Patterns of aquatic decay and disarticulation in juvenile Indo-Pacific crocodiles (Crocodylus porosus), and implications for the taphonomic interpretation of fossil crocodyliform material

High levels of skeletal articulation and completeness in fossil crocodyliforms are commonly attributed to rapid burial, with decreasing articulation and completeness thought to result from prolonged decay of soft tissue and the loss of skeletal connectivity during ‘bloat and float’. These interpretations are based largely on patterns of decay in modern mammalian and avian dinosaur carcasses. To address this issue, we assessed the decay of buried and unburied juvenile Crocodylus porosuscarcasses in a controlled freshwater setting. The carcasses progressed through typical vertebrate decay stages (fresh, bloated, active decay, and advanced decay), reaching the final skeletal stage on average 56 days after death. Unburied carcasses commenced floating five days post-mortem during the bloated stage, and one buried carcass only commenced floating 12 days post-mortem. While floating, skeletal elements remained articulated within the still coherent dermis, except for thoracic ribs, ischia and pubic bones. The majority of disarticulation occurred at the sediment-–water interface after the carcasses sank during the advanced decay stage, ~ 36 days post-mortem. Based on these results we conclude that fossil crocodyliform specimens displaying high levels of articulation are not the result of prolonged subaerial and subaqueous decay in a low-energy, aqueous environment. Using extant juvenile C. porosus as a proxy for fossil crocodyliforms, rapid burial in an aquatic setting would have to occur prior to the carcass floating, and would also have to continually negate the positive buoyancy associated with bloating. Rapid burial does not have to be the only avenue to preservation of articulation, as other mechanisms such as physical barriers and internal physiological chemistry could prevent carcasses from floating and subsequently disarticulating upon sinking. The inference that a large proportion of skeletal elements could drift from floating carcasses in a low energy setting with minimal scavenging, thereby causing a loss of completeness, seems unlikely.


Experiment location
The experiment took place between September 2012 and February 2013 during Brisbane''s sub-tropical spring--summer. The experiment was situated in the UQ Palaeontology Lab open-air storage and preparation facility, in St Lucia, Brisbane, Australia (hereafter referred to as the 'study site'). Although the study site was located within a roofed pen which limited the amount of solar radiation affecting carcasses, it allowed for minimal influence of precipitation, which was deemed more likely to affect water quality and, potentially, change decomposition patterns.

Treatments
Three (3) treatments were designed for this experiment, to observe the rate of decay and disarticulation of crocodile carcasses compared with the rate at which carcasses are buried. All carcasses were placed in water filled aquaria after one day post-mortem. For Treatments 1 and 2 (see following description), sand was added to the water filled aquaria to simulate rapid burial in an aquatic environment. The water levels in these aquaria were then maintained by gently topping up the water once a fortnight using hose fitted with a fine shower nozzle so as not to disturb the carcasses, or in the case of buried carcasses, the overlying sand (2-3 cm of water above the sand layer was present for the duration of the decay experiment). These treatments are detailed below:

•
Treatment 1: Carcasses were buried under a 20 cm thick layer of fine-grained sand. This simulated a rapid burial scenario.
• Treatment 2: Carcasses were allowed to 'bloat and float' and decay subaerially. Once all parts of the carcass had sunk, they were gently covered in a 20 cm layer of fine-grained sand.
• Treatment 3: Carcasses were allowed to 'bloat and float' and decay subaerially. Once the carcasses had sunk, they were left undisturbed to decay subaqueously with no burial under fine-grained sand.
These treatments simulated 'best case scenarios' for subaerial and subaqueous decay under rapid burial and low energy environmental conditions in an aquatic setting. No current was induced in any of the aquaria in order to ensure that patterns of disarticulation observed at the end of the experiment were not a result of water movement.

