Use of tusks in feeding by dugongid sireniansObservations and tests of hypotheses.код для вставкиСкачать
THE ANATOMICAL RECORD 290:523–538 (2007) Use of Tusks in Feeding by Dugongid Sirenians: Observations and Tests of Hypotheses 1 DARYL P. DOMNING1* AND BRIAN L. BEATTY2 Laboratory of Evolutionary Biology, Department of Anatomy, Howard University, Washington, DC 2 Anatomy, New York College of Osteopathic Medicine, Old Westbury, New York ABSTRACT Most living and fossil sea cows of the subfamily Dugonginae (Dugongidae, Sirenia, Mammalia) are characterized by large upper incisor tusks, which are thought to play an important role (at least primitively) in feeding on seagrass rhizomes. Testing this hypothesis is difficult, because the only extant tusked sirenian (Dugong dugon) is morphologically and perhaps behaviorally aberrant. The tests attempted here involve examination of stomach contents of wild Recent dugongs, experiments using plastic replicas of diverse tusks to harvest seagrasses, gross anatomical observations on tusks and skulls, measurements of tusk tip geometry, and observations of microwear on tusks. We conclude that (a) male D. dugon (with erupted tusks) do not consume more rhizomes than females (without erupted tusks); (b) the tusks do not play a significant role in feeding in the modern dugong; (c) larger, more bladelike tusks are more effective at harvesting rhizomes, but the effect of shape was not experimentally separated from the effect of exposed tusk length; (d) some fossil dugongines show apparent cranial adaptations for downward and backward cutting motions of their large, bladelike tusks; (e) geometry of wear surfaces is consistent with use of at least the more bladelike tusks as cutting instruments; (f) preliminary observations of microwear in D. dugon do not indicate more than occasional use of the tusks in purposeful harvesting of rhizomes, and then only opportunistically by large adult males. The hypothesis of such tusk use by extinct dugongines (in contrast to the living species) is so far corroborated, but available data and tests do not suffice to establish this conclusively. Anat Rec, 290:523–538, 2007. Ó 2007 Wiley-Liss, Inc. Key words: Sirenia, Dugongidae; Dugong dugon; tusk; feeding; seagrass rhizomes; stomach contents; dental microwear A quantitatively significant but poorly understood aspect of mammalian history is the long relationship between herbivorous marine mammals and the plants they have eaten. This history extends over the 50 million years of the fossil record of these mammals (mainly those in the Order Sirenia: dugongs and manatees), and chiefly involves the tropical-to-temperate, cosmopolitan marine angiosperms or seagrasses (Hydrocharitaceae and Potamogetonaceae), which today constitute one of the Earth’s most productive ecosystems, as has presumably been true throughout the Cenozoic (Ivany et al., Ó 2007 WILEY-LISS, INC. 1990). The greatly diminished diversity of modern sirenians, however, is a major obstacle to exploring how feeding adaptations, niche partitioning, and other *Correspondence to: Daryl P. Domning, Department of Anatomy, Howard University, Washington, DC 20059. E-mail: email@example.com Received 2 March 2007; Accepted 6 March 2007 DOI 10.1002/ar.20540 Published online in Wiley InterScience (www.interscience.wiley. com). 524 DOMNING AND BEATTY aspects of the order’s ecology evolved (Domning, 2001; see also review in Uhen, 2007, this issue). The typical seagrass is a submerged marine plant that comprises parts both above ground (shoots and leaves) and below ground (roots and rhizomes). In most seagrasses, the rhizomes or underground stems compose at least 50% of the plant’s biomass and are a rich source of carbohydrates in the form of sugars and starch, although the exact distributions of biomass and nutrients within the plant seem to vary seasonally (Lanyon et al., 1989). Therefore, they are a potentially significant resource for marine herbivores able to extract them. Several seagrass herbivores (dugongs, sea turtles, fish, etc.) often occur sympatrically and have to compete for plant resources in a community typically comprising fewer than a dozen angiosperm species. Therefore, a species’ ability to extract seagrass rhizomes may be critical to its long-term survival in the system and to the overall pattern of niche partitioning among these competitors. Here, we examine this ability in the extant dugong (Dugong dugon) and its extinct relatives. Although green turtles (Chelonia mydas) are major seagrass consumers, they do not disturb the rhizomes, and they also eat large amounts of algae, in contrast to manatees (Trichechus spp.) and dugongs (Lanyon et al., 1989; André et al., 2005; cf. simulation study by Aragones and Marsh, 2000); so it is doubtful whether they compete significantly with dugongs or manatees at the present time. With the possible exception of some members of the extinct mammalian Order Desmostylia, which were restricted to the North Pacific Ocean (Domning, 1978), no other marine animals are known to extract significant amounts of seagrass rhizomes (McRoy and Helfferich, 1980; Klumpp et al., 1989). Sirenians, then, are the only known herbivores that have had the capacity to exploit rhizome resources in the tropical seagrass beds of most parts of the world at any time during the Cenozoic Era. Moreover, sirenians are almost exclusively consumers of angiosperms and eat algae in significant amounts only when angiosperms are scarce (e.g., after storms, Spain and Heinsohn, 1973; or in the range of the extinct Steller’s sea cow [Hydrodamalis], Domning, 1978); so far as is known, this has always been true. Sirenians also eat animals on occasion (e.g., Powell, 1978; Preen, 1995), but never, apparently, as a major part of the diet. We, therefore, seem justified in considering tropical marine sirenians as obligate seagrass consumers for all practical purposes, and as the only potential consumers of seagrass rhizomes. These conclusions apply in particular to the Dugongidae, which during the Holocene, and apparently throughout their history, have been exclusively marine and (except for hydrodamalines) tropical. Increasing evidence is emerging for the sympatric occurrence of three or more dugongid species at various times and places in the geological record (Domning, 1989a, 2001; Toledo and Domning, 1991; Aranda-Manteca et al., 1994). The presence of more than one sirenian species in a tropical seagrass community will likely prove to have been the rule and not the exception throughout the Tertiary history of seagrasses. This finding becomes problematical when viewed in light of the low taxonomic diversity of present-day seagrasses (only 12 genera and 50 or so species worldwide; Phillips and Meñez, 1988). Moreover, the fossil record of seagrasses, scant though it is, gives no hint that this diversity was ever significantly greater: all known Tertiary seagrasses, even as far back as the Eocene, are referred to living genera, and many to living species (Larkum and den Hartog, 1989; Ivany et al., 1990). There is evidence (Lumbert et al., 1984; Ivany et al., 1990) that the Caribbean seagrass flora was more diverse in the Eocene than it is today, but not more diverse than the Indopacific seagrass flora of today. Thus, we are faced with the general problem of explaining how it was possible for several sirenian species to coexist on such a narrow resource base. Several potential kinds of answers present themselves: (a) seagrass communities were more diverse in the past than their fossil record now reveals; (b) marine sirenians had more varied diets (including more algae, freshwater plants, and/or animals); or (c) it is in fact possible for several sirenian species to stably partition a seagrass assemblage of the level of diversity observed today. The main methodological problem with any such venture into paleoecology is that of maintaining some link with the realm of testable hypotheses. The more factors allowed to vary from the present in a reconstruction of the past, the fewer and more tenuous such links become. It has, indeed, been suggested (Sickenberg, 1934) that the extinct sirenian Miosiren fed on molluscs, but this possible exception by itself does not threaten to change the overall picture outlined above. This study will, therefore, proceed on firmer ground by first exploring (and, if need be, eliminating) the alternative that postulates the least divergence from modern conditions. An initial working assumption is that tropical seagrass diversity (at least at the generic level) has not fluctuated dramatically during the Cenozoic and that all the tropical dugongids of the past were like the modern Dugong in their reliance on a seagrass diet. Recent studies on the digestive system of the living D. dugon, together with much else that we know about their anatomy and physiology, have shown that modern Sirenia are extremely efficient in minimizing energy use (Aketa and Kawamura, 2001; Aketa et al., 2001, 2003). Despite what is known about their internal digestive efficiency, however, little is known about their means and efficiency of food acquisition. Large tusks formed by the first upper incisor teeth are a prominent feature of D. dugon and many of its extinct relatives in the Family Dugongidae. These tusks vary considerably within and among taxa in size, shape, and degree of eruption and wear. In cases where several genera of dugongids occur sympatrically in the fossil record, differences in tusk morphology are among their most obvious points of contrast in presumably adaptive features and may offer some of the best clues to strategies of niche partitioning among these animals (Domning, 1989a, 2001). However, little direct evidence of the actual function of sirenian tusks has been reported. Display as a major function seems inconsistent with the fact that most of the tusk is always embedded in (and supported by) the premaxilla, with only a small portion exposed, and even that portion mostly concealed by the upper lip—a design better adapted for forceful use than for show. The tusks’ past and/or present use in feeding, specifically for the extraction of seagrass rhizomes, has been conjectured (Domning, 1989a–c, 2001), but attempts to directly observe tusk use for feeding by wild USE OF SIRENIAN TUSKS IN FEEDING dugongs (e.g., Vosseler, 1924–25) have not succeeded, and this conjecture is still in need of empirical support. This study seeks to provide this support by bringing a diversity of indirect techniques to bear on the question. Only a single species of tusked sirenian (Dugong dugon) survives today to give us possible insight into this problem. Unfortunately, the modern dugong is atypical of its family in that its tusks are sexually dimorphic: although large in both sexes, they erupt and become worn only in adult males and in a few old, possibly postreproductive females (Marsh, 1980; Domning, 1995). This has led some writers (e.g., Anderson, 2002) to conclude that they serve only a social function. They are used by the males when attempting to mate, as weapons against rival males and possibly as instruments to roll females into position for mating; the pairs of parallel scars that are abundant on most dugongs are attributed to the tusks of