Differential responses by urban brown rats (Rattus norvegicus) toward male or female-produced scents in sheltered and high-risk presentations

Abstract

Wild rats (Rattus norvegicus) are among the most ubiquitous and consequential organisms in the urban environment. However, collecting data from city rats is difficult, and there has been little research to determine the influence, or valence, of rat scents on urban conspecifics. Using a mark-release-monitor protocol, we previously learned rats can be attracted to remote-sensing points when baited with mixed-bedding from male and female laboratory rats. It was thus essential that we disambiguate which scents were eliciting attraction (+ valence), inspection, a conditioned response whereby attraction may be followed by avoidance (–valence), or null-response (0 valence). We used radio-frequency identification tagging and scent-baited antennas to assess extended (>40 days) responses to either male or female scents against two risk presentations (near-shelter and exposed to predators). In response to male scents, rats (n = 8) visited both treatments (shelter, exposed) more than controls (0.2 visits/day treatment vs. 0.1/day; P < 0.05) indicating scents accounted for response more so than risk. Dwell-times, however, did not differ (1.2 s/visit treatment vs. 0.9 s/visit; P > 0.5). These outcomes are consistent with inspection (–valence). In response to female scents, rats (n = 7) increased visitation (5.02 visits/day vs. 0.1/day controls; P < 0.05), while dwell-times also increased 6.8 s/visit vs. 0.2 s/visit in both risk-settings. The latter is consistent with persistent attraction (+valence), but was also influenced by shelter, as runway visits (1.1 visits/day) were a magnitude more common than predator-exposed (0.1 visits/day). Further understanding and exploiting the mobility of city rats is necessary for improvements in basic and applied research, including city pathogen-surveillance and urban wildlife management.

Introduction

Rats (Rattus spp.) are among the most consequential organisms in the urban environment (Firth et al. 2014). Due to their propensity to gnaw and live alongside humans, they disable vehicles, start fires and damage infrastructure. They have also been referred to as ‘pathogen sponges’ due to their prolific ability to transmit disease (Himsworth et al. 2015). Because of these behaviors, rats cost the economy billions through food spoilage in restaurants and private dwellings, damaged wiring in motor vehicles, buildings and airplanes, and city-imposed fines on owners charged with having rats on their property (Pimentel, Zuniga, and Morrison 2005). Despite these impacts, little is known about the natural behavior of rats living in the city (i.e. city rats) (Banks and Hughes 2012). The social and logistic reasons for the lack of research have been well-documented (Parsons et al. 2017; Desvars-Larrive et al. 2018). This is primarily due to the difficulty of releasing pest animals back into the urban environment following capture and identification. Unlike many species that are available in captive and semicaptive conditions, pest species are not generally accessible for observation in zoos or wildlife parks. Thus, the only means to generate robust behavioral data is through protocols that include the capture, preidentification, release and repeated observation under natural conditions (Parsons, Sarno, and Deutsch 2015). The problem is intensified by the reluctance of private or municipal-owned properties to provide access to researchers (Dyson et al. 2019), and the challenges of using technology such as GPS in built-up environments (Chen et al. 2010; Byers et al. 2017). Another barrier for research is due to city rat-control programs. In most areas where rats are available in sufficient numbers for observation, they are also being actively exterminated. Thus, mortality is a barrier for studies to obtain repeated observations and longitudinal analyses (Himsworth et al. 2013).

Ironically, it is the frequent association between wild and laboratory rats that drives the misconception that researchers already have a comprehensive knowledge-base about city rats (Parsons et al. 2017). Most of our familiarity with rats is either based on repeated sightings of a few of the most gregarious animals or from genetically indistinct laboratory rats bred in close proximity to one another (Puckett, Micci‐Smith, and Munshi‐South 2018; Storsberg et al. 2018). Consequently, science is almost bereft of ethological data on rats living in areas of high human density (Desvars-Larrive et al. 2018).