Specimens
Eight male juvenile Indo-Pacific crocodile carcasses (Crocodylus porosus) were sourced from a previous experiment investigating influence of diet on bone growth and the isotopic composition of tissues. During the previous experiment these animals were initially fed either chicken (Gallus gallus domesticus) and kangaroo (Macropus sp.) mince supplemented with ProVet multivitamin supplement, or half chicken/kangaroo mix and half minced mullet (Mugil cephalus).
In the final six months leading up to euthanasia all the animals were fed chicken and kangaroo meats supplemented with beef (Bos primigenius), mullet, and whole dayold chicks, and Wombaroo Reptile Supplement. On the 18th of September 2012, the C. porosus were euthanized via thiopental sodium (1 mL/kg) administered intravenously into the cervical sinus, after which breathing stopped immediately.
Final weight and morphometric measurements were taken within an hour after death. The previous experimental conditions were deemed unlikely to have influenced the subsequent decay sequences, excluding, perhaps, post-mortem sampling that involved an incision through the right dorsal femoral muscles and removal of femoral bone mid-shaft at the caudal depression (the assumed attachment point for the m. adductor femoris in crocodilians; see Brochu, 1992;Klein et al., 2009). As all individuals underwent the same procedure on the right femur, and the left femur remained intact on each individual, the observations drawn from decay patterns in limbs would focus on the left and right forelimbs and left hindlimb. Carcasses were refrigerated overnight, remained unfrozen, and allowed to reach ambient air temperature 2 hoursh prior to the commencement of experiment.
Each crocodile carcass was randomly allocated the following treatment numbers:  Table 1.

Methods
Five glass aquaria with glass partitions (totaling 10 aquaria, eight of which were used for the experiment, each 140 L in capacity, 0.9 m long, 0.4 m wide and 0.4 m high) were placed at the study site. Each aquarium was filled with a 10 cm deep layer of very fine to fine-grained quartz sand. The aquaria were then filled with tap water (approximately 90 L). A YSI Model 85 Handheld water quality probe and a Eutech Ecoscan pH 5/6 meter were used to record water temperature, pH, conductivity, salinity, and dissolved oxygen prior to carcass addition (see Table 2).
Aluminium mesh wire was laid across the top of each aquarium to prohibit access to carcasses by macro-scavengers, but still allowed access for micro-scavengers (predominantly flying insects).

1.
Freshimmediately following death until the carcass shows signs of bloating and begins to float to water''s surface;

2.
Bloatedcarcass floats to water''s surface due to gas production from bacterial activity in the gastrointestinal tract, body cavity swells and becomes distended;

3.
Active decayinsect larvae are actively scavenging carcass, have penetrated skin surface and exposed some internal organs, but no bones visible;

4.
Advanced decayfirst bones are exposed. Eventually internal organs are penetrated, resulting in the release of gases. The carcass sinks. Decay continues underwater from aquatic organisms, with less than one half of the skeleton exposed;

5.
Remainsmajority of flesh has been removed exposing more than one half of the skeleton. Very few scavengers remain.
These observable decay stages for Treatment 2 and Treatment 3 crocodile carcasses (CR2A, − -2B, − -2C, and − -3A, − -3B, − -3C respectively) were monitored via photography and recording of pertinent information (such as location of insect larvae, and portion of carcass above water) approximately every second day until all carcasses sank. Decay could not be observed in Treatment 1 aquaria (CR1A and 1B) for the duration of the experiment due to carcass burial, and in Treatment 2 aquaria (CR2A, − -2B, − -2C) after each carcass sank and was buried. As water clarity in Treatment 2 and 3 aquaria decreased as decay progressed, after each carcass had sunk, water was gently syphoned and bailed by hand from each aquarium in order to photograph the carcass positions. This water was then replaced by fresh tap water for the remainder of the experiment. Although a filtration system may have increased water clarity, it was not emplaced due to the experimental requirements for minimal disturbance, and because its effectiveness could not be guaranteed (a similar filtration methodology was employed by Brand et al. (2003) that resulted in clogging of filtration systems by organic debris, such as scales and soft tissue).
The experiment was halted on 26th February 2013; approximately 2 months after the last unburied carcass had progressed to the final stage of decay (the remains stage). At this point, buried carcasses were exhumed, photographed, and examined for articulation patterns. All water was drained from the remaining aquaria, and carcasses were photographed and examined for degree of articulation. The position of carcasses after they had first sunk and at the end of the experiment were compared to confirm no movement of visible skeletal elements had occurred due to water syphoning: minimal (less than 2 mm) lateral movement of some manual and pedal digits in Treatment 3 carcasses was noted.
The degree of skeletal articulation for each carcass was recorded using a scoring system for fossil marine reptiles created by Beardmore et al. (2012b), who grouped skeletal elements into seven units: head (skull and mandible), neck (cervical vertebrae), trunk (prothoracic, thoracic, lumbar, and sacral vertebrae), tail (caudal vertebrae), ribs (thoracic ribs), left and right forelimbs (including pectoral girdles), and left and right hindlimbs (including hips and each pubis) (refer to schematic interpretation of CR1A in Fig. 8). In this system, the degree of skeletal articulation within each unit is scored from 0 to 4, defined as follows: • Score 0: no articulation in the head, neck, trunk, or tail units; 0--10% articulation for ribs; or, no articulated joints in the forelimbs or hindlimbs; • Score 1: limited articulation in the head, neck, trunk, or tail units; 10--25% articulation for ribs; or, 1 of 4 articulated joints in the forelimbs or hindlimbs; •  Although the temperatures recorded fluctuated due to changes in ambient temperature and time of day recorded, they remained relatively similar between and within treatments (see Fig. 2).

