The effect of collar weight and capture frequency on bodyweight in feral cats (Felis catus)

Introduction

Of critical importance to wildlife conservation and management is ecological knowledge of animal life histories, such as habitat requirements, foraging and reproductive behaviour, movement, dispersal, and home-range attributes. Such data provide stakeholders with the practical knowledge required to make informed management decisions. However, gathering this information from wild animals in their natural habitat can be challenging. Many species are cryptic, whereas others live in environments where direct observation is challenging or impossible (Cooke et al. 2013; Wilmers et al. 2015).

The development of animal-borne devices or biologgers to overcome these difficulties has dramatically expanded the range of animals that can be studied by researchers (Portugal and White 2018), as well as provided the opportunity to understand aspects of animal physiology and ecology that were not previously possible (Rutz and Hays 2009). These devices have revealed remarkable long-distance movements (McGregor et al. 2016; Pilfold et al. 2017; Pedler et al. 2018), exposed novel hunting behaviours (Sakamoto et al. 2009; Troscianko and Rutz 2015; Korpela et al. 2020), and allowed observation of environmental conditions in inaccessible habitats (Charrassin et al. 2008; Hussey et al. 2015).

However, in some studies, animal-borne devices have been found to change behaviour (Aldridge and Brigham 1988; Bruholt 2018), energetics (Godfrey et al. 2003), survivorship (Osborne et al. 1997; Rasiulis et al. 2014), and reproduction (Demers et al. 2003; Venturato et al. 2009), potentially introducing bias and compromising research validity (Casper 2009). Conversely, other studies have recorded no physiological or behavioural impacts on animals from biologgers (Hines and Zwickel 1985; Laurenson and Caro 1994; Thirgood et al. 1995; Bank et al. 2000; Golabek et al. 2008). Recognition of the potential for animal-borne devices to influence results is critical to making informed inferences and conclusions. Consequently, studies using animal-borne technologies commonly employ a general rule of thumb about the weight of the devices relative to the animal’s body size. For non-flying terrestrial animals, a commonly used rule is that weight of the device must be less than 5% (e.g. Tuyttens et al. 2002) or sometimes 3% (e.g. McGregor et al. 2017) of total bodyweight of the animal to reduce potential behavioural or physiological effects. However, the empirical basis and origin of these thresholds is unclear (Barron et al. 2010), and they have been frequently exceeded (see Portugal and White 2018). Negative effects have also been detected when the weight of the device is less than 5% of bodyweight (Aldridge and Brigham 1988; Coughlin and van Heezik 2015). For instance, several studies have demonstrated negative effects of biologgers on bodyweight. Adult red foxes (Vulpes vulpes) in captivity wearing collars of 1–2% bodyweight significantly decreased in weight relative to uncollared control foxes (Bruholt 2018), while San Joaquin kit foxes (Vulpes macrotis mutica) experienced a decrease in bodyweight shortly after collaring relative to uncollared foxes. Collars ranged from 3% to 7% of bodyweight, but decrease in bodyweight did not differ significantly among collar designs (Cypher 1997).

Fixed limits on device weight are commonly accepted by ethics committees and adopted by researchers because of the lack of more refined knowledge on species and device-specific effects. Ideally, guidelines should be based on empirical examination of the negative effects, or lack thereof, caused by device deployment, as determined for the species, device size and deployment duration required for a particular study (Casper 2009; Barron et al. 2010; White et al. 2013; Bodey et al. 2018; Portugal and White 2018).

Introduced cats (Felis catus) are one of the most harmful and widespread of invasive mammalian predators (Doherty et al. 2016), found on many offshore islands and on all continents bar Antarctica. Feral cats have had substantial impacts on wildlife populations (Loss et al. 2013; Nogales et al. 2013; Bellard et al. 2016; Duenas et al. 2021), and have been especially detrimental on islands (Burbidge and Manly 2002; Medina et al. 2011). Because of their cryptic nature, many studies have used biologgers to gather information on cat hunting behaviour (Moseby and McGregor 2022), home-range size and habitat use (Moseby et al. 2009; Recio et al. 2010, 2014; McGregor et al. 2014, 2017), long-distance movements (Guttilla and Stapp 2010; McGregor et al. 2016; Jansen et al. 2021), and predation rates (Loyd et al. 2013; McGregor et al. 2015), to better inform management of the invasive species. All of these studies used collars of <5% bodyweight and, in some cases, <3% or 4%. However, Coughlin and van Heezik (2015) put GPS collars between ~0.86% and 3.11% of bodyweight on 20 domestic cats, and found collars greater than 2% bodyweight could reduce cat home range by 15% relative to lighter collars. These tracked domestic cats had smaller home ranges (<2 ha) than those reported for feral cats (e.g. 50–12,300 ha Moseby et al. 2009). They were also fed by their owners so did not need to hunt to satisfy their energy requirements and could potentially afford to reduce their ranging and hunting behaviour. The authors suggest that this evidence for reduced mobility in domestic cats indicates that feral cats wearing devices may also be similarly affected, and recommend that collars on feral cats should be no more than 2% of the cat’s bodyweight (Coughlin and van Heezik 2015).

