Brain topography, skull thickness and electrode placement sites
Twenty-six head specimens were collected from captive-bred commercial crocodile farms after slaughter and were dissected along various coronal (n = 13) and sagittal (n = 13) plane cuts made by a bandsaw. Four of the heads underwent magnetic resonance imaging scans before dissection (Fig. 1). The scans were carried out using a 1.5 T machine (Ingenia MRI system; Phillips, South Africa) operated by experienced human radiologists. The skull progressively decreased in thickness from caudal (16.7 ± 1.2 mm) to rostral (4.5 ± 0.8 mm) over the brain cavity. The cerebral hemispheres of the brain were located ventral to the centre point of the cranial plate and this part of the brain was the largest portion that was closest to the surface.
E-stunning procedure
Animals
Fifteen healthy crocodiles, 30–32 months of age, 144–160 cm in total body length were randomly selected by farm management from a group of growers that were housed in a closed communal pen. The crocodiles were captured manually and moved from the communal pen into an experimental house adjacent to the procedure room 2 days prior to the electroencephalogram recordings. A total of five crocodiles were placed in each of the three indoor concrete-floored pens that contained a water pond. The use of the live crocodiles in this study was approved by University of Pretoria Animal Ethics Committee (V084-18) and all study methods were performed in accordance with the relevant guidelines and regulations. The study is reported in accordance with ARRIVE guidelines. Once the data collection was complete, all crocodiles were returned to the owners who reintroduced them into the grower communal pens of the farm.
Procedure
A 7-hole plastic skull-cap that was custom made by 3-dimensional printing to fit similar sized crocodiles and strapped to the head of the crocodile was used to hold the electrodes in a fixed position during the electroencephalogram recording procedure. On the day of the electroencephalogram recordings, the skull-cap was assembled by placing injection needles (35 mm long; 21 gauge) through the 7 holes and 6 mm of the needle tip was bent to form a 90-degree angle which improved contact with the scale surface. On the crocodile end, the seven electroencephalogram cables were attached to the hub end of the needles using metal clips. On the electroencephalogram recorder end, the seven cables were attached to the mobile electroencephalogram recorder (Sleepwalker, Lifelines Ltd) and laptop (Inspiron, DELL). The seven cables were wrapped together to form an umbilicus and a quick-release connector was positioned midway along the 1.2-m-long cables to allow for rapid disconnection and reconnection.
Once the electroencephalogram recorder was setup the crocodiles were manually captured, one at a time, the snouts immediately taped and blindfolded and transferred to the procedure room. They were placed on a low work bench and physically restrained by two experienced animal caretakers, one holding the neck and the other the tail. The assembled skull-cap was placed onto the cranial plate of the crocodile and a 30 mm thick sponge was placed over the top to push the bent needles firmly against the scales. The entire assembly was secured using two self-adhesive material (Velcro) straps wrapped around the head. Then the raw tracing was visually inspected using compatible software (TrackIt version 2.8.0.12; Lifelines Ltd) for stability (not wandering off the isoelectric line or exhibiting excessive course oscillations around the isoelectric line) and an impedance check was done to ensure a value of less than 16 kΩ for each of the 5 electroencephalogram recoding channels. Once a stable tracing was obtained, the cloth blindfold was removed from the crocodile and a pre-stun electroencephalogram recording commenced for 5 min at a 200 Hz sampling rate. Following the pre-stun electroencephalogram recording, the cable umbilicus was disconnected, and the crocodile was prepared for e-stunning.
A research custom-built e-stunner (I’Vimbi technologies) that was powered by single phase mains electricity (220 V, 50 Hz) was used for this study. The e-stunner included a stunning controller referred to as the stunner and stunning apparatus referred to as stunning wand. The stunning wand was a hand-held plastic pole (1.4 m long) with a push-button at the operator-end and a V-shaped studded (three studs on each electrode) stunning electrodes at the crocodile-end of the wand. The volt-out controller of the Stunner was set at 170 V, and the timer set at 5 s stunning period for the first three crocodiles and 7 s for the remaining 12 crocodiles examined.
The data logger was set to record volts, amps and hertz applied for each crocodile and duration of stunning. For the stunning procedure, the neck and head region of the crocodile was doused with water and the stunning wand electrodes were pressed firmly at approximately 50° angle to the horizontal plane, on the dorsal surface of the neck immediately behind the occipital bone. The push-button was activated and the electrodes kept in the same position on the crocodile until the timer had run out. Immediately thereafter, the EEG cable umbilicus was reconnected, and the electroencephalogram recording continued for 7 min during the post-stun period. Throughout the procedure, for each crocodile, the principal investigator digitally video recorded the pre- and post-stunning periods and recorded behavioural indicators.