Without any existing model or monitoring assay to follow, our approach to studying city rats is based on rats proclivity toward scents (Parsons, Sarno, and Deutsch 2016). We chose olfaction because, as nocturnal animals living in persistently noisy environments (Whishaw and Tomie 1989), rats depend heavily on scents (Bazhenov et al. 2014; Parmiani, Lucchetti, and Franchi 2018) to make mobility-related decisions. Olfaction research, however, is complex. Scents cannot be understood nor exploited for management, without in-depth appreciation of their functional value and the environmental contexts under which they are presented (Orrock, Danielson, and Brinkerhoff 2004; Orrock and Danielson 2009). The information transmitted from a scent is more than the mere presence or absence of the sender (Gosling and Roberts 2001; Hurst 2009). Rodents communicate ‘silent’ information such as genetic identity (Cheetham et al. 2007), social rank/threat (Lee, Khan, and Curley 2017), disease status (Zala, Potts, and Penn 2004), hunger level/diet (Wyatt 2010) or receptivity (Hurst and Beynon 2004). These scents are often comprised of common pheromones such as 2-heptanone (Gutiérrez-García et al. 2007) and major urinary proteins (MUPs) (Zhou and Rui 2010) in the urine, stress-induced pheromones secreted from the peri-anal region (Kiyokawa et al. 2009), and highly noxious methyl-sulfides from urine and feces (Wheeler 1977). These chemicals can be embedded in almost any animal tissue product or exudate. Common sources include sebum from the skin, porphyrin around the eyes, dander, glandular secretions, urine and feces (Wyatt 2010). Even tears have been shown to transmit pheromones (Tsunoda et al. 2018). Scents detected by a conspecific is referred to as a ‘pheromone’, whereas the same scent when detected by a member of another species is referred to as kairomone. We refer to scents and pheromones synonymously as they relate to rats responding to rat-produced scents.

Context of presentation and valence

Following a recent meta-review of the literature, Parsons et al. (2018a) identified three types of responses to scents: persistent attraction (+ valence); indifference (0 valence); or an immediate or delayed aversion (– valence). Identifying latency after detection is essential for experimental assessment and targeted deployment of the right scents for management purposes. This is because a potential aversive scent may at first attract an animal to the scent before later repelling it. This perplexing behavior is referred to as scent inspection (Lledo-Ferrer, Peláez, and Heymann 2011) or predator inspection (Fishman 1999) and has confounded our understanding of the valence of predator–prey scents.

When producing tools for wildlife management or urban biological assays, it is important to understand the functional ‘mechanisms’ and contexts of presentation in order to manage the response. For instance, we generally consider a predator scent as a deterrent for prey animals. But depending on contexts such as where the scent was deposited in relation to the prey animal, and the age and specificity of information in the scent, the ‘deterrent chemical’ may not immediately produce the expected outcome (Parsons et al. 2018a). Indeed a predator scent might first attracts prey animals that approach the scent seeking additional information about the sender (Lledo-Ferrer, Peláez, and Heymann 2011). A relevant example occurred in New Zealand when Stoats (Mustela furo) showed a strong attraction to sympatric and novel predator scents taken from cats (Felis catus) and African wild dogs (Lycaon pictus) (Garvey, Glen, and Pech 2016).

A proper valence assessment should seek to determine how long the scent remains active during chemical breakdown and changing environmental contexts, such as perceived vulnerability or shelter (Orrock and Danielson 2009). If a prey-animal is exposed i.e. away from shelter, when it detects a scent, then it takes more risks in order to sample scents (Supplementary Video 1) and locate the resource that the scent represents. Conversely, if an animal is in an enclosed sheltered environment (or under laboratory conditions), it may either feel more, or less, safe than in the natural environments. Recognizing these ‘mechanisms’ for response is a critical step in the deployment and management of wildlife scents for rodents.

Urban biological assay

With no alternative means to monitor the behavior of city rats, we developed a radio-frequency identification (RFID) assay to overcome the aforementioned logistic obstacles (Parsons, Sarno, and Deutsch 2016). In doing so, we learned that urban rats in an indoor/outdoor industrial facility (identity remains anonymous) were attracted to pooled, mixed-gender scents from laboratory animals (Parsons, Sarno, and Deutsch 2015). However, we did not know which pheromone-source among the composite scent was driving the response, nor did we recognize the overall valence or impact of presentation.

