Predation exerts a significant selection pressure on prey, shaping a multitude of traits that serve as antipredator defences. In turn, natural selection could favour combinations of antipredator defences with synergistic effects that enhance prey survival. An especially interesting antipredator defence is death feigning (DF), present in a wide variety of taxa and usually characterized by the prey lying motionless often along with defecation, musking, and auto-haemorrhaging. All these aspects of DF should work in conjunction with one another, intensifying the overall effect of the display and in turn facilitating quicker escape. To confirm this hypothesis, we tested 263 dice snakes (Natrix tessellata) directly in the field. We noted the occurrence of smearing faeces and musk and autohaemorrhaging and we measured the duration of DF, expecting to see a negative association between the occurrence of these behaviours and the duration of DF. Our results affirm our hypothesis: dice snakes that smeared themselves in musk and faeces before DF and had auto-haemorrhaging during DF spent significantly less time in DF. Our results highlight the functional integration of antipredator behaviors across different phases of predator-prey interactions, emphasizing the need for future research to prioritize studying the sequential display of behaviors.

Fieldwork was performed at Golem Grad Island (40°52′08′′ N, 20°59′23′′ E, Prespa Lake, North Macedonia). We searched for dice snakes in the open or sheltered under natural cover and quickly captured them by hand. At capture, each snake (N=263) was always grabbed approximately at mid-body with one hand since it has been previously shown that holding snakes at different parts of the body elicits drastically different behavioural responses.

During the first 30 seconds after capture, while holding the snake in the air, we noted the occurrence of smearing. Smearing was defined as the snake expelling the contents of the cloacal glands and defecating while at the same time using the tail to smear musk and faeces on its body. This was termed as the “capture test”. All authors captured the snakes and observed the occurrence of smearing as it was easily observable.

Immediately after the capture test, we performed the “post-capture test”, which included a series of generalized predator-like actions to test for post-capture antipredator response. We assume that, although humans were proxies for natural predators, the same general principles apply to natural encounters between snakes and their predators. To standardize this procedure only one person handled the snakes (VB). This predator-like handling procedure involved a 30-s handling sequence. Each snake was first held for 10 seconds with both hands, and two fingers were used to pinch the body at two fixed points, behind the head and above the cloacal region. Then, the snake was passed from one hand to the other for 10 s and finally slightly stretched for 10 s to imitate a generalist predator handling its prey. The snake was then placed on its back, and the person handling the snake stepped out of view of the snake, mimicking predator hesitation or latency to eat the prey. Another observer (MM) remained motionless at 1 m (crouching) to record the snake’s behaviours. From the moment when the snake was put on its back (i.e., the start of the post-capture test), we recorded measurements of the following sequence of variables: (1) occurrence of death feigning (DF) – if the snake had its mouth open with a protruding tongue while remaining completely immobile we defined this as DF; (2) occurrence of auto-haemorrhaging (AH) – while in DF, if the snake had visible blood around the mouth area we considered this a display of AH; (3) duration of DF – if DF occurred, duration of DF was measured as the time elapsed from the onset of DF until the snake closed its mouth.

After the snake completed both tests, we first measured the body temperature by inserting a probe thermometer into the cloaca while holding the snake by the tail. Then, we noted the dorsal colour morph (blotched, green or melanistic), sex (based on the shape of the tail or hemipenis eversion when necessary), measured the snout-vent length (SVL; performed by gently stretching the snake along a fixed tape measure) and determined the reproductive status in females (through palpation). In order to prevent pseudoreplication, animals were permanently marked with a unique scale clip ID (by notching the ventral scale with scissors) for later identification, and released at the site of capture. We determined size thresholds of sexually mature individuals (forming the variable age (adult/immature) based on a dataset containing over 6000 measured and marked Golem Grad dice snakes. The SVL of the smallest male found in copulation was 55 cm, whereas the SVL of the smallest gravid female in this population was 63 cm.

To analyse our data we used R statistical software (ver. 4.2.2). We determined the distribution data using the Shapiro-Wilks test and visual inspection of the data. Then we modelled the variation in the occurrence of AH using binomial generalized linear models (GLMs). As AH only occurs during DF we subset the entire dataset to individuals who displayed DF during post-capture tests (N=92). We first built two sets of models; a null model, containing only the response variable (AH~1) and a full model containing all factors of interest (Morph*State+SVL+Smearing+Temperature). We then used stepwise regression to find the best-fit models (with the step() function from the MASS package in R). The same approach was performed with the duration of DF, except we used Gamma GLMs.

After finding the best-fit models for all response variables we obtained model summaries (odds ratios, coefficients, confidence intervals and *p values) which we report in RSBL-2024-0058_ESM_model summaries. *We used the ggplot package to plot the figures. The reproducible code can be found in the accompanying R file as RSBL-2024-0058_script.