As well as the potential effect of wearing a logging device, study animals must also cope with the stress involved in the capture and fitting or checking of devices. Wild mammals exhibit acute stress responses in reaction to researcher interventions (see also Engelhard et al. 2002; Harcourt et al. 2010; reviewed by Cockrem 2013). Significant human-induced stress responses have been associated with reduced fecundity (Alibhai et al. 2001), lower survival (Breed et al. 2019), and altered behaviour (Rachlow et al. 2014). To cope with the energetic cost incurred by mounting a stress response, an animal must either use stored reserves, or divert resources used for other physiological functions (Moberg 2000). Energy reserves are usually stored as fat, and the biological cost model (Moberg 2000) predicts that stress should cause a loss of bodyweight as fat and other energy stores are expended. Consequently, change in bodyweight has been shown to be a practical and relevant proxy for chronic stress responses (McLaren et al. 2004; Moorhouse et al. 2007; Gelling et al. 2010).

In the present study, we used change in bodyweight to estimate the potential for adverse impacts of biologger collaring on feral cats, and investigated whether lighter collars reduced this impact. We used data from 50 free-living feral cats collared during two time periods in the same study area (McGregor et al. 2020a; Van der Weyde et al. 2023), with collars ranging between 0.29% and 4.88% of total bodyweight. We measured the effect of collar weight, number of recaptures, period of deployment, and environmental variables on the change in bodyweight over time.

Methods

Results

Over the two study periods, we caught and collared 50 feral cats and recorded a total of 98 recaptures. Cats carried collars for a minimum of 1 day and a maximum of 231 days between captures (mean = 54.8, s.d. = 44.3 days), and were recaught on average 1.91 times, with s.d. of 1.18 (min = 1, max = 6 times). Collars ranged between 0.29% and 4.88% of total bodyweight (mean = 2.04, s.d. = 1.22), and cat weight averaged 4.04 kg (s.d. = 0.93 kg), with a minimum of 2 kg and a maximum of 6.4 kg (Fig. 1).

Fig. 1.

Spread of cat weights (kg) and percentage weights of collar deployed on feral cats at Arid Recovery in 2016–2017 and 2021.

Of the 128 models compared, seven had delta AICc of ≤2 (Table 1), and were subsequently averaged in an information-theory framework.

The most important variables for explaining changes in cat weight on recapture were the preceding average maximum monthly temperature, the number of recaptures, and the number of days the collar had been on since last recapture (Table 2). Experimental period, rainfall in the previous 6 months, initial cat weight, and the percentage of bodyweight of the collar were also included in the model averaged output; however, these terms were not significant predictors of cat weight (Table 2).

Collar weight as a percentage of bodyweight (Fig. 2a) was not a significant predictor of change in cat weight on recapture (P = 0.506). Rather, our model predicted that cat weight change declined by 24 g (s.e. 10 g, P = 0.015) for each increase in temperature of one degree Celsius (Fig. 2b). After each successive recapture, predicted cat weight change declined by 116 g (s.e. 50 g, P = 0.022) (Fig. 2c). However, we also found a positive relationship between weight change and the amount of time that had passed since the cat was last captured, whereby weight change increased by 3 g per day (s.e. 1 g, P = 0.040) between successive recaptures (Fig. 2d).

Fig. 2.

Model output of predictor variables (a) collar percentage of bodyweight, (b) average maximum temperature, (c) number of times captured and (d) days since the last capture, versus change in cat bodyweight (kg) on recapture.