aggressive males (Anderson and Birtles, 1978; Anderson, 1979; Marsh et al., 1984; Preen, 1989). This, however, leaves two questions unanswered: do male dugongs also use their tusks in feeding, and did the tusks of their extinct, nondimorphic relatives function in feeding or only in social interactions? The following investigations were performed to answer these questions: (1) examination of dugong stomach contents to see whether adult males consume more or larger seagrass rhizomes than adult females; (2) experiments to determine the relative efficiency of different shapes and sizes of dugongid tusks as tools for rhizome extraction; (3) observations of the gross morphology of the tusks, rostral architecture, and other oral structures of Dugong and related fossil forms; (4) measures of tip geometry of tusks to determine their relative utility as tools; and (5) observations of microwear on Recent and fossil dugongid tusks. Tasks 1–3 were carried out by Domning, and tasks 4 and 5 by Beatty. The Recent dugongs studied were from northeastern Australia and Papua New Guinea, where there exists a diverse flora of seagrasses with rhizomes of several different sizes. Dugongs in this region reportedly feed largely on species of Halophila and Halodule (Johnstone and Hudson, 1981; Marsh et al., 1982), whose rhizomes are <2 mm in diameter. Less frequent in the dugong diet are species of Cymodocea, Zostera, and Syringodium (rhizomes 1–3 mm); Thalassia hemprichii (2–5 mm); and Enhalus acoroides (10–15 mm; den Hartog, 1970; Meñez et al., 1983; Phillips and Meñez, 1988). Dugongs use at least two different modes of feeding: cropping of seagrass leaves (e.g., when feeding on Amphibolis; Anderson, 1986), and rooting up the whole plant, during which their characteristic feeding trails are produced (e.g., Anderson and Birtles, 1978; De Iongh et al., 1995). These trails are typically approximately 10–25 cm wide (roughly the width of an adult dugong’s facial disk) and several meters long; 1–3 trails are produced on a single dive (Heinsohn et al., 1977; Anderson and Birtles, 1978; De Iongh et al., 1995, 1997; Anderson, 1998). Such trails appear to be made most often by dugongs eating the smaller, more delicate seagrasses (Halophila, Halodule) growing on soft substrates, and less frequently on larger seagrasses (Zostera, Syringodium). Dugongs feeding in this manner typically stir up conspicuous clouds of silt (Anderson and Birtles, 1978) and are probably less vigilant for predators; they harvest rhizomes in this manner when predation risks from 525 sharks are absent, and for short periods (11% of foraging effort) if risk of predation by sharks is minimal but present (Wirsing, 2005). It is thought that a protruding, soft-tissue knob on the upper jaw is used to plow through the upper 4–6 cm of sediment, turning up the rhizomes together with the shoots and leaves, which are then tucked into the mouth by the upper lip and its enlarged prehensile bristles (Gudernatsch, 1908; Marshall et al., 2003; H. Marsh and A. Preen, personal communication). This plowing through sediment can presumably cause incidental wear on the tusks, whether these are intentionally used for digging or not. Arguably, a third mode of feeding occurs as a limitcase of trail-making. While foraging on leaves and rhizomes of small seagrasses like Halophila and Halodule, which offer little resistance to the animal’s forward movement and require no tusk use for their excavation, dugongs are able to maintain headway, making feeding trails on the bottom. As the seagrass bed becomes denser, however, the feeding trails (each made on a single breath-hold dive) become shorter (Anderson and Birtles, 1978); and when dugongs feed on plants or invertebrates that are very resistant to excavation, they must remain in one spot during each dive and, hence, make a pit rather than a linear trail (Anderson, 1998; Domning, 2001). As the rhizomes of their food plants vary toward the larger and more fibrous (as well as more deeply growing), dugongs apparently tend to feed more exclusively on the leaves. Harder substrates, and greater density of rhizome mats, would also discourage extraction of rhizomes. Dugongs extracted 75% of rhizome-root biomass from a Halodule-dominated meadow in one study (De Iongh et al., 1995), and are probably capable of extracting even more, whereas Florida manatees (Trichechus manatus latirostris) feeding on Syringodium beds can remove up to 96% of the total biomass despite their complete lack of tusks (Packard, 1984). Therefore, it seemed necessary, in examining the stomach contents (the first study reported here), to concentrate on dugongs eating still larger seagrasses (Thalassia) to determine the limits of Dugong’s ability to excavate rhizomes. STOMACH CONTENTS Materials and Methods The hypothesis that male dugongs use their tusks in extracting rhizomes from the bottom was tested by examining mouth or stomach samples from dugongs that had been feeding on Thalassia. Johnstone and Hudson (1981) reported analyses of such mouth samples from Daru, southern Papua New Guinea. These samples are now stored at James Cook University of North Queensland, Townsville, Australia (JCU). Thalassia was the most abundant seagrass in the collection area and was also the most abundant species in 52 of the 102 mouth samples studied. Unfortunately, records no longer exist to show which 52 animals these were. Neither were the great majority of the mouth samples themselves preserved, nor were the skulls of most of the dugongs collected. However, stomach contents (preserved in 10% formalin–seawater) were kept from most of these specimens. Therefore, all the adult (hence, presumably tusked) males (n ¼ 35) from the list of 102 specimens studied by Johnstone and Hudson (1981) were selected 526 DOMNING AND BEATTY for study. Of these, 27 are represented by stomach contents. Adults (of either sex) were defined as animals 240 cm or more in body length, because sexual maturity appears to be reached by both sexes at variable body lengths between 220 and 250 cm (Marsh et al., 1984). To increase sample size, stomach samples were also examined from an additional 11 adult males in the same collection that had not been represented by mouth samples in Johnstone and Hudson’s study (namely, those that had been collected in 1978 and the first half of 1979). This period was selected because none of the eight samples examined that had been collected later was found to contain substantial amounts of Thalassia as defined below. Therefore, the total sample of 38 was selected to maximize the likelihood that the stomach samples examined would contain substantial Thalassia (Table 1). Five smears (slides) of each sample were made and scanned with a dissecting microscope at 310. Each stomach sample was assigned to one of four categories according to the approximate proportion of Thalassia leaf fragments present: none, small amounts (<5%), moderate amounts (5–10%), or substantial amounts (>10%). These visual estimates were quantified and calibrated by measurements on 19 of the samples, using a Weibel graticule and counting the leaf material seen at 50 points on each of the 5 slides. Thalassia rhizomes were searched for on the slides and in the remainders of the samples. Because this analysis produced little or no evidence of feeding on Thalassia rhizomes by males (see below), examination of samples from females was deemed superfluous for testing the hypothesis. To detect possible sexual differences in consumption of the next smaller category of rhizomes, stomach samples from Queensland, Australia, were examined. These were collected from animals either drowned in shark nets or killed by native hunters between 1968 and 1978, and are also stored at JCU. From 95 of these samples that were analyzed in detail by Marsh et al. (1982), the samples of adult males and females whose stomachs contained 10% or more of Cymodocea, Zostera, or Thalassia were selected. Unfortunately, these amounted to only 8 individuals: 5 males (at least 4 and perhaps all with erupted tusks) and 3 females (all with unerupted tusks). Five smears from each were examined at 310; using the Weibel graticule, plant material at 50 points per slide was classified as either rhizome of <2 mm diameter, 2–3 mm diameter, >3 mm diameter, or nonrhizome. Results The results of the analysis of stomach samples from Daru, southern Papua New Guinea, are shown in Table 1. No dugongs showed evidence of having consumed significant amounts of Thalassia rhizomes, not even those that had been eating substantial quantities of Thalassia leaves. Of the 38 stomachs, 8 contained rhizome fragments that appeared to represent Thalassia, and such fragments occurred most often in the stomachs containing the most Thalassia leaves, as would be expected if the rhizome material was correctly identified. However, these fragments were few in number, small in diameter for Thalassia (3 mm or even less), and constituted a negligible fraction of the stomach contents. The eight stomach samples from the Queensland, Australia, collection contained various amounts and mix- TABLE 1. Analysis of Thalassia content of stomach samples of adult male Dugong dugon from Daru, Papua New Guinea. Animals whose tusks are known to have been erupted are indicated by T; tusks are assumed to have been erupted in all cases. Proportion of Thalassia leaves as a percentage of plant material in the stomach is indicated where this was measured; in other cases assignment among the four general categories was by visual estimate. Stomachs in which one or more fragments of Thalassia rhizome were identified are indicated by R Specimen No. Body Length (m) Thalassia Tusks? NO THALASSIA 052 2.83 122 2.60 139 2.67 142 2.60 145 2.55 LITTLE THALASSIA 049 2.53 054 2.55 124 2.45 132 2.45 134 2.70 147 2.59 156 2.75 158 2.70 181 2.75 182 2.60 190 2.55 202 2.50 229 2.55 247 2.54 249 2.60 270 2.55 276 2.80 MODERATE THALASSIA 107 2.45 126 2.40 130 2.60 135 2.51 143 2.70 232 2.45 SUBSTANTIAL THALASSIA 051 2.45 055 2.55 068 2.48 071 2.57 075 2.47 094 2.50 120 2.45 127 2.78 177 2.60 183 2.67 T T T T ? ? ? T ? ? T T T T T ? ? ? ? T ? T T T ? ? T ? ? ? ? ? T ? T T T T % Leaves Rhizomes? 