Stable isotopes, hypsodonty, and the paleodiet of Hemiauchenia (Mammalia: Camelidae): a morphological specialization creating ecological generalization

—Morphological adaptations may indicate increased specialization (narrowing of ecological niche) or expansion of the suite of lifestyles available to an organism (increasing niche breadth). Hypsodonty in mammals generally has been interpreted as a specialization into a grazing niche from a browsing niche. Here I examine the feeding strategy of the extinct hypsodont camel Hemiauchenia through an analysis of stable carbon isotope values from its tooth enamel, which was used to clarify its feeding strategy and to resolve conflicting interpretations of dental versus muzzle attributes. The paleodiet of Hemiauchenia is then used to test whether hypsodonty correlates to grazing within fossil Lamini. This study focuses on fossils from Florida, which is geographically ideal because unlike other regions of the country almost all extant plants on which animals browse use the C3 photosynthetic pathway. In contrast, most of the grasses and sedges utilized by grazers use the C4 photosynthetic pathway. If Hemiauchenia was an obligate grazer, the stable carbon isotope values of tooth enamel should reflect primarily a diet of C4 grass and sedge (.21.3‰). If Hemiauchenia was mainly a browser, the isotopic value should be considerably more negative reflecting ingestion primarily of C3 browse (,27.9‰). The mean d13C values for Hemiauchenia during each time interval average more negative than 28.0‰, indicating a dominantly C3 browse diet, and there is no evidence for abandonment of the browsing niche from the Hemphillian through the Rancholabrean North American Land Mammal Ages. However, an increase in the range of isotopic values indicates a diet with a higher proportion of C4 grasses and sedges through time. This study therefore suggests that Hemiauchenia was a hypsodont intermediate feeder with preference for browse during the past 5 million years. Hypsodonty is not strictly associated with obligate grazing; instead it may, in this case, represent an adaptation to widen niche breadth that allowed grazing as well as browsing. Robert S. Feranec. Department of Integrative Biology and Museum of Paleontology, University of California, Berkeley, California 94720. E-mail: [email protected] Accepted: 31 October 2002 Introduction Determining the diet of ancient animals has important implications for interpreting paleocommunity structure. For example, determining whether extinct herbivores are browsers, intermediate feeders, or grazers is required to understand details of resource partitioning, which ultimately influences diversity in ecosystems by affecting competition and allowing for the co-occurrence of species (Hutchinson 1958). The interpretation of feeding strategy for ungulates is that there was a progression through a pathway from browsing to intermediate feeding to grazing, with the ancestral state in diet being browsing and the derived state being grazing (Perez-Barberia et al. 2001). What has been unclear, however, is whether increasing hypsodonty indicates increased specialization for abrasive foods at the expense of browse, or an actual broadening of niche breadth by simply adding the ability to graze. At issue is whether extinct hypsodont animals were obligate grazers, or whether they continued to utilize browse plants in significant quantities and added a grazing component to their feeding strategy. Because fossils are rarely preserved with any direct physical evidence of diet, indirect proxies are commonly used to assess feeding strategies. Dietary proxies include morphological characteristics of the skull, such as hypsodonty of teeth and shape of the muzzle. Such data, however, can lead to conflicting interpretations of diet (Webb 1974; Dompierre and Churcher 1996). A prime example of this conflict involves the feeding ecology of Hemiauchenia. Hemiauchenia is an extinct genus of llama that ranged from the late Barstovian (ca. 13 Ma) to the late Rancholabrean (ca. 0.01 Ma) North American Land Mammal Ages (NAL231 PALEODIET OF HEMIAUCHENIA MA) in North America (Honey et al. 1998). This genus has been suggested to comprise only browsers, according to morphological characteristics of the snout (Dompierre and Churcher 1996), or to include intermediate feeders with a preference for grasses, according to the presence of hypsodont teeth and abundant