In contrast to long bones, ribs have not been a focus of scientific study for decades. Nopcsa (1933), in an innovative study for that time, was the first to analyze dinosaur rib histology using ornithopod ribs. Only a few recent histological papers focused on ribs (e.g., de Buffrénil et al., 1990, D’Emic et al., 2015 and Waskow and Sander, 2014). Canoville et al. (2016) published the first quantitative exploratory study about the microanatomical diversity of amniote ribs. A few histological studies compared the dorsal rib histology of some dinosaurs with other skeletal elements (Erickson et al., 2004, Erickson, 2005 and de Ricqlès et al., 2008). Most papers that take rib histology into account concern marine vertebrates (e.g., Hayashi et al., 2013, Houssaye, 2009, Houssaye, 2013, Houssaye and Bardet, 2012 and Laurin et al., 2007). More recently, Klein et al. (2012) stated that sauropod neck ribs are ossified tendons. The first study focusing exclusively on sauropod rib histology was done by Waskow and Sander (2014). The lack of focus on ribs is primarily because humeri and femora, for example, allow growth curve calculations based on skeletochronology (caused by their nearly circumferential shape).

The study of Waskow and Sander (2014) shows that, in contrast to long bones that have most of the growth record preserved in the midshaft region (Currey, 2002 and Hall, 2005), ribs, due to their growth trajectory from proximal to distal end, have the most complete growth record at the proximal end of the rib shaft. According to Amprino's rule (Amprino, 1947), the primary bone structure is closely related to the bone apposition rate. This rule has been tested and confirmed in several studies related to vertebrate long bones (Castanet et al., 1996, de Margerie et al., 2004 and de Ricqlès et al., 1991), but in ribs, the relative bone apposition rate is also related to morphological changes during ontogeny, as determined by the histological rib section series of Waskow and Sander (2014). In addition, ribs grow much more slowly than long bones, which causes a difference in bone tissue type, remodeling rate, or number of preserved LAGs, even within the same individual (Horner et al., 2000, Waskow and Sander, 2014 and Woodward et al., 2014). The differences in the LAG number counted in different skeletal elements of a same individual can also be explained by different resorption rates during the expansion of the medullary cavity and the difference in the degree and rate of remodeling (Waskow and Sander, 2014). The fast-growing fibrolamellar bone is also more strongly influenced by Haversian remodeling than the slower-growing lamellar-zonal bone (Chinsamy-Turan, 2005 and Enlow and Brown, 1956). Moreover, the mechanical forces and loads that affect a bone differ in the individual skeletal parts and might therefore influence the primary bone tissue type (Currey, 2002 and Dumont et al., 2014). The fact that ribs, in contrast to long bones, do not have an isometric growth could also be an advantage for the growth record preservation. Due to the formation of the rib ridge, the medullary cavity seems to expand less on the posteromedial side during growth, resulting in lesser resorption of the primary bone tissue and therefore more recorded LAGs.

Institutional abbreviations – ML: Museu da Lourinhã, Lourinhã, Portugal; SMA: Sauriermuseum Aathal, Aathal, Canton Zürich, Switzerland.

The following specimens used for this study represent a phylogenetical broad but incomplete selection of archosaurs including saurischians (Theropoda and Sauropoda), ornitischians (Stegosauria and Ornithopoda) and crocodylomorphs. All samples (excluding the literature data of sauropod SMA 0002) are from the Lusitanian Basin, western Portugal. All but one (Baryonyx walkeri) are dated from the Kimmeridgian–Tithonian Lourinhã Formation according to regional stratigraphic context, vertebrate fauna and strontium isotope dating (Mateus et al., 2014). This formation yields dinosaur bones, crocodylomorphs, pterosaurs, mammals, amphibians and other Jurassic vertebrates as well as teeth, eggs and tracks (e.g., Araújo et al., 2013, Hendrickx and Mateus, 2014 and Milàn et al., 2005), providing a unique window into the terrestrial Late Jurassic fauna of Europe. This study includes dorsal rib histology from the proximal rib shaft of a Camarasaurus specimen used in Waskow and Sander (2014), as well as the following dorsal rib samples of specimens from the ML collections.