Here, as part of our overarching project to characterize the life history and behaviors of a colony of rats at a waste recycling center in Brooklyn, New York, we deconstructed the valence of each scent. Our primary RFID tool is passive micro-transponders that provide a positive identification for animals that otherwise appeared uniform in appearance. These transponders are not battery-powered and require the animal to visit a point (runway or corridor) where active antennas producing low-frequency radio waves are installed. Evocative scents can be placed on or near these antennas, allowing for the quantification of an individual animal’s attraction to particular pheromones.

Unlike food rewards from cafeteria trials which commonly attract insects, birds and multiple species (Parsons and Blumstein 2010), sex, alarm and territorial pheromones are more often species-specific (Stowers and Kuo 2015). Unless specifically eavesdropped upon by a predator seeking information on prey, they are ignored by other species or predators of the target species (Hughes, Korpimäki, and Banks 2010; Jones et al. 2016). The target animals’ chemical environment already consists of many mixed-plumes of scents constantly reforming and degrading (Müller-Schwarze 2016). Thus, mammalian scents are probably best deployed at a microscale because scents may not be detectable from a distance in the urban environment.

Objectives

Our aims were to separately determine the valence (+, 0, –) of male-produced and female-produced scents on wild rats in an industrial center using an RFID assay. We specifically asked whether microchipped rats were more likely to increase visitation (number) or dwell-time (s) at either male- or female-produced pheromones (used bedding for males or used cage boards for females) as compared with controls (wood chips from their environment). Additionally, we sought to understand how this valence differed when pheromones are placed near shelter versus 5 m away from their natural runway in an exposed environment. Our results address a knowledge gap about rat-scent preference that has been apparent since the 1980s (Hurst 1989; Banks and Hughes 2012) and could assist improved trapping, biological control options, or enhanced delivery of other urban wildlife management tools, such as immune-contraceptives (Liu et al. 2013; Burd 2014).

Methods

Study area

New York City (NYC), 40.71°N, 74.01°W, is the second largest city in North America and the largest in the United States. The population density for the 8.5 million inhabitants is 2000 people/square mile (Griffith and Wong 2007). The climate has warm, moist summers (summer highs average 27.8°C with 19.3°C minima) and cold winters (average maximum and minimum are 5.0°C and –1.5°C, respectively) with an annual precipitation of 50–200 cm (NOAA). Brown rats (Rattus norvegicus) likely arrived by ship from Europe between 1700 and 1750 (Puckett et al. 2016; Combs et al. 2018). Predators of rats include feral cats (Felis domesticus) (Parsons et al. 2018b), red-tailed hawks (Buteo jamaicensis), coyotes (Canis latrans) and red foxes (Vulpesvulpes).

We obtained permission to utilize an industrial waste transfer site (anonymous) that closely mirrored our previous research site ( Parsons, Sarno, and Deutsch 2015, 2016 ). This site consists of a large, partly enclosed building, with persistent strong smell to the human nose from collected rubbish. The site had regular human, and intermittent vehicle, traffic inside and outside and was generally noisy. The primary rat colony (Fig. 1) is located at the western end of a waste transfer station in a large, open room that separates a storage area from incoming wastes. The room is usually dimly lit or dark day and night. The primary food supply is located immediately to the west of the main burrows (Fig. 1). Rats were abundant inside the building and were reproductively active year-round due to the continual availability of food and shelter.

Figure 1:

Treatments and controls in sheltered (across a 30 m runway next to the wall) and exposed presentations (5 m away from the wall) at a rat (R. norvegicus) colony used in an RFID assay at an undisclosed waste treatment site in New York City from December 30, 2017–February 28, 2018 and July 4–August 28, 2018. Dark filled squares (numbers 1 and 4) are treatments for each presentation. Light colored squares (numbers 2 and 3) are controls. The active colony (colony 1) and inside the wall to the east of the colony. The principal food source is located to the west of the runway