Description of C.Crocodylus porosusdecomposition
The dates for each decay stage listed below indicate the average day on which each stage began, plus or minus the standard deviation, indicative of the variability between individuals. Table 3 shows the start date and length of time each carcass spent in each decay stage.

Fresh stage: day 1 ± 0
The carcasses (CR1A and CR1B) were placed in the aquaria dorsal side up, and immediately sunk to the bottom of the aquaria, without any assistance, in this same orientation prior to burial.

Bloated stage: exact date unknown
CR1A remained buried for the duration of this stage; therefore no observations could be made, although presumably it underwent bloat underground.
However, the same process of internal gas production and resulting buoyancy within the body cavity of CR1B overcame the downward pressure of the sand it was buried under, which resulted in it rising and floating to the surface ventral side up on day 12.
CR1B was re-buried on the day 13 of the experiment, with excess water drained from both Treatment 1 aquaria for two reasons: (1) so that both buried carcasses underwent the same experimental variables, and (2) to eliminate the possibility of CR1A also floating.

Active decay stage/advanced decay stage: exact dates unknown
Carcasses remained buried for the duration of these stages; therefore observations could not be made.

Remains stage: exact date unknown
The buried carcasses were unearthed on day 161 (26th of February 2013).
Both carcasses had already reached the remains stage, with extensive soft tissue decay leaving behind only the mummified dermis and cuticle, and some adipocere alongside the caudal vertebrae from the saponification of fatty tissue. Skeletal elements, including osteoderms, were all in their in-vivo positions. The exact position of skeletal elements is described in Section 3.4.1.

Fresh stage: day 1 ± 0
The carcasses were placed in the aquaria facing dorsal side up, and immediately sunk to the bottom of the aquaria, without any assistance, in this same orientation. Slight variations in limb orientations occurred for each individual; however, in most instances the limbs were protracted and abducted, and flexed at the elbow/knee, with the medial surface facing up.

Bloated stage: day 3.8 ± 0.75
After an average of 3 days the carcasses bloated, and 24-48 hoursh later began to rise through the water column rotating on their long axis to float ventral side up (excluding CR2C, which rotated on its long axis to float left lateral side up).
While floating, the head and neck of each carcass was initially at the water''s surface then began to flex dorsally such that they remained below the water line, whereas the medial (rarely cranial) surface of each limb and the ventral surface of the trunk was exposed above the water line --excluding CR2C, which only had the left-most ventral and lateral surfaces of the head and trunk and the lateral surface of each left limb exposed, with the head eventually flexing left laterally. Initially, the terminal half of the tail remained below the water line for all carcasses, progressively flexing more dorsally as decay progressed the entire length of the tail. Between day 5 and day 7, the cloaca of each carcass was visited by flying insects. The end of the bloat stage in this experiment did not necessarily correlate with each carcass sinking, but rather the onset of scavenging of portions of the body that were subaerially exposed, which in turn lead to the onset of active decay stage. Gas bubbles in the form of white foam were observed around the mouth and cloaca, and near the surgical wound in the right hindlimb (see Fig. 3) from the escape of endogenously produced gases, also marking the end of the bloat stage. (during bloated stage). Gas bubbles in the form of white foam can be seen in the water above the right hindlimb incision point. Scale bar represents 2 cm.