cementum (Webb 1974; Webb and Stehli 1995). This study uses stable carbon isotopes and measures of hypsodonty to determine the feeding strategy employed by Hemiauchenia during the past 5 million years. The results are then used to assess whether the attainment of hypsodonty within the Lamini evolved through the progression of browsing to intermediate feeding to grazing, and whether hypsodont animals are specialized grazers. Background Information As mentioned above, hypsodonty and the shape of the ungulate muzzle are but two morphological proxies used to interpret paleodiet. Generally, more hypsodont animals tend to feed on grasses and live in open habitats. Because the spread of grasslands during the Miocene appears roughly coincident with widespread hypsodonty, it has been hypothesized that hypsodonty evolved as a coevolutionary response to the spread of phytolithladen grasses, and as an adaptation to deal with grit and dirt that typically coats plants in open habitats (McNaughton et al. 1985; Janis 1988, 1989, 1995; MacFadden 1997). With respect to the shape of the muzzle in ungulates, grazers have a broad muzzle capable of feeding on large quantities of low-lying forage, and browsers generally have narrow muzzles for manipulating through branches to select choice, succulent forage. Intermediate feeders generally have muzzle widths in between (Janis and Ehrhardt 1988; Solounias et al. 1988; Dompierre and Churcher 1996). Modern Ungulate Feeding Categories. Hofmann and Stewart (1972) used forage composition and stomach structure, to characterize modern ungulate feeding strategy into three main types: concentrate selectors (browsers), which focus feeding on flora other than grasses and sedges (browse); bulk and roughage feeders (grazers), which concentrate feeding on grasses and sedges; intermediate feeders, which have a composite diet of grasses and sedges, and browse. These three categories have been adopted in much of the paleoecological literature for interpretation of paleoungulate diets, and for use in paleoenvironment interpretation (e.g., Janis 1988; Quade et al. 1992; Solounias et al. 1995; MacFadden and Cerling 1996; Janis et al. 2000). In using these three categories alone, much of the complexity in feeding in ungulates is obscured; for example, an intermediate feeder does not necessarily have a diet midway between a browser and a grazer. Hofmann and Stewart (1972) suggested further subcategories within the three main categories to reflect some of this complexity. Thus, concentrate selectors can be subdivided into tree and shrub foliage eaters and fruit and dicot foliage eaters. Bulk and roughage feeders include the subcategories roughage grazers, fresh grass grazers, and dry region grazers. Within intermediate feeder are the subcategories preferring grasses and preferring forbs, shrubs, and tree foliage. However, these subcategories are seldom used in paleobiological studies because it generally is difficult to determine the percent of a particular forage type within the diet of an extinct ungulate. Whereas one can observe the forage and droppings of modern ungulates, it is often difficult or impossible to link paleobotanical remains and/or coprolites to particular extinct animals. However, as is explained below, in a restricted geographical area in which the flora is relatively well known, the analysis of stable carbon isotopes in the tooth enamel of extinct herbivores provides a means for determining the percentage of certain forage in the diet. This proxy can be used to distinguish browsers, grazers, intermediate feeder with a preference for grasses and sedges, and intermediate feeder with a preference for browse. Carbon Isotopes and Diet. The carbon isotope values of fossil teeth are useful in paleodiet studies because different photosynthetic pathways impart different ratios of 13C/12C in different kinds of plants, which are ultimately reflected in the animals that eat them. Plants can be grouped into three main photosynthetic pathways: C3, C4, and Crassulacean Acid Metabolism (CAM). Tropical, warm-sea232 ROBERT S. FERANEC son sedges and grasses using the Hatch-Slack photosynthetic pathway (C4) are enriched in the heavy carbon isotope (13C). Conversely, Calvin cycle plants (C3), including most browse and high-latitude grasses, are enriched in the light carbon isotope (12C). The third pathway, the CAM pathway, is characteristic of succulents and incorporates intermediate