Experimental design

As part of a 1-year naturalistic study on city rat behavior, we ran trials with male scents from December 30, 2017–February 28, 2018 and female scents from July 4–September 17, 2018. The latter included a 4-week latency period where no scent was deployed at either treatment or control antennas (August 19–September 17, 2018). Our design was modified from our previous study at a different waste transfer station (Parsons, Sarno, and Deutsch 2015). We used four RFID antennas (Fig. 1). The first set of treatment and control antenna (#1 and #3) were placed 10 m apart in the runway. The second treatment and control (#4 and #2) were placed in the exposed open area (Fig. 1) at a 45° angle away from the runway, with 5 m between ‘exposed’ antennas. Male- and female-scent trials were performed the same, with the exception of the latency trials, whereby no scents were provided at either the treatment or control antennas. Treatments consisted of 20 g of used rat bedding from males or used cage boards from females from laboratory rats placed directly on antennas #1 and #4. We used 20 g of wood chips for controls. The latency period allowed us to assess whether rats would continue to differentially visit any antennas for any reason other than the presence of a scent.

Our protocols for anesthesia, microchip implantation, and remote monitoring have been well reported ( Parsons, Sarno, and Deutsch 2015, 2016 ) and are included as Appendices 1 and 2. We also mounted two Browning Strike Force 10.0-Megapixel digital infrared cameras with sound for video monitoring of rat and predator activity along the full length of the colony corridor (a preidentified and well-used runway).

Scents

For male scents, we collected used bedding (containing sebum, dander, feces and urine) from six adult, male laboratory rats (domesticated R.norvegicus, Wistar strain) obtained from the Icahn School of Medicine at Mount Sinai, NYC. This substance was placed directly onto the antennae and replaced every 2 weeks. We chose domesticated rat donors because laboratory rats have been previously used as effective lures for wild rats (Shapira et al. 2013). These animals were fed a standard diet, were healthy young adults and not known to have disease or parasites.

Female scents were provided by Bell Labs, Inc. They were also collected from females that were born in captivity and had been physically separated from males. Scent droppings (dander, urine and feces) were allowed to drop from the cage onto deotized animal cage boards (DACBs) which were placed under each row of three females. DACBs were folded in half, wrapped in plastic bags and transported to the field site for deployment. We maximized the freshness of all scents by storing in a cool, dark place under minimal head space in order to minimize the differential of volatile chemicals that advertise past or present state (Hegab et al. 2014). We did not freeze the scent to prolong its shelf-life. Freezing is known to change the conformational state of carrier proteins that are either compose the scent (Ferrer and Zimmer 2007; Tsunoda et al. 2018) or involved in scent dispersal (Hurst and Beynon 2004).

In our experience, it takes 2–3 years, on average, to locate metropolitan-based industrial site that will allow the capture and release of wild pest animals on their property. Therefore, we had to choose between running multiple scent trials at the same time within a single site, or risk conflating findings with seasonal impacts associated with longer duration studies (e.g. longer than 60 days). We chose to run trials that spanned more than one season because animals lived indoors, had a continual supply of food, and thus, reproduced year-round.

Statistical analysis

Data parsed from the RFID data logger included the date and time (hh: mm: ss: ms) a rat visited a specific antenna. Dwell-time (seconds) was calculated as the time elapsed between consecutive visits that are at most 4.99 s apart. This was to account for animals that moved on and off of antennas within a 5 s span. Any such activity was recorded as a single visit. We standardized ‘days elapsed’ by recoding the first day that the animal was RFID-tagged as day 1 until the last date a visit was logged. For count data, such as number of visits per day, we performed a longitudinal Poisson regression analysis with random intercept to model the number of visits per day and to account for repeated visits by rats over time. For continuous data (e.g. dwell time), we performed a longitudinal linear regression analysis with random intercept to model the average dwell-time per rat, per day. We used SAS 9.4 (Cary, NC) to conduct the longitudinal regression models using PROC GLIMMIX for the Poisson model and PROC MIXED for the linear model.