Active decay stage: day 14 ± 4.15
Visible larval insect scavenging of exposed soft tissue began on days 11-13, excluding for CR2A and CR3C where larval scavenging only began after day 17 and day 21, respectively. Larvae were smooth and pale, ranged in size between 8and 12 mm, and using the identification key outlined in O''Flynn and Moorhouse

Advanced decay stage: day 24 ± 4.52
As decay continued, skeletal elements (primarily thoracic ribs, the pubic bones and ischia) were exposed and became disarticulated at their respective joints.
Exposure and disarticulation resulted not only from the direct removal (consumption) of muscle, tendons and surrounding fascia by larvae, but also from the 'tugging' feeding action employed by many of the larvae, pulling soft tissue back and forth at least 0.5 to 1 cm (seeFig. 4). Between days 19 and 31, carcasses started to submerge and then slowly sink. Insect larvae evacuated the soft tissue as it sunk. Due to the turbidity of the water, continued sinking and eventual contact with the substrate could not be viewed first hand. As carcasses retained skeletal articulation while floating (excluding disarticulation of some thoracic ribs, the pubic bones and ischia as described above), and disarticulation was observed after the carcass had sunk and water was gently drained from the aquaria (so the degree of articulation could be documented), we deduce that the majority of disarticulation occurred when the carcasses contacted the substrate. Due to the lack of disturbance in the aquaria and the position of the carcasses during decay and while sinking (with the head and whole of tail submerged, creating an inverted U-shape profile), the mostly likely source of this disarticulation was the pressure placed on skeletal elements and decayed joints at the sediment--water interface as the body succumbed to gravitational forces, 'buckled' and pressed on other body parts, causing bones to move out of life position. Treatment 2 carcasses were buried during this stage, with a top layer of water added then drained 13 days after burial to replicate the Treatment 1 variables.

Patterns of decay and disarticulation in floating carcasses
All the crocodile carcasses in Treatments 2 and 3 floated after reaching the bloated stage. One crocodile carcass in Treatment 1, CR1B, also floated (see previous description in 'Treatment 1 --bloat'). On average, carcasses spent 32 days (± standard deviation of 12.79 days) floating: the average pattern and timing is outlined below.
• Three days post-mortem, the carcass showed visible signs of bloat (abdominal swelling).
• After the carcass began to bloat, it took 24-48 hoursh for it to float and rise 20 cm to the surface of the water. This marked the beginning of'bloat and float'.

•
The carcass floated for approximately 19 days, initially with the majority having their ventral surfaces subaerially exposed (from the tip of the mandibular rostrum to the terminus of the tail, the medial surfaces of the limbs, and palmar/plantar surfaces of the manus/pedes respectively). Sinking of the tail progressed from the tip cranially, with the terminal-most portion being the first to fall below the surface of the water. Similarly, sinking of the head progressed from the tip of the rostrum caudally. White foam was observed around the mouth, cloaca, and pre-experiment right hindlimb surgical wound (see Fig. 3). All carcasses stayed articulated during this initial float period, and during both the bloat and the active decay stages.

•
Over the next 12 days, the neck and the entire length of the tail submerged, with the head, limbs, and terminal half of the tail beginning to sink. The pubic bones, gastralia, and some thoracic ribs were exposed and disarticulated, coinciding with the start of advanced decay. As decay progressed, the trunk and tail base remained at the water''s surface while the head, neck, forelimbs, hind limbs, and the remainder of the tail continued to become more submerged, without disarticulating. This resulted in an inverted U shape profile for all carcasses (see Fig. 7).

•
The end of 'bloat and float' occurred when the carcass sunk over the next 11day period and settled on the bottom of the aquarium. It is important to note that the end of the bloated stage did not coincide with the cessation of floating: carcasses continued to float while progressing through the active decay stage and part of advanced decay stage.

