PNEUMATICITY, NECK LENGTH, AND BODY SIZE IN SAUROPODS Mathew Wedel, UCMP, Berkeley, CA 94720-4780 · [email protected] · http://www.sauroposeidon.net/ ABSTRACT UPDATED! 4. HOW ARE THE VARIABLES RELATED? Long necks are synapomorphic for sauropods, and additional neck elongation occurred independently Organisms and their traits are not statistically independent because they are united by in different sauropod clades. Increases to 15 or more cervical vertebrae occurred at least four times, in common descent. Phylogenetically independent contrasts must be used to remove the effect of mamenchisaurs, diplodocids, Euhelopus, and Rapetosaurus. Necks longer than 10 meters also evolved at phylogeny from studies of character correlation (Felsenstein 1985). I used the PDAP module in least four times, in mamenchisaurs, diplodocids, brachiosaurids, and giant titanosaurs like Puertasaurus. The Mesquite v1.12 (Milford et al. 2005, Maddison & Maddison 2006) to test correlations among the longest-necked sauropod for which a rigorous estimate is possible is Supersaurus, which had a neck at least following: absolute neck length (ANL), trunk length (TL), femur length (FL), cervical count (CC), 13.7 meters long. The sauropods with the longest necks, such as Supersaurus, Sauroposeidon, and maximum elongation index (MEI), proportional neck length (PNL; ANL/TL), air space proportion Puertasaurus, are also among the largest known terrestrial vertebrates. (ASP). Pneumatic vertebrae facilitated neck elongation in sauropods. The lightest sauropod vertebrae were I ran two sets of tests. The first set included only the 15 taxa for which ASP values are known 89% air by volume and had a specific gravity (SG) of 0.22. By comparison, cervical vertebrae of the or could be reasonably estimated (see Table 1). The second set of tests included all 19 taxa in Table giraffe have an SG of 1.3, and are scarcely lighter than other postcranial elements (SGs up to 1.7). The 1, but did not include the ASP variable. Both sets of tests were run using four different tree cervical column of Brachiosaurus is 8.5 meters long but the vertebrae would have totaled less than 600 topologies: (1) the topology of Wilson (2002), with Massospondylus, Plateosaurus, and Cetiosaurus kilograms—about the same mass as the animal’s paired humeri, which are each only two meters long. C added, and prosauropods monophyletic; (2) the same topology but with prosauropods paraphyletic; The relationships between body size, neck length, and pneumaticity can be evaluated statistically using (3) the topology of Upchurch et al. (2004), with Massospondylus and Plateosaurus added, and EI phylogenetically independent contrasts. I used the PDAP module in Mesquite v1.06 to test the correlations prosauropods monophyletic; and (4) the topology of Rauhut et al. (2005), with Massospondylus, among femur length (FL), trunk length (TL), absolute neck length (ANL), proportional neck length (PNL; 16 Plateosaurus, and Cetiosaurus added, and prosauropods monophyletic. These phylogenetic neck length/dorsal length), cervical count (CC), maximum elongation index (MEI), and air space proportion rearrangements had very little effect on character correlation. ANL and FL are only correlated when (ASP) in 14 sauropodomorphs. ASP and the size-independent measures of neck length (PNL, CC, and MEI) the topology of Upchurch (2004) is used; otherwise the same statistically significant correlations form a pool of mutually-correlated variables (at p<0.05), as do the variables related to absolute size (FL, TL, were detected under all topologies. In addition, the same relationships are found among the and ANL). However, few significant correlations link the two pools. Neck elongation in sauropods is tied to variables other than ASP in the tests with 15 and 19 taxa. Statistical relationships among the 2 pneumaticity but largely independent from size-related variables. variables are shown below; black lines indicate significant correlations (r > 0.65, p < 0.05). INTRODUCTION MEI NECK PROPORTIONS Sauropods were the largest and longest-necked terrestrial animals of all time. The presacral vertebrae of most sauropods were filled with pneumatic spaces that would have greatly reduced the mass of the Figure 3. At roughly 600 kg, the 8.5-meter cervical series of Brachiosaurus had about 12 FL PNL vertebral column. The goal of this project is to answer the following questions: 15 the same mass as the animal’s paired 2-meter humeri. C 1. How many times did characters related to neck length and pneumaticity evolve ANL independently in sauropods? EI CC 3. HOW MUCH DOES PNEUMATICITY MATTER? TL 2. How long was the longest neck of any sauropod? In contrast to prosauropods and basal sauropods, mamenchisaurids and neosauropods had extensive 2. Was pneumaticity important in reducing the mass of sauropod vertebrae? vertebral pneumaticity. Percent pneumatization of the presacral vertebrae—the number of pneumatic elements ASP divided by the total presacral count—rose to a maximum early