Results

Male-scent trials

Eight animals (six males, two females, all adult, Table 1) were captured and RFID implanted in order to measure the valence of male-produced pheromones. Rats averaged 1.07 visits per day (0.3 by females; 0.8 by males) across all antennas over 61 standardized days. There were treatment effects for both the sheltered and exposed antennas (Table 2; F₃=3.1; P = 0.028). Inside the runway (Fig. 2), the average number of visits per day by an individual was 0.4 ± 0.1 at antenna #1 (runway treatment) and 0.2 ± 0.1 at antenna #3 (runway control). In the outer area, antenna #4 (exposed treatment) received 0.3 ± 0.1 visits per day and antenna #2 (exposed control) received 0.2 ± 0.1. Antenna #1 had more visits as compared with antenna #2 and #3 (Table 2); but was not different from #4 (exposed treatment). There were no differences in average dwell time/visit per antenna between treatments or controls (F₃=1.54; P = 0.22; Table 2). Antenna #1 had an average dwell time of 0.8 ± 0.3 s; antenna #2 had 1.5 ± 0.9 s; antenna #3 had 0.2 ± 0.1 s and antenna #4 had 4.1 ± 2.4 s. Neither treatment nor controls differed according to treatment presentation (Figs 2 and 3).

Figure 2:

City rats (R. norvegicus) responses to male- (left) and female-produced (right) pheromones in scent trials at an undisclosed waste-treatment site in NYC from December 30, 2017–February 28, 2018 and July 4–August 28, 2018. (A) Average visitation per day/per rat per treatment. (B) Average dwell-time (s) for each visit. Numbers 1–4 refer to antennas (treatments or control in high and low shelter presentations)

Figure 3:

Histogram of number of visits by city rats (R. norvegicus) with 5-day bins at an undisclosed waste-treatment site in NYC from December 30, 2017–February 28, 2018 and July 4–August 28, 2018

Table 1:

City rats (R. norvegicus) identified for participation in scent trials at an undisclosed waste-treatment site in NYC from December 30, 2017–February 28, 2018 and July 4–August 28, 2018

Trial #	ID	Sex	Adult	Weight (g)	Tail	Body
1	EF71	M	A	298	206	176
1	BEB5	M	A	334	209	152
1	9B29	M	A	327	–	–
1, 2	EFEB	F	A	328	203	184
1	AB3A	M	A	413	206	182
1	D42E	F	A	431	180	162
1	CE45	M	A	382	227	175
1	C443	M	A	416	204	202
2	40E0	M	A	371	208	176
2	BE5D	F	A	316	170	220
2	DEDD	M	A	456	206	248
2	63FA2	M	J	120	175	195
2	C53D	F	A	352	193	215
2	5D129	M	A	328	182	248

Table 2:

Male scents: longitudinal Poisson regression analysis of number of visits per antenna per day at an undisclosed waste-treatment site in NYC from December 30, 2017–February 28, 2018 and July 4–August 28, 2018

Effect	Antenna	Estimate	SE	DF	t value	Pr > |t|
A. Visitationa
Intercept		–1.30	0.37	7	–3.53	0.001
antenna	2	–0.86	0.36	425	–2.39	0.017
antenna	3	–0.96	0.37	425	–2.57	0.011
antenna	4	–0.37	0.31	425	–1.20	0.231
B. Dwell-timeb
Intercept		2.41	0.93	7	2.58	0.0363
antenna	1	–1.23	1.22	15	–1.02	0.3259
antenna	2	0.06	1.52	15	0.04	0.9672
antenna	3	–2.20	1.56	15	–1.41	0.1793

a

Type 3 test of fixed effects for antenna: F₃ = 3.20; P = 0.023.

b

Type 3 test of fixed effects for antenna: F₃ = 0.94; P = 0.446.

