TAPHONOMY OF THE TERRESTRIAL MAMMALS OF LEISEY SHELL PIT lA, HILLSBOROUGH COUNTY, FLORIDA Ann E. Pratti and Richard C. Hulbert, Jr.2 ABSTRACT The terrestrial mammal fraction of the Leisey Shell Pit 1 A bone assemblage (early Pleistocene) is numerically dominated by medium-sized herbivores, with camelids and equids accounting for 45% and 22% of the individuals, respectively. Some taphonomic modifications to the bone assemblage look place at the site ofdeposition (scratchmarks, breakage, hydrodynamic sorting, and a limited amount of scattering); others occurred elsewhere, either in a nearby terrestrial setting (carnivore/scavenger activity, subaerial weathering both generally to a relatively low degree) or during fluvial transport to the site. There is no preferred orientation of the long bones at the site, suggesting the lack of a strong, directional current However, multidirectional tidal currents probably aided in carcass disassociation and winnowing of smaller, lighter elements. A significantly greater proportion of equid and proboscidean skeletal elements were weathered and damaged by mammalian carnivores than those ofthe camelids. Elements of the latter more frequently entered the site in an unmodified state as parts of carcasses (later disassociated) while those of the former were more often brought to the site individually in a modified condition. Population dynamics of Polaeolama mintica and Equus "lei*i" (the two most common mammals) also conform to a mixture of catastrophic and attritional mortality. The principal mammalian carnivores responsible f6r modifying the Leisey bone assemblage were Canis armbrusteri and Arctodus pristinus. The former caused the bite marks and chewed surfaces on the bones of medium-sized herbivores, while the latter scavenged on the carcasses of ground sloths and proboscideans. RESUMEN La fracci6n de mamiferos terrestres del ensamblaje de huesos del dep6sito dc conchuelas de 1£isey 1 4 es dominada num6ricamente por herbivoros de tamafio medio, con cam61idos y 6quidos representando respectivamente un 45% y 22% de los individuos. El ensamblaje de huesos sufri6 algunas modificaciones tafon6micas en: el sitio de deposici6n (rasguiladuras, rompimiento, reordenamiento 1 2 The senior author is an Assistant Professor of Biology at Georgia Southern University, Statesboro. Georgia 304604042. The junior author is Curator of Paleontology and Assistant Professor of Geology at Georgia Southern University, Statesboro, Georgia 30460-8149. PRATT, A.E., AND R.C. HULBERT, JR. 1995. Taphonomy of the Terrestrial Mammals of I.jeisey Shell Pit 14 Hillsborough County, Florida. Bull. Florida Mus. NaL Hist 37 Pt 1(7):177-250. 178 BULLETIN FLORIDA MUSEUM NATURAL HISTORY VOL. 37, Fr. I, No. 7 modificaciones tafonimicas en: el sitio de deposici6n (rasguaaduras, rompimicnto, reordenamiento hidrodinlimico y cierto grade de esparcimiento); en otras areas, tanto cerca como lejos del ambiente terrestre (actividad de carnivores/carroaerax efectos climbticos subacreos, ambos en un grado relativamente bajo); o durante transporte Buvial hacia el el sitio de deposicitn. El hecho de que los huesos largos no muestran una orientaci6n preferida en el sitio sugiere la ausencia de una corriente direccional fierte en el sitio. Sin embargo, , ' ' ... ' en la disociaci6n de carcasas y la dispersi6n de los elementos mas pequeaos y livianos. Una proporci60 significativamente mayor de elementos del esqueleto de 6qui,los y proboscideos fue afectada por el clima y daaa£la por mamiferos carnivores en comparacidn con elementos del csqueleto de camdlidos. Estos Oltimos entraron mas frecuntemente at sitio eomo Bm Completas no modificadas (mas tarde disociadas1 mientras que los elementos del esqueleto de 6quidos'y proboscideos fueron mas frecuentemente Ilevados individualmente al sitio, habiendo ya sido modificados. La dinhmica de las poblaciones de Pateolama mirvica y Equu3 "lei*" (los dos mamiferos mas comuncs) estan en conf6rmidad tambien con una mczcla de mortalidades por caustrofe y competencia. Los principales mamiferos camivoros responsables de la modificacidn del ensamblaje de huesos de 1.eisey fueron Canis armbrusteri y Arctodus pristinus. El primero caus,5 las marns de mordidas y superficies masticadas en los huesos de herbivoros de tamafo medio, mientras que el Oltimo carrofie6 las carcasas de perezosos terrestres y proboscideos. INTRODUCTION Leisey Shell Pit lA is a rich, early Irvingtonian vertebrate fossil assemblage from southwestern Hillsborough County, Florida (Hulbert and Morgan 1989). The bone-producing sediments of Lcisey Shell Pit lA (= Leisey lA) consist of a 5 to 30 cm thick layer of muddy sand and shells. Analysis of the Leisey invertebrates, as discussed by Portell et at. (this volume), shows that the depositional environment was nearshore marine, perhaps a grass or mud flat. Mean water depth probably did not exceed one meter and deposition occurred in a relatively low-energy environment (Hulbert and Morgan 1989). Invertebrate contributions from nonmarine environments include both freshwater and terrestrial mollusks. The environment of deposition is suggestive of a marine embayment fed by freshwater streams and open to the Gulf Coast. As is the case with the invertebrate fauna vertebrate remains were derived from several sources. Contributions to the vertebrate component of the shell bed originated in at least four different environments: 1) The environment of deposition (autochthonous). Well-preserved remains of nearshore and estuarine species of sharks and bony fish (Scudder et al. this volume) and wading birds (Emslie this volume) represent animals that were living in, or close to, the environment of deposition. 2) Offshore marine environments. Contributions from this area include fragmentary remains of sea tunles and cetaceans (Meylan this volume; Morgan and Hulbert this volume). 3) Freshwater environments. Emydid and trionychid turtles and alligator remains form the major component of the freshwater fauna at Leisey lA (Meylan this volume). Several species of freshwater PRATr & HULBERT: TERRESTRIAL MAMMAL TAPHONOMY OF LEISEY SHELL PIT 179 fish and water birds have also been recovered (Scudder et al. this volume; Emslie this volume). Freshwater mammals (Neochoerus, Lutra, Trichechus) are uncommon. 4) Terrestrial environments. The vast majority of the large vertebrate fossils found are those of land mammals. Large tortoises are also common. Based on the interpretation of Leisey as a shallow embayment fed by freshwater streams, it is not difficult to account for the presence of aquatic vertebrates. Therefore, this analysis deals primarily with the terrestrial vertebrates, in particular the mammals, because the reasons for their great abundance in a marine deposit are not readily apparent. Several possible scenarios will be discussed that could account for the presence of bones of terrestrial vertebrates in the shell bed. For example, the animals may have been killed and buried as the result of a single catastrophic event such as a hurricane or storm surge, or the assemblage may have formed attritionally over a number of years. The terrestrial vertebrates may have been living in the immediate vicinity of the deposit, or at some distance from it. We investigate which agents of transport if any, played a role in forming the bone-bearing layer, and the condition of the remains when they were introduced into the site of deposition. The intent of this paper is to reconstruct the taphonomic histories of the large mammals of Leisey lA to determine the sequence of events leading to formation of the terrestrial component of the fauna. ACKNOWLEDGMENTS Field data were collected by the authors. Gary S. Morgan. Steven D. Emslie, other Florida Museum of Natural History (FLMNH) personnel, and volunteers from the Tampa Bay Mineral and Science Club. We thank D. K McLain and D. R. Drake ibr qidisdical advice, Gary S. Morgan for hospitality and assistance in the FLMNH collection, M. Evans for loan of R. W. Allmendinger's Stereonet computer program, and A K. Behrensmeyer and P. Shipman fbr their opinions regarding the origin of sulface features on the Lcisey bones. Helpful reviews ofthe manuscript were provided by A R. Fiorillo and Gary S. Morgan. FIELD EXCAVATION METHODS The large size of the site (approximately 2,000 m2) and the enormous number of bones and bone fragments (> 50,000) at Leisey 14 coupled with the relatively short time period allowed to excavate the site, placed numerous constraints on the collecting methods and the amount of taphonomic data that could be taken. To accurately record the exact position and orientation of every bone was impossible. A procedural compromise was struck between this extreme, and its opposite, to not record any positional data at all. 180 BULLETIN FLORIDA MUSEUM NATURAL HISTORY VOL 37, PT. I, No. 7 N 1 P 2 meters 0 N REGION M EXCAVATED L IN 1983 K J 1 H H G G F F E E D D C C B B A A -A 1 2 3 4 5 6 7 8 9101112131415161718192021 Figure 1. Map of the Lcisey Shell Pit 1 A fossil site, showing the grid coordinate system that was established in April 1984. Each squareis referred tobyaletter-numeral combination, for example, 84. During the 1984 field operations at Leisey lA, the site was gridded into approximately 2x2 meter squares. The east-west running rows of squares were identified by letters, with the southernmost being row A (Fig. 1). Columns were given numerical designations, with the westernmost being column 1. Using this system, individual squares were named by a letter-numeral combination; for example the southwestern-most square was called Al. All bones collected from a particular square were kept together and labelled with its coordinates. When identifiable specimens were eventually cataloged into the FLMNH collection, the square's coordinates were incorporated into the collection's computerized catalog database. This practice proved very useful for retrieving data for sorting and counting analyses. The bone-bearing unit at Leisey lA originally lay about 6.5 m below the ground surface (Morgan and Hulbert this volume, fig. 1). Prior to the initial excavation of a portion of the site in 1983, the surficial sand unit and the upper PRATr & HULBERT: TERRESTRIAL MAMMAL TAPHONOMY OF LEISEY SHELL PIT 181 2 m of the upper shell bed were removed by heavy earth-moving machinery. About a meter of the upper shell bed and 0.5 m of the indurated "hard layer" then remained on top of the bone-bearing horizon. A small bulldozer cleared away most of the upper shell bed before the start of the 1984 field season. The remaining overburden was cleared manually with picks and shovels to expose the bone bed. It was primarily dug with small tools such as awls and screwdrivers, which proved to be more effective in the shelly matrix than standard trowels (Fig. 2). Although many of the bones were crushed, this is thought to be primarily the result of geologic compaction and not from the earth-moving equipment. The color along a freshly broken surface is different (lighter in tone) than along an ancient break. The latter has the same dark color as the surface of the bones. The color of the broken edges of the compacted bones indicated that the crushing was not modern. Ideally every piece of fossil bone was collected from a square. However, given the thousands of small fragments and gar scales, and the varied patience and experience of the collectors, squares were not collected with equal thoroughness. Undoubtedly almost every identifiable element of the large vertebrates was collected from each excavated square. To ensure completeness, several of the taphonomic analyses were limited to squares excavated by experienced FLMNH personnel. About one metric ton of matrix from the bone-bearing horizon at Leisey lA was screen-washed for small vertebrates through standard window screen mesh. Although the concentrate contained numerous bones of small fish, recovery of herps, birds, and mammals was rare. Small samples of matrix were saved for geologic and palynologic analysis. Positional and orientation data were taken on over 1300 bones before their removal from the sediment. We concentrated on obtaining data on mandibles, maxillae, cranig scapulae, innominates, and major limb elements of all the large vertebrates, as well as the elongated cervical vertebrae and proximal phalanges of the camelids. The large sample (N = 1367) is representative and unbiased with regard to distribution of bones within the site. Information taken on each element included location within the square (to the nearest centimeter), bearing, plunge, length, and which side faced "up" in the field. No measurement of (vertical) depth was taken as the bone-bearing unit was quite thin relative to its lateral extent. Bones appeared to be distributed throughout the layer and not concentrated at the top or bottom. The position/orientation data were later sorted and analyzed using the Quattro spreadsheet program. TAPHONOMIC ANALYSIS The contribution of various taphonomic processes to the formation of the bone assemblage will be determined by investigation of the following aspects of the 182 BULLETIN FLORIDA MUSEUM NATURAL HISTORY VOL. 37, PT. I, No. 7 A 1 1/ fAILia-.6- 7 -1.'-I , ..--- I 5 0 - 2,/4(7 244 Figure 2. Field photos of I,eisey Shell Pit l A taken during the 1984 field season. (A) View of six partially excavated squams showing abundance and distribution offossils. (B) G. S. Morgan taking the bearing of a camelid metapodial. PRATr & HULBERT: TERRESTRIAL MAMMAL TAPHONOMY OF LEISEY SHELL PIT 183 fossil deposit: bone distributions and associations, types and degrees of bone modification, and skeletal part representations. Each of these bone or assemblage features has been shown to provide information concerning the taphonomic pathways leading to formation of a fossil deposit (e. g., Voorhies 1969; Wolff 1973; Behrensmeyer 1975, 1978; Haynes 19802. 1985, 1988; Badgley 1986a; Blumenschine 1986, 1989; Pratt 1990). Examination of all these lines of taphonomic evidence will be employed to determine the events that culminated in the terrestrial vertebrate assemblage preserved at Leisey lA. ART[CULATION AND ASSOCIATION During excavations at the Leisey lA locality it quickly became evident that terrestrial vertebrate remains had not been buried as complete, articulated skeletons (Fig. 2). Even though bones were not found in articulation, in a number of instances elements or portions of elements of one individual were found within a few meters of one another. Evidence of association was most obvious for less common taxa, where remains apparently belonging to single individuals were found spread over a distance of a few squares. The large number of individuals of camelids, and to a lesser extent equids, precluded definite identifications of postcranial associations for these taxa; however, many matches were found between cranial elements. The most abundant non-mammalian vertebrates (excluding fish), freshwater turtles and land tortoises, were most frequently represented by individual bones of the plastron and carapace, although contacts could be found with isolated elements from other regions of the site. The total degree of skeletal association may never be known as the number of bones from the site is so large as to make finding all associations a nearly impossible task. Therefore, representative taxa and elements were selected to provide some estimate of association and distribution patterns. All equid mandibles, symphyseal portions, and lower cheekteeth from the two most common species, Equus "lei*i" and Equus (Hemionus) sp., were examined for matches. In addition, all cataloged tapir and proboscidean elements were examined for evidence of association. Of a total of 84 equid right or left mandibles, 18 symphyseal regions, and over 300 isolated cheekleeth, matches were found between 20 pairs of right and left mandibles, 11 jaws and symphyses, and 40 mandibles and isolated cheekteeth. Figure 3 illustrates the distribution of these associated elements within the site, and Figure 4 graphically demonstrates the range of distances separating the associated elements. Over 45% of associated cranial elements of equi(is were located within 2 m of one another, and nearly 75% were separated by a distance of 5 m or less. 184 BULLETIN FLORIDA MUSEUM NATURAL HISTORY VOL- 37, Fr. I, No. 7 ~11111111111 R Q N -1 { D\€ -P 0 \ T°=-2 N 2 meters ll TT M TD T\ L D-D \\ - K D-J H_ _H G_ f-D G F - DZ.- - E- 0-*U 4 01 -0 f -E DD_ , . --AD 1 - t_ 1' P - C 0- DN \8 _ X ~ DO-T A.11_1'''' .