Introduction

Norway and black rats (Rattus norvegicus and Rattus rattus) are commensal rodents with a virtually worldwide distribution [1]. They are highly adapted to coexisting with human populations, and are particularly ubiquitous in the urban environment [1]–[3].

Infestations are problematic in urban settings because rats are the source of a number of zoonotic pathogens (pathogens transmissible from animals to people) responsible for significant human morbidity and mortality in cities around the world [4], These pathogens include Leptospira interrogans, Rickettsia typhi, Yersinia pestis, Streptobacillus monilliformis, and Seoul hantavirus, among others [4]. Urban rat infestations and rat-associated public health risks have been shown to increase in association with the growth of cities and increased urban poverty [4], [5]. Given current unprecedented rates of global urbanization (over half the global population now resides in urban centers [6]), rat-related disease issues are likely to increase in the future.

Rodent ecology is a primary determinant of zoonotic pathogen dynamics in rodent populations and of the risk of pathogen transmission from rats and people [4], [7]. Yet many studies of rat-associated zoonoses (RAZ) do not account for the characteristics of rat populations during sampling or data analysis [8]–[12]. This is problematic because it hinders the development of a comprehensive understanding of pathogen ecology in rat populations, prevents comparability and synthesis of results generated by different studies, and can lead to erroneous conclusions with regard to RAZ dynamics.

The characteristics of rat populations can influence the ecology of RAZ and/or our understanding of RAZ ecology in a variety of ways. For example, the prevalence of many zoonotic pathogens among rats can vary remarkably even over a short geographic distance [13], [14]. The reasons for this are unclear but could be related to geographic variations in rat population density, independent of, or in combination with, environmental factors. Although past research has indicated that urban rats form tight colonies with small home ranges [15], there are very little data regarding how rats are distributed across the modern urban landscape.

The ecology of RAZ could also be influenced by temporal factors, particularly season. Seasonal changes in rodent population dynamics have been shown to influence the ecology of rodent-borne zoonotic pathogens in more sylvatic settings [16], [17]. However, there is considerable conflict in the literature regarding the impact of season on rats and RAZ in urban ecosystems [18]–[21], which, being largely under human control, would seem less prone to natural seasonal variations.

In addition to extrinsic factors, the characteristics of rats themselves can influence the likelihood of infection with a RAZ. Body mass is one of the most common variables used to analyze and interpret data on RAZ, with mass being used as a proxy for chronological age [13], [19], [22], [23]. However, there are a variety of factors other than age that could influence body mass, such as population of origin [24], [25], health status [26], and access to resources [21], [27], [28]. The factors contributing to body mass in urban rats have not been well described.

Rat behavior and social interactions are also likely to have significant impacts on disease transmission but are very difficult to capture and quantify during field research. One aspect of behavior that can be identified is intra-specific aggression resulting in visible cutaneous bite wounds. Past research has identified a positive correlation between Seoul hantavirus infection and the presence of cutaneous bite wounds in urban rats [29]. However, it is difficult to say whether this indicates that the virus is transmitted through biting, or whether bite wounds are a reflection of some underlying characteristic of the individual, behavioral or physiologic, which is driving infection. As with body mass, the factors affecting wounding in urban rats have not been fully investigated.

Finally, most studies of RAZ use data generated though trapping to make inferences about population and disease ecology [18], [19], [30]. These inferences are based on the assumption that all animals in a population are equally likely to enter any given trap at any given time (i.e., that trapping itself does not create any systematic bias in the characteristics of the study population when compared to the source population). However, there are a variety of factors that could influence if and when an animal enters a trap, including trapping methodology, trap environment, and characteristics of the individual rat [27]. Given marked variation among studies with regard to how trapping is carried out [19], [22], [31], [32], it is important to better understand how trapping might influence the number and characteristics of rats trapped.

The overall objective of this study was to describe the characteristics of rat populations from an inner-city neighborhood of Vancouver, Canada, with a focus on factors that could influence the ecology of zoonotic pathogens in these populations and/or our understanding of pathogen ecology. Specifically, we aimed to: 1) Describe rat population density and distribution across the study area; 2) Describe seasonal variations in rat population size and structure; 3) Identify the factors that contribute to body mass in urban rats; 4) Identify the factors that influence the presence and number of cutaneous bite wound in rats; and 5) Describe how factors associated with trapping can influence the success of rat collection and the characteristics of the rat population collected.