End of experiment articulation
The results are described below for each treatment, and intra-and inter-unit articulation stage scores are summarized in Tables 4 and 5  a The 'ribs' skeletal unit defines articulation between ribs and the thoracic vertebrae, and therefore is technically a measure of inter-unit articulation, not intra-unit articulation. However, as the 'ribs' unit was defined by Beardmore et al. (2012b), we have opted to keep it in the intra-unit articulation category. Table 5 Inter-unit articulation scores between skeletal units, where 'F' indicates full inter-unit articulation -in contact, no breaks (rotations or spaces), 'P' indicates partial inter-unit articulation -skeletal elements in contact or near-to, but rotated out of life position, and 'D' indicates inter-unit disarticulation -skeletal elements are not in contact and not in-vivo.
Treatment 1 carcasses showed highest inter-articulation scores due to burial during the fresh stage.
Treatment 2 and 3 carcasses showed lower inter-unit articulation scores due to soft tissue decay in water and gravitational forces upon sinking forcing skeletal elements out of in-vivo positions.
Treatment 2 carcasses overall scored higher than Treatment 3 carcasses, as they were buried during the advanced decay stage with some soft tissue still attached. Treatment 1  Treatment 2  Treatment 3   CR1A CR1B   CR2A  CR2B  CR2C  CR3A CR3B  CR3C Head --neck

Treatment 1 --buried one day post mortem
As seen in Fig  Scale bar represents 10 cm.

Treatment 2 --subaerial decay prior to burial upon sinking
In contrast to Treatment 1, skeletal elements of Treatments 2 carcasses

Treatment 3 --subaerial and subaqueous decay, no burial
As with Treatment 2 skeletons, and in contrast to Treatment 1 skeletons, skeletal elements of Treatments Treatment 3 carcasses (Fig. 10) were disarticulated or partially disarticulated. Individual variation in intra-and inter-unit articulation was observed, with carcasses on average scoring lower for intra-unit articulation than Treatment 2 carcasses. The skulls of CR3B and CR3C remained articulated with their mandible, both laying ventral side up. The exception was CR3A, with the mandible disarticulating from the skull, and both elements lying dorsal side up (scoring '0' for intra-unit articulation). In all three carcasses, the cervical vertebrae disarticulated from the base of the skull. The majority of prothoracic, thoracic, lumbar, and sacral vertebrate in Treatment 3 carcasses were rotated out of, but remained near to, their in-vivo positions (each carcass scoring '2' for trunk intra-unit articulation). The intra-unit articulation of caudal vertebrae varied from disarticulated to moderately articulated among all three carcasses (with intra-unit articulation scores of '0', '1', and '2' for CR3A, CR3B, and CR3C respectively). Among all three carcasses, the only full inter-unit articulation maintained was between the trunk and tail for CR3C.
Partial inter-unit articulation was noted between the neck and trunk in all three carcasses, the trunk and tail in CR3A and CR3B, and the left hindlimb and trunk of CR3A. Disarticulation was observed between all other units in all three carcasses.
Treatment 2 and 3 carcasses were most similar in intra-unit articulation for ribs (scoring '0'), and inter-unit articulation for left and right forelimb --trunk (with disarticulation between all excluding the left hindlimb of CR3A) (see Tables 4 and 5).
Similar to Treatment 2, Treatment 3 carcass forelimbs and hindlimbs disarticulated at the shoulder and hip joint, respectively. On average, the manus, pedes and caudal vertebrae of Treatment 3 carcasses had relatively lower intra-unit articulation scores than in Treatment 2 carcasses. No soft tissue remained on the manus and pedes; carpals/tarsals, metacarpals/metatarsals, and digits were disarticulated, rotated, and moved out of in-vivo position. CR3C retained some intra-unit articulation in the forelimbs and hindlimbs (scoring '2' for both forelimbs, and '1' for both hindlimbs), whereas CR3A and CR3C suffered greater degrees of disarticulation (with scores of '1' for the hindlimbs of CR3A, and '0' for the forelimbs of CR3A and all limbs in CR3C). For those limbs that retained some articulation, CR3A showed flexion at the right hindlimb tarsus, and extension at the left hindlimb knee, and CR3C showed flexion of the right forelimb at the elbow, flexion of the right hindlimb at the knee and tarsus, and extension of the left hindlimb knee and tarsus. Limb articulation patterns on average were mirrored between the left and right sides of each individual; when the left forelimb or hindlimb was articulated, so was the right forelimb or hindlimb.