ratios of 12C and 13C (O’Leary 1988; Ehleringer et al. 1991; Ehleringer and Monson 1993). Animals that feed upon these various kinds of plants will reflect the isotope ratio (13C/12C) ingested (DeNiro and Epstein 1978; Quade et al. 1992). The 13C/12C ratio of tooth enamel is further enriched by a consistent amount, which is about 114‰ for the tooth enamel of mediumto large-bodied mammals (Cerling and Harris 1999). Although post-depositional diagenesis can overprint carbon isotope values in bone (Schoeninger and DeNiro 1982), tooth enamel is not prone to diagenetic alteration and thus reliably reflects isotope values derived from feeding (Quade et al. 1992; Wang and Cerling 1994; Koch et al. 1997). As a general guide, grazers are predominantly C4 feeders, browsers are predominantly C3 foragers, and intermediate and CAM feeders utilize both C3 and C4 plants. Care must be taken when interpreting carbon isotopic values in herbivore tissue because some grasses, especially those above 438N latitude, use the C3 pathway; some browse uses the C4 pathway, especially in lower latitudes and drier climates; and animals that are strict CAM feeders can show values similar to either C3 or C4 feeders (Teeri and Stowe 1976; Stowe and Teeri 1978; O’Leary 1988; Tieszen et al. 1997). Because CAM flora is generally only dominant in xeric habitats (Ehleringer et al. 1991), it is assumed that Hemiauchenia did not use this flora as a significant proportion of its diet. Florida as a Model Location. Florida is ideal for this study because (1) the flora eliminates many of the discrepancies that can occur when interpretation is based on isotopic values, and (2) because Hemiauchenia is abundant in late Tertiary through Quaternary deposits. Nearly all present-day browse in Florida is C3 (.97% [Stowe and Teeri 1978]), whereas most of the grasses and sedges are C4 (.63% for Poaceae, and up to 43% for Cyperaceae [Teeri and Stowe 1976; Teeri et al. 1980; Sage et al. 1999]). Two studies of modern floral composition on sites near fossil localities within this study show that even though C3 grasses and sedges may be present and represent up to 40% of the species richness of grasses, C3 grasses are generally confined to wet and marshy areas and are not widespread on the landscape (Easley and Judd 1990; Huffman and Judd 1998). Many of the fossil localities within this study are suggested to have floras similar to the interglacial floras of today (Webb 1991; Rich and Newsom 1995), and models of past floras during glacial stages would suggest the spread of grasslands and an increase in dominance of grasses and sedges using C4 photosynthesis due to decreasing CO2 levels (Webb 1991; Ehleringer et al. 1997; Cowling 1999). This would suggest that percentages of C3 grasses in Florida are unlikely to have been significantly greater during the past, and because C4 browse is rare, it is unlikely that these percentages have significantly changed. Materials and Methods Definitions and Feeding Strategies. Grazers are strictly defined in this study as animals that eat grasses and sedges (Poaceae and Cyperaceae) as at least 90% of their diet. Browsers are considered here to have diets consisting nearly exclusively of non-grasses and sedges, that is browse (.90%). Intermediate feeders eat both grasses and sedges, and browse (i.e., .10% grasses and sedges, and browse). This type of designation is appropriate because anatomical and physiological differences characterize animals that focus feeding on grasses and sedges and/or browse. Grazers, for example, must deal with abundant abrasive phytoliths and grit from feeding close to the ground, and accordingly develop strategies to deal with tooth abrasion. Browsers, on the other hand, ingest more anti-herbivory chemicals such as tannins (McNaughton 1991; Heckathorn et al. 1999), which necessitate strategies to deal with these secondary chemicals, but tooth abrasion offers less of a selective pressure than it does for grazers. The definitions used in this paper parallel the cate233 PALEODIET OF HEMIAUCHENIA FIGURE 1. Fossil localities in Florida of the Hemiauchenia samples used in this study. Items in parentheses refer to the number of specimens from the locality and to the North American Land Mammal Age. Abbreviations as follows: H 5 Hemphillian, B 5 Blancan, I 5 Irvingtonian, R 5 Rancholabrean. gories of Hofmann and Stewart (1972) for ease in comparison with modern ungulate analogs and