--1 1111111111111 1 2 3 4 5 6 7 8 9 101112131415161718192021 Figure 3. Map of the area ofthe Lcisey Shell Pit 1 A excavated in 1984 (see Fig. 1) showing the spatial distribution of associated Equus dentaries and lower teeth. Locations of isolated or unmatched specimens are not shown. Lines connect associated remains. D represents a cheektooth-bearing portion of the dentary, S a mandibular symphysis; and T an isolated lower tooth Compared to camelids and equids, tapir (Topirus haysii) elements are relatively uncommon, but matches were found between right and left mandibles and lower cheekteeth, as well as between mandibles and maxillae. Four individuals are represented by associated cranial and postcranial material. The majority of associated right and left mandibles were found within 2 m of one another, while maxillae and mandibles were separated by distances of up to 20 m. Definitely associated postcrania were spread over distances from 0.5 to 15 m. Based on femora, it is estimated that at least 5 proboscidean individuals are preserved at the Leisey lA locality (analysis of mandibles produced an MNI of 8, see Table 11). However, no evidence of association of proboscidean postcrania was evident with the exception of a pair of juvenile tibiae and humeri, separated by approximately 12 m and 25 4 respectively. Presence of unassociated PRATT & HULBERT: TERRESTRIAL MAMMAL TAPHONOMY OF LEISEY SHELL PIT 185 proboscidean remains raises the question as to the state in which terrestrial vertebrates reached the environment of deposition: primarily as carcasses; as isolated skeletal elements; or as a mixture of the two. It is possible that proboscidean carcasses decomposed and were scattered at the site in a subaqueous environment. However, the fact that proboscidean elements show signs of weathering (see section on weathering modification) indicates that they were exposed in a terrestrial setting. This finding, and the lack of skeletal association suggests that these large elements were transported separately to the environment of deposition, perhaps by moving water. This hypothesis will be dealt with in following sections. There are numerous matches between isolated shell elements of turtles or tortoises. In a few instances disarticulated but nearly complete carapaces or plastra were spread over an area of 16 m2 or less. At least 4 Trachemys scripta and 3 Hesperotestudo crassiscutata shells were found that were over 50% complete when reassembled. These shells must have reached the site as complete units, and subsequently disarticulated. Although it has been suggested that the site formed over a short period of time (Hulbert and Morgan 1989), there was obviously sufficient time for the turtle shells to disarticulate. Based on decomposition studies on Recent marine turtles, plastral and carapace elements may have begun to separate within two weeks of their arrival at the site (Meyer 1991). The fact that the terrestrial vertebrate remains show no degree of articulation suggests that they entered the site of deposition as individual elements. Although the agent of transport has not been identified, it is reasonable to speculate that bones from carcasses that decomposed and were scavenged further inland were carried into the embayment when heavy rains flooded freshwater creeks and surrounding low-lying areas. The presence of freshwater vertebrates and invertebrates, terrestrial gastropods, and transported grains of terrestrial pollen (Rich and Newsom this volume) support this suggestion. However, the occurrence of some associated skeletal remains (e.g., equid crania, tapir and sloth crania and postcrania, turtle carapaces) indicates that a significant fraction of the elements were introduced to the site of deposition as components of larger units, in some cases as complete or partial carcasses. The possible mechanisms by which complete remains may have been added to the assemblage will be discussed further in following sections. The large number of associated elements that were recovered in close proximity to one another essentially rules out the possibility that these bones had been disarticulated or broken prior their arrival at the site of deposition. Therefore, disarticulation, breakage, and scattering of the elements in question took place at the location where final deposition and formation of the bone bed occurred. The relatively limited range of distances between associated elements, as illustrated by the equid jaws and isolated teeth (Fig. 4) and the associated turtle shells, indicates that any transport mechanisms operating within the final depositional environment itself apparently were not highly effective. 186 BULLETIN FLORIDA MUSEUM NATURAL HISTORY VOL 37, PT. L No. 7 15- S C E O Jaw-Tooth 0 W 10- O Jaw-Sym ~ Jaw-Jaw 6. E n 2 51»]1 r 0 0000006000006666 4, 0 to 0 U) 0 w O M O w 0 10 0 60 1 -009 -098 ~ -0 0 6 r=z~--096 -000&-- CY¢\If)(9**lou)©(DA'- -0 9 0 1 < Distance Apart (cm) Figure 4. Histogram of distances scparating the associated Equus dentaries and lower teeth from Leisey 1 A shown in Figure 3. The final column includes all associated elements separated by more than 1025 cm. Note that while all but one pair of dentaries was separated by less than 6 m, it was not uncommon for teeth and symphyses to be separated from their corresponding dentary by more than 6 m Disarticulation and scattering of remains within the site may be attributed to several factors. Predators and scavengers have been reported to scatter elements of prey carcasses (Haynes 19804 1982; Blumenschine 1986). Given the marine nature of the site, scattering of this type in situ could only have taken place during periods of extreme low tide if terrestrial predators and scavengers were involved. It is also possible that aquatic scavengers such as sharks or crabs may have been responsible for some disarticulation and scattering of associated remains. Trampling by large herbivores has also been demonstrated to cause breakage and scattering of elements (Andrews and Cook 1985; Olsen and Shipman 1988; Fiorillo 1989), and Recent elephants are known to actually pick up, move, and carry remains of other elephants (Coe 1980; Conybeare and Haynes 1984; Haynes 1991). While bones in aquatic habitats may be trampled (Behrensmeyer and Boaz 1980; Haynes 1991), water depth at the site of trampling must be relatively shallow to allow the agents of trampling to wade rather than swim. This condition PRATT & HULBERT: TERRESTRIAL MAMMAL TAPHONOMY OF LEISEY SHELL Prr 187 evidently was met at the Leisey lA site (Hulbert and Morgan 1989). The third factor that could account for scattering of associated remains is the action of moving water (Voorhies 1969; Behrensmeyer 1975; Hanson 1980). It is fairly certain that tidal currents were features of the site of deposition, but it is not clear whether the current strengths were sufficient to cause movement of large, dense elements of large mammals. The possible role of moving water as well as other taphonomic factors will be considered more fully in the following sections. Bone Orientation In the case of disarticulated remains, orientations and positions of skeletal elements within a fossil site can provide information concerning agents responsible for bone concentration and dispersion. Presence of a preferred or dominant direction of bone orientation is often considered evidence that moving water was a feature of the depositional environment (Voorhies 1969; Hunt 1978, 1990; Shipman 1981; Pratt 1990), as water currents have the capacity to cause alignment of skeletal elements. There are several interpretations that may be made if a significant orientation pattern is lacking, including the absence of a current of sufficient duration and velocity to cause alignment. It is also possible that other factors that resulted in random patterns of bone dispersion were acting in opposition to the effects of moving water. During the field collection phase of this study, bearing and plunge data were obtained on 1447 fossil bones. However, as a number of elements do not routinely show predictable preferred axes of orientation in running water (Voorhies 1969; Pratt 1990), directional analysis was restricted to a total of 995 limb bones, mandibles, and rib fragments. These elements possess elongate axes that show predictable alignments relative to current direction and therefore are considered the most informative indicators of the presence and direction of paleocurrent (Voorhies 1969; Pratt 1990). Figure 5 is a mirror-image rose diagram of long axis orientations of these elements, and Figure 6 a stereographic projection of the bearing and plunge of 650 long bones. A %2 test for significance shows that there is no preferred directionality of long bone orientations (Table 1). As some types of long bones are more reliable indicators of current direction than others (Voorhies 1969; Pratt 1990), bearings of various types of elements were also examined for presence of any orientation pattern. For example, long axes of tibiae consistently align in the direction of a prevailing current, provided that water depth is sufficient to completely cover the bone (Voorhies 1969) and that current velocity is greater than the minimum speed necessary to move the bone into this position (Pratt 1990). At current velocities below those required to cause tibial 188 BULLETIN FLORIDA MUSEUM NATURAL HISTORY VOL„ 37, Fr. 1, No. 7 Number of Bones 0 40 Ill Figure 5. Mirror-image rose diagram showing the directional bearing of 995 long bones collected from Leisey Shell Pit lA in 1984. Eachinte,val represents 10°. alignment proximal ungulate phalanges may show a bimodal distributional pattern (Pratt 1990). Although over 25% of the tibiae (23) were aligned from 160(340) to 180(360) degrees east of north, the overall pattern of orientation was not significantly different from that of a uniform, non-preferred pattern (Table 1). Proximal phalanges also showed no preRrred axis oforientation (Table 1). Within the total area excavated, bones were primarily concentrated in two regions. One area was in the southwestern portion of the site, a roughly rectangular region bounded by squares B5 and B10 on the south and F5 and F10 at the north. The second area was located north of the first, from L10 to L14, and 010 to 014 (Fig. 3). Long bone orientations from each of these two regions of the site were analyzed to determine if elements within these separate concentrations PRAIT & HULBERT: TERRESTRIAL MAMMAL TAPHONOMY OF LEISEY SHELL PIT 189 were aligned to any extent. Although some elements from the more northern bone concentration show alignment in a northeasterly to southwesterly direction, in neither of the regions were orientations significantly different from a uniform, non- preferred pattern (Table 1). The relative positions of associated elements to one another also provides information concerning mechanisms responsible for dispersion of skeletal remains within the site. If separation of two associated elements occurred as a result of the lighter portion being removed from its counterpart by the action of water flowing in one predominant direction, then the direction in which this element moved from its original location would be indicative of current direction. Analysis of the directions in which associated jaws and teeth moved apart (Table 1), indicates that the dispersion pattern of these elements is not significantly different from a random pattern. The lack of a preferred orientation pattern does not necessarily imply that elements at the site were unaffected by moving water. Coastal manne environments, such as that represented at Leisey 14 are characterized by tidal currents. In a narrow channel, flow completely changes direction as tide goes in and out, although a 180° change in direction of flow could conceivably cause alignment of bones similar to that of a unidirectional current. However, in a large Table 1. Chi-squared values for bone orientations compared to those expected in a uniform, non-preferred orientation. All samples have 17 degrees of freedom (df.) except for phalanges and orientations between associated jaw elements. Data were combined in the latter two tests to accommodate for low numbers of bones pertreatment (d.f = 5). N = numberofelements, p =probability, considered significant at 0.05. Element N %2 P Long bones 995 23.43 0.11 Humeri 76 16.37 0.50 Radii 91 16.40 0.50 Femora 70 12.82 0.75 Tibiae 81 22.78 0.15 Metapodia (Camelidae) 89 14.76 0.65 Mandibles 216 13.33 0.60 Lnng bones, L10-014 203 25.50 0.07 Long bones B5-F10 332 14.99 0.65 Proboscidea, Long Bones 87 12.93 0.75 Phalanges 48 4.25 0.51 Orientations between Associated Jaw Elements 39 0.85 0.97 embayment, tidal currents are often extremely complex; direction of flow not only changes with different stages of the tidal cycle, but may also vary spatially and chronologically. Under such conditions, no preferred alignment of long bones 190 BULLETIN FLORIDA MUSEUM NATURAL HISTORY VOL 37, Fr. L No. 7 would be evident. It is also possible that a current of one predominant direction was present, but was not of sufficient velocity to cause alignment of the majority of large vertebrate limb elements, most of which belong to Voorhies transport groups III and II (Voorhies 1969; Korth 1979; Pratt 1990). Sediment (Hulbert and Morgan 1989; Hulbert and Morgan this volume) and pollen (Rich and Newsom this volume) analysis indicate that the environment of deposition at Leisey lA was generally low-energy. The low angles of plunge of the majority of the elements (Fig. 6) suggest that most bones came to rest on relatively level surfaces and were buried in those positions. There is no directional component to bones that plunged at angles of greater than 10 degrees (Table 1). Voorhies (1969) suggested that groups ofbones exhibiting dips in a predominant direction may have been imbricating upstream and dipping against the direction of a prevailing current. The lack of a preferred axis of plunge of bones at Leisey substantiates previous observations that a unidirectional or strong current was not a typical feature of the deposit. The only feature of the bone positions that indicates the presence of moving water is the side of the bone found facing upward in the field. When exposed to flowing water, some elements routinely assume a predictable, stable position (Voorhies 1969; Pratt 1990). One of the most reliable elements in this regard is the innominate (isolated) which often comes to rest with the lateral (acetabular) side up when exposed to running water (Pratt 1990). Of 47 camel and equid innominates from Leisey for which positional data were obtained significantly more (36) were recovered with the acetabular surface facing upward as opposed to facing downward (%2 = 6.65, p < 0.01 at 1 d.f.). This suggests that nowing water moved these elements into their fuvially-stable resting positions. The absence of a preferred orientation for long bones lends credence to a hypothesis that some of the scattering of bones within the site may have been caused by scavengers or by large herbivores. Recent experimental studies have shown that a random pattern of orientation results when large herbivores trample and kick bones (Fiorillo 1989). However, steeply plunging bones have also been described as a characteristic of trampled assemblages (Hill and Walker 1972; Behrensmeyer and Boaz 1980; Fiorillo 1989); some elements may end up in vertical orientations if they are driven lengthwise into soft sediment. No bones were found that plunged at angles of greater than 53° at Leisey lA (Fig. 6), suggesting either that trampling did not affect bone orientations and distributions, or that the sediment was relatively hard. Trampling will be considered further in the section on scratchmarks on bones. Small, light elements may have been PRKIT & HULBERT: TERRESTRIAL MAMMAL TAPHONOMY OF LEISEY SHELL PIT 191 N 4... - I 0 ./0.e : 0 3: ·*34 I./ 49 e :i.:i.74'/0, ': . ..•6 00 :. 00:%\. •*Ake · °47:01 51 0 0 e •1 ix. : 0 + i:/:• • . /69.... ..0 -0,8191.: .14*6 0 ..: 4;37, 1 S Figure 6. Stereonet projection ofthe bearing and plunge of 650 long bones collected from L.cisey Shell Pit lA in 1984. The vast majority of the fossils had plunges ofless than 15°. transported in various directions, or removed from the site entirely as the tide ebbed and flowed, while larger, denser elements were periodically scattered by the effects of trampling. Bone Modification Bone modification results from any activity that alters the original appearance of bone. The causative agents of bone modification may be instrumental in causing the animal's death (predators), or can cause alteration at any point between death of the animal and recovery of the fossils (e.g., scavenging, weathering, trampling, transport abrasion by sediments, epibiont activities, post-depositional breakage, reworking). 192 BULLETIN FLORIDA MUSEUM NATURAL HISTORY VOL 37, Fr. I, No. 7 Identifiable bones of large mammalian taxa (Edentata, Proboscidea, Artiodactyla, Perissodactyla, and Carnivora) were examined for modification, as were shell elements of the freshwater turtle Trachemys scripta and the large tortoise Hesperotestudo crassiscutata. The majority of identifiable elements examined in this phase of the study were collected from squares B4-B7, C5-C7, C9, D6, D7, D9, D10, and E6. The equid sample was supplemented with bones from squares Al-(4. The comparison of identifiable versus fragmentary bone was tabulated using material collected from B4, a square from which all fossil specimens were collected. Two hundred ninety-two unidentifiable fragments from both this square and from square D9 (another totally sampled square) were randomly selected for analysis of types of bone modification on fragmentary remains. Bone Completeness.