During the Late Jurassic/Early Cretaceous, the Tendaguru area was located at a paleolatitude of 30° south of the equator (Scotese, 2001). Using sedimentary and fossil floral and invertebrate evidence, Aberhan et al. (2002) concluded that the Middle and Upper Saurian Beds, for which the new terms “Middle Dinosaur Member” and “Upper Dinosaur Member” were suggested by Bussert et al. (2009), represent deposits on siliciclastic tidal flats of a lagoonal system, intercalated with small tidal channels. In the upper part of the Middle Saurian Bed, occasional sabkha (salt flat)-like coastal plains with brackish lakes and ponds formed (Aberhan et al., 2002). These environments tend to be poorly vegetated. However, fossil floral elements such as Cheirolepidiaceae, Podocarpaceae, Araucariaceae and conifers, suggestive of a forest, and shrub-like elements such as pteridophytes and pteridosperms (Aberhan et al., 2002; Rees et al., 2004) most likely originate from the vegetated hinterland. Cheirolepidiaceae are widely considered to be a dominant element in forests bordering the sea in the Late Jurassic and Early Cretaceous (Alvin, 1982; Gee, 2011). The paleoenvironment of the hinterland can be best visualized as a savannah, covered by low shrubs, with higher vegetated areas around water courses (see Heinrich, 1999 for references; Rees et al., 2004; Heinrich et al., 2011). Dinosaurs that used the vegetated hinterland as feeding grounds may have died there and, after decomposition, their bones were transported by water courses, together with the plant remains, to the coastal plains where they were deposited and finally buried. Alternatively, dinosaurs, in search of water and food during dry seasons, may have died on the coastal plain itself being trapped in the mud and/or due to weakness as a result of water and food scarcity. The accumulation of mainly isolated large bones belonging to almost 40 individuals of Kentrosaurus in the ‘St’ quarry suggests some transport or reworking of the carcasses and/or sorting of the bones by tides or floods, during which smaller and, thus, lighter bones are carried farther by the water than large, heavy ones. The lack of abrasion indicates short to medium transport distances or reworking, during which disarticulation of the skeletal remains especially that of the peripheral body parts such as the skulls and most distal limb elements occurred. Trample marks on the bones and signs of scavenging may have been caused prior to or after transport/reworking and final deposition. The latter are rare and this observation as well as the lack of signs of pre-burial weathering indicate relatively rapid transport/little reworking after death and rapid burial after deposition of the isolated bones or partial skeletons. The stratigraphically younger quarry (‘X’ in the Upper Dinosaur Member) contains mass accumulations of manual and pedal elements of Kentrosaurus. This observation suggests two explanations: (a) Bones may have been sorted according to their size and weight. (b) Members of a Kentrosaurus herd might have died trapped in the mud and the decomposing carcasses were then transported by tides often leaving the stuck feet behind (Janensch 1914). The occurrence of manual and pedal elements only but lack of other similar-sized skeletal elements in the “X” quarry support the latter hypothesis.
Bone Histology and Ontogenetic Variation
The primary bone in all samples consists predominantly of FLB, which is mostly well vascularized. For comparative purposes, since we studied both cores and cross sections, histological characteristics and ontogenetic variation were assessed for the anterior regions of the femoral shaft for all samples (which is the standard sampling location for drill-core samples).