0.0 0.0 4.4 4.4 1.6 2.8 2.4 2.4 0.4 4.8 0.4 1.2 0.4 R 7.2 7.6 R R 8.0 R R R 16.8 R R 14.4 25.6 tures of Cymodocea, Zostera, and/or Thalassia in addition to smaller amounts of other seagrasses and algae. The proportion of plant material in these samples previously identified by the JCU workers as leaves of one or more of these three relatively large seagrasses averaged 45.8% 6 30.88 (SD) in the five male samples and 44.3% 6 1.53 in the three females. They also classified as rhizomes (of all species) 40.2% 6 26.30 of the male stomach contents and 49.0% 6 3.61 of the female stomach contents. Domning’s own classification of the rhizomes in these samples into USE OF SIRENIAN TUSKS IN FEEDING three categories gave the following results: rhizomes <2 mm in diameter: males, 15.7% 6 8.70, females, 19.7% 6 7.43; rhizomes 2–3 mm in diameter: males, 3.3% 6 3.69, females, 5.6% 6 7.99; rhizomes >3 mm in diameter: males, 0.3% 6 0.72, females, 0.5% 6 0.92. Discussion Although Thalassia has more distinctive rhizomes than most seagrasses, identification of chewed-up rhizomes in general is problematical, and it is conceivable that unrecognizably macerated Thalassia material was overlooked in the Daru, southern Papua New Guinea, samples; however, if this was present to any important extent, then recognizable fragments should have been more in evidence. (Cymodocea rhizomes in the stomach samples were readily recognized.) The fragments present were consistent with what might have been ingested incidentally to feeding on Thalassia leaves plus leaves and rhizomes of other species in a mixed-species bed, especially if current scour had exposed occasional patches of Thalassia rhizomes on the surface. It was concluded that the male dugongs collected at Daru were not consuming significant amounts of Thalassia rhizomes, and, therefore, were not using their tusks for this purpose. Next considered was the possibility that Thalassia rhizomes might be beyond the ability of Dugong to extract, but that feeding on seagrasses with somewhat smaller rhizomes might reveal a difference between the sexes. The collection of Queensland, Australia, dugongs provided a set of stomach samples whose species content was approximately known. This allowed selection of those with the desired species content (Cymodocea, Zostera, and/or Thalassia), but this yielded only a very small sample of individuals (n ¼ 8), as the Queensland dugongs had been feeding mostly on Halophila and Halodule. The total proportion of rhizomes in the Queensland stomach samples was consistently underestimated, in comparison with the estimates of earlier workers, probably because of some difference in our working definitions of rhizome fragments. However, it is clear that the males in no way surpassed the females in proportion or size of rhizomes consumed; rather the contrary. Therefore, neither the Australian nor the Papua New Guinean samples of dugongs showed evidence of greater consumption of large rhizomes by the males. Although these samples are not as large or as well suited to answering the question as could be desired, in the absence of other evidence, it appears that the tusks of the males do not play a significant role in feeding. This, of course, was to be expected from the fact that the females lack erupted tusks (unless diet is sexually bimodal as well). However, the failure of present-day dugongs to use their tusks for feeding does not prove that they or their relatives did not do so in former times. Hence, the seagrass-harvesting experiments were designed to answer the question: If the tusks were to be used for rhizome removal, how well would they work? SEAGRASS-HARVESTING EXPERIMENTS Materials and Methods To evaluate the relative effectiveness of different types of dugongid tusks as tools for excavating seagrass rhi- 527 zomes, plastic casts of fossil tusks were used by hand as digging tools. Casts (Fig. 1) were made of the following specimens: one halitheriine, Metaxytherium floridanum (USNM 356686; middle Miocene, Florida from U.S. National Museum of Natural History [USNM], Washington, DC), and three dugongines, Crenatosiren olseni (holotype, UF/FGS V6094; latest Oligocene, Florida from former Florida Geological Survey [FGS] collection, now housed in the Florida Museum of Natural History, University of Florida [UF], Gainesville), Dioplotherium manigaulti (ChM PV2633; late Oligocene or early Miocene, South Carolina, from Charleston Museum, South Carolina), and Corystosiren varguezi (USNM 425695; early Pliocene?, Florida; broken tip restored to match other specimens referred to Corystosiren and Rytiodus). These taxa were selected to span the spectrum of size and shape variation of dugongid tusks (cf. Domning, 2001). The original specimens are described in detail by Domning (1988, 1997, 1989b, 1990a), respectively. On the models, no attempt was made to reproduce the difference in hardness between enamel and dentine; only the effect on efficiency of overall size and shape was tested. An actual tusk of an Australian Dugong dugon was also used for comparison in each experiment. Although only one example of each of the fossil tusks was available, resulting in pseudoreplication at the level of sirenian species, this is unlikely to alter the results as the interspecific variation across the range of tusk morphologies used was vastly greater than observed intraspecific variation. The trials were conducted at two sites in northern Queensland, Australia: one stand of a seagrass with relatively thick rhizomes (Thalassia hemprichii), and one of intermediate thickness (Cymodocea serrulata). (As indicated above, and as confirmed by a pilot study in Florida [Domning, 1990b], all sizes and shapes of dugongid tusks are equally effective at extracting rhizomes of Halodule size or smaller.) The Thalassia bed was located on silty coral sand (7.4% silt, 89.2% sand, 3.5% larger material) on the reef platform at Clack Island, northern Great Barrier Reef (148 040 S, 1448 150 E); the seagrass cover at this site was visually estimated at 60% and the leaf height was approximately 7 cm. The Cymodocea bed was located on coarser coral sand (0.0% silt, 93.1% sand, 6.8% larger material) in Cockle Bay, Magnetic Island, Townsville (198 100 S, 1468 500 E). (Comparable beds of these species or genera growing on the same substrate could not be located, although this would have improved the design of the experiments.) The Cymodocea beds sampled unavoidably contained a mixture of C. serrulata and Halodule uninervis (wideleaved form). In each trial, a tusk or tusk model was used to excavate all seagrass material from a 25 x 25 cm area. The tusk was held in one hand, directed anteroventrad, with approximately the same length of its tip exposed as would have been outside the gum in the living animal (here termed ‘‘effective tusk length’’: Metaxytherium, 1.5 cm; Crenatosiren, 2–3 cm; Dioplotherium, 3–5 cm; Corystosiren, 8–9 cm; Dugong, 3–4 cm). Backward and downward (i.e., posteroventrad with respect to the tusk’s morphology) strokes roughly 10 cm long or less were used to penetrate the sediment and sever any rhizomes contacted. (The choice of backward rather than forward movement is justified below under Cranial Architecture.) 528 DOMNING AND BEATTY Fig. 1. Plastic replicas of fossil sirenian tusks used in seagrassharvesting experiments: a: Metaxytherium floridanum (USNM 356686; left tusk in premaxilla, medial view). b: Crenatosiren olseni (UF/FGS V6094, right tusk, medial view). c: Dioplotherium manigaulti (ChM PV2633, left tusk, medial view). d: Corystosiren varguezi (USNM 425695, right tusk, lateral view showing broad wear surface, partly restored). Scale bar ¼ 15 cm. The force applied was kept as constant as possible, whether this sufficed to sever a rhizome with one stroke or not. Shoots and rhizome fragments loosened by this process were removed by hand and collected. Rhizomes severed at one end that snapped off or pulled out of the sediment easily when encountered by hand were also removed; a rhizome felt to be still attached at both ends was severed with the tusk before removal by hand. This process was continued until all plant material in the 25 3 25 cm test quadrat had been removed to a depth of 8–10 cm; the number of tusk strokes needed to do this was recorded. The seagrass material collected was washed and sorted into above-ground (shoots þ leaves) and below-ground (rhizomes þ roots) biomass; the number of shoots was counted; and the rhizomes þ roots were blotted dry and weighed. Five such trials with each tusk design were made. At Clack Island, all of the test quadrats were located within a 6-m radius of each other. A 10-cm diameter core sample was also taken in this area, which revealed Thalassia rhizomes at depths of 6–12 cm. Therefore, it is likely that some of the deepest rhizomes in the other sampling areas were missed. However, careful exploration by hand in the bottom of each quadrat before termi- nation of sampling, and continuance of excavation until no more rhizomes were felt, ensured that all those with parts lying within approximately 10 cm of the sediment surface were collected. The aim of this study was simply to produce a rank order of rhizome-extraction efficiency for the tusk designs used, without any absolute values being assigned to these efficiencies. To avoid misunderstanding, it is essential to emphasize that this was not an attempt to model animal behavior, but rather to quantify a physical property of certain physical objects. Apart from the length of each tusk exposed, the position in which it was held, and the direction in which it was moved, there is no reason to think that the manner of tusk use described above bears any relation whatsoever to how any sirenian would have used these objects in extracting seagrasses. It is better to think of the tusks in these experiments not as parts of living organisms, but simply as tools manufactured for the purpose of cutting or digging, and whose relative efficiency is being tested under standardized, albeit artificial, conditions. The results of these experiments were analyzed using a two-way analysis of variance with sirenian species and seagrass species as fixed factors. Input data were num- USE OF SIRENIAN TUSKS IN FEEDING 529 Fig. 2. The mean wet weights of seagrass rhizomes removed per stroke by each tusk demonstrate the differences among the tusks in their ability to excavate Thalassia, but not Cymodocea þ Halodule. The means for Thalassia have been adjusted to compensate for the significant differences in the wet weight of rhizomes available to the various tusks. The least significant difference between the means is shown for each species of seagrass. MF, Metaxytherium floridanum; CO, Crenatosiren olseni; DD, Dugong dugon; DM, Dioplotherium manigaulti; CV, Corystosiren varguezi. ber of tusk strokes per quadrat and wet weight of rhizomes removed per stroke. The number of strokes was log-transformed to equalize the error variance. Unfortunately, although the quadrats were selected to be as uniform with respect to seagrass shoot density as possible, there was a systematic difference in the wet weight of rhizomes in the Thalassia quadrats selected for excavation with the different tusks. However, there was no such problem with the Cymodocea/Halodule quadrats. Accordingly, the analysis of variance was reanalyzed to test for the differences among tusks in their ability to excavate Thalassia with wet weight of rhizomes as a covariate. stroke. This was obviously the result of the much greater depth of burial as well as the greater size and strength of the Thalassia rhizomes. It was not attributable to sediment type, which was slightly finer and siltier in the Thalassia bed. When the results from the trials on both the seagrass types were considered together, there was a significant difference among the sirenian species in mean number of strokes per quadrat, but not in the weight of rhizomes removed per stroke. However, as clearly shown in Figure 2, when the different seagrasses were considered separately, there was no significant difference among the sirenian species in ability to remove Cymodocea þ Halodule. There was, in contrast, a marked difference in the case of Thalassia (adjusted for the differences in the weight of rhizomes available), with the larger and more bladelike tusks being the