Sensitivity to predator-related cues and performance of antipredator behaviors are universal among prey species. Rodents exhibit a diverse suite of antipredator behaviors that have been examined in both field and laboratory studies. However, the results from the laboratory have not always translated to the field. While laboratory studies consistently indicate strong fear-inducing effects of cat fur/skin odors, it is unclear whether this occurs in the field with wild rats. To address this issue, we tested the antipredator responses of wild brown rats (Rattus norvegicus) to predatory (domestic cat fur) and non-predatory (common brushtail possum fur) odor cues in a semi-natural experimental paradigm. Rats were housed in open air enclosures containing two feeding stations. Following several nights of acclimatization, the feeding stations were paired with cat fur, possum fur or no fur. Rats spent less time at a feeding station that was paired with cat fur. Duration of time spent at feeding stations increased across consecutive test days and across hours within individual test nights, although the rate of increase within nights was lower for cat fur paired stations. This overall increase might reflect habituation of antipredator behaviors, increasing hunger, or loss of cue potency over time. We suggest that wild brown rats recognize and respond to cat fur odor cues, but their behavioral response is highly adaptable and finely tuned to the trade-off between predation risk and starvation that occurs across short temporal scales.

Study species

The genus Rattus includes at least five of the most invasive agricultural pest rodents in Australia, as well as 14 species considered pests worldwide (Aplin et al., 2003). It also includes the two main species of urban pest rodents worldwide, R. norvegicus and R. rattus (Feng and Himsworth, 2014; Gerozisis et al., 2008). In this study, we focused on the ubiquitous urban rodent pest, the brown rat, R. norvegicus.

We caught rats by live cage trapping (40.64 cm L x 12.7 cm W x 12.7 cm H; model 602; Tomahawk, USA) using a mixture of peanut butter, rolled oats and honey as bait. Trapping occurred on four separate trapping sessions during April to September 2019, within the University of Sydney Camperdown campus, New South Wales (33.8886° S, 151.1873° E, Sydney, Australia). During each trapping session, 25 traps were locked open and baited for a minimum of two nights to allow the rats to familiarize themselves with the traps. On the third day, traps were set before dusk and checked around dawn. Once captured, we transported animals to the Fauna Park at Macquarie University, Macquarie Park, New South Wales, Australia, for housing and testing in outdoor enclosures.

A total of 45 rats were caught during the four trapping sessions (Supplementary Data SD1). Due to the number of testing arenas available, we initially kept five rats from each trapping session for testing and the rest were euthanized (Supplementary Data SD1). A single female died due to poisoning during the acclimatization period. A total of 19 rats (9 ♂ and 10 ♀) completed behavioral testing. All rats tested negative for toxoplasma infections.

Test stimuli

We presented each individual rat with two stimuli involving different animal furs as well as a control (no odor) condition. The stimuli used were cat (Felix catus) fur and common brushtail possum (Trichosurus vulpecula) fur.

Cat fur was obtained on an ongoing basis from pet groomers and veterinary clinics around the Greater Sydney area with collection of fur from a variety of different individual cats. Possum fur was collected directly from 31 animals captured on the University of Sydney Camperdown campus, from March 2017 to April 2019, by live cage traps (66 cm L × 24 cm W × 24 cm H; model 205; Tomahawk, USA) using apples and peanut butter as bait. Once captured, possums were transferred to a light-proof hessian bag and, while maintaining the animal’s head inside the bag, only a small section of fur (<5% of the animal’s surface area, ~5g) was shaved using a battery-operated clipper. The animals were then released at the same place of capture.

Possums are common in urban areas, including the University of Sydney Campus, where they scavenge in rubbish bins. Therefore, we expected rats captured on campus might have had contact with this species, hence representing a familiar non-predatory odor. The University campus also has some feral cats and is surrounded by a residential area where there are many domestic cats present. Thus, it is to be expected that the rats use in this study have experienced predation attempts by domestic and feral cats. Fur from individuals of both species were stored separately to avoid potential habituation of defensive responses to a single individual, as rodents have been reported to be able to discriminate odors from individual cats (Staples et al., 2008). All fur was stored separately (i.e. by individual) at -20°C for long term storage (up to 2 years), and at -4°C when not in use during testing (9 days). During testing, each individual rat was presented with 3 g of fur per feeder per night, with the fur sample representing fur taken from a different individual animal each time.

Arenas

Rats were housed individually in one of five outdoor enclosures (1.8 m L x 1.8 m W x 0.6 m H) made from aviary wire mesh (12 mm x12 mm openings with 0.7mm gauge) (Figure 1). The walls of each enclosure were covered with breathable reflective wall isolator wrapping which aided infrared camera recording, and the floors were lined with wood shavings (~5 cm). All enclosures were located within a 50 m² predator-proof aviary, open to the elements. Shade cloth was secured around (>1 m from the enclosures walls) and above (>2 m above the enclosure top) to offer protection from heat and avian predation.

On one side of each enclosure, a wooden nest box (23 cm L x 30 cm W x 40 cm H) with a single circular entrance (7 cm diameter) containing enrichment (coconut fiber and wood block) was secured (Fig. 1, box d). Two open ended feeding stations, made from transparent red perplex (30.5 cm L × 28 cm W × 25 cm H), were secured at the opposite end of the enclosure from the hide box. Each was fitted with a food hopper made from open mesh wire, secured to the underside offering protection from the weather (Fig. 1, boxes b and c). The food hopper’s mesh wire opening was smaller than the diameter of the food pellets offered (standard laboratory rodent feed), thus forcing the rats to feed by chewing at the pellets through the mesh opening, ultimately preventing rats from harvesting and storing any food items during the trials. A spherical metal mesh tea strainer (5 cm diameter) was secured to each feeding station, behind the food hopper, within which the relevant fur was located (Fig. 1, box a). Four water bottles were secured to the walls of the enclosure, at either side of the hide box. Food and water were accessible ad libitum.