Discussion

We found no significant relationship between collar weight and cat weight change in recaptured feral cats. However, our experimental design does not allow us to rule out a potential effect of wearing any collar on effecting weight change, because we did not have a control sample of uncollared feral cats.

Coughlin and van Heezik (2015) described reductions in movement from domestic cats wearing collars of approximately 3% bodyweight. Likewise, it is possible that our cats reduced movement but that this did not lead to weight loss. Coughlin and van Heezik (2015) indicated that the impact of heavier devices on home-range size and distances travelled suggested that devices worn by cats may influence energy expenditure, and to accommodate increased energy expenditure cats could reduce high-energy behaviours, such as ranging and hunting. However, unlike feral cats, domestic cats are fed by their owners and do not need to hunt to satisfy their energy requirements; so, they can potentially afford to reduce their ranging and hunting behaviour, whereas feral cats may not. Unfortunately, as we did not directly measure movement or energy expenditure, we were unable to monitor for any behavioural change that may have offset the increased energy requirements. Alternatively, just as other research has found no influence of animal-borne devices on energy expenditure or foraging behaviour (Berteaux et al. 1996; Golabek et al. 2008), feral cats may simply be more tolerant of heavier collars. Cats are primarily ambush predators, relying on a ‘sit-and-wait’ strategy that relies on stalking and camouflage (e.g. Judge et al. 2012; Moseby and McGregor 2022); so, carrying the extra weight of a collar may not have a material impact.

Monthly maximum temperature was found to have a significant negative effect on feral cat bodyweight. This may occur because juvenile rabbits are less abundant in late summer, because rabbits breed from late winter to early summer in the study area (Bowen and Read 1998). Small juvenile rabbits are likely to be easier for smaller non-breeding female cats and/or younger inexperienced cats to hunt, because smaller cats may be less effective at catching larger prey (Read and Bowen 2001; Fleming et al. 2020). This is supported by Moseby et al. (2021), who found a significant interaction between bodyweight and temperature on rabbit consumption by feral cats, with smaller cats feeding on rabbits only in the cooler months whereas larger cats fed on rabbits regardless of temperature. Approximately 215 g of fresh rabbit is capable of meeting a feral cat’s daily energy requirements (Plantinga et al. 2011), and so a reduction in the availability of a high-energy resource would likely affect feral cats’ ability to maintain weight. In addition, higher temperatures could cause thermal stress in cats. Feral cats experience rapid increases in evaporative water loss and core body temperature when ambient temperatures rise above 30°C, increasing energy requirements to support thermoregulation, while also decreasing potential foraging time as cats seek refuge from extreme heat (Briscoe et al. 2022).

Rainfall in the previous 6 months was included in the model-averaged output; however, there was no support for this variable predicting cat bodyweight. Other studies have shown relationships between rainfall and small rodent abundance because they respond to the abundance of food provided by heavy rains (Read 1992; Brandle and Moseby 1999; Letnic et al. 2005). When numbers are high, rodents can be a key food resource for cats (Read and Bowen 2001; Plantinga et al. 2011; Doherty et al. 2015), and their availability is likely to influence change in cat bodyweight. Because we were unable to directly measure rabbit and rodent activity across the entire study period, it is possible that a relationship between bodyweight and prey availability was missed, or that prey abundance was always high enough to sustain the weights of feral cats in our study. However, it should be noted that although experimental period (2016–2017 or 2021) was not found to be a significant predictor of change in bodyweight, during the 2016–2017 study, there was a large and abrupt experimental reduction in rabbit numbers in the Experimental Paddock, which did cause a significant decline in survival and condition of feral cats (see McGregor et al. 2020a). In total, 8 of 21 cats survived by hunting native rodents along the southern fence line, where their abundance was likely to be greatest because of the proximity to the rest of the fenced reserve (McGregor et al. 2020b; Moseby et al. 2020), potentially causing a decline in small mammals in the Experimental Paddock also (McGregor et al. 2020a). In contrast, native rodent numbers were reported to be high throughout the 2021 period (Arid Recovery, unpubl. data) following good summer rains.

The number of times a cat was recaptured had a significant negative effect on bodyweight. After each successive recapture, our model predicted that change in feral cat weight was reduced by 116 g (s.e. 50 g). However, of 98 recaptures, only 25 represented cats caught three or more times. Although our model predicted a linear reduction in weight as cats were successively recaptured, confidence intervals around model predictions increased greatly after more than two recaptures, likely owing to small sample sizes. Without more data, we cannot conclusively demonstrate that change in weight was monotonic, and could potentially have been a single-step reduction in weight change that was maintained after multiple recaptures, possibly owing to cats acclimating to capture.