SMA 0002: Camarasaurus sp. (Sauropoda): specimen SMA 0002 is one of the most complete and articulated Camarasaurus skeletons ever found. Its assignment to Camarasaurus based on several diagnostic features mentioned by Ikejiri (2004), McIntosh (2005), and Upchurch et al. (2004), but this attribution has been recently challenged by Mateus and Tschopp (2013), who identified it as Cathetosaurus, based on the morphology of the pelvic girdle and dorsal transverse processes. In more recent studies, however, it was again claimed to be a Camarasaurus specimen, based on its autopodial and preliminary unpublished phylogenetic analysis (Tschopp et al., 2015 and Tschopp et al., 2016). The SMA 0002 specimen was found in the Upper Jurassic Morrison Formation (Kimmeridgian), which is situated in the central northern part of Wyoming, USA. It is the stratigraphically oldest Camarasaurus in the whole formation (Ayer, 2000 and Waskow and Sander, 2014). With three-dimensional preservation, the small but mature individual lacks only the vomers, the splenial bones, the distal end of the tail, and one terminal phalanx of the right pes (Tschopp et al., 2015 and Waskow and Sander, 2014). Former histological studies of the long bones and ribs suggest a senescent age of this individual (Klein and Sander, 2008 and Waskow and Sander, 2014), with a growth record of 37 preserved LAGs (Waskow and Sander, 2014).

All samples were taken as complete cross-sections from well-preserved regions of the proximal rib shaft of the different taxa. First, molds of all sampled rib parts were made to enable the reconstruction of the missing bone slice after sampling. These molds were made with a two-component silicone “Provil novo” consisting of the base putty and its catalyst (proportion 1:1). Each rib was protected with a solution of Technovit 4071, a technical polymer. The two components of this polymer (powder and liquid) are usually used in a ratio of 2:1. Here we used a 3:1 ratio to make the liquid suspension more viscous to properly mold the whole sampling region of the rib shaft. The protected area then was cut with an automatic rock saw and cut into standard paleo-histological thin sections (see Chinsamy and Raath, 1992, Enlow and Brown, 1956, Lamm, 2007 and Wells, 1989). After sawing, the hardened Technovit polymer was removed with acetone. Rib pieces were reconnected by putting the rib sections back in the mold and refilling the missing bone material with plaster. The authors decided to use whole cross-sections instead of core drilling (see Stein and Sander, 2009) to be sure that the most complete growth record preserved could be detected. Thin sections were analyzed under a Leica DML P polarizing microscope. The drawings of the sections were made with a Camera Lucida (Schott KL 1500 LCD). Photos were taken with a Leica DFC420 digital camera and processed with Imagic ImageAccess software.

In some cases, the center of the medullary region, where the bone growth initiated, is not located in the center of the section itself. Thus, the center of the medullary region was determined following the method of Waskow and Sander (2014) by the intersection of the two imaginary lines connecting the two greatest diameters of the rib. Depending on the region of a section, the distance between two same LAGs vary laterally. We chose to measure the LAG interdistances in the thickest part of the cortex, on the posteromedial side of each section. The percentage of the preserved growth record was calculated by measuring the distance from the medullary region to the surface of the bone at the thickest part of the cortex, and the distance from the first visible LAG to the bone surface. By assuming that the first measured distance equals 100% of the growth record, preservation was calculated after the rule of proportion. Based on the assumption that the LAGs are generally more widely spaced in the inner part of the bone (Chinsamy-Turan, 2005), the distance between the center of the medullary region and the first visible LAG was measured and divided by the widest growth cycle to obtain the maximum number of resorbed LAGs.

The primary bone tissue type and the low vascularization in this specimen suggest a relatively low deposition rate. The EFS indicates that the animal reached skeletal maturity before death. Using the method of retrocalculation, we can assume a maximum number of three to four missing growth cycles. Together with the 22 counted annuli, this would raise the age of skeletal maturity to about 25 years. If the EFS is excluded, the animal would have reached full size after 21 to 22 years of growth. Because of the relative completeness of the growth record preserved in the ribs, this is the most complete estimation for skeletal maturity in stegosaurs to date. Previous studies on this group have obtained a less complete growth record, or were only able to erect a general histologic ontogenetic stage (HOS) erected by Klein and Sander (2008) (Hayashi et al., 2009, Redelstorff and Sander, 2009 and Redelstorff et al., 2013). Here, a significant decrease in growth is observed after the 11th counted LAG (Fig. 6). This is interpreted as the point of sexual maturity, which would be reached at 14 to 15 years old, that is, long before the animal was skeletally mature.

The preserved EFS indicates that the specimen must have been a fully grown individual. A retrocalculation suggests that a maximum number of 1-2 growth cycles are missing. Therefore, the animal reached its skeletal maturity after 27 to 31 years of growth, depending on whether the EFS is included or not. Like in the stegosaur M. longicollum (see above), a decrease in the rib growth rate is observed (Fig. 7), but it is not as prominent. Between the 11th and 16th counted LAG, this growth rate decrease is observed, and therefore, sexual maturity can be estimated to be reached between the ages of 12 and 17.