The spacing of growth marks changes from wide, as in MB.R.3605, to more narrow, as seen e.g., in MB.R.3572 (Fig. 4B) and St 72 (Fig. 8A), and finally to very narrow spacing as seen e.g., in MB.R.3583 (Fig. 7B) and MB.R.3593 (Fig. 9A). The first decrease in spacing, i.e., slow-down in growth, is probably linked to attainment of sexual maturity due to resource allocation from growth to sexual reproductivity (Andrews, 1982; Shine and Charnov, 1992; Sander, 2000). The second decrease occurs when the animal is somatically mature (Sander, 2000). Non-avian dinosaurs reach sexual maturity at a fraction of their asymptotic size (somatic maturity; Sander, 2000; Erickson et al., 2008; Lee and Werning, 2008). Thus, only specimen MB.R.3605 (345mm femur length) is interpreted as a sexually immature individual; specimens MB.R.3572, St 72 and possibly also MB.R.3571 (482mm, 644mm and 519mm femur lengths, respectively) are sexually mature, and specimens MB.R.3583 and MB.R.3593 (609mm and 689mm femur lengths, respectively) appear to be somatically mature. Growth marks are absent in the scapula Ig436 and it is thus considered to be sexually immature, which is supported by the juvenile appearance of the overall tissue including large, almost unfilled vascular canals. Secondary remodeling is almost absent in this scapula sample as well as in the sexually immature specimen MB.R.3605. In the sexually mature, somatically immature specimens, secondary osteons replace primary tissue up to the outer cortex (up to the mid cortex in St 72), and up to the outermost cortex in somatically mature specimens. Secondary reconstruction in the deep and mid cortex in specimen MB.R.3571 (Fig. 5B) supports our interpretation that it is a sexually mature though somatically immature specimen. The prevalent reticular vascularization changes into a more longitudinal orientation of the vascular canals in the outermost cortex in sexually mature, somatically immature specimens. The outermost cortex in somatically mature specimens is almost avascular and consists of lamellar bone, which is indicative of an EFS.
Ontogenetic changes are well correlated to increasing lengths of sampled femora (Table 1, Fig. 1), with the exception of specimen MB.R.3583 and St 72. The latter specimen is sexually mature but lacks an EFS, i.e., it is somatically immature, although its femur length of 644 mm exceeds the length of the somatically mature specimen MB.R.3583 (609 mm). Either, specimen St 72 is a particularly large specimen or MB.R.3583 a particularly small one. Specimen St 72 appears even younger than the next smaller, somatically immature specimen MB.R.3571 as secondary remodeling is less extensive and the outermost cortex is higher vascularized than in the latter. We therefore assume specimen St 72 to be an exception. This variation in size in relation to growth rate and timing may be an indication of either strong developmental plasticity in the species (Sander and Klein, 2005; Sander et al., 2011) or perhaps of sexual dimorphism.
Growth marks are common but surprisingly variable in the Kentrosaurus specimens sampled. (1) LAGs, annuli and shifts in vascularization patterns (see discussion below) can occur in a single bone, e.g. in a single cross section of specimen MB.R.3571; (2) a growth mark can change from an annulus to a LAG or can be reflected as a vascularization shift on different sides of a single cross section of the bone, e.g., in specimen MB.R.3571; and (3) a single growth mark can separate into two growth marks on just one side of the cross section as in e.g., sample St 72. All these variations influence the count of the number of growth marks. Canal organization, which is used herein as a proxy for vascularization (even though vascular canals contain blood vessels, connective tissue etc.; Starck and Chinsamy, 2002; Hurum and Chinsamy-Turan, 2012), varies within each sample. Vascularization shifts occur in two specimens only, the smallest and youngest individual, MB.R.3605, and the third smallest, sexually mature, somatically immature specimen MB.R.3571. In MB.R.3605, the two shifts in vascularization patterns from reticular to longitudinal are the only growth marks present. On the posterior side of sample MB.R.3571, the mainly reticular vascularization pattern changes three times to mainly circumferential vascular orientations between LAGs 1 and 2, accompanied by switches from woven to PFB. Two of the vascularization shifts change laterally into two annuli on the anterior side, one still accompanied by a vascularization shift. The innermost shift is completely remodeled anteriorly. All shifts are not evident on the postero-medial side of the bone; lateral changes also affect the other growth mark types in this specimen, e.g., the three LAGs on the anterior and posterior side become three annuli on the postero-medial side. These changes are possibly the result of localized higher relative growth rates due to a muscular attachment site in the postero-medial region, resulting in a thicker bone wall.
Growth variation in specimen MB.R.3572 includes the occurrence of LAGs and annuli, as well as a LAG accompanied by an annulus. Four annuli and four LAGs alternate in specimen St 72. Lateral changes include separation of an annulus into two annuli on the anterior side and separation of a LAG into two LAGs on the lateral side of the bone. Specimens MB.R.3583 and 3593 experienced the least growth variation, showing regular LAGs and/or annuli, which may, however, be a result of the growth record being reduced by secondary remodeling.