Female scents

A new set of six rats (four males, two females, all adult except one juvenile male, Table 1) was captured and RFID-tagged to assess valence of female-produced pheromones. A seventh female rat that was tagged months earlier during the male scent trial and remained alive at the initiation of the female scent trial was also included. Rats averaged 5.7 visits per day (3.9 by females; 1.7 by males) across all antennas over 46 standardized days. There were significant differences between both sets of treatments and controls (F₃ = 63.15; P < 0.001; Table 3 and Fig. 3). Antenna #1 (runway treatment) had 1.1 ± 0.72 visits per rat per day and antenna #2 (runway control) had 0.02 ± 0.02. In the exposed block, antenna #4 (exposed treatment) had 0.1 ± 0.06 while antenna #3 (exposed control) had 0 visits. Antenna #1 was different to antenna #2 and #3 (P < 0.05), but not #4. There were significant differences in the dwell time/visit per antenna (F₃ = 21.97; P ≤ 0.0001; Table 2). The average dwell time per visit at antenna #1 was 6.77 ± 0.8 s, antenna #2 was 3.49 ± 2.5 s and antenna #3 was null due to lack of visits. Antenna #4 averaged 2.28 ± 1.11 s.

Table 3:

Female scents: longitudinal linear regression analysis of visits per antenna per day at an undisclosed waste-treatment site in NYC from December 30, 2017–February 28, 2018 and July 4–August 28, 2018

Effect	Antenna	Estimate	SE	DF	t value	Pr > |t|
A. Visitationa
Intercept		–0.77	0.48	6	–1.60	0.160
antenna	2	–3.65	0.41	1098	–8.83	<.0001
antenna	3	–	–	–	–	–
antenna	4	–2.22	0.21	1098	–10.56	<.0001
B. Dwell-timeb
Intercept		1.13	0.29	6	3.87	0.008
antenna	2	–0.79	0.12	1098	–6.61	<0.0001
antenna	3	–	–	–	–	–
antenna	4	–0.74	0.12	1098	–6.17	<0.0001

a

Type 3 test of fixed effects for antenna: F₃ = 63.15; P ≤ 0.0001.

b

Type 3 test of fixed effects for antenna: F₃=21.97; P ≤ 0.0001.

Latency period and risk

There were five visits across all antennas during the 4-week latency period when no scents (treatment or control scents) were deployed at any of the four antennas. Videos showed that six identifiable feral cats patrolled the rat colony nightly (Parsons et al. 2018b), while waste transfer professionals and exterminators were active throughout the study.

Discussion

We report the first data on differential responses toward male- and female-produced scents by city rats. Because context drives scent response, we do so while examining the influence of two risk presentations. Rats visited male pheromones irregularly at 0.2 visits/day, as compared with female scents (5.02 visits/day). Interestingly, these values are on opposite ends of the number we previously reported for mixed-gender scents (2.7 visits/day) (Parsons, Sarno, and Deutsch 2015). The dwell time at male scents, 1.2 s/visit, was also less than that at female scents, 6.8 s/visit. Again, these values fall on either side of the numbers we previously reported for mixed scents (3 s/visit), under the same conditions, and in the same type of environment for the same length of time (Parsons, Sarno, and Deutsch 2015).

Male scents elicited a treatment effect in both the sheltered and exposed areas, but visits were infrequent and dwell times were not different between treatment and control scents. This implies that animals may have been motivated to periodically sample a potentially threatening scent of a genetically disparate intruder (Fishman 1999). The lack of persistence (e.g. continual visitation) indicates inspection (–valence) (Lledo-Ferrer, Peláez, and Heymann 2011) rather than attraction (+valence), even though animals first approached the scent. One explanation could be that male scents are likely to induce visitations—either by males eavesdropping on a competitor (Garvey, Glen, and Pech 2016), or females seeking the best genes (Kumar et al. 2014)—irrespective of the level of risk to do so. However, it ostensibly does not take much time (Fig. 3) to interpret the significance (risk or reward) inherent in the scent by males or females, and thus need for revisitation is infrequent.

As with male scents, female-scents elicited treatment effects across both near and exposed presentations. However, the dwell times also increased significantly. When considering these responses, we might consider female scents as more evocative, and more consistent with a true attractant (+valence), than male scents for this population of rats, in these contexts of presentation. However, we also note that the treatment effect for dwell time was lost in the exposed presentation. On the infrequent occasion that rats did visit the external antennas, they were just as likely to dwell at the control scent as the treatment. We considered the possibility of ‘handedness’ or side-preference (the preference of animals to favor one side over the other) as a driver for this finding. However, this is unlikely given that we controlled for side-preference placing the treatment-scents on opposite sides for runway versus high-risk presentations (Fig. 1). We also considered whether the rats were simply staying inside their runways, and ignoring scents on the outside. However, this is not supported by the data. Among visitations, all presentations elicited treatment effects between treatment and control antennas.

