Validating osteological correlates for the hepatic piston in the American alligator (Alligator mississippiensis)

Anatomical modeling and Microcomputed Tomography (µCT)

To illustrate the diaphragmaticus muscle, a µCT scan of one sub-adult 3.4 kg A. mississippiensis (identified as Hamilcar Barca for this project) was obtained from a privately owned facility (Scales and Tales Utah) for clinical purposes (unrelated to this study) and donated for this project. The entire alligator was scanned in a prone position, during a natural apnea, at a resolution of two mm slices (80 kVp, 60 mA). The skeleton and lungs of this individual were segmented into a surface model using Avizo 7.1 (Thermo Fisher Scientific) and followed the methods of Lawson et al. (2021) and Schachner et al. (2023). The diaphragmaticus muscle and liver were then illustrated on to the model using Adobe Photoshop (Figs. 1A and 1B).

Ultrasound data collection

A total of 170 individual 16-second ultrasound videos were collected from three juvenile, two sub-adult, and two adult A. mississippiensis (n = 7) (see Table 1 for the names and snout-vent lengths [SVL] of the specimens). Individuals were scanned at Louisiana State University Health Sciences Center (LSUHSC) in New Orleans and at the Rockefeller Wildlife Refuge in Grand Chenier, Louisiana. Two individuals were part of the collection of Dr. Raul Diaz (then Southeastern Louisiana University) and five individuals were collected by Dr. Ruth Elsey and staff (Rockefeller Wildlife Refuge). All research conducted on living individuals was approved by the LSUHSC Institutional Animal Care and Use Committee (IACUC #6341).

The animals were not sedated and were scanned in an at-rest, prone, stationary position. While at rest, alligators can enter prolonged periods of apnea. Recordings were taken both during natural breathing, when the individual was compliant and breathed deeply on its own, and during induced breathing. Induced breathing was achieved by giving individuals a 5% CO₂, balance N₂ gas following the procedures outlined in Douse & Mitchell (1992) and Uriona & Farmer (2006). Induced breathing reduced the duration of the apneas and allowed for more breaths to be recorded within the video duration. A recording time of 16 s was selected to allow adequate time for the animal to take several breaths. Gas was pumped through a polyethylene tube to a bag that was fitted over the nares and mouth of each alligator. Individuals did not breathe the 5% CO₂, balance N₂ for longer than three minutes at a time. All measurements were taken with the GE LOGIQe Ultrasound using a 12-L probe and penetration depths ranged from two cm for the juveniles to seven cm for the adults. The measurements were collected at four locations (axillary, mid-thoracic, hepatic, and post-hepatic) on the left lateral side of the animal in the coronal plane using consistent anatomical landmarks (Fig. 1C). Individual animals were named for ancient military figures from Carthage, the Roman Republic, and one fictional British MI6 agent for identification.

Ultrasound measures

Displacement was quantified first by visually establishing the most cranial and most caudal positions of the pleural-hepatic margin during the breath cycle in each video. Frames corresponding with the extremes of the breath cycle were then exported into Adobe Illustrator (Adobe Inc., San Jose, CA, USA) and superimposed. The distance between the most cranial and most caudal positions of the cranial hepatic margin was measured by drawing a line segment between them, the length of which was recorded and scaled to the image size using the scale bars in the ultrasound videos. To ensure consistency, this process was repeated three times and averaged for each video. Results were reported both as total length in centimeters and as a percentage of SVL. This technique, while accurate, could be used only at the hepatic margin position where clear anatomical landmarks could be identified visually.