It is widely accepted today that bone depositional rates and hence, relative growth rates can be inferred from the organization of histological tissues in the bone (Amprino, 1947; Chinsamy-Turan, 2005, 2012; Erickson, 2005; Cubo et al. 2008; Montes et al., 2010). It is further well recognized that FLB with frequent vascular canals, found especially in birds, mammals and many dinosaurs, reflects higher growth rates than in most reptiles and all amphibians which regularly have lamellar-zonal tissue with sparse vascularization reflecting slow growth (e.g., de Ricqlès, 1980; Reid, 1987; Castanet and Smirina, 1990; Chinsamy-Turan, 2005; Erickson, 2005; Cubo et al. 2008; Montes et al., 2010). Varying vascularization patterns in the FLB indicate different bone depositional rates, sometimes even within a single bone tissue or a single individual (Castanet et al., 1996, 2000; Starck and Chinsamy, 2002; de Margerie et al., 2004; Klein and Sander, 2007; Hübner, 2012; Cubo et al., 2012). Experimental studies on penguin chicks documented that the FLB with a radial organization of canals has the fastest depositional rates, while FLB with reticular and longitudinal organization of canals have intermediate depositional rates, and FLB with circumferential (laminar) canal organization is formed with the slowest depositional rates. These growth rates are, however, relative and not absolute as they vary in different skeletal elements (de Margerie et al., 2004:876).
A vascularization change from reticular to longitudinal canals and change in tissue to lamellar bone in the outermost cortex of the somatically mature specimens, representing an EFS, suggest slow relative growth rates for the longitudinal pattern in Kentrosaurus. Shifts in vascularization from reticular to circumferential or longitudinal in MB.R.3605 and MB.R.3571, accompanied by a shift to PFB, indicate a periodic decrease in the relative bone depositional rate and hence, growth rate. A combination of a vascularization shift with an annulus in MB.R.3605 and a lateral change of a vascularization shift into an annulus in MB.R.3571 support our interpretation that the changes in vascularization reflect growth marks.
The increasing number of growth marks throughout the specimens correlates with the ontogenetic changes observed in the histology as well as with their femur lengths (Table 1). The largest specimen, MB.R.3593, shows the fewest growth marks but the overall compacta shows extensive secondary reconstruction that has obliterated evidence of earlier growth marks.
Cyclic variations in the tissue, such as LAGs or annuli, are assumed to reflect seasonal/annual changes in the ambient temperatures, precipitation and/or vegetation and, thus food supply, by many authors (e.g., Peabody, 1961; Castanet et al., 1993; Sander and Andrassy, 2006). This assumption is strongly supported by a recent study of growth marks in modern wild ruminants from varying climatic environments (Köhler et al., 2012), showing that (a) growth marks occur in ruminants from all regions and (b) growth is arrested during the unfavorable season (low temperature and rainfall and thus, low food supply) for energy conservation, resumes with progressively faster rates (represented by circumferential or reticular vascularization) peaking at the height of the favorable season, and declines towards the next unfavorable season.