The latency period trial, during which antennas were not baited with anything, helps to provide confidence in our findings. Indeed, in the 4-week latency period there were only five visits to any nontreated antenna. However, we stop short of broadcasting categorical findings. In scent trials, context is everything. Some of the most evocative scents are predator scents, and even the most vulnerable rodents may not respond to these risks unless exposed under full moon, or when their detection and defenses have been compromised by adverse weather conditions (Orrock, Danielson, and Brinkerhoff 2004; Orrock and Danielson 2009). Additionally, our 20 g treatment strength was somewhat arbitrary. Ironically, Jackson, Keyzers, and Linklater (2018) and Takahashi et al. (2005) have found that deploying too much of a scent can be as detrimental as having too low a concentration. Thus, more studies are necessary to determine the true valence of these scents.

It should also be noted that our scents were replaced only every 2 weeks, meaning that during parts of our data acquisition these scents were likely degraded. Differential volatility of various scent components likely leads to degradation of the signal and potentially changed valence (Apps, Weldon, and Kramer 2015). However, small mammals such as rodents likely need the same specificity in a scent as larger animals (Apfelbach et al. 2015). This is also supported by the Jackson et al. work, where single molecules generated a response (Jackson, Keyzers, and Linklater 2018).

Among contexts, we also note that this noxious environment, though natural to the animals that live there, might cause preferential detection of some scent types over others. Further, we are uncertain what impact the omni-present feral cats had on rats. Rats at this waste-recycling center are under heavy duress from people, machinery, noise and noxious conditions, as well as exterminators and other predators (Parsons et al. 2018b). Therefore, our findings will need to be reproduced in differing contexts and locations to understand their broad impact.

Because male and female scents were trialed during different seasons, we consider the possibility that seasonality effected our findings. However, among small, nocturnal animals, scents play an essential role in mate selection and social dynamics (da Silva et al. 2015). Given that reproduction was taking place year-round, and that scents were confined to an indoor/enclosed environment, we think any seasonal impacts would be minimal. However, if multiple urban sites are available to researchers at the same time, then we might recommend male and female trials be ran concurrently.

Ironically, finding urban research sites and obtaining behavioral data from city rats may be even more challenging than obtaining that of an endangered species. However, further research is essential if we are to characterize wild rat behaviors. This approach may be especially useful in conjunction with molecular protocols (Puckett et al. 2016; Combs et al. 2018). If it is determined that female scents are more ubiquitously preferred, then these scents could be further utilized in biological assays, pathogen surveillance (Firth et al. 2014; Frye et al. 2015), and potentially, urban pest management (Byers et al. 2017).

Further research will also be necessary to determine the valence of mixed-male versus individual-male scents. For instance, social rank may cause animals to differ in their response according to specific information in the presented scent (Hurst and Beynon 2004). Of particular interest would be whether scents from the most territorial or alpha males could induce a more prominent valence than scents from lower ranking omega males. Future research can also continue to disambiguate this finding by assessing female pheromones with and without mixed-male scents. A better understanding of this information is essential if we are to understand the value of pheromones for influencing mobility. As society becomes more urbanized (Cohen 2003), and climate change more predominant, our need to learn about and manage this species will only become greater.

Acknowledgements

We thank the Pest Management Association (NPMA) for helping ‘jump-start’ urban rat research through their generous funding, patience and high level of continued support. We also thank Arrow Exterminating Company Inc., of Lynbrook, NY (Rudy Hofler and Tom Jordan) for sponsorship and collaboration that exceeded expectations. We hope outcomes from this project inspire other academics and pest management professionals to work together. We thank Matt Ruiter and Craig Jordan of UID Identification Solutions for designing the all-weather RFID system for urban rats. We thank Bell laboratories, Inc. (Brian Lindgren Simone Jeans and Troy Ryba) for providing female rat scents. We also sincerely thank Robin Nagle of New York University for her support. Bob Schrock of Vet Equip designed the mobile, anesthesiology unit. Beverly Shelton from Columbia University Institute for Comparative Medicine provided advice for the protocol. We thank two referees for constructive comments that improved our paper.