- As a number of taphonomic processes are known to result in bone breakage, analysis of degrees and types of breakage may provide valuable insight into events that may have influenced the formation of a fossil assemblage. One very obvious feature of the Leisey 1 A site is that the majority of bone is composed of unidentifiable fragments. Table 2 compares the numbers and proporlions of complete (more than 90% of the element preserved) and partial (less than 90% of the element preserved) identifiable terrestrial mammalian elements with those of unidentifiable bone shards found in square B4. Unidentifiable fragments comprise over 90% of the bone sample from this region, although this value is probably inflated due to post-depositional compaction breakage and the presence of non-mammalian fragments. Carnivores and scavengers are known to cause bone splintering in Recent bone assemblages. Binford (1981) reported that over 90% of the specimens at wolf kill sites were unidentifiable fragments and splinters. Felids also cause splintering of bone (Brain 1981). In addition, other activities such as trampling of bones by large ungulates (Haynes 1983b; Fiorillo 1989) and to a lesser extent transport by running water (Behrensmeyer et al. 1989) may also cause breakage and presumably, splintering of bones. Therefore the presence of high percentages of bone splinters is not informative except in the context of sedimentary particles, as will be discussed later. The majority of identifiable elements (not including shards) both from 84 and from the site as a whole are complete rather than broken (Table 2). The presence of numerous complete podials, phalanges, and teeth is not surprising, as these compact, dense elements are not generally modified by taphonomic events (Behrensmeyer et al. 1989). Of the identifiable limb elements, vertebrae and mandibles, approximately two-thirds are incomplete, a situation similar to that reported by Behrensmeyer et al. (1989) at Miocene vertebrate fossil localities in Pakistan, in which 75% of the limb elements were incomplete. As pointed out by Behrensmeyer et al. (1989), the high incidence of breakage denotes activities of taphonomic agents and processes. PRATT & HULBERT: TERRESTRIAL MAMMAL TAPHONOMY OF LEISEY SHELL PIT 193 Table 2. Bone completeness of kisey lA mammals. Category A includes cranial elements. girdies. lintb elements, and vertebrae. Category B includes podials, phalanges, and teeth. A Elements from 84. N = 656. Identifiable elements Unidentifiable fragments Grand Complete Partial Total A B (all A) Identifiable >10 cm 5-10 cm <3cm total N 14 15 26 35 4 41 556 601 % 2.1 2.3 4.0 8.4 0.6 6.3 84.8 91.6 B. Elements from entire Lcisey sample ofidentifiable bone examined (not including fragments). N = 566. Complete Grand Partial A B Total (all A) N 142 164 306 260 % 25.1 29.0 54.1 45.9 C. Breakage types on partial limb elements ofmammals. Breakage type N % of total Spiral 8 5.5 Step 14 9.7 Combination Spiral/Step 30 20.7 Gnaw 17 11.7 Recent 76 52.4 Total 145 194 BULLETIN FLORIDA MUSEUM NATURAL HISTORY VOL 37, Fr. L No. 7 In this study, partial limb bones were categorized according to the type of breakage each possessed. The categories used are as follows: 1) Recent breakage: any type of post-fossilization (or prefossilization compression) breakage, including compaction breaks and collection- related breaks. 2) Spiral fractures: prefossilization breaks on the bone shaft that run oblique to the bone's long axis, as described by Binford (1981), Haynes (1983b), and Oliver (1989). 3) Step, or longitudinal fractures: prefossilization breakage that occurs as series of alternating longitudinal and transverse fractures to produce a step-like appearance on the broken edge of the bone (Hill 1979; Haynes 1983b). Breaks tend to follow the line of drying cracks that develop in the bone in the course of weathering. 4) Combination spiral/step fractures (interrupted spiral fractures): the line of the spiral fracture is broken or interrupted by step fractures (Hill 1979; Haynes 1983b). 5) Gnaw breakage: the broken edge of the bone has a scalloped or irregular outline caused by gnawing (Binford 1981; Haynes 19834 1983b). ~,~, _»-- FI , 571 / spiral (5.5%) u 1, ,»*=- i A--- gl step (9.7%) 1 1-1 -ttzy " =1'~in Il spiral/step (20.7%) -4 , _ ©_4 ,- 0 gnaw (11.7%) 0 Recenvunknown (52.4%) Figure 7. Pie diagram depicting relative frequency of different bone hakage types from a sample of 145 randomly selected, partial limb bones of large mammals from L£isey Shell Pit l A PRATr & HULBERT: TERRESTRIAL MAMMAL TAPHONOMY OF LEISEY SHELL PIT 195 The proportions of bones assigned to each category are listed in Table 2 and Figure 7. The most common type of break,ge seen on partial limb elements is post-depositional. The breaks appear to have been caused primarily by naturally- occurring sediment compaction, although it is also possible that heavy mining equipment and large trucks transversing the area contributed to bone brepkage prior to the site's discovery. Unfortunately, if the bone was partially fragmented in this manner, in many instances the fragments were not collected with the element and therefore could not be reattached to it. The result is that over half the partial mammal limbs in the collection have post-fossilization breaks, and the original type of pre-fossilization breakage cannot be discerned (Table 2, Fig. 7). The majority of pre-fossilization limb bone breaks are of the combination spiral/step fracture variety (Table 2). This type of breakage, as well as spiral fracturing is primarily caused either by carnivore/scavenger activity or by trampling. However, in the absence of other types of bone modification, breakage patterns alone cannot identify the causative agent of modification. The percentage of bones in an assemblage that are broken in spiral or combination fractures has been shown to vaiy with the size and condition of the bone, and the size and type of predator, or trampler. Fractures caused by carnivores in recent studies range from 8% of the total limb number to 100% (Haynes 1983b; Haynes and Stanford 1984). Up to 50% of long bones in an assemblage may be fractured by trampling (Haynes 1983b; Haynes and Stanford 1984; Agenbroad 1989). At Leisey, about 25% of the total mammalian limb elements examined were broken in this way, but this number is probably low as undoubtedly some elements in the Recent breakage category originally possessed prefossilization fractures. In addition, the majority of the elements examined (camelid and equid) are relatively large in size and were not as subject to breakage as smaller elements from deer and peccaries (Haynes 1983b). Regardless of whether bone breakage is caused by a carnivore or by trampling, in either case the exact breakage type is related to the amount of weathering the bone has experienced. "Fresh," or green bone tends to fracture in a spiral pattern (Behrensmeyer et al. 1989). Although bones broken in this manner are most often attributed to carnivores (Haynes 198Ob, 1983b; Hill 1980; Binford 1981) it can also occur when "fresh" or even slightly weathered bones are trampled (Myers et al. 1980; Haynes 1983b; Haynes and Stanford 1984; Fiorillo 1989). In addition bones deposited in moist or aquatic habitats retain characteristics of "fresh" bones for long periods of time and may fracture in a spiral fashion as long as they remain moist (Haynes and Stanford 1984). Bone that is slightly to moderately weathered most often breaks along combination spiral/step fractures when scavenged or trampled (Haynes 1980b, 1982, 1983b; Olsen and Shipman 1988). If a bone is broken after prolonged exposure to the forces of weathering, then a step fracture commonly results (Conybeare and Haynes 1984). Figure 8 shows that the type of breakage on mammal limb elements from Leisey lA is 196 BULLETIN FLORIDA MUSEUM NATURAL HISTORY VOL. 37, PT. 1, No. 7 80 - ~'iii'.1.illU 70 - /0 . . (60 50 - 40-&. - 1 1111,111 , 111'lilll"ll3 30- C 8 20- ~ 10 - -If.llilillillillillill i. llilllilli, lilillillliillith0 spiral combo step Breakage Type Figure 8. Percentage of broken mammalian limb bones subdivided by breakage type, showing features evident ofslight to moderate weathering at Leisey Shell Pit lA related to extent of weathering. Only one spirally fractured bone is even slightly weathered, while over half of the elements with combination spiral/step fractures are discernibly weathered, and 75% of the bones possessing step fractures are weathered to some degree. Investigation of surface features is required in order to ascertain if broken bones were primarily modified by carnivores or by trampling; however a few observations may be made concerning bone breakage. Based on breakage type and percentage of bones broken, it is not possible to determine the agent(s) of modification; however, the low percentage of "green" spiral breaks indicates that most bones were not broken by carnivores or trampling immediately following the death of the animal. It also indicates that any fresh bones immediately deposited in an aquatic habitat were probably not broken while wet. The presence of combination and step fractures indicates that most bones that were broken were probably modified after they had been exposed to some extent to drying and weathering, either by scavenging or by trampling. However, based solely on the breakage pattern, it is not possible to determine if these partial elements were broken prior to, or following their arrival at the site of deposition. It has been suggested in the section on articulation and association, and will be shown in PRATr & HULBERT: TERRESTRIAL MAMMAL TAI'HONOMY OF LEISEY SHELL Prr 197 following sections, that while some breakage did occur at the site of deposition, many elements were broken before they were deposited at the site of fossilization. Bone Surface Features.- Bones were examined for evidence of modification caused by predation and/or scavenging, exposure to subaerial weathering, water- borne sediments or other abrasive factors, and trampling by large herbivores. Degrees of modification were assessed using the following scales: 1) Evidence of predation was determined by characteristic breakage patterns, puncture marks, and depressed fractures (Haynes 19804 1983a; Hill 1980; Binford 1981). Light modification includes at least one of the following; one set of puncture marks, one chewed or gnawed edge, or obvious scoring marks: moderate modification is indicated by at least one of the following; two to five sets of puncture marks, two gnawed or chewed edges, and numerous scoring marks: heavily damaged bone has completely gnawed ends, epiphyses, or numerous tooth marks. These stages are roughly comparable to the utilization stages described by Haynes (1983a). Elements for which definite assignments to modification categories could not be made were scored as "possibly modified or unknown." 2) Weathering stages were assessed using the scale proposed by Behrensmeyer (1978) ranging from stage 0, or unmodified bone, to stage 5, completely weathered bone near final stages of destruction. 3) Evidence ofwater-wear, polishing or erosion was determined using a scale proposed by Pratt (1990) ranging from unmodified bone (stage 0) through a series of progressively more severely stages categorized as minimal (stage 1), moderate (stage 2), and finally extreme water- wear (stage 3) characterized by rounding and loss of diagnostic processes (comparable to "bone pebbles" of Andrews and Ersoy 1990). 4) Scratch marks were quantified by estimating the percentage of the bone surface covered by scratches, as described by Fiorillo (1989). Features Caused by Predation/Scavenging.- Bone damage caused by predation or scavenging is one of the most difficult types of modification to identify. Unless the type of damage is clearly recognizable, it is often not possible to definitely attribute it to a carnivore (Haynes 1983b). In addition, some carnivores, even in the process of breaking bone, may not leave identifiable marks on either the shaft or fragments (Haynes 198Ob, 1982, 1983a, 1985, 1988; Haynes and Stanford 1984). For these reasons, a large number of elements (in particular partial bones with post-depositional breaks) from Leisey lA could not be definitely categorized with regard to carnivore modification, and therefore are listed as "possibly modified." Figures 9-10 and Table 3 compare the extent of carnivore- 198 BULLETIN FLORIDA MUSEUM NATURAL HISTORY VOL 37, Fr. I. No. 7 Table 3. Extent ofcarnivorc damage to Leiscy elements. Man = mandibles; IX} = limbs/girdia; Ver = vertebrae; Pod = podials; GT = grand total; P = possible. A. Complete elements from entire Lcisey sample examined Modification Man IJG Ver Pod Total % Of StaRe N N N N GT None 5 48 41 164 258 84.3 Light 0 20 5 0 25 8.2 Moderate 0 0 0 0 0 0.0 Heaviy 0 0 0 0 0 0.0 P 14 9 0 0 23 7.5 GT 306 B. Partial elements from entire Lcisey sample examined. Modification Man IJG Vcr Pod Total % of Staae N N N N GT None 3 22 11 0 36 14.3 Light 0 24 16 0 40 15.9 Moderate 0 4 1 0 5 1.9 Heavy 0 0 0 0 0 0.0 P 21 101 49 0 171 67.9 GT 252 C. Unidentifi die fragments from 84 and D9. N % of GTModification Staae None 0 0.0 Light 4 1.4 Moderate 0 0.0 Heavy 0 0.0 P 288 98.6 Total 292 D. Carnivore damage to complete long bone and girdle elements ofselected tan Camelidae Equidac Proboscidea Modification Staee N % N % N % None 15 51.7 15 65.2 17 70.8 Light 6 20.7 7 30.4 7 29.2 Moderate 0 0 0 0 0 0 Heavy 0 0 0 0 0 0 P 8 27.6 1 4.3 0 0 Total 29 23 24 PRATT & HULBERT: TERRESTRIAL MAMMAL TAPHONOMY OF LEISEY SHELL PIT 199 Table 3 Continued. E. Carnivore damage to partial long bone and girdle elements of selected taxi Camelidae Equidac Proboscidea Modification Stage N % N % N % None 6 7.6 10 24.4 6 24.0 Light 14 17.7 3 12.2 4 16.0 Moderate 3 3.8 1 2.4 0 0.0 Heavy 0 0.0 0 0.0 0 0.0 P 56 70.9 25 61.0 13 60.0 Total 79 41 25 100- Cl complete bones (n = 142) 0 partial bones (n = 252) 90 - m fragments (n = 294) 80 - 70 - Pe rc en t 60 - 50 - 40 - 30 - 20 - 10 - ~ r-7 01,1,1 none light moderate heavy unknown Carnivore Damage Figure 9. Comparison ofthe degree ofbone modification by carnivores/scavengers on complete, partial, and fragmentary bones from Lzisey Shell Pit lA Elements included in this analysis were dentaries, limb bones, girdle elements. and vertebrae oflarge herbivores (Table 3). 200 BULLETIN FLORIDA MUSEUM NATURAL HISTORY VOL. 37, PT. f No. 7 A 80 - 70 - 0 Camelidae 0 Equidae 60- I Proboscidea none light moderate heavy unknown Carnivore Damage B 80 - 70 - 60 - 4- 50- 1 2 40- . 96 30- 1 1 none light moderate heavy unknown Carnivore Damage Figure 10. Comparison of the degree of bone modification by carnivores/scavengers on complete (A) and partial (B) bones from Lcisey Shell Pit 14 sorted by tan PRATT & HULBERT: TERRESTRIAL MAMMAL TAPHONOMY OF LEISEY SHELL Pri 201 inflicted damage on complete, partial, and fragmentary skeletal elements. Many of the identifiable bones of the most abundant large herbivores (proboscidean, camel, horse) show no evidence of being modified to any extent by carnivores. Over 83% of complete elements (including podials) are unmodified. Complete podials are least damaged (Table 3) which is not surprising considering that unless they are ingested (Haynes 1985; Behrensmeyer and Boaz 1980), they are not significantly modified by carnivores (Haynes 198Ob; Behrensmeyer et al. 1989). Most complete mandibles and isolated dentaries are broken into numerous pieces by post- depositional compaction, and it is difficult to assess modification, particularly breakage patterns, on those that have been repaired. In the case of equids, many mandibles have breaks at the diastema just posterior to the symphysis on either one or both sides. Camelid mandibles are more fragmentary, and separation of right and left dentaries may be either at the symphysis, or just posterior to it. Breakage of a herbivore jaw in the diastema region is a feature often attributed to carnivores, which pull off the upward4acing dentary of their prey in order to feed on its tongue (Behrensmeyer and Boaz 1980; Haynes 198Ob, 1982; Hill 1980). However, there are several features of Leisey mandibles that suggest that much of the jaw breakage was caused by other means. First of all, there is no clear evidence of carnivore- inflicted damage on jaws as described by Haynes (1980a). Many of the horse jaws are complete, with no pieces missing from the ascending ramus. Second, some mandibles possess pre-depositional breaks on both left and right mandibles, rather than on just one side. Third, in many cases, the break is a longitudinal fracture rather than the spiral fracture or v-shaped lever fracture that would result as the upward-facing jaw was forced open (see Haynes 1983b: fig. 4). In many instances, the breakage pattern indicates that the break occurred as a result of downward pressure, rather than upward tension. Finally, the close proximity of associated right and left mandibles to one another would mandate that these elements separated from one another at the site of deposition. Given the marine nature of the site, there are very few explanations that could account for mammalian predation within the actual environment of deposition. It is possible that carcasses that accumulated in the site were scavenged during low tide, or live animals trapped in quicksand or mud were attacked by predators. The latter explanation is not supported by the geology of the site (Morgan and Hulbert this volume). Approximately 17% of the vertebrae (including sacra) examined possess canine and cheektooth impressions on the centra, or gnawed processes. Unfortunately, the majority of vertebral remains are too fragmentary to allow positive identification of surface modification features. The clearest examples of carnivore modification are seen on girdle elements and long bones (Figs. 10-11; Tables 3-4). Approximately 25% of complete and nearly complete limb and girdle elements examined are modified (Fig. 10A), with damage consisting primarily of gnawed or bitten-off epiphyses and processes, depressed fractures and crenulated 202 BULLETIN FLORIDA MUSEUM NATURAL HISTORY VOL.