Ultrastruct. Res., 30, Luckett WP, Hartenberger J-L (eds.). Pp. 64- 77. 133-150. Plenum Press, New York. Marcus H. (1931). Zur Phylogenie der Schmelz­ Sigogneau-Russell D, Frank RM, Hemmerle J. prismen. Z. Zellforsch. 12, 395-429. (1984). Enamel and dentine ultrastructure in the early Jurassic therian Kuehneotherium. Zool. J. Martin SP. (1981). Enamel prism patterns of plesiadapid primates and paramyid rodents. Amer. Linn. Soc. 82, 207-215. J. Phys. Anthro. 54, 249. Simpson GG. (1929). American Mesozoic Mammalia. Mem. Peabody Mus. Nat. Hist., Yale Univ., l, McKenna MC. (1975). Toward a phylogenetic 1-235. classification of the Mammalia. In: Phylogeny of Simpson GG. (1937). Skull structure of the the Primates, Luckett WP, Szalay FS (eds). Pp. Multituberculata. Bull. Amer. Mus. Nat. Hist., 21-46. Plenum Press, New York. n. n7-763. Moss ML. (1969). Evolution of mammalian dental Sloan RE. (1979). Multituberculata. In: The enamel. Amer. Mus. Novitates 2360, 1-39. Encyclopedia of Paleontology, Fairbridge RW, Moss ML, Kermack KA. (1967). Enamel structure Jablonski D (eds.). Pp. 292-298. Dowden, in two Triassic mammals. J. Dent. Res. 46, Hutchinson and Ross, Stroudsberg. 745-747. Sloan RE, Van Valen L. (1965). Cretaceous Osborn JW. (1974). The relationships between mammals from Montana. Science 148, 220-227. prisms and enamel tubules in the teeth of Didel­ Van Valen L, Sloan RE. (1966). The extinction phis marsupialis, and the probable origin of the of the multituberculates. Syst. Zool. 15, tubules. Arch. Oral Bio. 19, 835-844. 261-278. Osborn JW, Hillman J. (1979). Enamel structure Vianey-Liaud M. (1979). Les Mammiferes Montiens in some therapsids and Mesozoic mammals. Cale. de Hainin (Paleocene moyen de Belgigue). Part I: Tissue. Inter. 29, 47-61. Multitubercules. Palaeovert. 1, 117-131. Petran M, Hadravsky M, Boyde A. (1985). The Vrba ES, Grine FE. (1978). Australopithecine Tandem Scanning Reflected Light Microscope. enamel prism patterns. Science 202, 890-892. Scanning 1, 97-108. Poole DFG. (1956). The structure of the teeth of some mammal-like reptiles. Quart. J. Microsc. Sci., 97, 303-312.

1606 Enamel Ultrastructure of Multituberculates

Discussion with Reviewers

M. Fortelius: What are your reasons for accep­ ting the geologically oldest character states as the most primitive? Authors: While it is true that the geologically oldest character states are usually the most primitive, we fully realize that such is not always the case. In constructing our cladogram for intergeneric relationships of Late Cretaceous and Early Tertiary genera of multituberculates, we have used plagiaulacoids as the outgroup, not because they are geologically older (although they do precede the Late Cretaceous forms by over 50 million years) but because they exhibit a more primitive morphology. Plagiaulacoids have more premolars than are known for any Late Cretaceous and Early Tertiary genera. Teeth are frequently lost in mammalian evolution and rarely added. Thus, it seems reasonable to hypothesize that plagiaulacoids are an appropriate outgroup, not because of age but because of the documentation of primitive morphology in independent charac­ ters.

M. Fortelius: Saying that arc-shaped prisms are primitive implies either that decussation (in the sense of relative ameloblast movement during secretion) arose prior to prisms (Tomes' proces­ ses), or that Boyde's fairly generally accepted scheme is wrong. Authors: Boyde and Martin (1984. The micro­ structure of primate dental enamel. In: Food Acquisition and Processing in Primates, Chivers DJ, Wood BA, Bilsborough A (eds), pp. 341-367. Plenum Press, New York) have recently demon­ strated that prisms need not necessarily be "arc-shaped" (as in prism packing patterns 2 and 3) in order to decussate. They observed "very well marked prism decussation" in Pattern 1 enamel in a New World monkey (Callithrix) and therefore implicitly modified "Boyde's fairly generally accepted scheme."

M. Fortelius: Considering that decussation must have arisen several times independently in mammals and that prism packing is severely constrained by geometric relationships, is it likely that any prism pattern homologies can be reliably established between higher taxa? Authors: Yes, at least such appears to be the case within the Multituberculata. Other cases must be evaluated independently within the context of explicit phylogenetic hypotheses based on characters other than prism packing patterns. Before we can use ultrastructural characters to infer relationship, we must have compelling evidence to support the claim that structural similarity is due to common origin. If different prism types can be shown to be distributed randomly across a phylogenetic tree constructed with a large number of independent characters, then it is unlikely that they are homologous. But if the same prism patterns are not distrib­ uted randomly, a parsimonious interpretation would hold that a hypothesis of homology is supported. A determination of the polarity of prism patterns is based on the relative positions of the patterns on the phylogenetic tree.

1607