PIVlab

To assess movement at other locations where visual assessment and direct measurement were not possible, we utilized a modified version of PIVlab (Thielicke & Sonntag, 2021), an open-source program for digital particle image velocimetry (DPIV). DPIV is an analysis technique that can be used to map the spatially resolved velocity or displacement of particle images. DPIV is increasingly used in research areas other than fluid dynamics such as measuring the displacement of diverse textures in digital image data (Thielicke & Sonntag, 2021). In the modified version of PIVlab, one or multiple user-generated region of interest (ROI) rectangles were created of any size in any position on the video. Once the ROI was selected, pre-processing image enhancements, Contrast Limited Adaptive Histogram Equalization–CLAHE (Pizer et al., 1987) and background subtraction (Thielicke & Sonntag, 2021) were performed to increase the signal-to-noise ratio. Cross-correlation was performed in the spatial domain (direct cross-correlation), and as repeated correlation, where the correlation matrices of slightly shifted evaluation windows were multiplied to maximize the signal-to-noise ratio. The displacement of the ROI was determined frame by frame and the displacement was used to offset the ROI in the following frame, enabling the user to lock on the selected pixel pattern in the ultrasound images over time.

Of 170 videos recorded, we were able to measure displacement in 53 videos using PIVlab. The remaining videos could not be processed due to issues with the ultrasound (e.g., dark shadows deep to bone that interfered with the program’s ability to track movement) or due to animal movement. It was also evident that tracking boxes frequently lagged behind the tissue in the target window in successful videos, sometimes dramatically. Although these videos demonstrated tissue movement, they underestimated the amount of movement.

Measuring displacement

The main aim of this work was to evaluate if the lung pleura was moving across the smooth interior ceiling of the thoracolumbar cavity during ventilation with the hepatic piston, and if so, to establish associated osteological correlates for this ventilatory mechanism in A. mississippiensis. The lung pleura sliding against the smooth ceiling of the thoracolumbar cavity was evident in all videos where deep breaths occurred, with maximum liver displacement ranging from 4.24% to 15.78% SVL. It is likely this study does not capture maximum inspiratory capacity nor the full range of displacement possible under varying physiological conditions. A. mississippiensis likely breathes more deeply during or after exercise or while diving (Farmer & Carrier, 2000a; Munns et al., 2005; Uriona & Farmer, 2006; Uriona & Farmer, 2008). Additionally, four of seven individuals were wild caught and thus were not fasted before analysis: Eating a large meal (>15% of body mass) can decrease the animal’s vital capacity by up to 25% (Uriona & Farmer, 2006), which would potentially decrease the maximum displacement measured in these wild individuals. However, it is unlikely these wild-caught individuals ate a large meal at the time of capture, they were caught during the fall season when feeding frequency decreases substantially (Lance, 2003). Other external physiological variables that may have an effect on inspiratory capacity include ambient temperature (Munns et al., 2012), if the animal is on land or in water (Gans & Clark, 1976; Uriona & Farmer, 2006), and if the animal is standing (Uriona & Farmer, 2006).

There are also many other internal physiological variables that can lead to substantial changes in total displacement via the hepatic piston. Crocodilians can switch between costal/abdominal dominant or hepatic dominant ventilation, greatly varying the amount the hepatic piston contributes to inspiration (Brainerd & Owerkowicz, 2006; Brocklehurst et al., 2017; Claessens, 2009; Farmer & Carrier, 2000b; Gans & Clark, 1976; Munns et al., 2012). Further, crocodilians can enter long periods of apnea (Gans & Clark, 1976; Uriona & Farmer, 2006), and can increase their total tidal volume (rather than breathing frequency) during periods of hypercapnia (Claessens, 2009; Wang & Warburton, 1995). Due to the propensity for apnea, we used 5% CO₂, balance N₂ gas to induce deep breaths (Douse & Mitchell, 1992; Uriona & Farmer, 2006) but did not find a statistical relationship between the amount of displacement and the use of 5% CO₂, balance N₂ gas (Fig. 5A). If breathing the 5% CO₂, balance N₂ gas was putting the individuals in a state of hypoxia we would expect to see an increase in displacement when the individuals breathed the 5% CO₂, balance N₂ gas. Instead, we observed no measurable difference when animals breathed the gas or did not. Nevertheless, the ability for crocodilians to change the dominant muscles of inspiration and change the frequency and intensity of apnea suggests that the displacement measured does not represent the displacement possible under various internal conditions.