Due to the expansion of the large medullary cavity of the specimen ML 370, only 57% of the growth record is preserved. Depending on the distance between LAGs used for retrocalculation, there are approximately five to eight LAGs missing in the inner cortex. An EFS is missing. The age at death of this skeletally immature individual is suggested to be between 14 and 17 years. Unfortunately, the LAGs counted in this section show up in different parts of the cortex. Some LAGs are only visible posteriorly, while others are only visible anteromedially. In contrast to all other sections, not all counted LAGs are visible in the thickest part of the cortex (the posteromedial side of the section), which means that it was not possible to measure an accurate distance between the single LAGs to generate the growth graph. Therefore, no evidence for the point of sexual maturity is apparent.

The specimen ML 1190 shows a clearly marked decrease in growth in the last 7 LAGs. Therefore, it is most likely that the individual was close to its maximum size. Including two to three missing LAGs computed by retrocalculation, it can be assumed that the animal died between the ages of 23 and 25 years, nearing its skeletal maturity. This is contrary to the anatomical evidence, that is, the neurocentral sutures are unfused, which suggests a rather earlier ontogenetic age. The occurrence of mature traits (based on the rib tissues) and subadult traits (partially unfused neurocentral sutures) could be interpreted as a possible paedomorphic trait for this taxon. Paedomorphosis has already been suggested in various aquatic or subaquatic tetrapods such as spinosaurids (Ibrahim et al., 2014), plesiosaurs (Araújo et al., 2015 and Storrs, 1993) and temnospondyls (e.g., Steyer, 2000), and could be related to a swimming mode of locomotion. A more detailed analysis including a comparison of bone compactness using software such as Bone Profiler (e.g., Houssaye et al., 2016, Laurin et al., 2011 and Steyer et al., 2004) would be useful for a more detailed lifestyle interpretation, but is beyond the scope of this study. A significant decrease in the growth rate after the 11th or 12th countable LAG suggests that the point of sexual maturity was reached at an age of 13 to 15 years (Fig. 8).

The bone tissue shows a slow deposition rate. This specimen did not reach its skeletal maturity before its death. The missing EFS indicates that the individual was indeed still growing. Considering one to two growth cycles are missing based on retrocalculation, the animal died at an age of 19 to 20 years. After the 9th counted LAG, a decrease in growth is observed (Fig. 9) and interpreted as the point of sexual maturity. In recent crocodiles and alligators, a significant sexual dimorphism has been observed (Chabreck and Joanen, 1979, Platt, 1996, Platt et al., 2008 and Wilkinson and Rhodes, 1997). While the smaller females reach sexual maturity after nine to ten years, the larger males need about 20 years to reach the reproduction age. The estimated age of death together with the observed decrease in growth suggests that the small individual ML 426 most likely was a female that reached sexual maturity after 10 to 11 years.

While in the past ribs were studied very little, this study shows that dorsal ribs have a reliable growth record containing important information about life history traits. The reason to neglect dorsal ribs for histological studies until now may be because ribs have been historically claimed to be more strongly remodeled with secondary osteons than long bones (Enlow and Brown, 1958), given that even juvenile dinosaur individuals were initially claimed to show high amounts of Haversian bone (Nopcsa, 1933). Also, the strong histological variability within ribs has led to their exclusion from life history studies. In the comparative histology study by Horner et al. (2000), the ribs were found to be the worst skeletal elements for skeletochronology, showing the lowest LAGs count of all sampled bones. This can easily be explained by the fact that the authors sampled the wrong part of the rib. All of these reservations have been addressed by the study of Waskow and Sander (2014), who demonstrated that ribs in general have the most complete growth record at the proximal end of the rib shaft where growth initiated, and that (in contrast to the observations in long bones) the younger distal parts are more strongly remodeled than the older proximal ends. This current study further supports these findings. Even if only the proximal end of the rib shaft was sampled for each taxon, the rib histology observed confirms the findings of Waskow and Sander (2014). Furthermore, this study extends their observations to more taxa and different ontogenetic stages. Irrespective of the taxon, the rib position within the ribcage, the locomotion type, or the ontogenetic stage of the individual, all sampled taxa show a comparable growth record at this position in the rib. None of the samples were completely remodeled or lacked cyclicity. Even though this study validates ribs to be useful for histological studies, the major problem of varying bone apposition rates in ribs during ontogeny (also discussed in Waskow and Sander, 2014) still remains unresolved. Therefore, mass based growth curves based on the measurements between cycles of one individual is not reliable without calculating a correction formula. This work is currently in progress by Waskow and Griebeler (in prep.).