other paleoecological studies. Where possible the subcategories prefers grasses and sedges and prefers browse similar to Hofmann and Stewart (1972) are also used for intermediate feeders. Stable Isotope Analysis. Hemiauchenia molars were obtained from the Vertebrate Paleontology collection at the Florida Museum of Natural History (FLMNH) and include four species: H. vera, H. blancoensis, H. seymourensis, and H. macrocephala. The teeth range in age from the latest Hemphillian (ca. 5 Ma) to the latest Rancholabrean (ca. 10 Ka) NALMAs and were taken from sites extending through the entire peninsula of Florida (Fig. 1). The ages of the included localities were determined biochronologically by assignment to North American Land Mammal Ages (Lundelius et al. 1987; Woodburne 1987; MacFadden and Cerling 1996). The Leisey Shell Pit locality has also been paleomagnetically dated at 1.5 Ma (MacFadden 1995). The general method for stable isotope sampling followed MacFadden and Cerling (1996) and Koch et al. (1997). Because the goal of the study was to recognize the feeding strategy of adult animals, sampling focused on but was not confined to the second through fourth premolars and third molar. These teeth are among the last ones to develop, mineralize, and erupt (Hillson 1986; Bryant et al. 1996 a,b). Sampling involved drilling 20–30 mg of pristine enamel powder off the tooth along a non-occlusal surface parallel to the growth axis and across its entire length using a 0.3mm inverted cone carbide drill bit and a variable speed Dremel y rotary tool. The powder was first collected and treated with 35% hydrogen peroxide for 24 hours to remove organics. It then was decanted and washed with distilled water and soaked in 0.1 N acetic acid for another 24 hours to remove any diagenetic carbonate. The following day it was again decanted and washed with distilled water, rinsed with 100% ethyl alcohol, and dried overnight. After treatment, the samples were analyzed by using an ISOCARB automated carbonate preparation system attached to a Micromass Optima gas source mass spectrometer within the Department of Earth and Ocean Sciences at the University of California, Santa Cruz. The ;1-mg samples were dissolved in 100% phosphoric acid at 908C to create CO2. The results are expressed in standard d-notation: X 5 [(Rsample/Rstandard) 2 1] 3 1000, where X is the d13C value, and R 5 13C/12C. All isotope values are reported relative to V-PDB. The precision for the analysis was 60.1‰. A total of 31 samples were collected, prepared, and analyzed along with the data for 19 samples obtained from the published literature (MacFadden and Cerling 1996; Feranec and MacFadden 2000). Mean differences in Hemiauchenia between NALMAs were compared by post hoc LSD tests, which are similar to t-tests but take into account multiple comparisons. Statistical analyses were run on Microsoft Excel 2000 and SPSS Student Version 8.0 for Windows, with significance set at p , 0.05. Carbon Isotope Values in Paleoecology. As noted above, herbivores will reflect the carbon isotope values of plants ingested. Modern C3 plants—trees, shrubs, and high-latitude grasses—have a carbon isotope value of 227.0‰ 6 3.0‰, whereas modern C4 plants—tropical sedges and grasses—have a mean isotopic value of 213.0‰ 6 2.0‰ 234 ROBERT S. FERANEC (O’Leary 1988; Quade et al. 1992; Koch et al. 1998). Because of the 114.0‰ fractionation from plant material to tooth enamel in large herbivores (Cerling and Harris 1999), modern C3 feeders will show enamel carbon isotope values between 216.0‰ and 210.0‰, whereas modern C4 feeders display isotopic values between 21.0‰ and 13.0‰. These enamel isotope values are based on the d13C value of the modern atmosphere, which has decreased from about 26.5‰ to about 28.0‰ as a result of fossil fuel burning over the last two hundred years (Friedli et al. 1986; Marino and McElroy 1991; Marino et al. 1992). Because d13C values of past glacial and interglacial periods are different from those of today, fossil enamel isotope values are about 0.5‰ to 1.3‰ more positive than in the modern taxa (Koch et al. 1998). For fossil enamel then, a diet of pure C3 plants would not have an isotopic value more positive than 28.7‰, and a diet of pure C4 plants would not have an isotopic value more negative than 20.5‰. Using these values as a continuum between 100% C3 feeding and 100% C4 feeding, values more negative than 27.9‰ are interpreted to indicate