Discussion
This experiment has demonstrated that juvenile Crocodylus porosuscarcasses decomposing in undisturbed fresh water progress through the five recognizable stages of vertebrate decay (fresh, bloated, active decay, advanced decay and remains), reaching the final stage on average 55 days post-mortem. Based on the results of this experiment, we can conclude that preservation of articulated skeletons of extant C. porosus is more likely to result from 'rapid burial', that is, burial during the fresh stage, and less likely to result from decay in a low-energy aqueous environment or burial during the advanced decay stage. The degree of articulation that we observed in different treatments related to the amount of soft tissue still present and holding skeletal elements in place, which in turn was related to the decay stage that the carcass was in. Floating was found to be integral to the disarticulation process: not only did it provide terrestrial micro-scavengers with access to soft tissue, but also enabled endogenous and exogenous soft tissue decay to progress to a point where, upon sinking, gravitational forces forced bones out of their in-vivo position as the parts of the carcass settled on the substrate. It is pertinent to note that the timeframes discussed herein are based on a 'best case' scenario for juvenile C. during the bloated stage. Floating carcasses did not start to sink immediately after gases started to escape via the mouth, cloaca, and right hindlimb surgical wound, or even after larval insects penetrated the intestinal and stomach walls during the active decay and advanced decay stages. Although bloat might be responsible for the carcass initially floating, the end of the bloated stage did not coincide with the end of the float time. Anderson and Hobischak (2004)similarly noted that pig carcasses in aquatic settings could float for weeks due to trapped pockets of intestinal gas, which was not to be confused with 'true' bloat where the integrity of the stomach and intestinal walls have not been disrupted, and gases have not been released through natural orifices. We noted that the white foam forming around carcass orifices and the right hindlimb surgical wounds was indicative of the release of some endogenous gas, but conclude that some gases must have remained trapped in the intestinal tract which allowed the C. porosus carcasses to continue floating after the bloat stage ended. The degree of saturation of soft tissue may also affect flotation time: as water enters the carcass either through intact epidermal and dermal layers that are permeable to water or disrupted epidermal and dermal layers, it will fill interstitial spaces in bone and displace endogenous gases in soft tissue. This will increase the overall density of the carcass, and if its density becomes greater than that of water, the carcass will sink.
While the C.Crocodylus porosus carcasses floated, the majority of skeletal elements stayed articulated or partially articulated within the still intact epidermis and dermis. This phenomenon has been observed in small lizards (Brand et al., 2003b), and large marine mammals (Schafer and Craig, 1972), and has been attributed to the presence of skin with relatively high durability and low permeability (Brand et al., 2003b;Cambra-Moo et al., 2008). Brand et al. (2003, pg 32) postulated that this could lead to misinterpretation of the fossil record, where articulation is maintained upon sinking and burial, and mistakenly attributed to rapid burial while the carcass was in the fresh stage. However, we observed that after were on average mirrored between the left and right side within each carcass. It would therefore also seem reasonable to assume that a fossil with disparate articulation between paired limbs had suffered non-necrolytic pre-, peri-, or postmortem soft tissue trauma.
Disarticulation of vertebrate carcasses in low-energy freshwater settings as seen in our experiment has also been observed in other experiments: both for multiple goldfish carcasses (Hellawell and Orr, 2012) and a single iguanid (Oplurus cuvieri) carcass (Richter and Wuttke, 2012). Hellawell and Orr (2012) found that skeletal disarticulation of goldfish occurred in shallow fresh water in petri dishes in the absence of any outside disturbance. They concluded that decay in a low-energy aqueous environment did not account for the high levels of articulation seen in fossil fish of the Green River Formation (Hellawell and Orr, 2012). InRichter and Wuttke's (2012) freshwater decay experiment of one iguanid carcass, they conclude that while some