most effective. Results Results are summarized in Figure 2. The seagrass beds harvested using the various tusks were uniform with respect to shoot density at both Clack Island (Thalassia: F ¼ 0.76; df ¼ 4, 20; P ¼ 0.56) and Magnetic Island (Cymodocea: F ¼ 0.70; df ¼ 4, 20; P ¼ 0.60; Halodule: F ¼ 0.09; df ¼ 4, 20; P ¼ 0.98; Cymodocea þ Halodule: F ¼ 0.20; df ¼ 4, 20; P ¼ 0.94). Cymodocea þ Halodule was easier to excavate than the Thalassia, as evidenced by the number of strokes needed per quadrat (F ¼ 1074.26; df ¼ 1, 40; P < 0.001) and the wet weight of rhizomes removed per stroke (F ¼ 24.70; df ¼ 1, 40; P < 0.001). Combined results from both excavation sites show there was a significant difference among the sirenian species in mean number of strokes per quadrat (F ¼ 12.25; df ¼ 4, 40; P < 0.001), but not in the weight of rhizomes removed per stroke (F ¼ 1.68; df ¼ 4, 40; P ¼ 0.1737). A difference was also observed in the wet weight of rhizomes in the Thalassia quadrats selected for excavation with the different tusks (F ¼ 5.39; df ¼ 4, 20; P ¼ 0.004). Discussion Overall, the Cymodocea þ Halodule was easier to excavate than the Thalassia (Fig. 2). This finding was demonstrated in both the number of strokes needed per quadrat and the wet weight of rhizomes removed per REVIEW OF DUGONG ANATOMY Gross observations on tusks and skulls of Dugong dugon were based on specimens collected by the dugong research project at JCU and deposited in the Museum of Tropical North Queensland, Townsville. Cranial Architecture The architecture of the skull provides further clues to the possible mode of use of the tusks, at least among the more derived dugongines having large, bladelike tusks. Some of the distinctive characteristics and evolutionary trends of the latter are: shortening of the nasal process of the premaxilla, broadening and thickening of the posterior end of this process, and development of a relatively flat and vertical transverse joint surface between this process and the frontal (Domning, 1989b,c). This contrasts markedly with the primitive condition found in most sirenians (including Dugong and Metaxytherium; see Domning, 1994 for phylogenetic analysis), in which the premaxillary–frontal joint surface slopes gently anteroventrad and the nasal process is relatively long and thin and overlaps the dorsal surface of the frontal. 530 DOMNING AND BEATTY Fig. 3. Lateral view of skull, mandible, and masseter muscle of a derived dugongine, illustrating the hypothesized manner of tusk use in harvesting seagrass rhizomes. With the front of the lower jaw braced against the substrate (a), contraction of the masseter to close the jaw (b) drags the bladelike tusk backward (c), severing rhizomes with the posterior edge of the tusk. The resistance of the substrate results in a compressive force field along the dorsum of the skull (d), which is countered in part by a butt joint between the premaxilla (PM) and frontal (F). The development of a premaxillary–frontal butt joint reaches an extreme in Xenosiren, where it appears to reflect a compressive stress field along the dorsal side of the skull (Domning, 1989c) (Fig. 3). If it is true that such a joint surface is better adapted to resist anteroposteriorly applied compressive stress than is a gently sloping oblique surface, then the progressive evolution of the butt joint in some dugongines may well reflect a large and unprecedented increase in such stresses in this group of dugongids. Posterad- or posteroventrad-directed stresses at the premaxillary–frontal joint would most naturally result from anterad- or anterodorsad-directed forces applied to the tip of the sharply deflected rostrum. This is because the premaxilla tends to pivot about its attachments to the maxilla, as can easily be seen by manipulating a dry dugong skull with loose sutures. Forward and upward stresses on the rostrum would, in turn, result from attempts to force it backward and downward against resistance, as would occur in closing the jaws on material between the symphysial pads (Fig. 3), or in depressing the head against resistance at the tip of the rostrum. Because the progressive evolution of the premaxillary– frontal butt joint in dugongines occurs together with increase in size and in bladelike form of the tusks and not with any other dramatic osteological change in the oral region, it is logical to conclude that this stress resulted from use of the tusks, which consequently were being applied most forcefully in a downward and backward direction and with their posterior edges as the leading edges. Such use of the jaw-closing muscles to cut with the tusks would be analogous to the canine shearbite postulated for the sabertooth cat Smilodon by Akersten (1985). This behavior can be envisioned as an elaboration or special case of the third (pit-making) mode of feeding described above for the modern dugong, i.e., it would more or less require the animal to work in one place during a dive. If efficient enough, however, it might enable the animal to maintain some headway, making at least a short feeding trail while extracting larger rhizomes than Dugong dugon. This conclusion in regard to the most derived dugongines, however, may not apply to other dugongids that also have enlarged tusks but did not develop a premaxillary–frontal butt joint (e.g., Dugong, Metaxytherium subapenninum ¼ M. forestii ¼ M. gastaldi; for synonymy see Pilleri, 1988). These (or in the case of Dugong, its ancestors) conceivably used their tusks in a different manner, i.e., in a forward and upward direction. More likely, perhaps, the frequency of large rhizomes in their diets may have been lower, although their large tusks gave them some capacity to extract such rhizomes when needed. In any case, this line of reasoning argues for a frequently downward-and-backward direction of tusk movement in at least one group, the advanced, bladetusked Dugonginae. For these reasons, this direction of movement was chosen for use in the seagrass-harvesting experiments described above. Anatomy and Use of the Dugong Feeding Apparatus The intact oral region of Dugong dugon (Fig. 4) has been most carefully described by Gudernatsch (1908), Gohar (1957), and Marshall et al. (2003). In brief, a broad, disk-shaped surface of the upper lip, covered with tactile bristles, lies anterior to the mouth opening. Specially enlarged bristles at the posterolateral corners of this oral disk have a prehensile function, much the same as in manatees (Marshall et al., 1998). Against the posterior margin of the oral disk lies a median knob-like projection (Fig. 4, hp), which is clearly visible in the intact animal when the mouth is closed, and which (as noted above) is thought to be used in rooting for rhizomes. This knob is made of tough connective tissue and is an anterior extension of the flat, trapezoidal pad that covers the palatal surface of the premaxillae. It is, therefore, an integral part of the upper jaw, but is not directly supported by bone. The tusks are longitudinally curved so that they diverge distally, and are slightly flattened in planes that diverge posteriorly, so that the posterior edges of the worn surfaces are farther apart than the anterior edges (Fig. 4). The wear surfaces, however, face more laterad than anterad. Both anterior and posterior edges of the tusk tips are usually sharp (see edge sharpness data given below). In animals with erupted tusks, the knob lies posteroventral to the tusks and extends a couple of centimeters beyond their tips. The posteromedial sides of the tusks lie against the knob, but are not enclosed within it, as shown by the fact that the dark staining that covers nonocclusal surfaces of the cheek teeth outside the gums also covers all sides of the tusks except the worn surface (Fig. 4). The limits of this staining accurately indicate the gumline, which lies less than 1 cm outside the margin of the tusk’s bony alveolus (see USE OF SIRENIAN TUSKS IN FEEDING 531 than with the anteromedial side. Hence, a potentially effective cutting edge is formed along the posterior edge of the tusk much sooner than on the anterior edge. The orientation of the cross-section in tusks of this type is well suited to rapid development and maintenance of a cutting edge useful in posterad tusk movement, but much less suited to anterad cutting. TUSK TIP GEOMETRY Materials and Methods Fig. 4. Oral region of adult male Dugong dugon, ventral view, showing part of lower jaw (lj), mouth opening (m), and knob-like horny pad on upper jaw (hp), with oral disk (od) pulled forward to expose tusks (t). Note prominent clusters of prehensile vibrissae (U2 bristle fields of Marshall et al., 2003) on corners of oral disk just lateral to tusks. Photo courtesy of Paul K. Anderson. also Fernand, 1953). The anterior and posterior edges and lateral side of the tusk’s tip are exposed to wear. The cross-sectional shape of the unworn tusk is, in some cases, suggestive of the direction of movement to which the tusk was adapted. In Crenatosiren, Dugong, Dioplotherium, and Corystosiren, the tusk is roughly oval or polyhedral in cross-section, with two fairly distinct flattened surfaces on the medial side, respectively facing anteromediad and posteromediad (cf. Domning, 1978, 1989b, 2001). These are the surfaces covered by enamel if any is present, and they form approximately equal angles with the median plane. The paper-thin enamel, confined to the medial side, provides for a selfsharpening edge analogous to the chisel point of a rodent incisor. However, because the posteromedial side of the cross-section is (except in Crenatosiren) longer than the anteromedial, the long axis of the cross-section lies at an acute angle to the median plane. As a result, as the lateral side begins to wear, the worn surface forms a more acute angle with the posteromedial side The study of tip geometry of tusks was based on tusks of Dugong dugon (isolated and in skulls) in the University of Kansas (KU) Natural History Museum Mammals Collection, Lawrence, Kansas (three isolated tusks cataloged as a lot, KU 130105) and the Mammals Collection of the Department of Zoology, Field Museum of Natural History (FMNH), Chicago, Illinois. A tusk referred to cf. Corystosiren by Domning (1990a) from the Dept. of Vertebrate Paleontology, Florida Museum of Natural History, University of Florida (UF), Gainesville (UF 18826, the only one with a complete tip) was also studied in this way. Casts were made of opaque urethane resin from microwear molds (described below) and sectioned for the study of geometric profile. Evans and Sanson (2003) have shown that dentitions can be described in terms of geometric details that indicate their utility as tools in food comminution. Although dugongid tusks are not brought into occlusion with opposing teeth, they can be considered in an analogous way with respect to a transverse plane parallel to the long axis of the tusk, here termed the transverse axial plane (this plane being analogous to the occlusal plane of the carnassial teeth of a carnivoran). This is because whether they are dragged through the substrate passively in anterad or anterodorsad motion during grazing, or during forceful posterad or posteroventrad motions in digging, their contact with other objects will