primarily browsing (,10% C4 diet), values more positive than 21.3‰ are interpreted to signal primarily grazing (,10% C3 diet), and values between 27.9‰ and 21.3‰ result from intermediate feeding strategies. Hypsodonty Measurements. Hypsodonty measurements for Hemiauchenia teeth were taken on specimens at the FLMNH and compiled from the published literature (Webb 1974; Webb and Stehli 1995). Most measurements come from second lower molars, and when possible the same teeth were sampled isotopically. Measurements of crown height and width were taken on teeth that were relatively unworn, and represent the greatest height and width for each tooth. Tooth crown measurements (height and width) for extant camelids, Camelus, Lama, and Vicugna, were acquired from the Museum of Vertebrate Zoology (MVZ) at the University of California, Berkeley. The Hypsodonty Index (HI) was calculated as HI 5 Crown Height/ Crown Width. Even though relatively unworn teeth were measured when possible, the HI represents a conservative estimate because any wear or incomplete eruption reduces the numerator, crown height. Because of wear and small sample sizes, these measurements are intended to show that Hemiauchenia was hypsodont but not to show the level of hypsodonty for any particular species or at any particular time. Relative degree of hypsodonty among species is provided by Webb (1974). Modern Camelids and Modern Analogs. Understanding the diets of modern camelids aids in determining the feeding strategy of Hemiauchenia. Modern camelids can be placed within three genera and up to six species: Camelus (including C. bactrianus and C. dromedarius), Lama (including L. guanicoe, L. glama, and L. pacos), and Vicugna (including V. vicugna). It is still uncertain whether the two identified species of Camelus or the three species of Lama are conspecific (Nowak 1999). Because of these taxonomic similarities and the similarity in morphology and ecology between the modern species (Nowak 1999), I treat them on the generic level. Studies on modern Camelus suggest that it is mainly a browser feeding predominantly, if not solely, on trees and shrubs, but is capable of feeding and subsisting on a significant quantity of grasses (.30%) when necessary (Gauthier-Pilters 1984; Newman 1984; Migongo-Bake and Hansen 1987; Kohler-Rollefson 1991; Nowak 1999). Camelus is therefore classified as an intermediate feeder with preference for browse. Modern Lama has been suggested to be an ‘‘adaptable mixed feeder’’ (Puig et al. 1996: p. 221) showing a preference for feeding on grasses, but fully able to feed on browse (Puig et al. 1996, 1997; Fraser 1998, 1999; Nowak 1999). Lama is classified here as an intermediate feeder with preference for grasses and sedges. Koford (1957) suggested that the main food of the Vicugna is grass, but that other forage such as forbs and the legume Astragalus was a prized food when it was, though rarely, available. Vicugna is classified here as a grazer. Isotopically, all modern camelid genera would be expected to exhibit a C3 diet. This is because C3 grass genera predominate where camelids live (Koford 1957; Puig et al. 1996, 1997; Fraser 1998, 1999; Sage et al. 1999). This study also uses a modern analog comparison similar to the Extant Phylogenetic Bracket (EPB) method to determine if HI is an 235 PALEODIET OF HEMIAUCHENIA FIGURE 2. Stable carbon isotope values of Hemiauchenia from the Hemphillian through the Rancholabrean of North America. osteological correlate that may be ‘‘causally related’’ to obligate grazing (Witmer 1995: p. 23). The comparison in this study differs from an EPB comparison in that a cladistic analysis with hypsodonty characters was not completed and is beyond the scope of this study. The modern analog comparison should nonetheless work in the same fashion as the EPB methodology. Therefore, if hypsodonty is a good osteological correlate for obligate grazing within fossil Lamines, then all taxa that have high-crowned teeth (the osteological correlate) should be grazers, or at least intermediate feeders with preference for grasses and sedges (the behavior or ecology). The stable isotope value of fossil enamel can be used as an independent indicator for grazing as suggested above. In this study, I use the phylogeny proposed by Honey et al. (1998), (Camelus, (Hemiauchenia, (Lama, Vicugna))), where the fossil genus Hemiauchenia is bracketed by the extant genera Camelus and Lama-Vicugna.