Beside LAGs and annuli, we observed vascularization shifts in the primary bone of two Kentrosaurus samples. These are very similar to the textural shifts documented in the polar Edmontosaurus (Chinsamy et al., 2012), in which periodic shifts from circumferential to reticular FLB are observed. Shifts are less distinct in Kentrosaurus than in the polar Edmontosaurus, but they are more distinct and frequent than in the temperate Edmontosaurus (Chinsamy et al., 2012). Spacing between the cycles varies from the anterior to posterior side of the femur according to the cortical thickness, and cross-sectional anatomical shape of the bone. The occurrence of various growth mark types in Kentrosaurus may indicate variable responses by the animal to environmental changes or other paleoecological influences, such as breeding season. We further postulate that these changes were seasonal since they tend to interrupt periods of rapid bone deposition and the modulations in growth rate appear similar to those recorded for other dinosaurs (Chinsamy et al., 2012; Hübner, 2012). Thus, a LAG forms when bone deposition is paused, while an annulus, often having lamellar bone tissue, suggests slowed-down rates of bone deposition (e.g., Chinsamy-Turan, 2005, 2012). In the case of two Kentrosaurus specimens and the polar Edmontosaurus, it appears that during the period of slow growth, growth was still reasonably fast to assure the deposition of PFB and FLB, respectively, but changes in the organization of vascularization suggests a slight change in the rate of bone deposition. In the sympatric ornithopod dinosaur Dysalotosaurus lettowvorbecki, LAGs are very rare to absent in femora (Chinsamy, 1995; Hübner, 2012) but a remarkably similar pattern of vascularization shifts as in Kentrosaurus occurs in the femur, tibia and humerus and have been interpreted as seasonal growth marks by Hübner (2012: Figs. 5C–H, 8E–H, 10). However, this type of growth marks is only evident in a restricted part of the cortex in Dysalotosaurus long bones (Hübner, 2012: Fig. 1). Interestingly, LAGs are also rare in the sympatric sauropods Brachiosaurus (Giraffatitan) brancai, Dicraeosaurus spp., Janenschia robusta and Diplodocinae indet. (=“Barosaurus africanus”) (Sander, 2000; Sander and Tückmantel, 2003; Sander et al., 2011), and when they occur, they developed in the EFS. In contrast, the so-called ‘polish lines’, which are growth lines in FLB that are visible in polished section but usually not in thin-section, are much more abundant and more regularly developed in these taxa. They were interpreted as growth marks demonstrating a strong slow down but not a complete cease of growth (Sander, 2000).
Varying growth marks may reflect varying periodic stress, either in duration or severity of the unfavorable season, and thus, the co-occurrence of vascularization shifts and/or LAGS and/or annuli in the Tendaguru dinosaurs may suggest variable responses to fluctuations in the seasonal changes of this environment. Köhler et al. (2012) documented that arrested growth is of longer duration in high latitudes (longer duration of unfavorable season) than in low latitudes and is expressed by simple LAGs but not in variations of the type of growth marks as seen in Kentrosaurus. This may suggest that Kentrosaurus reacted differently to fluctuations in seasonal duration and/or severity than modern ruminants.
Klevezal (1996) suggested that populations living in strongly seasonal environments develop regular growth marks whereas populations inhabiting less seasonal environments (even if both populations belong to the same species, e.g., wolf or Edmontosaurus) show much more variation in the development and regularity of growth marks (see also Hübner, 2012). Sedimentary evidence and the occurrence of fossil charcoal in the Tendaguru Formation support pronounced dry seasons alternating with seasonal rainfalls in a subtropical to tropical, semi-arid paleoclimate (Hallam, 1985, 1993; Valdes and Sellwood, 1992; Valdes, 1994; Aberhan et al., 2002; Rees et al., 2004). Seasonal rainfalls are indicated by the presence of Glyptostrobus- and Podocarpus-related wood (Figueiral et al., 1999), fungi infesting and degrading the wood (Alvin et al., 1981) as well as caliche nodules (Heinrich, 1999; Rees et al., 2004). Modern savannahs in similar latitudes and of similar climate experience one rainy season per year during the summer months (the 'high-sun' period from October to March; Rees et al., 2004) of the southern hemisphere with an average of 4–6 months of rainfall, which can decrease to < 4 months during drier years and increase to > 6 months during wetter years (e.g., Nicholson, 2000). Fluctuations of rainfall were remarkable throughout the African continent from the late Pleistocene to the last decades and even large yearly fluctuations were measured in the last approximately 100 years (Nicholson, 2000). Climate fluctuations of the Mesozoic cannot be directly compared to today's climate as the continental distribution differed and would have affected the circulation patterns of water currents, winds and atmospheric pressure that result in different precipitation patterns during the Mesozoic. However, as the presence of evaporites and microphyllous flora and conifers indicate generally arid to semi-arid conditions, and caliche nodules an alternating dry-wet climate (Heinrich, 1999; Rees et al., 2004), this savannah-like climate with one rainy season per year and fluctuations of rainfall in the rainy season are likely for the Mesozoic Tendaguru environment. Growth in Kentrosaurus may have been arrested, resulting in the deposition of LAGs, when vegetation was restricted to a few water courses, i.e., in a dry season (winter), and resumed at its usual rapid rate at the beginning of the following rainy season (summer) when vegetation was flourishing. Fluctuations in precipitation would presumably result in varying responses to bone depositional rate (i.e., FLB with circumferential canals indicating a slightly decreased rate of bone formation, even slower growth indicated by an annulus, or an arrest of growth reflected in LAGs). If the observed variation of growth marks in Tendaguru dinosaurs indeed reflects fluctuations in annual precipitation and food supply, it would indicate that growth of these animals was remarkably susceptible to environmental/climatic changes.