~ 37, PT. I, No. 7 ir€sy /A OF 65858 Figure 11. Carnivore damage on mammalian fossils from Lkisey Shell Pit lA (A) UF 82379, a partial innominate of Equus with a crenulate ilium border and two puncture marks, probably caused by a canid. (B) UF 67575, the distal end of a camelid femur, showing numerous puncture marks. (C) UF 65858, the proximal end of a left ulna of a juvenile Paramytodon harlam with a punture mark probably caused by Arctoduspriztinus. Scale bar equals 3.2 cm for 4 1 cm for B, and 2.6 cm for C. Table 4. Carnivore damage to mammalian elements. Categories: 1= minimal punctures, gnawing, 2 = more extreme, 3 = bone cylinders. # of Modification bones in Modification Example Bone type in sample category description (UF #) Scapula 1 5 blade and cranial border chewed 83374 2 1 crenulated, scalloped edges Huments 1 4 head and trochlear ridge gnawed 82327 furrows, tooth impressions Radio·uina 1 6 olecranon chewed, moon-shaped 82485 and ulna bite Radius 1 2 distal epiphysis gnawed PR ATT & H U LBER T: TER R ESTR IAL M A M M A L TAPH O N O M Y O F LEISEY SH ELL PIT 203 Innominate 1 16 ilium blade chewed. punctured 82379 crenulated. scalloped edges tooth furrows on ischial crest Femur 1 5 head. trochanter, or condyle 82261 ridge gnawed, canine impressions on con dyles 3 2 epiphyses removed-bone cylinders Tibia 1 4 cnernial crest chewed, removed 81572 2 2 pro)amal epiphysis removed Vertebrae Axis 1 2 canine puncture holes 85240 Cervical 1 3 puncture holes, processes chewed 83483 Thoracic 1 7 puncture holes, processes chewed 84217 Lumbar 1 4 puncture holes, processes chewed 82271 Sacral 1 3 puncture holes, processes chewed 83368 2 1 partially gnawed 204 BULLETIN FLORIDA MUSEUM NATURAL HISTORY VOL 37, PT. I, No. 7 edges, and occasional tooth impression marks (punctures and furrows). Due to the lack of recognizable surface features on many of the partial elements, the percentage (17%) of incomplete girdles and limb bones possessing obvious carnivore damage is somewhat lower than that of more complete long bones. Most modification consists of gnawed ends, tooth punctures and scoring marks. Several bone cylinders (sensu Binford 1981) were recovered, but not from the completely sampled squares that form the basis of the taphonomic analysis. If prefossilization breakage of partial elements was caused primarily by carnivores or scavengers, it seems reasonable to suggest that many fractured elements would possess features indicative of carnivore modification. About 20% of limb elements with prefossilization fractures (recent breaks not included) possess surface features interpreted as carnivore-caused. This value is much lower than the value reported by Haynes (1983b) in which 70% of broken limb elements at a bison kill site were visibly modified by wolves. While the Leisey value might be greater if a number of the unidentifiable bones belong in this category, the value would probably not approach the 70% level. The relationship of carnivore modification to breakage may also be examined by comparing the ratio of carnivore-modified to unmodified complete elements with the ratio of carnivore modified to non-modified partial elements. The "possibly modified" bones are not included in these calculations. The ratio of modified to unmodified complete elements is 0.42, while the ratio for partial elements is 1.27. The differences between these ratios for complete and partial elements suggests that bone breakage was in some way related to carnivore modification, either directly or indirectly. Some elements possessing evidence of carnivore modification may have actually been broken by other means, such as trampling, after being initially weakened by chewing and gnawing (Haynes 1983b). Bones were also analyzed to determine the relationship between stage of weathering and presence of carnivore-inflicted damage. Although reports in the literature suggest that most scavenging and accompanying breakage of limb bones generally occurs before the bones have begun to weather extensively (e.g., Haynes 1982), some scavenging may take place as much as 6 months after death of the prey, when elements have entered visible stages of weathering (Haynes 1982: fig. 3). Fossils showing evidence of exposure to weathering were accessible to scavengers for the period of time that these elements retained some nutritive value. In contrast, some, if not all, of the unweathered elements may have been made unavailable to terrestrial scavengers, either by immediate entry into the subaqueous environment of deposition, or by initially coming to rest in a moist habitat with dense ground cover (see following section on bone weathering). The ratio of modified to non-modified bone increases with weathering stage for both complete and partial elements (Fig. 12), showing that remains exposed to weathering also PRAIT & HULBERT: TERRESTRIAL MAMMAL TAPHONOMY OF LEISEY SHELL Prr 205 1.8 - 0 Complete Elements / Partial Elements 1.6- 1.4- P ro po rti on o f C ar i vo re -M od ifi ed 1.2- 1.0- o U nm od ifi ed B on e 0.8 - 0.6 - 0.4 - 0.2 - 0.0 0 1 2 Weathering Stage Figure 12. The relationship between weathering stage and carnivore modification on complete and partial bones from Leisey Shell Pit 1 A experienced more damage due to scavenging than bones that were not obviously weathered. This is indicative of different taphonomic histories for the unweathered, unchewed elements and the modified bones. The relative rarity of highly utilized or modified elements at Leisey may also be in part explained by the lack ofbone-crushing specialists at the site. Among the mammalian carnivores at Leisey, there are only a few that were large enough to have to preyed upon large herbivores and have caused bone modification; two felids, Smilodon gracilis and Homotherium sp., one cad< Canis armbrusteri, and an ursid, Arctodus pristinus. One problem in assessing the amount of destruction that these carnivores may have caused is that they are all extinct species, and in three cases extinct genera. Therefore, direct comparisons with living analogs cannot be made. However, some basic inferences can be drawn based on the predator's dental morphology and types of modification seen on bones of prey. In terms of relative numbers, the most abundant carnivore at Leisey lA was the saber-toothed felid Smi/odon gracilis (Table 11). While it has been shown that 206 BULLETIN FLORIDA MUSEUM NATURAL HISTORY VOL. 37, PT. I, No. 7 this cat was probably an active predator (Gonyea 1976), capable of killing camelids and equids, it is very unlikely that this species, or any other saber-toothed predator, was able to modifying bone to any great degree (Brain 1981). Van Valkenburgh et al. (1990) showed that Smilodon and other saber-cats avoided bone contact in order to protect their sabers from breakage. Although some of the types of bone modification at Leisey 14 such as bitten-off epiphyses, are similar to types of damage caused by Recent felids such as lions (Haynes 1983a; 1985), it is also possible that modification was caused by another predator such as a bear, which, as pointed out by Haynes (1985) may cause bone damage indistinguishable from felid damage. Homotherium sp. is also known from the site, although remains of this felid are rare. Homotherium was probably capable of killing large mammalian prey, but it is doubtful that this saber-toothed felid caused extensive bone modification. The majority of carnivore-inflicted bone damage appears similar to that caused by canids, in particular wolves (Haynes 1980a; 198Ob; 1983a; Binford 1.5- 1.4- I 1.3- 0 1.2- / 1.1 A I 1.0- IP Bi te L en gt h (c m ) ¤ 0.9- a O camel0.8 - O horse0.7- ~0~ I proboscidea A sloth 0.6 -0. 5 0.5-¤ 0 0.4 - · , · , - , · , · , · , · , · , · , · , · 1 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5 Bite Width (cm) Figure 13. Size distribution of the length and width of puncture impressions ("bite marks") on large mammalian herbivores from Leisey Shell Pit 1 A caused by the canine teeth of predators or scavengers. Dashed lines indicate the ' (basal) canine dimensions of Canis annbrusten. Larger marks fall within the range of size of the canines of Arcrodus pristinus. The larger bite marks (presumably caused by A. pristinus) were found predominantly on prey species with greater body size (ground sloths and proboscidcans), while C armbruster, apparently preyed or scavenged primarily on equids and camelids. PRATT & HULBERT: TERRESTRIAL MAMMAL TAPHONOMY OF LEISEY SHELL PIT 207 1981). The only canid at the site of sufficient size to have caused significant bone modification was Canis armbrusteri. This canid was slightly larger in size than C /upus, the extant gray wolf, and had an extremely wolf-like dental morphology (Van Valkenburgh 1991). It was probably able to kiltlarge herbivores, particularly if it was a pack hunter. In addition, C armbrusteri may have scavenged carcasses, including those abandoned by Smilodon graci/is, although it also had limited bone- crushing or bone-breaking abilities (Van Valkenburgh 1991). Much of the recognizable carnivore damage on camelid and equid remains includes canine and cheektooth impressions and furrows, the size and spacing of which match the dentition of C armbrusteri (Figs. 11, 13). The semi-lunar shaped bites on olecranon processes (such as UF 82485) correspond well with the size and shape of the anterior toothrow in this species. The final large mammalian carnivore represented at the Leisey lA locality is the ursid Arctodus pristinus. The diet of the better-known late Pleistocene Arctodus simus has been discussed at length by several workers. Kurt6n (1967) and Van Valkenburgh (1988) suggested that it was a specialized carnivore or scavenger. Emslie and Czaplewski (1985) noted similarities of the dental morphology and jaw structure of this ursid to more herbivorous forms, but suggested that it may have been primarily omnivorous with scavenging tendencies. Agenbroad and Mead (1986) and Gillette and Madsen (1992) provided evidence that A. simus was scavenging mammoth carcasses. Kurt6n (1967) suggested that the older, smaller A. pristinus may have been more omnivorous and less predaceous thanA. simus. However, carnivore-damaged elements of proboscideans and sloths from Leisey indicate that A. pristinus, while perhaps not capable of killing adult proboscideans, was definitely feeding on their carcasses and in some instances damaging bones. Canine and cheektooth impressions of large size are perfect matches for the dentition measurements of A. pristinus from the site (Figs. 11, 13). The majority of damage to sloth bones is seen on specimens of Nothrotheriops texanus, a small ground sloth. Nearly all of the identifiable tooth marks identified on N texanus elements were apparently caused by Arctodus. However, at this point it cannot be determined if A. pristinus was an active predator, or only a scavenger. It is interesting to note that no true bone-crushing specialists are present at Leisey. This is the only gap of this "niche" in Florida over the last 12 million years. At earlier Irvingtonian sites in Florida, such as Inglis 14 the hyaenid Chasmaporthetes ossgragus is present (Berta 1981). Miocene and Pliocene sites almost always contain borophagine canids (e.g., Webb et al. 1981; Van Valkenburgh 1991), widely regarded as ecological analogs of modern hyenas. Recent hyenas are known to cause extensive damage to bones from both fresh- caught and scavenged carcasses (Brain 1980; Haynes 19832; Blumenschine 1989). Later Pleistocene faunas contain the dire wolf, Canis dirus, which may also have had bone-crushing capabilities (Haynes 1983b, 1985; Van Valkenburgh et al. 208 BULLETIN FLORIDA MUSEUM NATURAL HISTORY VOL 37, Fr. L No. 7 1990). The absence of a bone-crushing specialist from the Leisey lA fauna may account in part for the low degree of modification seen on prey elements Results of the preceding analysis indicate that predation and scavenging were taphonomic events that to some extent influenced and contributed to the Leisey lA terrestrial mammal portion of the bone assemblage. The relative importance of these factors in forming the bone accumulation can be assessed as follows. At Leisey, innominate bones are the most abundant elements of equid and camel (see following section on bone representation). As evidence of predation is relatively easy to discern on this element, the ratio of carnivore-modified to intact innominates can be used to indicate the percentage of individuals of these tan either killed by carnivores or scavenged after death. Based on the innominates examined (N = 20 camelid, 15 equid), an estimated 25% of the camels and 36% of the equids found in the deposit were either killed by carnivores and/or fed upon by scavengers. Based on femora (the most common element of proboscideans [N =10] and sloths [N = 71), about 33% of the proboscideans, and 43% of the small ground sloth Nothrotheriops were modified to some extent by carnivores. While these percentages may seem high, it is important to stress that this is the percentage of dead animals, not the percentage of the total living population. Given that the overall death rate for Recent zebra is about 22% per annum (Western 1980), and that the predation-caused death rate is approximately 8 to 10% per annum, or in some cases as high as 16% (Foster and Coe 1968; Haynes 1988), then the remaining deaths (normally, 12-14% of the total population per annum) are not caused by predators. Therefore, in an ideal attritional death assemblage it is expected that from 36 to 45% of prey individuals would show evidence of predation. Scavenging of carcasses of individuals that died of natural causes would lead to a greater than predicted percentage of carnivore-modified bone, but in an attritional assemblage this increase would probably be offset by two factors: a greater tendency for bone damaged by carnivores to be subsequently completely destroyed, and by periodic occurrences of increased mortality due to drought or other catastrophic events (Coe 1980; Conybeare and Haynes 1984; Haynes 1988). The likelihood that minor catastrophic events may have contributed to the Leisey lA bone accumulation is supported by several lines of evidence, including the types and levels of bone modification (Table 4) which are generally consistent with "light to moderate" utilization of carcasses (Behrensmeyer and Boaz 1980; Haynes 19804 198Ob; Conybeare and Haynes 1984; Blumenschine 1989). In Recent environments, low utilization is seen during periods of high prey-to-predator densities, when prey species are weakened and die in large numbers as a result of drought or other environmental factors, and thus provide more food than carnivores can exploit (Behrensmeyer and Boaz 1980; Coe 1980; Haynes 198Ob). This appears to have been the case for the Leisey camelids, for which the proportion of modified individuals is lower than expected, and lower than the other large herbivore tan. The percentages of other prey species that were either killed by carnivores or scavenged following death fall within the ranges PRATT & HULBERT: TERRESTRIAL MAMMAL TAPHONOMY OF LEISEY SHELL PIT 209 expected in a typical population. An abnormally high amount of carnivore activity is not indicated for any of the tan examined. Therefore it is unlikely that the deposit represents primarily a carnivore-formed accumulation. It is also unlikely that the site was a mammal trap, such as a mudhole or quicksand, that attracted a high number of predators to the region. This finding is in agreement with the results of the population dynamics analysis. Bones of terrestrial mammals accumulated attritionally as a result of both natural deaths and predation, with perhaps some input from catastrophic events. Weathering.