Methods

Results

Population Density and Distribution

Relative rat abundance varied remarkably among blocks, with trap success ranging from 0 to 0.94 (median 0.04) and geographic clusters of high and low rat abundance were observed (Fig. 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1.

a. Spatial distribution of trapped Norway rats (Rattus norvegicus) and trap success in each city block. b. Spatial clusters of relatively high and low rat abundance on Getis-Ord GI* analysis. For the GiZScore, a high z-score indicates spatial clustering of high values and a low z-score indicates spatial clustering of low values. A z-score near zero indicates no significant clustering.

https://doi.org/10.1371/journal.pone.0091654.g001

Seasonality

Of the 43 blocks included in the study, 9 were trapped in the fall, 10 in the winter, 14 in the spring and 12 in the summer. There was no significant association between season and rat abundance.

Pregnant and lactating rats, as well as sexually immature rats, were found throughout the year, suggesting that reproduction and juvenile recruitment occur year round. However, there was temporal variation in the proportion of the population composed of pregnant/lactating and sexually immature rats. The proportion of immature rats in the population was lowest in the fall, increasing through the winter, spring, and summer (Fig. 2). The proportion of pregnant/lactating rats was highest in the fall, decreasing through the winter and spring before increasing again towards the summer. Parity closely mirrored pregnancy, with the proportion of parous females also being highest in the fall and decreasing through the winter and spring before increasing in the summer.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Seasonal variation in the proportion of immature, parous, and pregnant/lactating Norway rats (Rattus norvegicus).

https://doi.org/10.1371/journal.pone.0091654.g002

After controlling for the effect of block, compared to the fall, the odds a rat being sexually immature were 4 times greater in the spring (OR = 3.68, 95% CI = 1.17–11.56) and 5 times greater in the summer (OR = 5.03, 95% CI = 1.17–21.44). There was no significant difference between the fall and winter. The odds of a sexually mature female being pregnant or lactating in the spring was 1/4 that of the fall (OR = 0.28, 95% CI = 0.10–0.82). There was no significant difference in the proportion of pregnant and lactating sexually mature females in the fall compared to summer or winter. The odds of an adult female being parous in the spring was 1/5 that of the fall (OR = 0.20, 95% CI = 0.07–0.59). There was no significant difference in the proportion of parous females in the fall compared to summer or winter. There was no association between season and number of embryos in pregnant females.

Discussion

The results of this study show that urban rat populations have a variety of ecologic features that should be taken into account when studying RAZ.

For example, our data show that that Norway rat population density varies remarkably over a short geographic distance. Not only did density vary among adjacent city blocks, but rats appeared to be distributed non-uniformly along the length of the alley. This heterogeneous distribution is presumably a result of variation in the microenvironment of a city block, particularly as it pertains to resource availability [44], [45]. Given that urban rats are territorial and have a small home range [28], [42], small spatial variations in the abundance of suitable food and harborage could create marked variation in carrying capacity across the urban landscape [21].

The heterogeneous distribution of rats across the urban landscape may, in part, be responsible for observed spatial variations in the presence and prevalence of RAZ [46]. This, in turn, suggests that clustering of rats by city block, or whatever unit best approximates the colony, is an important consideration when designing sampling and analytic protocols for research. For example, it suggests that intensive sampling of rats across a study area is likely necessary in order to achieve adequate representation of populations of interest. This approach is in contrast to many past studies of RAZ, which appear to use less rigorous and/or more opportunistic sampling methods [10], [11], [20], [23], [47]. It also suggests that it is important to control for geographic clustering of rat populations during any data analysis, which does not appear to occur in most studies of rats and their pathogens.

In contrast to observed spatial variation, relative rat abundance did not vary significantly with season. Additionally, pregnant/lactating, parous, and immature animals were present throughout the year, which suggests a lack of strict reproductive seasonality. However, relative population composition did appear to vary significantly among seasons. Specifically, the proportion of mature females that were parous or pregnant/lactating was higher in the summer and fall and lower in the winter and spring, while the proportion of sexually immature rats in the population was lowest in the fall and higher in the winter, spring, and summer. Seasonal variations in reproduction and population demography could influence the ecology of RAZ in a number of different ways. For example, behavioural changes related to mating could influence pathogen transmission, while seasonal changes in the relative proportion of immature vs. mature animals could alter pathogen prevalence [17], [48], and thus the risk of pathogen transmission to people. For example, infection with a some of RAZ, including Leptospira spp. [13], [49] and Seoul hantavirus [18], [29], is more common in older rats, suggesting that juvenile recruitment might dilute the prevalence of these pathogens in certain seasons.