An infrared CCTV camera (Panasonic WV-CP300 Series 650 TVL Day/Night IR Dual Voltage Fixed Camera, with a computer CS-Mount 2.9-8.2mm Varifocal Lens) was set up directly above each enclosure. Two automatic infrared spotlights (Long Range Infrared Spotlight, Jaycar, Australia) were fitted above each enclosure to supplement lighting. The infrared spotlights were automatically activated when environmental illuminations were lower than one lux. All cameras were connected to a computer, where the ANY-maze Video Tracking System (Stoelting Co. 1999-2019) was used to track the movements of rats across the arena during testing. As brown rats are nocturnal (Apfelbach et al., 2005), tracking started at sunset, and recordings ran for 10 hours. Tracking was occasionally disrupted by heavy rain or strong wind and the data from such nights were removed from analysis.

Experimental design

Four experimental sessions were carried out between April and September 2019. A maximum of five rats were tested per session. All rats were acclimatized in their individual enclosures for five consecutive nights. This acclimatization period allowed the animals to familiarize themselves with the enclosure, as well as allowing for a quarantine period to assess the health of the animals. As in most urban areas, pest control by rodenticide is undertaken routinely at the University of Sydney Campus. The rodenticides used on campus have a physiological half-life of three to four days during which animals will show symptoms such as hemorrhage and bleeding from mucous membranes or fatigue (Van den Brink et al., 2018). Only one female showed such symptoms during this period, being excluded from the tests and euthanized. Following the acclimatization period, each rat was tested across nine consecutive nights.

The testing period consisted on consecutive exposure to different combinations of two out of three odor cues: cat fur, brushtail possum fur and control (no fur). For the cat and possum odor stimuli, three grams of fur wrapped around a Hothands® hand warmer were used, and control consisted of the Hothands® hand warmer by itself. All fur stimuli were enclosed in a tea strainer (Fig. 1, box a) and presented within an hour before sunset by hanging the tea strainer inside one of the feeding stations (Fig 1, box c), immediately next to the food hopper (Fig. 1, box b). The tea strainer ensured that the predator cue was contained and also that the stimulus was primarily olfactory, rather than visual or tactile in nature. The hand warmer was used to mimic animal body heat, which has been reported to increase fear response reported in laboratory experiments (Bowen et al., 2013). The Hothands® hand warmer produces a maximum temperature of 40^°C, and remains warm for up to 10 hours, the duration of the entire testing period. The fur stimuli were removed the following morning and tea strainers were cleaned with 70% ethanol solution.

The combinations of odor cues, between the two feeding stations and over the nine consecutive testing nights, followed a Latin-square fully-factorial cross-over design (Ratkowsky et al., 1993). Odor cues were changed every night, with every cue (cat fur, possum fur, no fur) present at each of the two feeding stations over the nine nights. Every possible combination of stimuli was presented at least once across the two feeding stations including those involving the same cue being presented at each feeding station (e.g. cat fur versus cat fur). Combinations of odor stimuli differed between the rats tested each night. The experimental design allowed us to determine any potential carryover effects: laboratory rats show long-term conditioned avoidance of locations were cat fur/skin has been encountered in the past (Bowen et al., 2013; May et al., 2012; McGregor et al., 2002) and so there is the potential for exposure to cat fur on one night to affect behavior on subsequent nights.

To prevent potential recognition and habituation to individual cats (Staples et al., 2008), and/or possums each rat was presented with fur from an individual cat or possum only once. Moreover, to prevent any effect of aging in the odor cues (Bytheway et al., 2013; Parsons et al., 2018a), the fur used each night was discarded and fresh fur was used the following night.

At the end of the experimental trials, all rats were euthanized, and blood samples were taken by cardiac puncture. Samples were used to test for toxoplasma infections using a toxoplasma commercial latex agglutination kit (Toxo-Screen DA, bioMerieux). This test was deemed necessary, since toxoplasma infection has been reported to cause changes in the antipredator behavior of rats (Berdoy et al., 2000; Vyas, 2015).

Data collection

We used ANY-maze Video Tracking System (Automated Software, ANY-maze, Stoelting Co., USA) to live track the position of animals within an area previously demarcated in a video feed. ANY-maze is a video tracking system designed to automate data recording of behavioral experiments. The software compares pixels across frames from a video feed, to detect and record the position of an animal within the video frame and relative to defined fixed objects (e.g. feeding stations). Therefore, this system does not require post-processing of video recordings by human observers. The system is able to identify different parts of the animals’ body - i.e. head, body and tail - ultimately allowing for the recording of each animal’s movements across the arena, as well as more fine-grained behaviors, e.g. sniffing and contact with the odor source.

For this study, we focused on movements and behaviors associated with each feeding station and food hopper, according to the stimuli assigned to them. Specifically, we focused on the amount of time animals spent under each feeding station and the amount of time the animals spend feeding from the food hopper i.e. amount of time their snout remained in contact with the food hopper. We then calculated the proportion of time the animals spent feeding while in the feeding station i.e. time in the feeding station / time feeding – as a proxy of the level of vigilance while in the feeding station. This measurement assumes that animals cannot be vigilant and feed at the same time, thus if an animal spends a higher proportion of time feeding while in the feeding station, their vigilance is reduced and vice versa.