- Figure 14 and Table 5 illustrate the extent of weathering on m,mm,lian elements from Leiscy lA. There is a significantly greater incidence of weathering on partial, identifiable bones as opposed to complete elements (Table 8). This result agrees with the observation made by Haynes (1991) that breakage increases with the amount of time a bone is exposed to the effects of weathering. The majority of bones possessing weathering featurds are only slightly modified; suggesting tlut conditions favoring extensive weathering (extremely dry climate, long periods of subaerial exposure) were not in operation. The extent of weathering overall is consistent with that ofbone that has weathered for a period of four years or less (Behrensmeyer 1978; Andrews and Cook 1985; Fiorillo 1989), or bone that has weathered in relatively protected environments, including damp habitats, or regions with ground cover. It has been shown that the amount of ground cover as well as microclimatic conditions influence the degree of bone modification caused by exposure to climatic factors (Behrensmeyer 1978; Brain 1980; Conybeare and Haynes 1984; Haynes 1991). If bone modification due to weathering is considered for each of the major large herbivore tan, a more informative pattern emerges (Fig. 15, Table 8). For all three taxa examined, a greater but non-significant proportion of partial limb and girdle elements are weathered compared to complete bones. In between-taxa comparisons, a significantly greater proportion of equid and elephant remains, both complete and partial, possess exposure-related modification than do camelid bones. The difference in extent of weathering between equids and proboscideans is not significant (Table 8). The greater incidence of weathering on equid and proboscidean remains as opposed to camels suggests different taphonomic histories for these groups. A larger proportion of camelid remains as opposed to both equids and proboscideans were either deposited immediately as carcasses into the aquatic depositional environment, or decomposed in terrestrial areas where rates of weathering were retarded for various reasons. The former explanation seems more acceptable, and it is supported by other data, as discussed previously (see section on carnivore modification). As bones of all taxa were found in close proximity to one another (Fig. 2), similar patterns of weathering for all taxa would be expected if weathering exposure had occurred at the site of deposition. The differences in degrees of weathering between tan indicate that weathering did not occur at the location 210 BULLETIN FLORIDA MUSEUM NATURAL HISTORY VOL 37, PT. I, No. 7 Table 5. Weathering modification on Lcisey mammalian elements. Stages described in the text Man = mandibles; IX} = limbs and girdlcs; Vcr = vertebrae; Pod = podials; T = total; GT = grand total. A Weathering stages on complete elements from entire LEisey sample examined Man L/G Vcr Pod % of Stage N N N N Total GT 0 6 47 33 155 241 78.7 1 8 24 12 9 53 17.3 2 3 6 1 0 12 4.0 3 0 0 0 0 0 0.0 GT 306 B. Weathering stages on partial elements from entire Leisey sample examined. Man IJG Vcr Pod % of Stage NNWN Total GT 0 6 76 40 0 122 48.4 1 14 54 34 0 102 40.5 2 2 19 3 0 24 9.3 3 2 2 0 0 4 1.6 GT 232 C. Weathering stages on unidentifiable fragments from squares 84 and D9. % of Stage N GT 0 176 60.2 1 84 28.8 2 32 11.0 3 0 0.0 T 292 D. Weathering stages on complete long bones and girdle elements of selected taxi Weathering Camel Horse Proboscidea Stage N % N % N % 0 23 79.3 12 52.2 11 45.8 1 5 17.2 9 39.1 10 41.7 2 1 3.4 2 8.7 3 12.5 3 0.0 0 0.0 0 0.0 T 29 23 24 PRATT & HULBERT: TERRESTRIAL MAMMAL TAPHONOMY OF LEISEY SHELL PIT 211 Table 5 Continued. E. Weathering stages on partial long bones and girdle elements ofselected taxa. Weathering Camel Horse Proboscidea Stage N % N % N % 0 49 62.0 16 39.0 7 28.0 1 22 27.8 19 46.3 11 44.0 2 8 10.1 6 8.7 5 20.0 3 0 0.0 0 0.0 2 8.0 T 79 41 25 70 - I complete bones 60 - O partial bones S fragments 50 - 40 - Pe rc en t 30 - :*.. .*:.:.: 20 - im·: 10- ri01,,, 0123 Weathering Stage Figure 14. Percentage of complete, partial, and fragmentary mammalian bones from Lcisey Shell Pit 1 A showing different weathering stages. where the bones were finally deposited. The more weathered elements undoubtedly represent carcasses that were skeletonized and weathered elsewhere before the bones were introduced into the site (see previous sections on disarticulation and orientation). 212 BULLETIN FLORIDA MUSEUM NATURAL HISTORY VOL- 37, PT. I, No. 7 A 80- .F 0 CamelWae 0 Equkdae ~ ProboscIdea 60 - ~ so - ~ Pe rc en t 40 ~ ~ 30-20~ ~102 0 %FFIh0 0 1 2 3 WeatherIng Stage B 80 - 70 - 60~ ~ so - ~ 10 - 0 1 2 3 WeatherIng Stage Figure 15. Percentage of complete (A) and partial (B) limb bones and girdle elements from L£isey Shell Pit 1 4 sorted by tan showing different weathering stages (Table 5). PRATI' & HULBERT: TERRESTRIAL MAMMAL TAPHONOMY OF LEISEY SHELL PIT 213 Surface Erosion.- Figure 16 and Table 6 show the proportions ofmnmmalian elements and podials possessing some degree of polishing or rounding. This type of modification is referred to by Olsen and Shipman (1988) as surface erosion, and on mammalian elements from Leisey 1 A consists primarily of slight rounding and smoothing of processes and polishing of broken edges. As is the case with other types of modification discussed previously, the majority of limb and girdle elements (62%) are unmodified (Fig. 17). A significantly greaterproportion of complete bones show this type of modification than partial elements and unidentifiable fragments (Table 8). The large percentage of fragments lacking any evidence of polishing is due mainly to the fact that a number of these pieces resulted from post-depositional breakage, however, even polished broken edges of fragments do not show any appreciable degree of rounding of sharp edges. Iffragments are eliminated from the comparison, the percentage of abr*led partial elements is significantly greater compared to complete elements. This differenceindicates that partial elements were more frequently exposed to agents of erosion. It is also possible that this finding reflects the fact that it is easier to discernpolishing on a broken edge than on a complete element. Surface erosion is most often attributed to polishing caused by water-borne particles in a fluvial environment however, polishing and rounding of bones may also occur as a result of trampling (Behrensmeyer 1978; Brain 1981; Haynes and Stanford 1984; Oliver 1989; Haynes 1991) or licking and manipulation by carnivores (Binford 1981; Haynes and Stanford 1984; Haynes 1991). In the case of the skeletal elements from Leisey lA, there is evidence that the primary agent of polishing was sediment entrained in running water. Examination of podialelements shows that almost half (significantly more than the percentage of abraded complete limb elements) are rounded to some degree (Table 8). This modification most likely occurred as these small elements were rolled from place to place within the site by the action of moving water. In addition, among the limb elements, there is no significant difference in levels of abrasive wear between taxa (Table 8). Within taxa, although in all cases a greater proportion of partial as opposed to complete elements are abraded (Fig. 17), the difference is significant only in the Equidae, and is due to the large number of complete elements that do not show evidence of any abrasion. The general similarities in extent of bone erosion between taxa suggests that the agent or agents responsible for the modification were acting at the site of deposition and affecting limb bones of all mammalian taxa more or less equally. Given the aquatic nature of the site, it is reasonable to assume that water-related polishing occurred. Although some polishing of bones may be attributed to trampling of skeletal remains in the depositional environment, there is no apparent correlation between the amount of carnivore damage and amount of abrasion. It therefore is unlikely that carnivores or scavengers were main agents of surface abrasion and polishing. 214 BULLETIN FLORIDA MUSEUM NATURAL HISTORY VOL 37, Fr. I, No. 7 Table 6. Surface erosion or watenvear on Lkisey mammalian elements. Abbreviations as in Table 5. A Erosion or watcnvcar on complete elements from entire Leisey sample examined Man llc Vcr Pod Total % of 6T Stage N N N N None 12 55 31 87 185 60.5 Minimal 7 21 14 68 110 35.9 Moderate 0 1 1 9 113.6 Extreme 0 0 0 0 0 0.0 GT 306 B. Erosion or watenvear on partial elements from entire Lkisey sample examined Man L/G Vcr Pod Total % of GT Stage N N N N None 10 74 47 0 131 52.0 Minimal 14 61 28 0 103 40.8 Moderate 0 15 2 0 17 6.8 Extreme 0 1 0 0 1 0.4 GT 252 C. Erosion or waterwear on unidentifiable fragments from squares 84 and D9. Stage N % of GT None 208 71.2 Minimal 84 28.8 Moderate 0 0.0 Extreme 0 0.0 T 292 D. Erosion or waterwear on complete long bones and girdle elements of selected tan. Weathering Camel Horse Proboscidea Stage N % N % N % None 18 62.1 18 78.3 18 75.0 Minimal 11 37.9 5 21.7 5 20.8 Moderate 0 0.0 0 0.0 1 4.2 Extreme 0 0.0 0 0.0 0 0.0 T 29 23 24 PRATT & HULBERT: TERRESTRIAL MAMMAL TAPHONOMY OF LEISEY SHELL PIT 215 Table 6 Continued E. Erosion or waterwear on partial long bones and girdie elements ofselected taxi Weathering Camel Horse Proboscidea Stage N % N % N % None 34 43.0 21 51.2 15 60.0 Minimal 15 44.3 17 41.3 7 28.0 Moderate 10 12.7 3 7.3 2 8.0 Extreme 0 0.0 0 0.0 1 4.0 T 79 41 25 80 - 70 - I complete bones 0 partial bones60 - m fragments 0 podials 50 - Pe rc en t 40 - 30 - 20 - 10 - 0 07n none minimal moderate extreme Waterwear Stage Figure 16. Percentage of mammalian fossils from Leisey Shell Pit LA, sorted by completeness showing evidence of different degrees ofwatenvear (surface erosion; Table 6). 216 BULLETIN FLORIDA MUSEUM NATURAL HISTORY VOL 37, FT. I, No. 7 A 80- 70 - 0 Carnelidae 60- 0 Equidae I Proboscidea 50- P er ce nt P er ce nt 40- 30- 20- 10- l/0 none minimal moderate extreme Waterwear B 80- 70 - 60- 50- 40- 30 - 20- none minimal moderate extreme Waterwear Figure 17. Percentage of complete (A) and partial (B) limb bones and $die elements from Leisey Shell Pit 1 4 sorted by taxa, showing different waterwear stages (Table 6). Based on the levels of surface erosion seen on terrestrial elements from Leisey 14 a number of conclusions may be drawn concerning bone transport into the depositional environment, and the length of time bones were exposed before burial. The absence of highly polished or rounded bones typically recovered from nuvial PRATT& HULBERT: TERRESTRIAL MAMMAL TAPHONOMY OF LEISEY SHELL PIT 217 sites such as the Verdigree quarry (Voorhies 1969), the Love site (Webb et al. 1981) and the Pgwlar Turkey, bone assemblage (Andrews and Ersoy 1990), indicate that elements were not carried long distances by a high energy fluvial system. Bones may have been transported into the site by the action of running water, but many came from regions near to the site of deposition perhaps during brief periods when current velocities of feeder streams were greatly augmented by floods or stomis. Once the bones reached the depositional environment, they evidently were not exposed to strong current action for long periods of time. It is likely that moderate currents caused the majority of polishing before the bones were buried in the sediment, an event that occurred not long after the bones were initially introduced into the site. These conclusions are supported by features of the site geology, palynology, and invertebrate fauna (Morgan and Hulbert; Portell et al.; and Rich and Newsom, this volume). Scratch Marks.- The most obvious type of physical damage to the elements from Leisey lA is fairly extensive scratching on the surface of many bones (Fig. 18; Table 7). The scratches appear as relatively shallow fine striations, from less than a centimeter to several centimeters in length. Commonly, a series of scratch marks of similar size, depth and direction are present on a portion of the bone. On extensively scratched bone, much of the surface is criss-crossed by numerous, multidirectional series of scratches. The marks are most evident on flat, shiny surfaces, and on limb bones the extent of surface covered ranges from 0 to 8054 with an average scratched surface area of about 35% (Fig. 19). Podials, bodies of vertebrae, and articular surfaces of limb elements are not noticeably scratched. The marks are very similar to scratches produced by sedimentary abrasion (Olsen and Shipman 1988: fig. 5), which occurs as abrasive sediment either moves across a bone surface, or as a bone is dragged across sediment. The matrix must be scraped against the bone surface with some degree of force in order to produce scratches. Sedimentary abrasion is generally ascribed to three types of agents; trampling by large hertivores (Andrews and Cook 1985: fig. 33; Haynes and Stanford 1984: fig. 6; Behrensmeyer et al. 1986; Fiorillo 1989; Haynes 1991: figs. 4.36-4.40), moving water (Behrensmeyer et al. 1989, Olsen and Shipman 1988), and post-depositional, post-fossilizational shifting of bones and sediment (Olsen and Shipman 1988; Behrensmeyer et al. 1989). The fossil matrix surrounding the bones at Leisey lA consists primarily of sharp fragments of mollusk shells and angular quartz sand which both would have been abrasive enough to cause the scratchmarks in question. In order to determine if this was in fact the case, it is necessary to ascertain whether scratchmarks were applied before or after the remains reached the site of deposition. This aspect of the study is restricted to long bones, as other elements seldom show evidence of scratching. -1 218 BULLETIN FLORIDA MUSEUM NATURAL HISTORY VOL 37, Fr. I, No. 7 Table 7. Scratchmarks on Lcisey mammalian elements. Man = mandibles; L/G = limbs and girdles, Ver = vertebrae; Pod = podials; T = total; GT = grand total. A Scratchmarks on complete elements from entire Leisey sample Surface Man I./G Vcr Pod Total % of Coverage N N N N GT <5% 2 24 40 139 205 66.9 >5% 17 53 6 25 101 33.1 GT 306 B. Scratchma,ks on partial elements from entire Izisey sample examined. Surface Man UG Vcr Pod Total % of Coverage N N N N GT <5% 7 72 68 0 147 58.3 >5% 17 79 9 0 105 41.7 GT 252 C. Scratchmarks on complete long bones and girdie elements ofselected taxi Surface Camel Horse Proboscidea Coverage N % N % N % <5% 8 27.6 11 47.8 4 16.7 >5% 21 72.4 12 52.2 20 83.3 T 29 23 24 D. ScratchmarkB on partial long bones and girdle elements of selected taxi Surface Camel Horse Proboscidea Coverage N % N % N % <5% 38 48.1 22 53.7 9 36.0 >5% 41 51.9 19 46.3 16 64.0 T 79 41 25.0 PRATT & HULBERT: TERRESTRIAL MAMMAL TAPHONOMY OF LEISEY SHELL PIT 219 Figure 18. Ventral view of a portion of the plastron of the turtle Trachemys scripta from Leisey Shell Pit 1 X showing the size and extent of scratch marks found on many of the bones. The arrow indicates the suture between the hyoplastron, found in square F5, and the hypoplastron which was found in F6. The scratch marks do not cross this suture, indicating that they were produced at the site of deposition, and not in son)e other environment and subsequently transported to the L~isey site. Scale bar is 0.5 cm 220 BULLETIN FLORIDA MUSEUM NATURAL HISTORY VOL- 37, PT. I, No. 7 50 - 40 - ~ Complete Limbs 0 Partial Limbs P er ce nt 30 - 20 - 10 - ° i <6 6-20 21-40 41-60 61-80 Percentage of Surface Area Scratched Figure 19. Percentage ofcomplete and partial mammalian limb bones and girdle elements from L,eisey Shell Pit 1 A showing proportion ofthcir surface area covered by scratch marks. Samples consisted of 77 complete and 151 partial specimern Several lines of evidence suggest that bones were scratched after reaching the fossil site. First, if scratches had been applied in a terrestrial setting, bones exposed for the longest periods of time to the causative agents should be the most scratched. It has been shown previously that an increase in terrestrial scavenger damage is correlated with increased bone weathering however, such is not the case for scratchmarks (Table 8). In fact, a greater percentage of complete bones in weathering stages 1 or 2 are unmarked when compared with complete, unweathered elements, although these differences are not significant according to a %2 test (Table 8). Fiorillo (1989) noted that scratches on bone surfaces may be obscured or obliterated in late stages of weathering, but weathering damage to bones in this study is so slight that scratches are still clearly visible. Second, in most cases, there is no significant difference in presence or absence of scratchmarks on bones of different herbivore tan (Table 8). A notable exception is seen for complete probosci(lean elements, a significantly greater proportion of which are scratched when compared to complete elements of camelids or equids. PRATT & HULBERT: TERRESTRIAL MAMMAL TAPHONOMY OF LEISEY SHELL PIT 221 A significantly greater proportion of complete elements are scratched compared to partial elements for all taxa combined (Table 8; Fig. 20). This may be due to the fact that flat, easily scratched surfaces are more common on complete elements than partial elements. The overall similarities between taxa and between whole and partial bones in terms of presence or absence of scratchmarks suggests that most of the bones were subjected to scratching under similar conditions, after they had reached a similar environment (i.e. the environment of deposition). Complete proboscidean limbs may be more scratched than limb elements of smaller hetbivores particularly if the latter were not buried quickly due to their large size, and thus were subjected to this type of modification in the aquatic environment for a longer period of time than smaller elements. Further support along these lines is provided by an investigation of elements of organisms that lived in completely different environments than the terrestrial herbivores. Isolated plastron and carapace elements of the freshwater turtle Trachemys were examined for scratches. Although scratchmarks could not be discerned on the majority of the elements examined, in part due to the rugose surface of the carapace elements, nearly 30% (primarily smooth-surfaced piastral elements) were scratched to some degree, and some were scratched extensively (45% of surface covered). This issue is complicated somewhat by the fact that a number of turtle shells undoubtedly were introduced into the depositional environment in a complete state, and therefore all surfaces of the shell were not equally exposed to agents of scratching. Nevertheless, the presence of scratches on nearly a third of turtle elements examined suggests that they were scratched in the same environment where large terrestrial herbivore bones were scratched, that is, the site ofbone accumulation. Third, if bones were scratched in a terrestrial environment and then transported to the environment of deposition, no correlation should be seen between the side facing up in the field and the most scratched surface, given that final resting position of an isolated bone after transport is not related to its position in a decaying, articulated carcass prior to transport. Such is not case for Leisey 1 A limb elements and mandibles examined; of bones having unequal surface scratch distributions, a significantly greater number (65 of 71 elements) were recovered with the more scratched side facing upward, according to a %2 test for significance (p < 0.001). This finding suggests that the upward-facing surface of the bone was most prone to scratching, and that many bones did not move appreciably after they were scratched. Therefore it is unlikely that bones were transported any great distance from the location where they were scratched. A final, convincing line of evidence for in situ scratching is provided by examination of those elements that reached the final depositional environment as complete bones and were broken at a later point in time (see preceding section on articulation and association). When contacts between the broken pieces are 222 BULLETIN FLORIDA MUSEUM NATURAL HISTORY VOL. 37, PT. I. No. 7 Table 8. Chi-squared test results for differences in bone surface modification factors. Statistically significant differences are signified by: 0 - significant at p 5 0.05. ** = significant at P 5 0.01. 