disarticulation may have resulted from the withdrawal of the carcass multiple times for X-radiographs over a 2-month period, the majority of disarticulation occurred in the absence of outside disturbance. We, too, similarly conclude that in the absence of any outside disturbance, disarticulation of skeletal elements in an aqueous setting can occur. Richter and Wuttke (2012) also observed that the iguanid fore-and hindlimbs separated from the axial skeleton but remained articulated within each limb: as per the terminology used in our study, this would be classified as full intra-unit articulation and inter-unit disarticulation. We observed a similar pattern of high intra-unit articulation scores for the fore--and hindlimbs of our  Weigelt, 1989, Plates 27 and 30), (2) the carcass might pass through the bloat period in a terrestrial setting, then may either be covered with water and sediment in-situ, or be transported into a body of water and buried (as has been proposed for the holotype Susisuchus anatoceps by Salisbury et al., 2003); (3) bloat not occurring due to species specific anatomical and physiological characteristics (Schoener and Schoener, 1984;Brand et al., 2003b;Reisdorf and Wuttke, 2012)--but we are unsure as to what these might be or if there are any examples of taxa in which floating is not known to occur--or (4) low water temperatures reducing gas production (Elder and Smith, 1984;Elder et al., 1988). Any of these scenarios that do not invoke rapid burial as the cause of articulated preservation must inhibit the access of aerial or aqueous macro-scavenger to the carcass. Scenario (1) could occur in tandem with 'rapid burial', extending the amount of time needed to sufficiently weigh a carcass down with sediment. Skeletal elements in a Scenario (2) carcass could be buried in a flood plain or riverbank under a relatively thin amount of sediment, initially in an aquatic setting, but with water levels receding and reducing water in interstitial spaces in the sand during the bloat stage. If a carcass underwent bloat unburied in a terrestrial setting to be later buried in an aquatic setting, the carcass would retain some intact soft tissue to avoid disarticulation during water and sediment influx in-situ, or transport to the final burial site. Scenario (3) would not require 'rapid burial', as the carcass would not float and disarticulate on sinking; however, soft tissue trauma would increase the likelihood of disarticulation occurring. In relation to Scenario (4), Elder et al. (1988) and Elder and Smith (1984) propose that water temperature controls the production of gases internally and the resulting buoyancy of a decaying carcass, with cooler temperatures (below 16 °C) limiting flotation of fish carcasses. In our experiment, carcasses began to float on days 5 to 7 (22nd--24th of September) when water temperature ranged between 21.2 °C and 23.6 °C with an average temperature of 22 °C (see Fig. 2), so it appears that the minimum fresh water temperature which still allow for juvenile C. porosus carcasses to float is at least 22 °C. It is feasible that lower water temperatures could have the same flotation inhibition effect on crocodyliform carcasses.
Using Crocodylus porosus decay patterns as an analogue for extinct crocodyliform decay patterns, it can be inferred that preservation of high intra-and inter-unit articulation in fossil crocodyliforms preserved in fresh water sediments was a result of the inhibition of floating via burial during the fresh stage, trapping under physical barriers, removal from water, or enough trauma to prevent bloat (under our experimental conditions, for juvenile C. porosus this would need to occur within the first week post mortem). High degrees of cranial and post-cranial articulation including vertical preservation of manus and pedes, similar to that seen in Treatment 1 carcasses, would be indicative of burial during the fresh stage with soft tissue and burial medium holding skeletal elements in-vivo. A lack of pelvic girdle elements in fossil crocodyliforms would be expected for carcasses of juvenile animals (with only partially fused joints between the ilia and sacral ribs) that floated during decay and that were buried only after sinking (seeSalisbury et al., 2003).