be, on average, in a direction perpendicular to the long axis of the tusk and, therefore, perpendicular to that transverse axial plane. In the case of a tusk digging rhizomes, the resistance is from root attachment in the substrate and the mechanical properties of the rhizome itself, not from the cusp of an occluding tooth passing by it. Tusks used in either way function essentially like knives, so the meaning behind tusk tip geometry is slightly different in comparison with how occluding cheek teeth break food (which is more like a mortar and pestle or pair of scissors). Therefore, tusk profiles were measured for variables important to blades (sensu Evans and Sanson, 2003) with respect to the transverse axial plane. Certain details, such as rake and relief angles, are important in slightly different ways (below) from the occluding cheek tooth models of Evans and Sanson (2003), simply due to the fact that clearance and orientation with respect to an opposing tooth are not issues for tusks. The variables measured included edge sharpness (measured in radians), rake and relief angles, and approach angle (measured in degrees; Fig. 5). Edge sharpness, as well as rake, relief and approach angles, were measured for both anterior and posterior tusk ‘‘blade’’ edges in Dugong to better assess effectiveness of tusks as cutting tools in either anterad or posterad movements. Some of these measures were not possible (or necessary) for 532 DOMNING AND BEATTY Fig. 5. A: Schematic of a right tusk of Dugong in lateral view, with longitudinal axis of tusk (equivalent to transverse axial plane seen edge-on, represented by the parallel vertical lines) oriented vertically, depicting anterior and posterior approach angles between longitudinal axis of tusk and long axes of blade edges. B: Schematic of right tusk of Dugong in apical view, illustrating posterior relief (<08) and rake angles (>908), as well as anterior rake (08 < rake angle < 908) and relief angles (08< relief angle < 908). The vertical lines represent parasagittal planes; the horizontal lines represent the transverse axial plane. cf. Corystosiren, because only a single blade edge (the posterior blade edge) actually exists. The edge sharpness of a blade is the radius of curvature of the blade edge. A blade with a smaller radius of curvature will have higher edge sharpness. Because the tusks of Dugong tend to wear at an oblique angle, but are pointed more or less ventrally when feeding, the edge sharpness of the anterior, posterior, and apical edges was measured for comparison of edge utility in cutting. Higher edge sharpness will decrease the area of contact and so increase stress in the food, and is indicative of a more effective cutting tool. The rake surface is that which follows the blade edge and is worn by occlusion or abrasion from food. Thus, the rake surface is first identified as the worn surface of the tusk. The rake angle is the angle at which this surface lies with respect to the transverse axial plane (Fig. 5B). The magnitude of a rake angle is indicative of its effectiveness. If a rake angle is larger, food will have greater clearance to pass by the tusk after being cut, and edge sharpness will be more easily maintained. Just as with a knife edge, the closer a rake angle is to 908, the more effective it will be as a cutting tool (the difference between cutting with a knife made with thick metal and one made of thin metal). If a rake angle is larger than 908, the rake surface is not facing in the direction of motion. In summary, a rake angle (RA) greater than 08 and less than or equal to 908 characterizes a blade edge (with the most effective blade closest to 908). A blade edge is not used at a rake angle greater than 908; hence, the rake angle of an edge can indicate whether the edge is used in a particular direction of motion (Fig. 5B). The relief angle is the angle between the surface of the tusk following the blade edge and a parasagittal plane abutting the blade edge; it is a measure of clearance posterior to the blade edge. A relief angle is described as negative if the portion of the tusk following the blade edge projects into the path of food. That is, if the relief angle is negative (like the posterior relief angle in Fig. 5B), after being cut a seagrass rhizome will still need to pass by more dental materials, increasing friction. A positive relief angle is the state in which the medial side of the tusk is deflected away from the path of food, permitting food clearance once it is cut, without wear on the medial side (like the anterior relief angle in Fig. 5B). Thus, the relief angle can theoretically be used to estimate whether tusks were used in an anterad or posterad manner (cf. Fig. 3): optimally, the more effective cutting direction would be the one making use of the tusk edge with the more positive relief angle. We can compare these angles for the alternate feeding modes and, thereby, estimate which mode is/was more likely used in modern and fossil taxa. The approach angle of a blade is the angle between the long axis of the blade (the tusk’s leading edge) and the transverse axial plane (Fig. 5A). The mechanical advantage (MA) of a blade will depend on the approach angle (a) of the blade, where MA ¼ 1/cos(a), so that a larger angle will have a greater mechanical advantage. This keeps a food item forced along the blade edge so that it experiences point contact with the entire length of the blade until it is cut completely or slips off the end of the blade edge. Additionally, the length of the blade edge (cutting edge) was measured. A longer blade edge maximizes the possible duration of a cutting event for every sequence of ingestive movements, so a longer blade edge would also enhance the effectiveness of a tusk as a cutting tool. Likewise, a longer blade edge permits further efficiency in cutting by permitting the blade edge to move along the rhizome for a distance (as opposed to a short or pointed blade that would only puncture or briefly jab at a rhizome). This may seem obvious, but studies of the mechanics of ‘‘pressing and slicing,’’ such as the motion of cutting cheese with a knife while using a slicing movement as opposed to simply pushing the knife in like a wedge, have proven that this movement is mechanically advantageous (Atkins et al., 2004). Similarly, the rake surface area was measured (essentially the same as the area of the worn facet). Just as with the rake angle and its role in minimizing friction USE OF SIRENIAN TUSKS IN FEEDING with the rake surface, minimization of the rake surface area itself would also minimize friction. A more effective blade should have a smaller rake surface area. Results Tusk tip geometry comparisons show that the edge sharpness varies greatly along the perimeter of worn tusk tips of Dugong and cf. Corystosiren. The radius of curvature for the anterior edge of the tusk tip is greater for cf. Corystosiren (1.39 radians) than Dugong (0.524 radians), while the radius of curvature of the posterior edge of the tusk tip in cf. Corystosiren (0.349 radians) is much less than that of Dugong (0.576 radians). The edge sharpness of the very extreme tip (apical end) is greater in Dugong (radius of curvature ¼ 0.401 radians) than in cf. Corystosiren (0.873 radians). The anterior rake angles were recorded as 268 for Dugong and 68 for cf. Corystosiren. The posterior blade rake angle was 1448 for Dugong and 758 for cf. Corystosiren. The rake surface of the posterior blade of Dugong is not facing in the direction of posterad motion (RA > 908). The anterior relief angles were 348 for Dugong and approximately 658 for cf. Corystosiren. The posterior relief angles were 258 for Dugong and approximately 108 for cf. Corystosiren. Anterior approach angles were 228 for Dugong and approximately 338 for cf. Corystosiren. Posterior approach angles were 408 for Dugong and approximately 558 for cf. Corystosiren. Blade edge lengths for Dugong were 33 mm (posterior), 35 mm (anterior), and approximately 3 mm (apical). The posterior blade edge of cf. Corystosiren was 71 mm. The apical and anterior edges of the tusk of cf. Corystosiren are not worn into blades and are, thus, difficult to differentiate and measure in this respect. The rake surface is 742 mm2 for Dugong and 870 mm2 for cf. Corystosiren. Discussion Tusk tip geometry comparisons show that the worn tusk tips of Dugong and cf. Corystosiren are geometrically very different. Edge sharpness varies greatly along the perimeter of a tusk tip in both taxa. The radius of curvature for the anterior edge of the tusk tip, presumably the edge that would encounter wear during passive substrate encounters during grazing, is much greater for cf. Corystosiren than for Dugong, meaning that the anterior edge of the cf. Corystosiren tusk is less sharp than that of Dugong. In contrast, the radius of curvature of the posterior edge of the tusk tip in cf. Corystosiren is much less than that of Dugong. The edge sharpness of the very extreme tip (apical end) is greater in Dugong than in cf. Corystosiren. This finding may be in part because only one edge (the posterior edge) of the tip in cf. Corystosiren is used for cutting and consequently gets sharpened. The apical edge of Dugong is significantly sharper due to wear on its anterior edge, and shorter because the overall wear produces a roughly elliptical rake surface, of which the apex is one of the narrow, more strongly curved ends. The straighter posterior blade edge of cf. Corystosiren permits the self-sharpening effect to maintain a long blade edge. The anterior rake angles of Dugong and cf. Corystosiren are drastically different, which reflects the betterdeveloped anterior-facing blade edge of the tusk of 533 Dugong as compared with lack of an anterior blade edge in cf. Corystosiren. The posterior blade rake angle of Dugong is large because the rake surface does not face the direction of motion. This strongly indicates that the posterior blade edge of Dugong is unlikely to have been used in cutting. The rake angle of the posterior blade edge of cf. Corystosiren is very positive, and clearly indicative of being a more effective tool when used in a posterad motion. This would provide not only slightly greater fragment clearance for cf. Corystosiren, but also help in maintaining the edge sharpness throughout its wear life. The anterior relief angles appear to be different between Dugong and cf. Corystosiren. The horny pad of Dugong blocks the potential passage for foods in an anteroposterior direction, and, therefore, negates the importance of this angle to some degree. For cf. Corystosiren, this angle indicates lesser use of the anterior edge for cutting. Although it is uncertain how much of a horny pad surrounded the tusk of cf. Corystosiren, it seems clear that the tusk protruded proportionately farther from the alveolus (and hence, presumably, from the gingiva) than in Dugong. The posterior relief angles between Dugong and cf. Corystosiren also appear to be different in magnitude, but are even more fundamentally different because the rake surface of the posterior blade of Dugong is not facing in the direction of posterad motion (RA > 908), effectively increasing the relief angle by pushing the entire posterior edge away from the sagittal