Crocodilians can also change the tension of the abdominal wall and gastralia (Farmer & Carrier, 2000b; Gans & Clark, 1976), have a rotating pubis that changes the potential caudal extent of visceral displacement (Claessens, 2004; Claessens, 2009; Farmer & Carrier, 2000b), and can maintain or change the pressure gradient on either side of the post-pulmonary septum (PPS) (Cramberg, Greer & Young, 2022). Klein & Owerkowicz (2006) discuss that the “diaphragmaticus-endowed PPS” can assist with cranial visceral displacement to avoid lung collapse. It should be noted that Klein & Owerkowicz (2006) often reference the crocodilian diaphragmaticus and PPS functionally together, but they are distinct anatomical structures (Cramberg, Greer & Young, 2022). Despite the known ability of crocodilians to drastically change how much the hepatic piston contributes to inspiration, our measures of displacement are consistent with observations from Schachner et al. (2022) and Claessens (2009).

µCT is a useful method for evaluating the potential of lung movement and of inspiratory capacity in crocodilians. Schachner et al. (2022) imaged multiple specimens of live and deceased Cuvier’s dwarf caiman (Paleosuchus palpebrosus) in different states: post-prandial, fasting, open to atmosphere, and at total lung capacity. Post-prandial individuals had cranially displaced lungs up to five vertebral segments compared to a fasted state. Individuals of deceased hatchling P. palpebrosus that had fully inflated lungs had a caudal displacement of 1.5 vertebral lengths relative to individuals scanned that were open to atmosphere. This demonstrates the large variability in crocodilian lung volume and position and is consistent with the results found here for A. mississippiensis. In the ultrasound videos at the pleural hepatic margin the pleura and liver are routinely sliding across one to three vertebral lengths (see supplementary materials on Data Dryad for all ultrasound videos). This is also consistent with Claessens (2009), who reports, using cineradiographic techniques, an average displacement of the cranial surface of the liver of 1.4 vertebral lengths in juvenile individuals of A. mississippiensis.

We did not find a correlation between size and displacement (Fig. 5B). Gans & Clark (1976) report, using EMG data in Caiman crocodilus, that breath frequency decreases in larger animals. This study includes 20 individuals that range in size from 400 g to 7.5 kg. Claessens (2009) in a cineradiographic study of five juvenile A. mississippiensis, ranging from 55–115 cm total length, reported liver displacement is larger in larger individuals. Munns et al. (2012) reported personal observations that the diaphragmaticus muscle in juvenile crocodilians are thin and translucent and thicker and more well-developed in adults. We confirm this observation (pers. obs., CGP and ERS) in dissections of A. mississippiensis. The limitations of correlating size and displacement include that Claessens (2009) and Gans & Clark (1976) studies contain no adults. Further, all studies, including this one, are limited in that crocodilians are known to vary their reliance on the hepatic piston during ventilation, therefore capturing the total variation is difficult for any one study. This is not a physiologic study on all possible variation in the displacement via the hepatic piston. To vigorously study the relationship between size and hepatic piston use we suggest a mixed methodology that includes XROMM and electromyography on many individuals ranging in size from hatchling to adults in numerous physiologic conditions.