Because ribs were not the focus of histological studies, one could argue that the growth record preserved in ribs might be different from that observed in well-studied long bones. However, Woodward et al. (2014) clearly showed that the growth record in different skeletal elements can be correlated using the example of an extant A. mississippiensis. All samples (even if not including dorsal ribs) show the same number of LAGs with the two closely spaced LAGs at the same relative position. Another more recent research paper by Wiersma et al. (2016) supports a good correlation between the growth record of ribs and long bones. The study of Waskow and Sander (2014) additionally shows that it is possible to correlate the growth record between ribs of a single individual independent of the extremely variable outer morphology.

Another point of criticism might be that there are no studies on modern animals verifying that ribs are equally good indicators for overall skeletal maturity as long bones. Because they are not weight-bearing bones, the possibility that their growth might stop earlier due to different restrictions compared to long bones cannot be excluded. However, even though ribs are not weight-bearing bones, to our knowledge, there is no significant allometric change during ontogeny between long bones and the rib cage in archosaurs. This justifies the assumption that these elements have no significant difference in their respective age at skeletal maturity. Additionally, the definition of skeletal maturity is the point where the animal's growth considerably slows down or stops. Histologically, this is reached at the beginning of an EFS. At this point, the lines of arrested growth are spaced significantly closer to each other without primary osteons, vascularization, etc. in between. For these reasons, these LAGs observed in the rib sections can be identified as the EFS, as it has been done before, for example, by Erickson et al. (2004) and Waskow and Sander (2014). Due to the variation in distance between two given LAGs all around the cortex one could argue that the identification of an EFS is difficult. Nevertheless the EFS in ribs can be identified doubles and independent of the asymmetric deposition of bone material between the single LAGs, that is caused by morphological changes during ontogeny by looking at the area where the bone apposition rate is highest. If the distance between the LAGs here shows the conditions of an EFS its existence can be confirmed. Also, a comparison between the EFS in long bones and ribs of the same individual has been done (at least for sauropods) in the study of Waskow and Sander (2014) with all elements showing an EFS with the same number of LAGs. This additionally reinforces the assumption that skeletal maturity was reached simultaneously in ribs and long bones. However, the aim of this study is only to show the completeness of the growth record in different archosaur taxa to highlight the research potential of these undervalued skeletal elements. More comparative histological work on bone elements of the same individual, similar to the work of Werning and Nesbitt (2016), Wiersma et al. (2017), and Woodward et al. (2014), needs to be done for extinct archosaurs to underline the importance and comparability of ribs with limb bones.

Sexual maturity is a major event in life history, and age estimation at sexual maturity in extinct animals is controversial (Chinsamy-Turan, 2005, Curry, 1999, Erickson, 2005, Erickson et al., 2007, Horner et al., 1999, Klein and Sander, 2007, Klein and Sander, 2008 and Sander, 2000). For ecological reasons of survivorship, Dunham et al. (1989) theorized that large non-avian dinosaurs must have required 5 to 20 years to reach their sexual maturity. For sauropods, Sander (2000) argued that sexual maturity can be detected on the basis of a marked decrease in growth, well before the animal was fully grown. This significant decrease in growth corresponds to the inflection point of the growth curve (Griebeler et al., 2013). At this point, it is thought that the animal starts to allocate more resources for its reproduction, which are then no longer available for growth (Case, 1978a, Case, 1978b and Sander, 2000). It is reasonable to assume that sexual maturity in non-avian theropods and ornithischians may occur well before full adult size (Erickson et al., 2007). This is coincident with the findings of Lee and Werning (2008), who reported on medullary bone (found only in breeding females) that occurs in specimens showing the same decrease in growth. This reduction of growth has been directly observed in large, recent reptiles such as turtles and crocodiles (Lance, 2003). In our sampled taxa from Portugal, as well as in the sauropod data point from the literature (Waskow and Sander, 2014), the inflection point mentioned above is reached at the halfway point of skeletal maturity (Fig. 10 and Fig. 11). We conclude that the decrease in growth rate marks the acquisition of sexual maturity in our study of a heterogeneous group of archosaur taxa including Ornithischia (stegosaurs and ornithopods) as well as Saurischia (theropods and sauropods) and crocodylomorphs.