plane. Ultimately, rake and relief angles of the posterior blade edge of Dugong are only important in illustrating their lack of use in cutting. The relief angles of the posterior blade edge of cf. Corystosiren are very small and would easily permit clearance of food past the tusk. Anterior approach angles for Dugong are comparable to that of cf. Corystosiren. Posterior approach angles for Dugong are smaller than that of cf. Corystosiren, although this could be easily modified by a small amount by adjustments to head orientation. The larger posterior approach angle would have given cf. Corystosiren a significantly enhanced mechanical advantage in cutting seagrass rhizomes during digging. Blade edge length is the other, more obvious, feature that distinguishes the mechanical utility of these tusks. Dugong has relatively short apical, posterior, and anterior blade edges. In contrast, the posterior blade edge of cf. Corystosiren is much longer, reflecting not only the depth to which it could dig during feeding, but also its shape when considered along with its rake surface. The apical and anterior edges of the tusk of cf. Corystosiren are not worn into blades and were thus difficult to differentiate and measure in this respect. The rake surface of Dugong is large in comparison with its blade edge length when compared with that of cf. Corystosiren. This finding reflects the narrow blade of the latter, optimal for maximizing cutting and minimizing friction with food after cutting. MICROWEAR ON TUSKS Materials and Methods The same tusks (Dugong dugon, KU 130105, and cf. Corystosiren, UF 18826) used in the study of tip geometry (above) were analyzed for microwear data. In addition, wear patterns were analyzed on a dugong skull from Arabia (JCU uncataloged) and two specimens from 534 DOMNING AND BEATTY Fig. 6. Lateral view of Dugong dugon skull and tusk, with microwear features of different edges illustrated by micrographs at right (oriented with the skull, and all at 703, original magnification). Data are reported as average numbers of pits (p), fine scratches (fs), and coarse scratches (cs). Arrows indicate average orientations of fine scratches (narrow arrows) and coarse scratches (thicker arrows) for each edge location, presumably caused by grit particles and/or seagrass rhizomes crossing the tusk during feeding. Queensland (MM 083 and MM116, from the James Cook University marine mammal collection [MM], now housed in the Museum of Tropical North Queensland, Townsville, Australia). Specimens were cleaned with acetone and cotton swabs. Once dry, each tooth was molded twice using a high-precision, polyvinylsiloxane dental impression material (President Jet Microsystem1; Coltene/ Whaledent). The first mold was discarded as a final cleaning step, and in the case of the fossil, this included remnant matrix left over from preparation. Casts were made using clear urethane resin: CC200 Crystal/Water-Clear1 ultraviolet-stable urethane resin and hardener from Eager Plastics, Inc., Chicago, IL. Once the urethane was mixed and poured, the molds were placed in a pressure pot to remove air bubbles and were then left to cure for 2 days. Casts were examined under a stereo light microscope at 335 and 370 magnification using the Solounias and Semprebon (2002) technique. A 0.3 mm 3 0.3 mm area was examined in several locations along the perimeter of the tusk tip (Fig. 6). On each tusk, four continuous microwear variables were documented: the number of pits, the number of light and coarse scratches, and the orientation of each scratch with respect to the tangent of the tusk edge. Pits were defined as those features that are generally circular. Scratches were features with greater lengths than widths and with parallel sides. Scratches were categorized as fine (narrow and shallow) or coarse (wider and deeper). Scratch orientations were then adjusted to relate to the anteroposterior axis of the whole tusk. onto the posterior edge of the worn surface, whereas the anterior edge is polished clean. On the other hand, MM 083 has apparent wear on the posterior edge of the tusk’s medial surface. Likewise, MM 116 has the stain removed from the posteromedial and not the anteromedial surfaces of both tusks. The majority of specimens examined, however, show no such indications of preferential movement in either direction. Dental microwear was not adequately preserved for cf. Corystosiren (UF18826) due to its extremely thin enamel and the effects of postdepositional processes on its dentine. Microwear for modern Dugong dugon (KU 130105) was well preserved and supported some of our expectations (Fig. 6), although a larger sample size will be needed to corroborate these conclusions. The number of pits and coarse scratches was small in comparison with the number of fine scratches on average for all three locations on the tusk edge, although notable variations exist in the proportions and orientations of those scratches, the number of light and coarse scratches, and the orientation of each scratch with respect to the tangent of the tusk edge. The average number of pits did not significantly increase along the blade, although scratches were on average different in number and orientation. Scratch orientation is reported here with respect to the coronal plane, so that comparisons between edges facing in different directions can be made with minimal assumptions about feeding posture (one can just reorient and add 908 for horizontal plane relationships). Fine scratches showed an average bimodal distribution in orientation on the anterior and apical edges, and a mostly unimodal distribution on the posterior edge. On both anterior and posterior edges, the bimodal distributions of fine scratches appear to be biased toward those from anteroposterior movements rather than those from Results The dugong skull from Arabia (JCU uncataloged) shows staining on the sides of the tusk that extends USE OF SIRENIAN TUSKS IN FEEDING dorsoventral movements. On the posterior edge the fine scratches are oriented predominantly anteroposteriorly, as would be expected if the fine scratches were being created as the tusk coursed through the substrate during anteroposterior head movements during feeding. Coarse scratches were much less numerous, but their average orientations may be more informative. The few coarse scratches found on the anterior and apical edges were inclined almost dorsoventrally with a slight inclination dorsoposteriorly. In contrast, coarse scratches on the posterior edge were, on average, oriented almost dorsoventrally but with a slight inclination inferoposteriorly. If indeed these coarse scratches result from forceful digging for rhizomes in the substrate, then their orientation appears to support the notion that this digging motion could include anterior and posterior cutting actions. If digging for rhizomes were a regular behavior, presumably a much larger number of deep scratches would be expected. Discussion It is most unlikely that the macroscopic wear on tusks of Dugong is produced merely by contact with the upper lip, as suggested by Fernand (1953). Neither can the (presumably gender-specific) social uses of the tusks account for the wear, which also occurs on erupted tusks of old females, as pointed out by Pocock (1940). It is, however, plausible that the forward movement of the snout through the substrate while making feeding trails would account for the wear, without any purposeful use of the tusks. This would suggest that the anterior edge of the tusk tip is the leading edge during wear. This appears to be supported by staining on the worn posterior edge and a cleanly polished anterior edge in a dugong skull from Arabia (JCU uncataloged). In contrast, the posterior edge of the tusk’s medial surface has apparent wear on a specimen from Queensland (MM 083), suggesting that the posterior edge was at least sometimes used as the leading edge. Similarly, the stain is removed from the posteromedial and not the anteromedial surfaces of both tusks in MM116. The majority of specimens examined, however, show no such indications of preferential movement in either direction. Given possibly idiosyncratic individual behavior, together with the lack of evidence that tusk wear in Dugong is anything but incidental to feeding, no firm conclusions could be drawn from macroscopic tusk wear about the manner of use of the snout in feeding by Dugong, let alone by extinct dugongids. Microwear was studied to see if directions of scratches on the tusk surfaces could resolve this question. Major scratches caused by sand grains were expected to be roughly perpendicular to the tusk’s long axis. Incidental wear was assumed to result mainly from forward movement of the head and tusk. If the tusks were deliberately used for rooting by forward movements of the snout, or if tusk wear were simply incidental to such movements, then the scratches should be in the same general direction as those incidentally created while making feeding trails. Rooting might also involve backward movements of the rostrum, leaving some incidental scratches in the opposite direction. If, however, the sharp posterior edges of the tusks were persistently and forcefully used as cutting instruments, the tusk would 535 have to be moved backward, making scratches indicating backward movement that might be deeper and more numerous than any of the incidental scratches. The data on dental microwear from Dugong dugon seem to support the expectation of wear if tusks are worn from passing through the substrate during normal feeding bouts, with occasional events of vigorous cutting of rhizomes by anterior and posterior movements of the tusks in the substrate. The deep gouges on both the ventral and dorsal edges of the Dugong tusk suggest that cutting may have been done by both motions, maybe using neck or jaw-closing movements. The gouges on the dorsal edge seem too large to be incidental to passing through the substrate during normal grazing, like the scratches on the tusk tip. Conclusions Tropical dugongids of the past likely resembled today’s dugong in eating seagrass. How, on the other hand, did these extinct dugongids differ from the modern one? One category of differences with adaptive implications is that of rostral deflection, which seems to reflect the average position of food items in the water column (Domning, 1977, 1980, 1982, 2001). This is not very helpful, however, in distinguishing among extinct tropical dugongids, all of which (like Dugong) seem to have been bottom-feeders with more or less strongly downturned snouts (an observation that supports the assumption of seagrass diets). Another category of conspicuous differences is tusks. The most impressive development of tusks in the Sirenia is found in the dugongid subfamily Dugonginae. Highly derived, extinct dugongine genera, such as Rytiodus, Corystosiren, and probably Xenosiren, have broad, mediolaterally compressed, bladelike tusks whose roots extend the full length of the premaxillary symphysis and whose medial surfaces are covered by thin enamel which provides a self-sharpening edge. It seems impossible to explain the morphology of such tusks, or their evolution from more primitive, subconical incisors, apart from the supposition that they were used to cut something. The narrower and less flattened but likewise selfsharpening tusks of Dugong are used as weapons against other animals of the same species (or even, reportedly, against sharks; Promus, 1937). However, it is not clear why this use would require any more bladelike tusks than Dugong now displays; cutting food items seems a likelier use for large blades. In opposition to this view, Anderson (2002), analyzing the evolution of sirenian mating systems, argued that ‘‘[t]usks may have had a social function throughout dugongid evolution and their social function(s) may have been primary. . . .Evolution and retention of tusks exclusively, or even primarily, as foraging structures would be unique among mammalian herbivores.’’ This hypothesis would, however, lead us to expect sexual dimorphism of tusks to have been prevalent among dugongids—which the fossil record does not support. With specific reference to D. dugon, he argues that ‘‘[i]f tusks were primarily foraging structures, the sex with the greatest nutritional requirements and the most to gain from a wide range of foraging options (females) should have retained them’’ (Anderson, 2002:77). This would be true if those females were capable of efficiently chewing and digesting fibrous 536 DOMNING AND BEATTY rhizomes harvested by the tusks; but (for reasons that admittedly remain mysterious; Domning, 1995) modern dugongs of both sexes lack the multicusped, enameled cheek teeth that this would require (Lanyon and Sanson, 2006). So while it seems clear that neither sex has a motive, at present, to use the tusks in any but social roles, we consider this an exceptional and derived situation in Dugong dugon that tells us little about primitive conditions among dugongids in general. In our view, therefore, the social functions of dugongid tusks are historically secondary to the more universal and primitive use of incisor teeth in gathering food. And if we conceive of dugongids in general as seagrass eaters, then the obvious and only food items needing to be cut by large tusks are the more or less tough and deeply buried rhizomes of the larger seagrasses. How can this hypothesis be tested? First, by seeing whether Dugong tusks today have alimentary in addition to social functions. The stomach-content data presented here suggest that males do not eat more rhizomes than females. This, of course, was expected simply from the fact that females lack erupted tusks; sexual divergence of diet in large mammals would be highly unusual. However, the fact already noted that almost no fossil dugongid is known to exhibit tusk dimorphism shows that the sexual dimorphism (and, hence, possibly the social role of tusks) seen in D. dugon is not representative of the family as a whole. While the stomach-content data do not rule out the possibility that D. dugon or its relatives used their tusks for feeding in the past, they do deprive us of the one uniformitarian key to the past that a living species might provide, and reduce us to evaluating the tusks on their own evidence: How, and how well, might such instruments work as cutting and/ or digging tools if they were in fact so used? As noted above, when dugongs feed on Halophila and Halodule, they typically uproot and consume the entire plant and make a linear trail. Because a trail is made in the few minutes between successive breaths, it is necessarily made by an animal maintaining more or less steady forward motion. Hence, such trails are usually not made on seagrass beds that offer significant resistance to this sort of plowing or rooting, due to toughness and/or deep burial of the rhizomes and/or hardness of the sediment. If a dugong were to feed on rhizomes under the latter conditions, it would have to spend its available time on any one dive working on a comparatively small area of bottom, and would presumably clear only a small patch or make a circular pit rather than a linear trail (as is sometimes observed; Anderson, 1998). It is not obvious a priori how a dugong might use tusks in feeding on seagrasses with larger, stronger rhizomes such as Thalassia—whether cutting them with a forward and upward stroke, using a toss of the snout, or with a downward and backward movement, perhaps using the jaw muscles, as suggested for Xenosiren by Domning (1989c) (Fig. 3). Once again, Dugong dugon fails us as a guide, partly because its mode of feeding is poorly understood and partly because specimens can be found to seemingly support either model. The evidence from cranial architecture, tusk geometry, and tusk wear adduced above, however, tends to support the downward-and-backward alternative. The relatively crude seagrass-harvesting experiments reported here did not separate the effects of tusk length from those of tusk shape. In fact, it seems likely that the advantage of the Corystosiren tusk over the smaller tusks at excavating Thalassia rhizomes from depths of 6 cm or more was largely due to its effective length of 8–9 cm. When effective tusk length is less than the minimum depth of rhizome burial, several strokes are required merely to penetrate and remove the sediment before the rhizomes begin to be disturbed. As for the smaller tusks that were within a centimeter or two of each other in effective length and which differed mainly in shape, the experimental technique was apparently not sensitive enough to reveal significant differences in digging/cutting efficiency if such exist. Janet Lanyon (personal communication) has, in fact, stressed sediment disturbance as a potential alternative function of tusks, separate from cutting of rhizomes. She visualizes the tusks being used to loosen and stir up the sediment surrounding the rhizomes, shaking them free and making them accessible to breakage by other mouthparts. This could logically involve a forward and upward toss of the snout, which would also serve to pull intact rhizomes up toward the sediment surface. Small, conical tusks with no cutting edges, such as those of Metaxytherium floridanum (Fig. 1a), could well have served mainly for sediment disturbance long before the evolution of larger ones better suited for cutting. As for D. dugon, sediment disturbance may now be the principal way in which tusks remain useful to the males for feeding, if indeed they are still helpful in feeding at all. Regarding the larger tusks, several possibilities remain: their more bladelike shape may in fact make a direct contribution to digging and/or cutting efficiency distinct from the contribution of tusk length (by increasing cutting edge and minimizing drag-inducing lateral bulk); or the mediolateral flattening of the tusk may serve to stiffen it in the plane of its action (much as a deep mandibular body prevents dorsoventral bending) or anchor it more strongly in the alveolus (by increasing surface area for periodontium); or the flattening may be in some way a developmental or structural byproduct of enlargement of the tusk. These alternatives are not mutually exclusive. The last, however, can probably be eliminated, because in at least one dugongid (Metaxytherium subapenninum), and in the miosirenine trichechid Miosiren kocki, the tusk became lengthened to the full length of the premaxillary symphysis without losing its subconical shape, suggesting that tusk thickness is not limited by lateral dimensions of the premaxilla or other factors. While our evidence generally supports the hypothesis that the tusks of most large-tusked dugongids served in feeding (most probably to excavate and sever large seagrass rhizomes, by means of a backward, jaw-closing movement), there are intriguing exceptions: two dugongids, Dugong dugon and Metaxytherium subapenninum, (a) independently evolved large tusks (extending the length of the premaxillary symphysis) that are (b) subconical rather than conspicuously flattened and bladelike, (c) lack a premaxillary–frontal butt joint, and (d) show (or possibly show) sexual dimorphism. This combination of character states is unique to these two forms. Did they alone use the tusks primarily for social purposes? Other dugongids likely used their tusks socially as well as for feeding, as Anderson (2002) emphasizes; many mammals in which both sexes use tusks for feed- USE OF SIRENIAN TUSKS IN FEEDING ing, such as elephants, use them also for social purposes (Spinage, 1994). Sexual dimorphism in elephant tusks is mostly restricted to their age of eruption, which does little to alter their utility in feeding. The nearest known relative, and possible ancestor, of D. dugon is represented by an undescribed adult skull from the late Pliocene of Florida (cast, USNM), which has large, subconical, enameled and erupted, somewhat worn tusks, together with enameled molars, comparable to the conditions seen in M. subapenninum. If this specimen represents the ancestry from which D. dugon was derived, then the latter’s somewhat more flattened, selfsharpening tusk evolved subsequent to the Pliocene, as did its loss of functional enamel crowns on its molars (cf. Lanyon and Sanson, 2006). Domning (1995) speculatively attributed the latter development to a shift away from large rhizomes in the diet—rather than toward such a diet, as the tusk modifications would suggest. The relative timing and ecological context of these two changes, seemingly adaptations in different directions, remain to be clarified. In contrast, tusk enlargement in Mediterranean Metaxytherium took place from the late Miocene to the middle Pliocene, was not associated with molar degeneration, and seems connected with the late Miocene Mediterranean salinity crises (Bianucci et al., 2004). Only in the later (Pliocene) stages of this process did sexual dimorphism possibly develop; in the single putatively male specimen, the tusk is somewhat flattened and possibly self-sharpening, although it is unclear if enamel is present. Do Dugong lack a premaxillary–frontal butt joint as a result of evolutionary reversal? If so, this could have come about through neoteny, which seems reflected in other aspects of the feeding apparatus (Domning, 1995). This hypothesis would presuppose that a butt joint was absent in juveniles of taxa whose adults possessed it; but juveniles of the latter fossil species have not yet been found. In any case, such a joint does not seem to have existed in any of the direct ancestors of Dugong (or Metaxytherium; Domning, 1994). In conclusion, available evidence seems to support (albeit less than conclusively) our hypothesis that extinct dugongines with large, bladelike tusks used them to feed on robust seagrass rhizomes. The modern dugong, however, is atypical of its subfamily in having evidently evolved away from this sort of specialization; it now appears to use its tusks only in historically secondary social roles. It is, thus, of only limited help in resolving the complexities of dugongine feeding ecology. ACKNOWLEDGMENTS Katherine Rafferty fabricated the model tusks. Warren Lee Long and the Northern Fisheries Research Centre, Queensland Department of Primary Industries, Cairns, loaned field equipment and analyzed sediment samples. Jeff Miller of the Department of Primary Industries allowed Domning to join his field team and made possible his work at Clack Island. Access to collections under their care was kindly provided by T. Holmes (University of Kansas Natural History Museum Division of Mammals), W. Stanley (Field Museum of Natural History Mammals Collection), and R. Hulbert (Florida Museum of Natural History Dept. of Vertebrate Paleontology). 