After completion of the post-stun, meloxicam (0.5 mL; Metacam 20 mg/mL; Boehringer, South Africa) was administered subcutaneously and the skull-cap assembly were removed, the crocodile was weighed and then returned to their experimental pen. The next crocodile was captured and brought to the procedure room. This same procedure was followed until all crocodiles were examined in 1 day. Crocodiles were returned to their communal pen the following day.
Data analysis
Electroencephalogram recording data
Raw electroencephalogram recordings, consisting of five channels per recording, were uploaded to an open-source software (Brainstorm software package, freely available for download online under the GNU general public license at http://neuroimage.usc.edu/brainstorm) for processing and analysis. The raw electroencephalogram recordings were reviewed manually for signal artefacts and excessive movement away from the isoelectric line. A band-pass filter (1 Hz and 35 Hz) was applied to the raw electroencephalogram tracing and was reassessed visually to detect the three best performing channels out of the five. If at least three channels could not be identified because of poor signal quality, low amplitude (< 5 μV), excessive movement artefacts or excessive shifting away from the isoelectric line then the entire crocodile electroencephalogram recording was discarded and not analysed further nor included in the statistical analysis. Crocodiles 2, 3 and 5 were removed from the study because the electroencephalogram recordings could not be analysed.
A series of seven 4-s-long epochs were identified on stable, artefact free electroencephalogram recordings and marked, as follows: 1 pre-stun epoch (period 0); 1 post-stun epoch (period 1) identified within 1–7 s after reconnecting the cable umbilicus and 5 additional post-stun epochs (periods 2–6) exactly every 60 s after the start of the first post-stun epoch. First, each epoch was visually inspected for stability and to ensure they were artefact free. Then the three-channel epoch was standardised by applying an average reference montage. The three-channel epochs were then analysed using the Welch method of power spectral density analysis (time window: 4 s; Hamming window length: 0.5 s; window overlap ratio: 50%; decomposition into delta [2–4 Hz], theta [5–7 Hz], alpha [8–14 Hz] and beta [15–30 Hz] frequency bands). The three spectrum graphs were normalised and then averaged using root mean square averaging. The final averaged PSD analysis was used to obtain power values (μV2/Hz) for each frequency band of interest for each epoch12. The total power was calculated by adding all frequency band powers. The percentage of power within each frequency band was calculated by dividing the power within the band by the total calculated power and converted to a percentage. Additionally, 40 s epochs were extracted from the filtered electroencephalogram recording during the pre-stun (40 s before disconnecting electrode cable umbilicus) and post-stun (40 s after reconnecting the electrode cable umbilicus) periods for visual comparisons and to identify tonic–clonic seizure activity.
The PSD data were assessed for normality through plotting of histograms, inspecting descriptive analysis and the Anderson–Darling test for normality. The pre-stun power within each frequency band was compared to post-stun power using Kruskal–Wallis test. Crocodiles with electroencephalogram data that demonstrated a post-stun increase in delta power and a decrease in alpha power were considered unconscious and therefore an effectively stunned. Crocodile with an absence of these electroencephalogram findings were considered to be conscious but electro-immobilized and therefore not effectively stunned. The PSD analyses were compared and matched descriptively to their behavioural indicators that were recorded and interpreted with the aid of reviewing the digital video recordings. Data was analysed using commercially available software (MiniTab 18.1; MiniTab Inc) and significances interpreted at p < 0.05.
E-stunner data
The data recorded by the stunner data logger was downloaded to a laptop (Inspiron, DELL) for analysis. The volts, amps and hertz were captured per second for the duration of the stun for each crocodile. The area under the curve (AUC) for amps versus time was calculated using the Trapezoidal method (AUC1−n = ∑{\((\frac{Cp1+Cp2}{2})(t2-t1)\}+\{(\frac{Cp2+Cp3}{2})(t3-t2)\}+\dots \)) (1) where Cp is the amps and t is time (seconds). Normality and homogeneity of variance of the stunner data was assessed using the respective Shapiro-Wilks and Levene’s tests. The demographic data (total length, head length and weight) of the crocodiles and stunner data of crocodiles were compared between effectively stunned and not effectively stunned crocodile using Student’s t-tests. Statistical analyses were performed using IBM SPSS v26 (IBM Corp, USA).