Those that show vertical preservation of disarticulated portions of the axial or appendicular skeleton, especially the manus and pedes (resembling that of CR2A), were likely to have been buried during advanced decay upon sinking (for juvenile C. porosus, approximately three to eight weeks post mortem), with higher degree of articulation in the manus and pedes resulting from burial with manual and pedal soft tissue still intact. High degrees of articulation between caudal vertebrae could result from the presence of adipocere prior to burial, given that many species of extant crocodylians typically have large adipose tissue deposits medial to both the epaxial and hypaxial muscle masses in the base of the tail (Gadow, 1882;Frey, 1988a). Those that have a greater degree of disarticulation of the forelimbs and hind limbs, especially skeletal elements of the manus, pedes, and tail, are most likely indicative of carcasses having reached the remains decay stage (for juvenile C. porosus, greater than approximately eight weeks post-mortem) prior to burial, where extensive subaqueous soft tissue decay has occurred. Disparate articulation in paired limbs would therefore be indicative of pre-, peri-, or post-mortem soft tissue trauma excluding necrolysis and micro-scavenging. In C. porosus, as occurs in all other extant crocodylians, the osteoderms that form part of the dorsal shield (paravertebral shield + accessory osteoderms) sit within the corium, with sagitally adjoining osteoderms bound together via interosteoderm ligaments (Schmidt, 1914;Frey, 1988a). In many taxa, laterally adjoining dorsal osteoderms are also joined to each other by serrated sutures (Frey, 1988a;Salisbury and Frey, 2001); but we note that such joints to not occur in C. porosus, which has very reduced dorsal osteoderms relative to other extant taxa. These interosteodermal joints (interosteodermal ligaments and interosteodermal sutures) could result in groups of articulated osteoderms detaching during decay and becoming preserving as intact sections.
Although osteoderms were not observed in their in-vivo positions for the juvenile C.
porosus carcasses at the end of the experiment, the additional tight tendinous integration of the paravertebral shield with the vertebral column of the trunk and tail base via the epaxial musculature (see Frey, 1988a,b;Salisbury and Frey, 2001) could result in the additional maintenance of anatomical coherence of these parts of the trunk skeleton until after the carcass has sunk; this is not true for the cervical and more terminal caudal vertebrae, which are not tightly as integrated with the dermis and epaxial musculature (see Frey, 1988a,b;Salisbury and Frey, 2001). In our experiment, the vertebral column of each carcass remained articulated during floating, with only two specimens suffering vertebral disarticulation upon sinking (CR2A and CR2B); even so, the vertebral column separated into articulated sections, demonstrating some remaining anatomical coherence between sections of the vertebral column and the epaxial musculature.
The need to inhibit floating in order to maintain skeletal articulation must also be taken into account when considering the taphonomic histories of vertebrate fossils associated with Lagerstätten. Articulation of skeletal elements and preservation of soft tissues in Lagerstätten fossils are thought to have resulted from carcasses in good condition decaying in low energy aquatic environments that are anoxic, acidic or hypersaline, all of which limit opportunities for scavenging and destruction of soft tissue (Allison, 1988;Allison and Briggs, 1991;Taylor, 1995;Behrensmeyer et al., 2000). Although rapid or catastrophic burial can form Lagerstätten, other 'stagnation' mechanisms can also result in the formation of Lagerstätten, such as (1) sinking into a soupy substrate (Smith and Wuttke, 2012), (2) microbial mat formation (Hellawell and Orr, 2012;Schwermann et al., 2012;Iniesto et al., 2013), and (3) adipocere formation (O''Brien and Kuehner, 2007;Ubelaker and Zarenko, 2011;Schwermann et al., 2012). However, these stagnation scenarios do not account for the likelihood that even in low energy conditions, vertebrate carcasses may float and then disarticulate upon sinking. And even after sinking, prolonged subaqueous decay prior to burial might still allow for partial or total disarticulation. For example, in our experiment, adipocere formed along the caudal vertebrae of the juvenileCrocodylus porosus, but those left to decay subaqueously (Treatment 3) still suffered disarticulation of caudal elements. In future taphonomic analyses, assumptions that involve any of these four mechanisms must also take into account processes that stop carcasses floating, or that reduce flotation time such that the carcass sinks immediately at the end of the bloat stage, or in the early stages of active decay, before extensive soft tissue decay and skeletal disarticulation can occur.