2 values not so indicated have p> 0.05. There is 1 degree offreedom in all cases. A.Weathering Complete Bones vs. Identifiable Paftial/ Bones/ all taxa combined Podials Included Podials Excluded 55.98" 5.38* Complete Limb and Girdle Bones vs. Partial Limb and Girdle Bones/ all taxon Camel Horse Proboscidean 2.99 1.15 2.59 Comparisons between Taxa-Complete Limbs and Girdles Camel vs. Horse Camel vs. Proboscidea Horse vs. Proboscidea 4.16* 6.34* 0.25 Comparisons between Taxa-Partial Limbs and Girdles Camel vs. Horse Camel vs. Proboscidea Horse vs. Proboscidea 5.03* 10.13** 1.87 B. Surface erosion or waterwear Complete Bones vs. Identifiable Partial Bones and Fragments/ all tan combined 6.32* Complete Bones vs. Identifiable Partial/ Bones/ all taxa combined Podials Included Podials Excluded 4.04* 10.82** Complete Bones vs. Podials/ all tan combined 7.31** Complete Bones vs. Partial Identifiable Bones/ all taxon Camel Horse Proboscidean 0.89 4.08* 1.44 Comparisons between Taxa-Complete Bones Camel vs. Horse Camel vs. Proboscidea Horse vs. Proboscidea 1.55 0.76 0.74 Comparisons between Taxa-Partial Bones Camel vs. Horse Camel vs. Proboscidea Horse vs. Proboscidea 0 01 0.08 0.18 PRATr & HULBERT: TERRESTRIAL MAMMAL TAPHONOMY OF LEISEY SHELL Prr 223 Table 8 Continued- C. Scratchmarlm Complete Bones vs. Partial Identifiable Bones/ all taxa Podials Included Podials Excluded 4.45* 5.14* Complete Bones vs. Partial Identifiable Bones/ by taxon Camel Horse Proboscidean 108 0.20 2.30 Comparisons between Taxa-Complete Bones Camel vi Hone Camel vi Proboscidea Horse vs. Proboscidca 2.23 5.14* 5.91* Comparisons between Taxa-Partial Bones Camel vs. Horse Camel vs. Proboscidea Horse vs. Proboscidea 0.58 0.824 1.92 D. Weathertng vs. Scratches Complete Bones Partial Identifiable Bones 1.33 0.11 rejoined, in most instances (90% of associated elements examined), scratchmarks do not cross the break (Fig. 18). This finding demonstrates that most scratches were applied following breakage of these elements, which almost certainly occurred at the site of deposition. Based on the results discussed above, it is highly likely that the shell-sand matrix ofthe Leisey lA locality was involved in producing the numerous striations on bone surfaces. The question remains as to whether the agent responsible for exerting the force required to cause abrasive damage to the bone surface can be determined. Post-depositional, post-fossilization scratching by sediment shifting can be ruled out for several reasons. The scratches are found predominantly on the upper bone surfaces. In post-depositional sediment shifting, other surfaces should be scratched as well. Behrensmeyer et al. (1989) suggest that during compaction events, vertically-oriented rather than horizontally-oriented surfaces would be most heavily scratched. There is no predominant directionality to the marks on the Leisey elements (Fig. 18) as might be expected if sediment moved or shifted in one direction. In addition, the marks are darkly colored and obviously were made on 224 BULLETIN FLORIDA MUSEUM NATURAL HISTORY VOL 37. Pr. L No. 7 A 60 - u Camelidae 0 Equidae 50 - I Proboscidea 40- Pe rc en t 30-201 1,111 ~1 8-10- 0 , <6 6-20 21-40 41-60 61 -80 Percentage of Surface Area Scratched B 60 - 50 - 0 C wd mfi <6 6-20 21-40 41-60 61-80 Percentage of Surface Area Scratched Figure 20. Percentage of complete (A) and partial (B) limb bones and girdle elements from Leisey Shell Pit 14 sorted by tan showing proportion of the surface area covered by scratch marks. Samples consisted of 29 complete and 79 partial camelid specimens. 23 and 41 ofequids, and 24 and 25 ofproboscideans. PRATr & HULBERT: TERRESTRIAL MAMMAL TAPHONOMY OF LEISEY SHELL PIT 225 the bones before fossilization occurred. Post-fossilization scratches on Leisey elements are generally light in color and easy to discern from pre-fossilization scratches. Support for damage by trampling is provided by the great similarity of the marks to those caused by large herbivores (Fiorillo 1989; Haynes 1991). The types and proportions of bones characterized by scratching, and the average degree of scratching are also similar to values reported in studies on trampling of Recent and fossil bones (Olsen and Shipman 1988; Behrensmeyer et al. 1989; Fiorillo 1989; Oliver 1989). As the scratching occurred in a marine environment, a trampling scenario mandates that water depth must have been relatively shallow, at least periodically, so that terrestrial mammals would have had no difficulty crossing the area by walking. The invertebrate fauna of the bone bed indicate that water depth was shallow (Hulbert and Morgan 1989; Morgan and Hulbert this volume; Portell et al. this volume). Several other features suggest that attributing scratchmarks to trampling may be erroneous. A major problem is that scratches do not cross breaks on associated partial elements. If these elements were broken by trampling impact, it seems that scratches applied at the time of breakage would transverse the fracture; however more comparative studies on modern bone are needed to substantiate this supposition. In addition, the rarity of steeply plunging long bones within the site argues against the presence of a force sufficient to drive elements into soft or moist sediment (Fig. 6). It has been noted that trampling of bones lying on wet or moist muddy sediments often results in vertically-plunging bones (Hill and Walker 1972; Behrensmeyer and Boaz 1980; Fiorillo 1989), although the amount of plunge may vary with sediment consistency (see Olsen and Shipman 1988). The Leisey lA sediment, with its high sand and shell content, probably provided a more resistant surface than fine-grained muds. The locations of scratchmarks on bone surfaces suggests that perhaps water- related abrasion rather than trampling was the major agent of scratching, however, the findings do not conclusively support this hypothesis. In an analysis of location of scratchmarks on mandibles, it was noted'that in the majority of cases, a right or left mandible (complete with symphysis and diastema of the opposite side) possessed scratchmarks predominantly on the side found facing up in the field. Its associated mandible, broken posterior to the diastema and often separated from its partner, was equally scratched on both the upper and lower surfaces. Complete limb elements for which these data were recorded also possess the majority of scratches on the side facing up in the field, while incomplete bones tend to be equally scratched on both sides. Only a very small proportion of elements have more scratches on downward-facing surfaces. These findings suggest, as previously noted, that most scratches were applied to the upward4acing surface of the bone. Larger bones did not move or flip over before final burial, unlike smaller or incomplete elements that bear scratches on both upward and downward facing flat surfaces. Behrensmeyer et al. (1986) indicated that the upward-facing surface 226 BULLETIN FLORIDA MUSEUM NATURAL HISTORY VOL 37, M. 1, No. 7 of the bone is scratched as sediment lying on top of the element or attached to the foot is driven across the bone surface when the foot steps on the bone. Olsen and Shipman (1988) showed that scratches are also applied to the downward-facing surface of the bone if it moves slightly across an abrasive surface when it is stepped on. The large number of elements with scratches predominantly on the upper surface is therefore best explained by the movement of water-driven sand and shells acro= these surfaces, although there is some question as to whether the pressure applied would have been sufficient to cause scratchmarks. It has been noted previously in this paper that cumnt velocities within the environment of deposition were for the most part relatively slow. Bones with equally-scratched upper and lower surfaces may either have been trampled and kicked on several occasions, or transported by moving water within the site, turning over several times before burial. Based on the information currently available, it is unclear as to which agent moving water or trampling, was ultimately responsible for causing the scratchmarks. Summary: Bone Modification.- The findings from the investigation of bone modification may be summarized as follows: 1) The presence of associated elements indicates that some individuals were introduced into the site of deposition as complete or partial carcasses. Breakage and scratching of associated elements took place within the site, and may have been the result of trampling. Isolated and broken elements with no associations were introduced individually after decomposition, disarticulation, brealcage, weathering, and scavenging occurred elsewhere, probably in a terrestrial habitat. 2) Bones that were exposed on terrestrial surfaces belong primarily to equids and proboscideans, as evidenced by the extent of weathering and scavenger damage. Camelid bones show lower proportions of these types of modification, suggesting their introduction into the environment of deposition in more complete condition. 3) The main scavengers/carnivores on large mammalian herbivores were Smilodon gracilis, Canis armbrusteri, and Arctodus pristinus. The relatively low percentage of carnivore-damaged bone indicates that the assemblage was not formed primarily by carnivores. It also indicates that prey (or prey carcass) to predator ratios may have been high, perhaps on a seasonal basis. The population dynamics of the equids and camelids resemble those of an attritional assemblage with periodic catastrophic additions. Relatively low degrees of weathering indicate that even weathered bones were not exposed to the effects of climatic deterioration for long periods of time. Periodic floods of feeder streams may have cleansed the banks of remains and carried them to the site of deposition on a regular basis. PRAIT & HULBERT: TERRESTRIAL MAMMAL TAPHONOMY OF LEISEY SHELL PIT 227 4) Relatively little sedimentary erosion or polishing of bones suggests that the site was not characterized by a sustained current of high velocity, that agrees with the lack of a preferred orientation of the long bones. 5) Location and position of scratchmarks shows they were applied in the depositional environment as the bones were scraped against the shelly-sand matrix. The actual agent of scratching may have been either trampling herbivores or moving water. BONE RELATIVE REPRESENTATION Relative representation involves the numerical assessment of the types of skeletal elements that comprise the fossil deposit. Generally, taxa are grouped by size in relative representation studies, as bone abundance is often a size-related phenomenon. In most cases small taxa elements are more subject to destruction by transport and various agents than those of larger vertebrates (Dodson 1973; Wolff 1973; Behrensmeyer and Boaz 1980; Andrews 1990). In the case of Leisey mammals, as total bone counts could only be undertaken in those squares that had been completely sorted and curated, abundances of mammals other than camelids and equids were not represented by sufficient numbers of elements to comprise a reliable sample. As these taxa were of similar body size, analysis of different patterns of bone relative representation due to differences in body size cannot be addressed at this time. However, as evidence presented in previous sections suggests different taphonomic histories for the camelids and equids, their bone relative representation patterns were analyzed separately. It was not possible to identify all camelid and equid postcrania to species, so the analysis is limited to the generic level for equids, and the tribal level (Lamini) for the camelids. Based on jaws, it is likely that the majority of the camelids were Palaeolama miritica and the equid, Equus "lei*i." Due to the difficulty in assigning ribs and post-cervical vertebrae to family, these were not used in family level assessments. Bones of camels and equids from squares B4, B5, B6, B7, C5, C6, C7, D6, D7, D9, D10, E6, E7 and F5 were counted. These squares, located in the most fossiliferous region of the site, were primarily or exclusively excavated by FLMNH personnel, insuring maximum bone recovery and minimum collecting bias. In addition, a large number of these squares were contiguous, so the likelihood of counting possibly associated remains was high. There are two generally accepted methods of assessing bone abundance: relative representation or percentage skeletal preservation (Voorhies 1969; Wolff 1973; Korth 1979; Pratt 1990), which is determined by dividing the number of one type of bone found by the number expected based on the minimum number of individuals (MNI) represented; and relative skeletal part frequencies (SPF) 228 BULLETIN FLORIDA MUSEUM NATURAL HISTORY VOL 37, Fr. I, No. 7 (Behrensmeyer and Boaz 1980; Badgley 1986a), a value obtained by dividing the number of elements of one type by the total number of all elements recovered. Although the latter method was used in order to compare Leisey lA bone abundances with Recent assemblages analyzed in this fashion, relative representation is deemed the most useful, reliable, and informative method, for the following reasons: 1. It is based on MNI (minimum number of individuals), which provides a reliable indication of the actual number of individuals of a taxon preserved in the fossil site, especially in cases where there is evidence of skeletal articulation or association. In these instances, the relative representation values can be used to determine the numbers of each element type that have been lost from the complete skeleton. The proportion missing of each bone is equal to one minus the relative representation of that element. 2. Each bone relative representation value is calculated independently of those of other elements. In contrast, SPF calculations have no clear meaning unless they are compared with the percentage representation of that element in a complete skeleton. SPF also does not provide an independent assessment of the bone's abundance, because each value is a percentage of the total. A high value for one bone will mean that another bone will have a low percentage value. 3. SPF underemphasises losses of bones that are present in large numbers in a skeleton (ribs, vertebrae, phalanges) and overemphasises losses of elements that are the only element of that type in a skeleton, such as the skull. 