537 Helene Marsh very kindly did the statistical analysis of the seagrass-harvesting experiments and gave other assistance. William Akersten, George Heinsohn, Janet Lanyon, Helene Marsh, and Tony Preen provided valuable discussions and/or comments on earlier drafts of the manuscript. Paul Anderson, Kate Baldwin, Ray Bernor, and Irina Koretsky assisted with illustrations. LITERATURE CITED Akersten WA. 1985. Canine function in Smilodon (Mammalia; Felidae; Machairodontinae). Nat Hist Mus L A County Contrib Sci 356:1–22. Aketa K, Kawamura A. 2001. Digestive functions in sirenians (review). Bull Faculty Bioresources, Mie University 27:85–103. Aketa K, Asano S, Wakai Y, Kawamura A. 2001. Apparent digestibility of eelgrass in dugongs (Dugong dugon). Honyurui Kagaku (Mammal Sci) 41:23–34. Aketa K, Asano S, Wakai Y, Kawamura A. 2003. Apparent digestibility of eelgrass Zostera marina by captive dugongs Dugong dugon in relation to the nutritional content of eelgrass and dugong feeding parameters. Mammal Study 28:23–30. Anderson PK. 1979. Dugong behavior: on being a marine mammalian grazer. Biologist 61:113–144. Anderson PK. 1986. Dugongs of Shark Bay, Australia – seasonal migration, water temperature, and forage. Natl Geographic Res 2:473–490. Anderson PK. 1998. Shark Bay dugongs (Dugong dugon) in summer. II. Foragers in a Halodule-dominated community. Mammalia 62: 409–425. Anderson PK. 2002. Habitat, niche, and evolution of sirenian mating systems. J Mammal Evol 9:55–98. Anderson PK, Birtles A. 1978. Behaviour and ecology of the dugong, Dugong dugon (Sirenia): observations in Shoalwater and Cleveland Bays, Queensland. Aust Wildlife Res 5:1–23. André J, Gyuris E, Lawler IR. 2005. Comparison of the diets of sympatric dugongs and green turtles on the Orman Reefs, Torres Strait, Australia. Wildlife Res 32:53–62. Aragones L, Marsh H. 2000. Impact of dugong grazing and turtle cropping on tropical seagrass communities. Pacific Conservation Biol 5:277–288. Aranda-Manteca FJ, Domning DP, Barnes LG. 1994. A new Middle Miocene sirenian of the genus Metaxytherium from Baja California and California: relationships and paleobiogeographic implications. In: Berta A, Deméré TA, editors. Contributions in marine mammal paleontology honoring Frank C. Whitmore Jr. Proc San Diego Soc Nat Hist 29:191–204. Atkins AG, Xu X, Jeronimidis G. 2004. Cutting, by ‘pressing and slicing’, of thin floppy slices of materials illustrated by experiments on cheddar cheese and salami. J Material Sci 39:2761– 2766. Bianucci G, Carone G, Domning DP, Landini W, Rook L. 2004. PeriMessinian dwarfing in Mediterranean Metaxytherium (Mammalia: Sirenia): evidence of habitat degradation related to Mediterranean desiccation? (Abstract) Sedimentary Basins of Libya, 3rd Symposium, Geology of East Libya, November 21–23, 2004, Benghazi, Libya: 19. De Iongh HH, Wenno BJ, Meelis E. 1995. Seagrass distribution and seasonal biomass changes in relation to dugong grazing in the Moluccas, East Indonesia. Aquat Botany 50:1–19. De Iongh HH, Bierhuizen B, van Orden B. 1997. Observations on the behaviour of the dugong (Dugong dugon Müller, 1776) from waters of the Lease Islands, eastern Indonesia. Contrib Zool (Amst) 67:71–77. Domning DP. 1977. An ecological model for Late Tertiary sirenian evolution in the North Pacific Ocean. Syst Zool 25:352–362. Domning DP. 1978. Sirenian evolution in the North Pacific Ocean. University of California Publications in Geological Sciences 118:xi þ 176. Domning DP. 1980. Feeding position preference in manatees (Trichechus). J Mammal 61:544–547. 538 DOMNING AND BEATTY Domning DP. 1982. Evolution of manatees: a speculative history. J Paleontol 56:599–619. Domning DP. 1988. Fossil Sirenia of the West Atlantic and Caribbean region. I. Metaxytherium floridanum Hay, 1922. J Vertebrate Paleontol 8:395–426. Domning DP. 1989a. Fossil sirenians from the Suwannee River, Florida and Georgia. Southeastern Geological Society Guidebook No. 30:54–60. Domning DP. 1989b. Fossil Sirenia of the West Atlantic and Caribbean region. II. Dioplotherium manigaulti Cope, 1883. J Vertebrate Paleontol 9:415–428. Domning DP. 1989c. Fossil Sirenia of the West Atlantic and Caribbean region. III. Xenosiren yucateca, gen. et sp. nov. J Vertebrate Paleontol 9:429–437. Domning DP. 1990a. Fossil Sirenia of the West Atlantic and Caribbean region. IV. Corystosiren varguezi, gen. et sp. nov. J Vertebrate Paleontol 10:361–371. Domning DP. 1990b. Sirenian rhizivory studies. Appendix in: Lefebvre LW, Powell JA, Manatee grazing impacts on seagrasses in Hobe Sound and Jupiter Sound in southeast Florida during the winter of 1988–89. U.S. National Technical Information Service Document No. PB90-271883:34–36. Domning DP. 1994. A phylogenetic analysis of the Sirenia. In: Berta A, Deméré TA, editors. Contributions in marine mammal paleontology honoring Frank C. Whitmore Jr. Proc San Diego Soc Nat Hist 29:177–189. Domning DP. 1995. What do we know about the evolution of the dugong? (Abstract) Mermaid Symposium: First International Symposium on Dugong and Manatees. November 15–17, 1995, Toba, Mie, Japan. Abstracts. Toba, Toba Aquarium. p 23–24. Domning DP. 1997. Fossil Sirenia of the West Atlantic and Caribbean region. VI. Crenatosiren olseni (Reinhart, 1976). J Vertebrate Paleontol 17:397–412. Domning DP. 2001. Sirenians, seagrasses, and Cenozoic ecological change in the Caribbean. In: Miller W III, Walker SE, editors. Cenozoic palaeobiology: the last 65 million years of biotic stasis and change.Palaeogeogr Palaeoclimatol Palaeoecol 166:27–50. Evans AR, Sanson GD. 2003. The tooth of perfection: functional and spatial constraints on mammalian tooth shape. Biol J Linnean Soc 78:173–191. Fernand VSV. 1953. The teeth of the dugong. Ceylon J Sci (B) 25:139–147. Gohar HAF. 1957. The Red Sea dugong. Publ. Marine Biol. Station Al-Ghardaqa Red Sea, No. 9:3–49. Gudernatsch JF. 1908. Zur Anatomie und Histologie des Verdauungstraktes von Halicore Dugong Erxl. I. Mundhöhle. Gegenbaurs Morpholog Jahrbuch 37:586–613. den Hartog C. 1970. The sea grasses of the world. Verh. K. Nederl. Akad. Wetens., Afd. Natuurk. 2, 59(1):1–275. Heinsohn GE., Wake JA, Marsh H, Spain AV. 1977. The dugong (Dugong dugon (Müller)) in the seagrass system. Aquaculture 12: 235–248. Ivany LC, Portell RW, Jones DS. 1990. Animal-plant relationships and paleobiogeography of an Eocene seagrass community from Florida. Palaios 5:244–258. Johnstone IM, Hudson BET. 1981. The dugong diet: mouth sample analysis. Bull Marine Sci 31:681–690. Klumpp DW, Howard RK, Pollard DA. 1989. Trophodynamics and nutritional ecology of seagrass communities. In: Larkum AWD, McComb AJ, Shepherd SA, editors. Biology of seagrasses: a treatise on the biology of seagrasses with special reference to the Australian region. Amsterdam: Elsevier. p 394–457. Lanyon JM, Sanson GD. 2006. Degenerate dentition of the dugong (Dugong dugon), or why a grazer does not need teeth: morphology, occlusion and wear of mouthparts. J Zool Lond 268:133–152. Lanyon JM, Limpus CJ, Marsh H. 1989. Dugongs and turtles: grazers in the seagrass system. In: Larkum AWD, McComb AJ, Shepherd SA, editors. Biology of seagrasses: a treatise on the biology of seagrasses with special reference to the Australian region. Amsterdam: Elsevier. p 610–634. Larkum AWD, den Hartog C. 1989. Evolution and biogeography of seagrasses. In: Larkum AWD, McComb AJ, Shepherd SA, editors. Biology of seagrasses: a treatise on the biology of seagrasses with special reference to the Australian region. Amsterdam: Elsevier. p 112–156. Lumbert SH, den Hartog C, Phillips RC, Olsen FS. 1984. The occurrence of fossil seagrasses in the Avon Park Formation (late Middle Eocene), Levy County, Florida (U.S.A.). Aquat Botany 20:121– 129. Marsh H. 1980. Age determination of the dugong (Dugong dugon (Müller)) in northern Australia and its biological implications. Reports of the International Whaling Commission (Special Issue 3):181–201. Marsh H, Channells PW, Heinsohn GE, Morrissey J. 1982. Analysis of stomach contents of dugongs from Queensland. Aust Wildlife Res 9:55–67. Marsh H, Heinsohn GE, Marsh LM. 1984. Breeding cycle, life history and population dynamics of the dugong, Dugong dugon (Sirenia: Dugongidae). Aust J Zool 32:767–788. Marshall CD, Huth GD, Edmonds VM, Halin DL. 1998. Prehensile use of perioral bristles during feeding and associated behaviors of the Florida manatee (Trichechus manatus latirostris). Mar Mammal Sci 14:274–289. Marshall CD, Maeda H, Iwata M, Furuta M, Asano S, Rosas F, Reep RL. 2003. Orofacial morphology and feeding behaviour of the dugong, Amazonian, West African and Antillean manatees (Mammalia: Sirenia): functional morphology of the muscularvibrissal complex. J Zool 259:245–260. McRoy CP, Helfferich C. 1980. Applied aspects of seagrasses. In: Phillips RC, McRoy CP, editors. Handbook of seagrass biology: an ecosystem approach. New York: Garland Publications. p 297–342. Meñez EG, Phillips RC, Calumpong HP. 1983. Seagrasses from the Philippines. Smithsonian Contrib Mar Sci 21:iii þ 40. Packard JM. 1984. Impact of manatees Trichechus manatus on seagrass communities in eastern Florida. Acta Zool Fennica 172:21– 22. Phillips RC, Meñez EG. 1988. Seagrasses. Smithsonian Contribs. Mar Sci 34:vi þ 104. Pilleri G. 1988. The Pliocene Sirenia of the Po Basin (Metaxytherium subapenninum (Bruno) 1839). In: Pilleri G, editor. Contributions to the paleontology of some Tethyan Cetacea and Sirenia (Mammalia). Ostermundigen (Switzerland): Brain Anatomy Institute. p 45–103. Pocock RI. 1940. Some notes on the dugong. Ann Magazine Nat Hist (Series 11) 5:329–345. Powell JA Jr. 1978. Evidence of carnivory in manatees (Trichechus manatus). J Mammal 59:442. Preen AR. 1989. Observations of mating behavior in dugongs (Dugong dugon). Mar Mammal Sci 5:382–387. Preen AR. 1995. Diet of dugongs: are they omnivores? J Mammalogy 76:163–171. Promus J. 1937. Netting dugong. Walkabout 3:40–41. Sickenberg O. 1934. Beiträge zur Kenntnis tertiärer Sirenen. Mémoires du Musée Royal d’Histoire Naturelle de Belgique, No. 63:1– 352. Solounias N, Semprebon G. 2002. Advances in the reconstruction of ungulate ecomorphology with application to early fossil equids. Am Mus Novitates 3366:1–49. Spain AV, Heinsohn GE. 1973. Cyclone associated feeding changes in the dugong (Mammalia: Sirenia). Mammalia 37:678–680. Spinage C. 1994. Elephants. T&AD Poyser natural history series. Princeton: Princeton University Press. Toledo PM de, Domning DP. 1991. Fossil Sirenia (Mammalia: Dugongidae) from the Pirabas Formation (Early Miocene), northern Brazil. Bol. Museu Paraense Emı́lio Goeldi, Sér. Cienc da Terra 1:119–146. Uhen MD. 2007. Evolution of marine mammals: Back to the sea after 300 million years. Anat Rec (this issue) Vosseler J. 1924–25. Pflege und Haltung der Seekühe (Trichechus) nebst Beiträgen zu ihrer Biologie. Pallasia 2:58–67,113–133,167– 180,213–230. Wirsing AJ. 2005.Predation-sensitive foraging behavior of dugongs (Dugong dugon). Unpublished PhD Thesis, Department of Biological Sciences, Simon Fraser University.