Funding

This work was sponsored by the Pest Management Foundation, Fairfax, Virginia.

Ethical approval

All procedures were in accordance with the guidelines for ethical conduct in the care and use of nonhuman animals in research (Fordham IACUC JMS 17-01).

Conflict of interest statement. None declared.

Appendix 1

Anesthesiology and transponder implantation

Rats were captured using 18 × 5 × 5 cm Havahart traps baited with bacon fat, peanuts, tuna, or bananas, placed overnight near or within colony runways. We modified a Tupperware lab cart into a mobile laboratory. This consisted of a calibrated TEC 4 table-top isoflurane vaporizer with dual procedure circuit. One circuit terminating into a 20 g aquarium that was modified into an induction chamber and the other circuit went to a rodent mask. All equipment was procured from VetEquip (Pleasantville, CA). After rats were captured, we set isoflurane to 5% with a flow rate of 2–3 ml/min of oxygen (O₂).

Upon capture, one researcher approached the trap, and covered it with a dark pillowcase to limit stress to the animal. The researcher then carried the filled trap by hand from the area it had been captured. The researcher used slow, steady footsteps while transporting the trap from the runway to the anesthesiology lab cart, about 50 m away. One researcher then placed the cage into the over-sized induction chamber for up to 20 min, until the animal displayed immobility with a regular slow breathing pattern.

To remove the trapped rat, a second researcher lifted the cage at a 45° angle and released the trap-door freeing the animal inside the chamber. The first researcher donned animal gauntlets, used the toe-pinch method to determine consciousness and when the rat was deemed deeply anesthetized (no response to toe pinch), lifted the animal out of the cage, and onto the surgical area. The second researcher then placed the rodent mask on the rat and reset the isoflurane dials to 3% rate on 3 ml/min O₂ flowrate (Balla et al. 2014). We recorded health condition, sex, weight, length and condition of tail, body and width of testes. A rat was considered adult if it was >179 g for females or >199 g for males (Firth et al. 2014), otherwise it was considered juvenile.

We then placed the animal onto the table ventral-side down and shaved a 2 × 2 cm section of fur between shoulder blades. The shaved area was prepped with a cotton swab dipped in betadine using an inside-to-outside circular motion. We inserted a pre-sterilized, lancet style microchip below the skin at the nape of the neck on the dorsal side. We pinched the skin between fingers to form a tent and inserted the needle bevel up, ejected the chip from the retractor, with fingers in place to grasp chip from the outside of the skin. The needle was then removed with a 180° twist. After checking for bleeding, we applied firm pressure and/or vet glue when necessary to seal the wound.

Following successful implantation, animals were returned from the mobile field laboratory to the site of capture inside their cages. Upon being returned to the point of capture, animals were allowed to sleep inside the open cages, and left freely once the anesthetic had worn off. Most animals retreated from the cage into the burrow (Fig. 1), but were free to visit scents immediately following release. Onsite regulations required ongoing baiting of rats with second generation rodenticides. Therefore, we proceeded with the knowledge that identified subjects may be few in number and perish within weeks to a few months (Parsons et al. 2017).

Appendix 2

Remote sensing and rfid-monitoring

We used the PADAR 4DLTH remote-monitoring system by UID Identification Solutions, Chicago, IL, USA. The unit was supplied in a weatherproof enclosure for use in temperatures from –20°F to 120°F and in all weather conditions. The PADAR System is a self-contained high-power RFID reader with four external antennas, data logger and cell uplink. The cell mobile hotspot uplink provides access to the data from the data logger in remote locations that are difficult to visit. Every event is recorded in the data logger and at a predefined time, the data are automatically uploaded to a secure cloud server system and the researcher is notified via email. The data are then downloaded into Excel format for analyses.