PIVLAB

PIVlab was useful in providing clear respiration curves (Fig. 4). Our measurements of displacement in PIVlab underestimated displacement compared to the manual measurements at the hepatic margin. This is not the result of an inherent flaw with PIVlab but is related to the nature of ultrasound data. Ultrasound data from the thoracic region commonly have shadows from ribs that interrupt tracking in PIVlab and have lower contrast between the target and background ultrasound scatter. The human eye and manual tracking can follow movement that an automated program cannot. Also, PIVlab functions by maintaining the tracking box in every frame of the video. A shadow that may darken halfway through the video is enough for PIVlab to lose its target: The PIVlab algorithm tracks the most statistically significant pixel pattern. If a shadow dominates the tracked region, then the algorithm will tend to track the shadow. The manual procedure needs a clear frame of only the most cranial and most caudal aspects of the breath to measure the displacement. However, this initial work and prior studies that have used DPIV with ultrasound (Farron, Varghese & Thelen, 2009; Korstanje et al., 2009; Korstanje et al., 2010; Korstanje et al., 2012; Loram, Maganaris & Lakie, 2006) demonstrate that there is promise in advancing PIVlab technology to work more consistently with ultrasound data.

Heterogeneity of lung movements

When describing the lung of the Nile crocodile (Crocodylus niloticus), Perry (1988) states that it will “tend to fuse with the parietal pleural surface” and therefore the lung tissue stretches instead of sliding during inspiration via the hepatic piston (particularly the cranial half). There are two main points in Perry’s (1988) work that are of interest: (1) the C. niloticus lung is more strongly attached to the parietal pleura cranially than caudally, and as a consequence, (2) in the caudal half of the lung, the pleural tissue is stretching instead of sliding. It is possible that the specimens evaluated by Perry with fusion between the visceral and parietal pleura were pathologic, but this needs to be investigated further in other adult specimens of C. niloticus. We qualitatively demonstrate in this study that the amount of pleural displacement increases moving from cranially to caudally in all individuals via the ultrasound videos in A. mississippiensis. There is minimal displacement in the axillary position, where the ribs are forked, but the visceral pleural tissues do not appear to be fused with the parietal pleural. The largest amount of pleural displacement is in the hepatic position where the thoracic vertebrae are wide and the ribs are flat, creating a smooth surface (on the interior thoracic ceiling) on which the lung tissue can either expand or slide freely. However, studies on the potential heterogeneous mechanical properties of the lung pleura may reveal more information related to the structural changes in the lung tissue that Schachner et al. (2022) and Perry (1988) observed and how this varies across Crocodylia. Schachner, Hutchinson & Farmer (2013) did not find any fusion between the visceral and parietal pleural described for C. niloticus by Perry (1988) and there may be intraspecific variation or pathologies associated with fusion observed in some of these animals. It is possible that adhesions occur variably and impact hepatic piston function. Additional investigation into differential movements of the alligator pulmonary tissues will also reveal if there are any functional relationships between differential regional mobility/immobility of the pulmonary pleura and the architecture of the bronchial tree.

Evolution of the hepatic piston

Of all the various morphologically and ecologically diverse pseudosuchian groups that thrived during the Late Triassic period, only small, terrestrial crocodylomorph pseudosuchians survived the Triassic/Jurassic (Tr-J) mass extinction ∼200 million years ago (Butler et al., 2011; Nesbitt, 2011; Wilberg, Turner & Brochu, 2019). These small gracile-limbed reptiles were the terrestrial ancestor to all crown-group semi-aquatic extant crocodilians (Nesbitt, 2011). Following the Tr-J extinction event, Crocodylomorpha quickly diversified and these animals were found in a wide array of niches by the Middle Cretaceous, ranging from fully marine to terrestrial localities, and potentially even arboreal habitats (Mannion et al., 2015; Wilberg, Turner & Brochu, 2019). It has been hypothesized that crocodylomorph archosaurs transitioned from terrestrial to semi-aquatic or to marine environments multiple times (Lessner et al., 2023; Schwab et al., 2020; Wilberg, Turner & Brochu, 2019). Given the hypothesis that the hepatic piston evolved in association with diving by controlling pitch and roll (Uriona & Farmer, 2008), establishing when the hepatic piston first evolved may refine our understanding of the complex terrestrial and aquatic evolution of crocodylomorphs. Our proposed osteological correlates for the hepatic piston offer a novel set of postcranial characters that can be used to evaluate this environmental history.