Regardless, reduction in weight owing to capture may be attributable to repeated stress caused by the capture and handling methods used. Wild mammals show an acute stress response when they are captured and handled (Engelhard et al. 2002; Harcourt et al. 2010; Breed et al. 2019), which incurs an energetic cost (Moberg 2000) that could translate into an impact on bodyweight. Repeated handling stress has been shown to cause permanent reduction in bodyweight of adult lab rats (Harris et al. 1998; Zhou et al. 1999). In addition, other felids such as cheetah, Acinonyx jubatus (Braud et al. 2019), European wildcats, Felis silvestris (Monterroso et al. 2022), Bornean leopard cats, Prionailurus bengalensis borneoensis (Nájera et al. 2014), and Sunda clouded leopards, Neofelis diardi (Nájera and Hearn 2022), have also been recorded exhibiting stress responses to capture and handling. However, standard error was considerable, potentially because of differences in duration of recapture attempts, or transport of cats to cooled sheds for collaring on hot days. The stress caused by our most commonly used trapping method burrow traps in warrens may also be different from that caused by other methods such as cage trapping, spotlight netting, or leghold trapping. Changes in capture methods used between recapture attempts could also have contributed to the significant variation described by our model; however, methods were recorded inconsistently between sessions and so we were unable to include this explanatory variable in our models. Additionally, we did not use anaesthesia during restraint and collar fitting so as to minimise handling and recovery times; however, possibly handling aware animals is more likely to cause acute stress.

Another contributing explanation for a reduction in weight as a result of capture could be that neophobia induced by handling resulted in behavioural changes, potentially reducing foraging activity. Neophobia (aversion to new or unfamiliar stimuli) can be a plastic trait within individuals, which can be induced by exposure to high-risk environments (Brown et al. 2013). Neophobia benefits animals in high-risk environments, where novel stimuli are likely to be hazardous and so avoidance is advantageous (Greenberg 2003). However, putting energy into avoidance of novel stimuli can incur other costs such as reduced access to resources (Sol et al. 2011). Reduction in cat weight as a result of capture may also have been related to reduced access to resources. The sudden, stressful experience of capture and handling may have triggered an alteration of the cats perceived environmental risk, thereby creating a sustained reduction in their capacity to forage effectively.

Bodyweight was also positively correlated with the amount of time the collar remained on, or the time between successive recaptures. This could be attributed to younger cats growing and gaining weight steadily over time (Jones and Coman 1982); however, the absence of a reliable ageing mechanism for feral cats makes it difficult to determine whether there was a demographic effect at play. Heavier cats are perhaps likely to be older, but although the previous weight of cats was included in the model-averaged output, the term had low importance and the estimate was not statistically significant. Additionally, longer periods between recaptures could also have allowed more time for cats to recover from associated stress, although an interaction between recapture number and time did not feature in any of the top-ranked candidate models. Regardless, because weight change was predicted to increase by 3 g each day post-capture, researchers seeking to minimise the physiological impact of multiple recaptures on feral cats could consider extending the intervals between captures to a minimum of 30 days. This approach could mitigate the adverse effects on cat weight attributed to the stress response induced by frequent recaptures.

Conclusions

Our results showed that captures affect bodyweight in cats. Research involving cat captures should factor that into their experimental design, especially if feral cat weight or condition over time is a parameter of interest.

Although we cannot conclude that wearing a collar had no effect on feral cat bodyweight, we did not find any evidence to suggest that increasing collar weight as a percentage of bodyweight influenced cat weight when collars were <5% of bodyweight. However, as we did not directly measure movement or energy expenditure, we were unable to detect any behavioural change that may have offset the possible increased energy requirements imposed by heavy collars. Future research should address this knowledge gap.

Given that increasing collar weight up to 4.88% did not affect cat weight, but number of recaptures did, it may be more prudent to increase collar weight to maximise data, and so reduce the need for successive recaptures. Alternatively, increasing the time between recaptures to at least 30 days could mitigate weight loss resulting from capture and handling. Future experiments should compare capture and handling methods (particularly with and without anaesthesia) to determine how they influence weight loss and stress levels in feral cats.