Many rodents demonstrate seasonal variations in reproduction, population density, and demography, including several species of bandicoots (Bandicota spp.), mice (Mus pp), and rats (Rattus spp.) [37]. However, there has been considerable conflict among studies regarding if and how urban rat populations are influenced by season [21], [50], and it remains unclear to what degree these discrepancies are a result of methodological inconsistencies vs. true geographic or temporal variation. Given that our findings are based on a single year of study, we cannot definitively conclude that the rat populations under study demonstrate reproductive seasonality [51]. However, it appears prudent to account for season in studies of urban rats and RAZ.

Body mass in rats is often used as a proxy for age in rats, but research has shown this relationship is only approximate and often inaccurate, especially for the youngest and oldest rats in a population [15]. Additionally, previous studies have found significant variation in the growth rate of rats among different habitats, which has been attributed to some combination of genetic heterogeneity and variation in resource availability among different locations [24], [25], [52]. The results of our study, particularly the relationship between mass and length, as well as sexual maturity, support the conclusion that, although age is important determinant of mass, a number of factors independent of age (such as body fat and block of origin) can also influence mass.

Similarly, the presence and number of bite wounds were associated with a variety of demographic factors, such as sex, mass/length, and season. Specifically, bite wounds appear to be most common in longer/heavier rats and male rats, which is consistent with past behavioral studies that suggest that intra-specific aggression is most common among mature males [28]. The association between bite wounds and season, while controlling for mass/length, suggests that there might be seasonal variations in the incidence of intra-specific aggression. Past research has suggested that aggression among male rats is often related to competition for oestrus females [15]; however, the temporal relationship between wounding and pregnancy was less clear in our data. It is interesting to note that in other rodent species, intra-specific aggression also appears to be more common in older males and in certain seasons, although this may vary according to species and location [53], [54]. As with rats [29], [55], intra-specific aggression among males appears to play a role in pathogen transmission among other rodent species [53], [54].

It is of note that the relationship between mass, wounding, and other demographic and environmental factors varied depending on whether they were included in bivariate or multivariate models. This suggests that mass and wounding are complex variables, and that this complexity should be taken into account when using these variables in epidemiologic and ecologic studies. Much of the past research on RAZ has been based solely on biviarate modeling techniques [19], [22]. However, given that ecologic factors have the potential to confound one another, as well as the probability of infection with a zoonotic pathogen, it seems prudent to use more complex statistical methods in the future (e.g., multivariate models) to avoid the identification of erroneous associations.

In this study, we used two different methods to assess nutritional condition. These included a published body condition scoring system utilizing external palpation of the rat’s body [26] and a fat store system (generated for this study) using visual assessment of internal fat stores at necropsy. Scores generated using these two methods were only weakly correlated, and while fat score was associated with a number of different variables mentioned above, far fewer variables were associated with body condition score. Overall, we believe fat score to be the superior variable as it is the most direct reflection of nutritional condition and we found it easier to asses in consistent manner. We are not certain why the body condition scoring system was not successful in our study, but it could be related to the fact that this system was developed for laboratory animals (vs. wild rats, who are generally leaner) or because of our relative inexperience in using this method.

This study employed a trap-removal methodology, which has been used commonly in the past for the study of rats and RAZ [36], [50], [56]–[58]. However, one of the issues common to all trapping-based studies is that it can be difficult to distinguish population characteristics from factors that affect trapability. In our study, we were concerned that trap placement and weather (temperature or precipitation) might affect our ability to trap rats. However, we found that none of these variables were significantly associated with trap success. It was interesting that the majority of rats were caught in the first few days of trapping. Indeed, the number of rats trapped dropped remarkably after the first day with 22.2% of all rats trapped on trap day 1 vs. 14.6% on day 2. This number dropped to <1% by trap day 10. This suggests shorter trapping periods (i.e., of 3–4 days in duration) might be sufficient to collect an adequate sample of urban rats.