4. When the SPF is used in statistical tests, it is less powerful than relative representation in determining differences. In a comparison of two bone assemblages using the Spearman rank-coefficient test, a test frequently used in these types of assessments, SPFs are less likely to indicate actual difference between bones assemblages than are relative representation values. Unless the assemblage has been drastically modified, the bones with the highest SPF values will always be those with the greatest number of elements in the skeleton, and the lowest will be those represented in a skeleton by only one or two elements. A total of 466 identifiable elements were counted for camelids and 212 for equids. Minimum number of individuals recovered from these squares was based on left innominates for both taxa. Cameli(Is were represented by 20 left innominates, and equids by 11. Relative representations are shown in Table 9 and Figure 21. In the figure, elements are arranged in descending order by quartz grain equivalent values (Behrensmeyer 1975; Korth 1979; Pratt 1986, 1990) and transport groups (Voorhies 1969; Korth 1979; Pratt 1990), which provide a general indication of each element's transportability in running water. In addition, because the grain equivalent value is related to the density and size of the bone, it is also to PRATT & HULBERT: TERRESTRIAL MAMMAL TAPHONOMY OF LEISEY SHELL PIT 229 some extent indicative of the element's resistance to destruction by other taphonomic factors. In a comparison of Figures 21A and 218 it is immediately evident that skeletal representations for camelid and equid elements are different (Tables 9-10). Representation of camelid elements is relatively high, with mean representation of 33.354 while mean representation for equid elements is 20.4%. The greatest difference is seen in Group II elements, in particular long bones. On average, approximately 58% of the expected number of camelid limb elements were recovered, as opposed to about 27% for equids. The pattern of skeletal preservation of camelids is similar to that reported for several Recent bone assemblages (Table 10), including caribou remains killed and fed upon by wolves (Binford 1981) and a surface assemblage in Amboseli Park, Kenya (Behrensmeyer and Boaz 1979; 1980). In the case of the former, although the pattern of skeletal preservation is similar, the actual relative abundances of elements are generally greater for the fossil camelids than for the caribou, showing that a lower proportion of the elements in the fossil assemblage were subject to loss or modification than the wolf-modified assemblage. The Leisey camelid assemblage is very similar in overall pattern of abundances to the Amboseli bone assemblage, but less so to accumulations from Ngorongoro Crater and Serengeti National Park, Tanzania (Blumenschine 1989). There are two reasons for this result. First of all, the Amboseli study is based on SPF, which as mentioned above, generally provides a high degree of similarity in bone abundance rankings between assemblages. In addition, there are some basic differences between the Recent bone assemblages. The Amboseli accumulation evidently formed in part as a result of mass deaths, which led to an increase in prey (carcass)-to-predator ratios and resulted in a lower degree of bone destruction than normally occurs (Behrensmeyer and Boaz 1980). The two assemblages in Tanzania were heavily scavenged by hyaerms, which cause high rates of bone loss from carcasses (Blumenschine 1989). The absence of a bone-eating scavenger at Lcisey probably contributed to the high representation of elements (see section on carnivore-modified bone) and the greater similarity of the Leisey camelid assemblage to the Amboseli accumulation. Greater percentages of camelid elements than those reported in Recent "landscape" studies suggests that some of the camelids were introduced into the depositional environment in a complete or nearly complete state. It has been shown previously that only about 18% of the camelid remains (not including podials) were modified by predation and scavenging. Periodic mass deaths occurring close to the site of deposition provided a possible source of more complete remains. The striking similarity of the camelid element representation pattern to that of a hypothetical lag deposit (Hanson 1980: fig. 9.3) suggests that moving water, rather than carnivores or scavengers, may have been the primary agent that removed camelid elements from the environment of deposition. It was previously shown that a constant unidirectional current was not a feature of site, 230 BULLETIN FLORIDA MUSEUM NATURAL HISTORY VOL- 37, rr. I, No. 7 Table 9. Bone abundances ofcamelids and equids. r ' '' ~ ' " =number of elements ofone type recovered/number expected based on MNI (x 100). In both cases. MNI is based on innominates. Number eqected Number based % Relative Element found on MNI Representation A CAMELIDAE (MNI= 20) Mandible (R or L) 30 40 75.0 Maxilla 23 40 57.5 Scapula 26 40 65.0 Humerus 26 40 65.0 Radiofulna 22 40 55.0 Innominate 37 40 92.5 Femur 26 40 65.0 Patella 3 40 7.5 Tibia 16 40 40.0 Cervical Veltebrae 42 140 30.0 Sacrum 12 20 60.0 Metapodials 22 80 27.5 Astragalus/Calcaneum 23 80 28.8 Fibula 11 40 27.5 Other Podials 84 440 19.1 Phalanges (Prox & Med) 48 320 15.0 Distal Phalanges 4 160 2.5 B. EQUIDAE (MNI =11) Mandible (R or L) 15 22 68.2 Maxilia 3 22 13.6 Scapula 5 22 22.7 Humerus 5 22 22.7 Radius 1 22 4.5 Ulna (prodmal end) 5 22 22.7 M 21 22 95.5 Femur 2 22 9.1 Patella 2 22 9.1 Tibia 4 22 18.2 Cervical Vertebrac 26 77 33.7 Sacrum 2 11 182 Medial Metapodials 5 44 11.3 Lateral Metapodials 22 88 25.0 Astragalus/Calcaneum 5 44 11.3 Other Podials 45 242 18.6 Phalanges (Prox. & Med) 13 88 14.8 Distal Phalanges 6 44 13.6 (Hooves) PRATr & HULBERT: TERRESTRIAL MAMMAL TAPHONOMY OF LEISEY SHELL PIT 231 A 100- III II I 90- % R el at iv e R ep re se nt % R el at iv e R ep re se nt at io n 70 60 50 40 30 :28333 I.,1lili 10 O,#*t~ 44%.rAM / / / / / / > AJ@ O 65:id 5:3EE22&5221 Z 02-£5 E-:ed EN X- JOOH ®r~-0='LLaa 2-JOOH B 100- 90- C 80- 113 40 - %$ P © a\\ E 2 2 5 5 1 8 1 z S / 5 / k\\\IB ~: 0 z U. 66 0 J -1 Figure 21. Percentage relative representations ofskeletal elements from Leisey Shell Pit 14 based on MNI (Table 9). A Camelidae (Palaeolama and Hemiauchenia combined). with an MNI of 20 based on left innominates. B. Equus, with an MNI of 11, also based on left innominates. Roman numerals and different shading patterns separate the elements into the hydraulic transport groups of Voorhies (1969). Element abbreviations: Den, dentary, Max, maxilla; Pei pelvis (innominate); Hum, humerus; R/U, radioulna (camelids only); Rad, radius; Tib, tibia; Scap, scapula; Met, fused metapodials 3 and 4 (camelids only); MMet, medial metapodial (of digit 3, Equus only); A/C, astragalus and calcaneum; Fib, fibula (camelids only); LMet, lateral metapodials (Equus only); Uln, ulna (Equus only); Pod, podials (carpal and other tarsal elements); Phi proximal and medial phalanges; Cerv, cervical vertebrae; Pat, patella. 232 BULLETIN FLORIDA MUSEUM NATURAL HISTORY VOL. 37, Fr. I. No. 7 Table 10. Values of r for Speannan rank-coefficent test d.f. = degrees offreedom * = significant at P E O.05, ** = significant at p E 0.01. RR= relative representation. SPF = skeletal part frequency. Comparisons df Camelidae Equidae Leisey Camelidae-RR 15 - 0.477 Wolf Kills of Caribou-RR 10 0.990** 0.521 (Binford 1981: Tab. 5.01) Size 3 Herbivores-RR 13 0.506 0.207 Ngorongo Crater, Africa (Blumenschine 1989: Tab. 5) Size 3 Herbivores-RR 12 0.611* 0.201 Acacia woods grass plains (Blumenschine 1989: Tab. 5) Size 3 Herbivores-RR 13 0.448 0.216 Riparian woodlands (Blumenschine 1989: Tab. 5) Surface Bone Assemblage--SPF 6 0.959** - An,boseli Basin, Kenya (Behrensmeyer and Boaz 1980: Tab. 5.6) Wolf Kills of Caribou-RR 13 - 0.563* (Binfbrd 1981: Tab. 5.01) Corrected for losses due to hydraulic transport Group III, II elements only 8 - 0.833** but other factors such as water-abraded bone. low preservation of terrestrial microfaunal species, and the transported aspect of pollen grains (Rich and Newsom, this volume), indicate the presence of flowing water. The current velocity of water flowing through the site may have been sufficient to transport many of the group I and some of the group II elements from the site of deposition. The relative representation of equid elements is less easily explained. The pattern of bone representation bears no statistically significant resemblance to patterns seen in Recent landscape assemblages (Table 10). Limb elements of Leisey Equus are generally not as highly represented as are those in modern assemblages, while podials and phalanges are more abundant. According to most modern bone assemblage studies, small, distal limb elements are lost in a number of ways, including burial (Behrensmeyer and Boaz 1980) and destruction or ingestion by carnivores (Haynes 1980a; Blumenschine 1989). It is also possible PRATr & HULBERT: TERRESTRIAL MAMMAL TAPHONOMY OF LEISEY SHELL PIT 233 that these small elements may be missed in surface bone counts, so that their low representation is in part due to sampling problems. Among the Recent bone samples, the equid assemblage is most similar to that of Blumenschine's description of a "stage 3" carcass (1989: fig. 6) represented primarily by the axial skeleton and the scapula. In this stage, limb elements have been removed from the carcass by predators and scavengers, and the axial skeleton remains articulated with the pelvis. It was previously demonstrated that equid elements were significantly more weathered than camelid elements, indicating that equid carcasses may have been available for modification by carnivores and scavengers while some camelid skeletons were not. The lack of similarity between the Leise equid bone representation and those of modern assemblages indicates that multiple taphonomic factors determined the final abundances of equid elements. Assuming that we accept the previous hypothesis that the bulk of camelid bone loss was through hydraulic transport from the depositional environment, then it is logical to assume that the equid bone assemblage that reached the site of deposition was also subjected to this type of loss. The major difference in the final bone accumulations for these two taxa would therefore result from differing bone frequencies in the assemblages of each taxon brought to the site (different input). Based on previously discussed data (see section on carnivore modification), the original equid bone assemblage may have initially resembled a modern caribou bone assemblage resulting from wolf predation and scavenging (Binford 1981: Table 5.09 Column 26); the carnivores present and the prey size are both similar to those at the Leisey site. If the same proportion of each element type is removed from this assemblage as was lost from a presumed complete skeletal assemblage of Leisey camelids, the skeletal relative representations shown in Figure 22 result. For example, if relative representation of a certain element type for Leisey camelids is 75%, then it is assumed that 25% of those elements were removed from the assemblage by hydraulic transport. Representation of the same element type is reduced by 25% in Binford's caribou assemblage. The resulting bone abundance ranks are correlated with those of the Leisey equids (Table 10). The similarity between Group III and II elements is the most startling, and the correlation between abundance ranks for these groups is highly significant (Table 10). The greater representation of equid Group I elements may be explained in several ways. First of all, as mentioned previously, it is possible that these elements may have been missed in counts of the modern assemblage. It is also possible that the fossil agents of bone destruction were operating differently and not destroying or removing these bones to the same extent as wolves. Finally, given that horse carcasses may have been decomposing along bodies of water that eventually fed the environment of deposition, during some periods smaller bones may have washed in when larger bones were not affected, thus increasing the number of bones of this type in the deposit. Results of this portion of the study substantiate other findings. While the possibility that factors causing mass deaths of camelids were also affecting equid 234 BULLETIN FLORIDA MUSEUM NATURAL HISTORY VOL 37, Yr. L No. 7 population dynamics is certainly likely, apparently camelids were living and dying closer to site of deposition than equids, and a lower proportion of camelid remains were exposed on terrestrial surfaces to agents of bone destruction than equids. The equid remains may have been carried to the depositional environment by moving water, and once there, the entire assemblage of all elements was modified by a relatively low velocity current, most likely less than 50 cm/second in velocity (Behrensmeyer 1975; Pratt 1986, 1990) and capable of transporting primarily Transport Group I elements from the depositional environment RELATIVE ABUNDANCES OF MAMMALIAN TAXA Relative abundances of the Leisey lA terrestrial mammals (i.e., Mammalia exclusive of Sirenia and Cetacea) were calculated using the MNI method (Table 11; Shipman 1981). Only specimens collected in 1984 were used in lhe analysis, as they represent the least biased sample for the entire site. Badgley (1986b) discussed the relative merits of various measures of relative abundance, and concluded that MNI was the most appropriate for sites containing disassociated skeletons. As shown on pages 190-193, there are numerous examples of these from Leisey 1 A. Fossils of large mammals (body mass > 1 kg) were observed and collected in the field, while those of smaller mammals were most often recoyered by screen-washing matrix. Since only a small percentage of the entire volume of the bone-bearing horizon was screen-washed, the overall relative abundances of small versus large mammals can not be directly compared. The most valid comparison is to use the contents of square D9, from which all of the matrix was collected and screen-washed (about 120 kg, dry weight). As shown in Table 12, the remains of large mammals from D9 far outnumber those of small mammals. The large terrestrial mammalian fauna of Leisey lA is dominated by herbivores with body masses ranging between 40 and 250 kg (Table 11): notably camelids, equids, tayassuids, and tapirids. Two families, comprising five species, the Camelidae and Equidae, together account for about two-thirds (68%) of the individuals. All artiodactyls and perissodactyls combined make up 81% of the individuals. Megaherbivores (body mass > 1000 kg) account for only about 7%. This group includes the three proboscideans and the larger of the two ground sloths, Paramy/odon harlani. Despite the richness of the Leisey mammalian fauna, it is not especially diverse, as the five most abundant species account for over 75% of the individuals, and the ten most abundant, over 90%. Altogether, the relative abundance of the nine members of the Carnivora is about 6% of the total (Table 11). Five of these taxa, Lutra canadensis, Procyon sp., Urocyon sp., Homotherium sp., and Lynx rufus, are represented by only one or two elements each and an MNI of 1. Although the MN[ of the canids Canis 100- 90 - % el at ive ep re se nt at io n 80 - I Leisey Equidae 70 - 0 transported kill site 60 50 P R A IT & H U LBER T: TER R ESTR IAL M A M M A L TA P H O N O M Y O F LEISEY SH ELL PIT 235 40 30 20 7- . M / - Den Max Pei Hum Fern Rad Tib Scap Mmet A/C Pod Phi Cerv Hoof Figure 22. Comparison between the percent relative representation ofskeletal elements ofthe Leisey Shell Pit lA Equus compared with that of caribou from a wolf kill site (Binford 1981) adjusted for hypothetical losses during hydraulic transport See text for discussion. Abbreviations as in Figure 21. 236 BULLETIN FLORIDA MUSEUM NATURAL HISTORY VOL 37, rr. I, No. 7 Table 11. Relative abundances of large terrestrial mammals at kisey Shell Pit 1 A Minimum numbers of individuals (MNI) were tabulated Rom mandibles collected during the 1984 field scason. Note that the MNI reported here do not necessarily reflect the maximum values from the site, as isolated teeth and postcrania were not considered. Mandibles were selected because they are all cataloged and can be identified to species. Eleven species (indicated by asterisks) were not represented by a mandible from the 1984 collection and were arbitrarily given an MNI of 1. Also reported is the percentage of the total MNI (N = 242) for each taxon and the percentage ofthe total number ofherbivores (N = 226) for the appropriate taxi Overall Herbivore relative relative MNI abundance (%) abundance (%) *Doopus bellus 1 0.4 *Dasypodidae, n.gen.andsp. 1 1 0.4 Holmesinajloridamis 1 0.4 0.4 Nothrotheriops texamts 8 3.3 3.5 Paramylodon harlam 8 3.3 3.3 *Castoroide: n. sp. 1 0.4 0.4 *Neochoerus sp. 1 0.4 0.4 Lepus sp. 1 0.4 0.4 *Carn~~~rusten 1 0.4 1 0.4 Canis edwardit 1 0.4 *Procyon toror 1 0.4 *Lutra canadens,3 1 0.4 Arctoduspristinus 2 0.8 Smilodon gracilis 5 2.1 *Homotherium sp. 1 0.4 *Lynx rufus 1 0.4 Tapins hays# 3 1.2 1.3 Equus "lei*r 36 14.9 15.9 Equus (Hemionus) n sp. 17 7.0 7.5 *Equus laternus" 1 0.4 0.4 Platygonus d P. vetus 22 9.1 9.7 Mylohyus d. M. fossilis 2 0.8 0 .9 Hemiauchenia macrocephala 1% 7.4 1.9 Palaeolama mirdica 92 38.0 40.7 Odocoileus virginianus 6 2.5 2.6 *Cuvieromus fropicus 1 0.4 0.4 Mammut amencanum 2 0.8 0.9 Mammuthus hayi 6 2.5 2.6 1 See Downing and White (this volume. Pt Ii, p. 378) for the descripion. PRATr & HULBERT: TERRESTRIAL MAMMAL TAI'HONOMY OF LEISEY SHELL PIT 237 armbrusteri and C edwardii is also 1 (based on mandibles), they are more common than the other rare carnivores. Each is represented by a number of specimens of various elements. Of the carnivores, only Smilodon gracilis is among the ten most abundant mammals at Leisey lA. The Smilodon sample includes several examples for most skeletal elements. Notably all are from fully adult individuals (indeed the carnassials in many of the jaws exhibit very heavy wear). The lack of juveniles or subadult Smilodon individuals contrasts sharply with the herbivores, in which younger age classes predominate (see below). Among the limited number of small mammalian taxa, Sylvilagus ~oridanus and Geomys pinens are most abundant (Morgan and White this volume). Table 12. Relative abundances of large (> 1 kg) and small (< i kg) terrestrial mammals from square D9 of Wsey Shell Pit l A. Both ' ' number of individuals (bIND and number of identifiable specimens (NISP) are reported. In oontrast to the five small mammal specimens, this square produced 712 identifiable fish specimens, with an MNI of 56 representing about 50 tan (Scudder et al. this vol.). MNI NISP Large Mammals Dasypodidae n.gen andsp.' 