Direct correlation of the varying growth marks in Kentrosaurus to climatic fluctuations is, however, problematic since variation across a single section can lead to different interpretations. Variation in type and regularity of growth marks is influenced by a combination of extrinsic factors, such as climatic fluctuations, and intrinsic factors, such as biomechanical properties of the bone (e.g., Chinsamy and Abdala, 2008; Hübner, 2012).
Comparison with Other Thyreophora
Bone histological studies on long bones of thyreophoran dinosaurs (Fig. 10) include that of the basal Scutellosaurus (Padian et al., 2004) and Stegosaurus (Hayashi et al., 2009; Redelstorff and Sander, 2009). Two Scutellosaurus femora sampled by Padian et al. (2004) consist of woven bone and are poorly vascularized by mainly longitudinal canals; vascularization is slightly denser in the internal cortex, however, remains less so than in most dinosaurs. Other long bones of Scutellosaurus (radius, tibia, ulna) even exhibit PFB with scarce vascular canals (and osteocyte lacunae). Padian et al. (2004) concluded that Scutellosaurus bone histology is characterized by a very slow growth rate, more similar to the bone histology of a captive Alligator (e.g., Lee, 2004; Woodward et al., 2011). Various long bones of small (femur length 233 mm) to large-sized (femur length 915 mm) Stegosaurus specimens were analyzed by Hayashi et al. (2009). They reported FLB with predominating radial and reticular canals with strong regional variation: radial canals on the posterior side and longitudinal canals grading into reticular ones on the anterior side. The medium- and large-sized specimens show similar bone histologies with mainly longitudinal vascular canals in FLB. In the studied specimens of Stegosaurus, longitudinal primary osteons are prevalent in the FLB that consists of a mixture of woven and PFB (Redelstorff and Sander, 2009). Although vascularization patterns are similar in Scutellosaurus and Stegosaurus (both being dominantly longitudinal), vascular canals are notably more scarce in the former and located within PFB, which implies even more slowly formed bone tissue than in Stegosaurus. Our finding of well vascularized, FLB in the phylogenetically intermediate Kentrosaurus is indicative of a more rapid growth rate, and suggests that the slow growth rates previously observed in other thyreophorans are not uniformly present throughout the clade.
An osteohistological analysis of a range of small- to large-sized dinosaurs by Padian et al. (2004) included the basal thyreophoran ornithischian Scutellosaurus (∼1 m adult body length), the basal ornithopod Orodromeus (1.5 m body length), the theropod Coelophysis (∼1.7 m adult body length), the basal ceratopsian Psittacosaurus (∼2 m adult body length) and the hadrosaur Maiasaura (∼7 m adult body length) and shows that large dinosaurs exhibit faster growth rates than small dinosaurs. Linear regression analysis of adult body mass and maximum growth rates quantatively supports this observation (e.g., Erickson et al., 2001, 2009; Lehman and Woodward, 2008; Hübner, 2012). This hypothesis is further supported when comparing the very slow growing, small Scutellosaurus (∼1 m adult body length) with the slightly faster-growing, large Stegosaurus (approximately 7–9 m). However, intermediate-sized Kentrosaurus (∼6 m) shows higher absolute growth rates than the large, more derived Stegosaurus, and contradicts the supposition by Padian et al. (2004). It is therefore possible that the plesiomorphic condition in the thyreophorans is (a) a rapid growth strategy, but that Stegosaurus for some as yet unknown reason secondarily reduced its growth rate or (b) a slow growth strategy and that Kentrosaurus exhibits an autapomorphic increase in growth rate.