That being said, it is important to consider that the characteristics of rats entering the traps may change over the course of the trapping period. For example, we found that sexually mature rats, heavier rats, and longer rats were more likely to be trapped on the first trap day vs. subsequent days. Given that many zoonotic pathogens are more common in older animals [4], short trapping periods might be appropriate for detecting RAZ in rat populations. However, more prolonged trapping might be necessary to obtain a representative sample of the population for epidemiologic studies, and to detect certain pathogens that are more common in young rats (e.g., Clostridium difficile [59]).

The aforementioned changes in the characteristics of the study population over the trapping period is interesting given that a week of pre-baiting was conducted in advance of active trapping. Pre-baiting is often used in studies of rats and other rodents in order to allow animals to overcome neophobic responses to traps and bait and thus ensure that all members of trappable population are equally likely to be captured once active trapping is initiated [27], [60]–[62]. The fact that older animals appear to have been trapped first might suggest that more mature, and thus more dominant rats were excluding immature and thus more submissive rats from the food resources within the traps (therefore younger rats began entering traps only when the older animals had been removed) [15]. Many studies of urban rats do not appear to employ pre-baiting [8], [22], [30], and it would be interesting to know how this might bias the characteristics of the trapped population.

Although the specific location of a trap within an alley did not appear to greatly affect the trap success for Norway rats, it is difficult to determine if an overall focus on ground-based outdoor trapping influenced our ability to capture black rats. Among the trapped population, Norway rats comprised 95% of the captures with the remainder being black rats. Most of the black rats were captured at the port area, where trapping was predominantly conducted inside structures. In contrast, 98% of the rats trapped in the city blocks were Norway rats, where traps were set on the ground outside of buildings. Norway rats are thought to reside primarily in underground burrows [44], [63], while black rats are more likely to live in human-made structures [63], [64]. Where these two species co-exist they are thought to remain segregated, with the larger and more aggressive Norway rats excluding the smaller and more timid black rats from ground-based resources [28], [65]. For these reasons our trapping methodology in the city blocks may have resulted in underrepresentation of black rats within the study area. That being said, in many cities, Norway rats are thought to have displaced black rats completely [63] so it is possible that black rats are actually rare within the majority of the study area. It should also be noted that rats in this study were identified to species based on external morphologic characteristics; therefore ‘cryptic’ Rattus spp. may have been overlooked. These cryptic species are morphologically indistinguishable from Norway and/or black rats, and therefore can only be recognized using genetic techniques [66].

Overall, the results of this study shed some light on the characteristics of urban rat populations that could influence the ecology and study of RAZ. However, there are still a variety of aspects requiring further elucidation. For example, although we were able to show that rat population density varies by city block, further research will be needed to identify the precise reasons behind this variation. There is some indication that the heterogeneous distribution of rats across the urban landscape is a result of variations in the environment with regard to resource availability [21], [44]. However, rat behaviuor and/or genotype, could also play a role in abundance and distribution [64]. Indeed, information on the genetic structure of rat populations, which has received little attention in the past, has the potential to deepen our understanding of both rat and RAZ ecology. Additionally, although we were able to identify seasonal variation in population structure, we could not account for seasonal variations in rat behaviuor, vector ecology, or rat-human interactions, all of which can influence RAZ ecology [48], [67], [68]. Perhaps most importantly, the use of a trap-removal methodology and the fact that the study took place over one year meant that we were not able to capture a variety of longitudinal changes in rat populations, for example, inter-annual variation and changes due to anthropogenic modification in the urban environment (e.g., urban decay and re-development). Scientists are well aware that long term environmental alterations, such as climate change, are likely to have significant impacts on the ecology of wild animals and their diseases [69]. However, little attention has been paid to the changes that occur in urban ecosystems, which would seem much more dramatic and rapid, particularly from the perspective of urban rats. For this reason we suggest that cities around the world should invest in surveillance and research that aims to understand and monitor local rat populations now and into the future.

Acknowledgments

We would like to thank the City of Vancouver (Murray Wightman and Stuart McMillan), the Urban Health Research Initiative, the Vancouver Injection Drug Users Study, and the Vancouver Area Network of Drug Users for their assistance with the fieldwork.