in sauropod evolution and was subject to only ABSOLUTE SIZE 3. Are variables related to pneumaticity, neck length, and body size statistically correlated? PNEUMATICITY one notable reversal, in dicraeosaurids (Table 1). The idea that pneumatic bones are lighter than marrow bones is intuitively appealing, but it needs to be tested. In particular, we need to know how much lighter pneumatic The variables related to absolute size (ANL, FL, and TL) and neck proportions (CC, MEI, 1. HOW MUCH HOMOPLASY? vertebrae are compared to apneumatic vertebrae if we want to understand any functional advantage. and PNL) share many correlations within pools but few correlations between pools: absolute neck Resolving characters on a phylogeny is a powerful tool for exploring evolutionary patterns. In The presacral vertebrae of most neosauropods were on average 60% air by volume (Wedel 2005, length is correlated with maximum elongation index and proportional neck length (and the latter particular, repeated evolution of characters in closely related lineages can elucidate shared developmental Woodward 2005, Schwarz & Fritsch 2006). The centrum walls, laminae, septa, and struts that composed the may be the result of autocorrelation, because PNL is based in part on ANL). Pneumaticity is processes and evolutionary trends. I collected character data by CT scanning and personal observation when Missing elements vertebrae are primarily made up of compact bone (Reid 1996). The specific gravity (SG) of compact bone is correlated with absolute and proportional neck length and cervical count. Importantly, in all possible, and from the literature when necessary (Table 1). The phylogeny shown here (Fig. 1) is broadly EI 2.0 in most tetrapods, so an element with an ASP of 0.60 would have an in vivo SG of 0.8. Some sauropod permutations none of the pneumaticity or neck proportion variables are correlated with femur length C vertebrae were much lighter. For example, Sauroposeidon had ASP values up to 0.89 and thus SG as low as or trunk length. congruent with several recent cladistic analyses (Wilson 2002, Upchurch et al. 2004, Rauhut et al. 2005, Increases in cervical count 17 Harris 2006), but agnostic about the relationships of several taxa that have proven to be unstable in those 15 0.22. On the other hand, many basal sauropods had ASPs of 0.30-0.40 and SG of 1.2-1.4. analyses (e.g., Cetiosaurus, Jobaria, and Haplocanthosaurus). To explore the impact of pneumaticity on skeletal construction, I estimated the in vivo mass of the CONCLUSIONS cervical skeleton of Brachiosaurus. From CT scans it is possible to calculate the volume of bone tissue in a The evolution of long necks in sauropods is marked by the repeated evolution of several important EI Elongation index > 4.0 1. Many characters related to neck length and pneumaticity evolved in parallel in different single vertebra and thus determine the mass of the element. I multiplied the mass of a single vertebra by scale characters, including the number of cervical vertebrae, the length of the individual vertebrae, the complexity 15 sauropod lineages. factors to determine the masses of the other vertebrae in the neck, and added the resulting values to estimate of vertebral internal structure, and the ratio of bone to air space. 2. Necks longer than 10 meters evolved independently in four sauropod clades. Supersaurus C Complex internal structure the mass of the whole cervical series. The mass of the cervical column of Brachiosaurus is 590 kilograms. For Cervical Count—If the primitive number of cervical vertebrae in Titanosauria is 13 (Gomani 2005), had the longest neck of any known animal: at least 13.7 meters, and possibly more than 16 the sake of comparison, I used graphic double integration (Hurlburt 1999) to estimate the mass of a then increases to 15 or more cervicals occurred at least four times: in the Mamenchisauridae, Diplodocidae, meters. Brachiosaurus humerus at 290 kg. The animal’s humeri are each 2 meters long but together they weighed as Euhelopus, and Rapetosaurus. Alternatively, 15 or more cervicals may be primitive for Somphospondyli, the 3. Pneumatic presacral vertebrae of neosauropods were, on average, 60% air by volume, and much as the 8.5-meter cervical series. The presacral vertebrae of Brachiosaurus have an average ASP of 0.67 lower counts in Malawisaurus and Saltasaurus may be reversals, and increases to 15 or more cervicals may therefore half as dense as the marrow bones of mammals. Figure 1. The evolution of long necks in sauropods. and a mean SG of 0.66. The cervical vertebrae of the giraffe are almost exactly twice as dense, with SG of 1.3 have only happened three times. The relatively primitive count of 13 cervicals in Brachiosaurus is 4. When the effects of phylogeny are removed by the use of independent contrasts, neck (Van Schalkwyk et al. 2004). If the cervical vertebrae of Brachiosaurus were built like those of a giraffe, they noteworthy; no other known sauropod had such a long neck with so few vertebrae.