1 1 Holmesina floridanus 1 4 Nothrotheriops texanus 1 3 Paramplodon harlani 2 7 Hemiauchenis macrocephala 1 2 Palaeolama mirijica 2 14 Camelidae, gen. indet - 23 Plaoyonus d P. vetus 1 2 Tayassuidae. gen. indi - 4 Tapirus haysil 1 2 Equus spp. 2 24 Proboscidea 1 4 Vertebrae - 12 Ribs - 22 Total 14 124 Small Mammals Sigmodon libitinus 1 2 Pedomys n. sp. 1 1 Rodent postorania - 2 Total 2 5 ' See Downing and White (this volume, PL 4 p. 378) for the description. 238 BULLETIN FLORIDA MUSEUM NATURAL HISTORY VOL 37, FT. L No. 7 The relatively very low representation of small species among the terrestrial mammals (Table 12) suggests one or more taphonomic filters acting on body size. Some of these were apparently not acting at the actual site of deposition, as remains of small aquatic birds and fish were abundant (Emslie this volume; Scudder et al. this volume). Apparently the bones of the smaller terrestrial mammals were preferentially destroyed either by weathering, carnivores, or trampling, far from the site. It is also possible that if small and large bones (or carcasses) were being transported together in a fluvial system, hydrodynamic sorting separated the two (Behrensmeyer 1976). The remains of the proboscideans may also be underrepresented for hydrodynamic reasons, as very large carcasses and bones would require higher than normal current velocities to transport them. This is supported by two types of data. Associated skeletons of proboscideans at Leisey lA (both Mammut and Mammuthus) belong to very young juveniles, not adults. Second, adult limb bones are significantly more weathered than those of the camelids (Table 8). This suggests they lay exposed on the ground for long periods of time until a sufficiently severe storm occurred to provide current velocities needed to transported them. Of the herbivores, grazers (Hemiauchenia, Equus, and Mammut/ms) make up 35% of the individuals, while browsers and mixed-feeders the remaining 65%. These numbers do not necessarily reflect the actual amal extent of open versus wooded landscape. However, the pollen and macrobotanical evidence corroborates the presence of at least some mixed hardwood/pine woodlands (Rich and Newsom this volume). In addition to body size, relative abundances of the Leisey 1 A mammals were influenced by habitat and behavior of the various taxa. Those that were more likely to die along or in rivers would tend to be more highly represented. However, other factors must have been involved, considering the re\adve rarity of Castoroides, Neochoerus, Lutra, and Trichechus. POPULATION DYNAMICS AND MORTALITY PROFILES Analysis of population dynamics and mortality profiles has proven to be a valuable tool in the taphonomic study of large mammals, especially ungulates (Voorhies 1969; Klein 1982). The large samples from Leisey lA readily allow such an analysis of the two most common species, Palaeolama mirifica and Equus "leia>i." Fossil populations are usually classified into two general categories, attritional or catastrophic. Attritional assemblages are thought to result when the cause of death includes predation, disease, and accidents. The key factor is that mortality rates (the qx of life tables) are very high for juveniles and old-age adults, low for mature adults, and very low for very young (prime) adults. In catastrophic assemblages, mortality rates are independent of age, resulting in much greater numbers of prime and mature adults, and relatively fewer very old adults. There is PRATr & HULBERT: TERRESTRIAL MAMMAL TAPHONOMY OF LEISEY SHELL PIT 239 often a mistaken one-to-one correlation inferred between whether a fossil population is attritional or catastrophic, and the amount of time (on an ecological time scale) that is represented by the sample. Attritional assemblages are often considered to represent more lengthy accumulations (hundreds to thousands of years), while catastrophic assemblages represent very rapid events (days to months). This may be true in many cases, but certainly there are various combinations of life histoty strategies and causes of mortality that provide numerous exceptions. For example, a snow-covered dead-fall trap (as described by Oliver 1989) should produce an assemblage in which mortality was independent of age (i.e, catastrophic), but which represents an accumulation of many years. Analysis of population dynamics should take geologic and other taphonomic data into account to reach conclusions on the amount of time represented by a fossil assemblage (e. g., Kurt6n 1983). In many cases fossil ungulate populations closely approximate expected distributions for either attritional (e.g., Van Valen 1964; Hulbert 1982, 1984; Klein 1982) or catastrophic (e.g., Kurt6n 1953; Voorhies 1969; Klein 1982) assemblages. The degree of underrepresentation of the juvenile age-class(es) varies depending on the taphonomic history of the individual site. Not all attritional assemblages in modern ungulates have a very high peak among the oldest age-classes (Klein 1982: figs. 3-4); thus the often reproduced example of Dall sheep from Mt. McKinley National Park (Deevey 1947) is an extreme form. Klein (1982) referred to these two types of attritional assemblages as "U-shaped" and "L-shaped." A more reliable indicator of an attritional assemblage is a low frequency of young adults relative to more mature and very old adults. Methodi- The two most common ungulate species at Leisey 14 Palaeo/ama miri#ca and Equus "tei*i," were chosen for analysis. The age at death was estimated using the methods of Klein (1982; Klein and Cruz-Uribe 1984). This involved measuring the crown height of either a deciduous or permanent cheektooth for each individual. The measurement was then placed in a quadratic equation along with estimates of initial crown height, age of eruption for the particular tooth, and potential longevity (or age when the tooth is shed if it is a deciduous tooth). For Palaeolama, the dp4 and m3 were measured at the protoconid (as illustrated by Klein 1982 for many artiodactyls). Using Wheeler's (1982) data on modern 11ama and alpaca, the dp4 was estimated to have been erupted at birth and shed at 4.5 years, while the m3 erupted at 3.2 years. Potential longevity was estimated as 18 years. For Equus, the dp2 and p2 were measured at the metaconid. Isolated dp45 cannot always be distinguished from dp35 in equids, so they were not used in the analysis. The crown height of the more hypsodont horse teeth usually could not be measured if the tooth remained in its alveolus. However, most of the horse jaws were extensively broken by compaction, so the crown heights were measured before they were repaired. In a few instances the crown height of the p2 in a jaw could 240 BULLETIN FLORIDA MUSEUM NATURAL HISTORY VOL. 37, pr. I, No. 7 not be measured, but other teeth were measurable. In these cases least-squares regression was used to estimate the p2 crown height based on the crown height of another tooth. By using either of these two methods, all but a few of the adult mandibles could be included in the analysis. Following Klein (1982), for E burche//i, the dp2 was assumed to be erupted at birth and shed at 3 years, while the p2 erupted at 3 years. Potential longevity was estimated as 22 years. Maximum, unworn crown heights were estimated using specimens in the samples that had little or no wear. For Palaeola,na, these were 16.0 mm for the dp4 and 24.5 mm for the m3. For Equus, these were 23.0 mm for the dp2 and 63.3 for the p2. After age determination, specimens were grouped into one of ten age- classes (each representing 10% of the potential longevity) to produce mortality profiles such as those used by Klein (1982). Comparisons between mortality profiles used the Kolmogorov-Smirnov test as illustrated by Klein and Cruz-Uribe (1984: 59-60). Life tables were constructed as described by Voorhies (1969) using dynamic analysis, primarily to estimate mortality rates and juvenile underrepresentation using the methods ofHulbert (1982). This method uses assumptions of birthrate, sex ratio, stable population size, and litter size to estimate the number of offspring produced by the analyzed population per year (Hulbert 1982). Usually this value far exceeds the number of observed individuals that comprise the first year age-class. Under the assumption that the population is stable (neither increasing nor decreasing in size), this discrepancy is accounted for by juvenile underrepresentation. That this is indeed the case is shown by unrealistically low first year mortality rates. Juvenile underrepresentation is common because the bones and teeth of young juveniles are more easily destroyed than older individuals. To estimate first-year juvenile underrepresentation in a life table, the number of individuals of the first age class in the d~ column is increased so that the number in the corresponding 1X column matches the number of expected offspring (the sum of the ixmx column). The juvenile underrepresentation factor is the number of individuals added to the first year dx column divided by the unadjusted value. For example, suppose in an attritional sample of 100 individuals, 20 are first year individuals, and (based on modern analogues) such a population would annually produce 160 offspring. In this case, the first number in the dx column is changed from 20 to 80, which increases the first value in the 1x column to 160. The sample would then have a juvenile underrepresentation factor of 60/20 or 3.0. Results.- The analysis of the Leisey lA samples of Palaeolama mirdica and Equus "lei*i" produced two important results (Fig. 23; Tables 13-14): (1) both samples are dominated by juvenile individuals, especially by those that died within three months after birth; and (2) neither sample produced a mortality profile that unambiguously matches either a "classic" catastrophic or attritional population. As expected in both models (but not usually observed due to juvenile underrepresentation), the first age-class contains the most individuals, about 40% PRATT & HUI.BERT: TERRESTRIAL MAMMAL TAPHONOMY OF LEISEY SHELL PIT 241 in both cases. The Palaeolama sample has successively increasing numbers of individuals in age-class 2 through 4 (Fig. 23A), that differs from catastrophic populations in which the numbers should progressively decline. However, unlike hypothetical attritional populations, the combined percentage of mature and old- age adults (age-classes 5 to 10) is less than those of prime adults (age-classes 2 to 4), 26% vs. 33%. The progressive decline in numbers in age-classes 6 to 10 is also unlike attritional assemblages, especially those with "U-shaped" mortality profiles (Klein 1982). Leisey Equus does have successively declining numbers in the prime adult age-classes, and the second age-class has the second highest number of individuals (Fig. 2313), both of which are characteristic of catastrophic assemblages. However, there are secondary peaks in the sixth and ninth age- classes, that is atypical of catastrophic samples. These are more characteristic of attritional assemblages, but, as was the case with the Palaeolama sample, the combined percentage of prime adults (37%) is greater than older adults (24%). Table 13. Life tables for the Leiscy lA population of Palaeolama mirvica calculated in the manner of Voorhics (1969) and Hulbert (1982. 1984). The following are reported for each annual age class: the number of deaths in age class x (dx); the number living at the beginning of the year Ox); the percent annual moltality rate (qx = 100dxAx); the estimated annual birth rate (mx); and the estimated number of newborns produced by members of age class x (Ix'nx) Age Ix qx mx 1 951 177 53.7 O.00 0 2 13 82 15.9 0.00 0 3 3 69 4.4 0.23 16 4 7 66 10.6 0.45 30 5 6 59 10.2 0.45 27 6 11 53 20.8 0.45 24 7 7 42 16.7 0.45 19 8 8 35 22.9 0.45 16 9 3 27 11.1 0.45 12 10 7 24 29.2 0.43 11 11 5 17 29.4 0.43 8 12 2 12 16.7 0.43 5 13 3 10 30.0 0.43 5 14 3 7 42.9 0.45 3 15 3 4 75.0 0.45 2 16 0 1 0.0 0.45 17 0 1 0.0 0.23 1 18 1 1 0.11 1 This value includes 55 individuals added to the or*inal 40 to account for juvenic tmderrepresentation (see text). This co~esponds to a juvenile underrepresentation factor of 1.375. 242 BULLETIN FLORIDA MUSEUM NATURAL HISTORY VOL 37, Fr. L No. 7 Table 14. Life table for the L,cisey lA population ofEquus "lei®i. Format as in Table 13. class 4 Ix qx mx 1 571 97 58.8 0.0 0 2 7 40 17.5 0.0 0 3 2 33 6.1 0.23 7 4 4 31 12.9 0.45 14 5 6 27 22.2 0.45 12 6 1 21 4.8 0.45 9 7 4 20 20.0 0.45 9 8 3 16 18.8 0.45 7 9 1 13 7.7 0.45 6 10 1 12 8.3 0.45 5 11 0 11 0.0 0.45 5 12 1 11 9.1 0.45 5 13 3 10 30.0 0.45 3 14 2 7 28.6 0.45 2 15 0 5 0.0 0.45 2 16 1 5 20.0 0.45 2 17 0 4 0.0 0.45 2 18 0 4 0.0 0.45 2 19 1 4 25.0 0.23 20 3 3 - 0.11 1 this value includes 43 mdividuals added to the original 14 to accolmt fof juvenile unde:representation (sce text). This corre3is of Animal Bones from Archeological Sites. Univ. Chicago Press, Chicago, 266 p. Klingel H. 1969. The social organ,dion and population ecology of the plains zebra (Equus quagga) Zool. Africana 4:249-263. Kont W. W. 1979. Taphonomy of microvertebrate fossil assemblages. Ann. Carnegie Mus. Nat- Hist. 48(15):235-285. Kult614 B. 1953. On the variation and population dynamics of fossil and Recent mammal populations. Ada Zool. Fenn. 76:1-122. 1967. Pleistocene bears of North America. 2. Genus Arctodus, short-faced bears. Acta Zool. Fenn. 117:146. . 1983. Variation anddynamics ofa fossil antelope population. Paleobiology 9(1):62.69. Meyer, C. 1991. Burial experiments with marine turtle carcasses and their paleoccological significance. Palaios 6(1):89-96. Myers, T. P., M. R Voorhies, and R. G. Corner. 1980. Spiral fractures and bone pseudotools at pateontological sites. Amer. Antiq. 45(3):483-490. Oliver, J. S. 1989. Analogucs and site context: bone damage from Shield Trap Cave (24CB91), Carbon County, Montana, U.S.A Pp. 73-98 in R. Bonnichsen and M. A SorF eds. Bone Modification. Inst for Quatemary Studies, Univ. Maine, Orono. Olsen. S. L, and P. Shipman. 1988. Surface modification on bones: Trampling versus butchery. J. Archaeo!. Sci. 15:535-533. Pratt A E. 1986. The taphonomy and paleoecology of the Thomas Farm local fauna (Miocene, Hemingfordian1 Gilchrist County, Florida. PAD, dissertation, Univ, Florida, 487 p . 1990. Taphonomy of the large vertebrate fauna from the Thomas Farm locality (Miocene, Hemingfordian). Gilchrist County, Florida. Bull. Florida Mus. NaL Hist 35(2):35-130. Schaller, G. B. 1972. The Serengeti Lion: A Study of Predator-Prey Relations. Univ. Chicago Press, Chicago, 480 p. Shipman„ P. 1981. Life History ofa Fossil: An Introduction to Taphonomy and Paleoecology. Harvard Univ. Press, Cambridge, 222 p. Sinclair, A R E. 1977. The African Buffalo: A Study of Resource Limitation of Populations. Univ. Chicago Press Chicago, 355 p. Spinage, C. A 1972. African ungulate life tables. Ecology 53(4):645-652. 250 BULLETIN FLORIDA MUSEUM NATURAL HISTORY VOL. 37, Fr. I, No. 7 Van Valen 1964. Age in two fossil horse populations. Ada Zool. 45:93-106. Van Valkenburgh. B. 1988. Trophic diversity within past and present guilds of large predatory mammals. Paleobiology 14(2):153-173. 1991. Iterative evolution of hypercarnivory in canids (Mammalia: Carnivora): evolutionary interactions among sympatric predators Pateobiology 17(4):340-362. -, M. F. Teaford. and A Walker. 1990. Molar microwear and diet in large carnivores: inferences concerning diet in the sabertooth cat, Smitodon/atatis. J. Zool. (London) 222:319-340. Voorhies. M. 1 1969. Taphonomy and population dynamics of an early Ptiocene vertebrate fauna, Knox County, Nebraska. Univ. Wyoming Contrib. Geol., Spec. Par. 1:1-69. Webb, S. D., B. J. MacFadden, and J. A Baskin- 1981. Geology and paleontology ofthe Love Bone Bed from the late Miocene of Florida. Amer. J. Sci. 281:513-544. Western, D. 1980. Linking the ecology of past and present inammal conununities. Pp 41-54 in A K Behrensmeyer and A P. Hill, eds Fossils in the Making, Vertebrate Taphonomy and Paleoecology. Univ. Chicago Press Chicago. Wheeler, J. C. 1982. Aging Ilamas and alpacas by their teeth. Llama World 1(2):12-17. Wolir, R. G. 1973. Hydrodynamic sorting and ecology of a Pleistocene mammatian assemblage from California. Palaeogeog- Palaeoclim., Palaeoecol. 13:91-101.