HARVARD UNIVERSITY Library of the Museum of Comparative Zoology MC2 OCCASIONAL PAPERS LIBRARY of the MAY OAjooq MUSEUM OF NATURAL HISTORY The University of Kansas Lawrence, Kansas UNIVERSITY NUMBER 130, PAGES 1-33 APRIL 27, 1989 COMPARATIVE ECOLOGY AND SOCIAL INTERACTIONS OF NORWAY RAT (RATTUS NORVEGICUS) POPULATIONS IN BALTIMORE, MARYLAND By G. E. Glass1, J. E. Childs1, G. W. Korch1 and J. W. LeDuc2 Despite its world-wide distribution and significance as an agricultural pest and reservoir of important human pathogens, there have been few recent ecological studies of wild Norway rats (Rattus norvegicus). This is particu- larly true in the United States, where little has been done since the early work by Davis and coworkers (Davis etal. 1948; Davis and Hall 1948, 1951;Emlen etal. 1949; Davis 1949, 1951 a-c, 1955). Recent studies have tended to focus on single populations. Farhang-Azad (1977a, b), and Farhang-Azad and Southwick (1979) examined rat populations and their parasites in the Balti- more City Zoo from 1972 to 1974. Stroud (1982) studied the population dynamics of sylvatic R. norvegicus in rice fields in Northern California, and Recht (1982) used radiotelemetry in suburban California to demarcate the fine structure of home ranges, activity patterns, and movements. One of the few comparative studies was conducted by Blanchard et al. (1985) who examined wounding among rats in different Hawaiian cane fields. Results from these and other studies (e. g. Bishop and Hartley 1976; Robitaille and Bovet 1976; Hartley and Bishop 1979; Pye and Bonner 1980) show similar trends in many aspects of rat ecology but differ in some 'Department of Immunology and Infectious Diseases, The Johns Hopkins University, 615 N. Wolfe St., Baltimore, Maryland 21205. 2United States Army Medical Research Institute of Infectious Diseases, Fort Detrick, Frederick, Maryland 21701. 2 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY significant respects. It is often difficult to determine if the differences simply reflect the various methodologies used or genuine differences in the popula- tions. For example, Davis et al. (1948), using a mark-recapture approach, reported that the home range diameters for rats in urban Baltimore rarely exceeded 25 meters. Using radiotelemetry, Recht (1982), on a University campus, and Taylor (1978), in an agricultural setting, obtained home range diameter estimates of 150-600 m. The size structures of rat populations also differ markedly. Blanchardetal. (1985) reported that 25%of their populations in Hawaii were under 110 g and less than one percent exceeded 250 g. In contrast, Davis (1951b) noted that 26% of his Baltimore population were less than 150 g but 63% exceeded 250 g. Farhang-Azad and Southwick (1979) reported an intermediate weight structure in their population, with 39% exceeding 250 g and 42% less than 150 g. While some of these differences may be due to geographic variation in genetic composition, even on a local scale populations may vary in average size by as much as 57% (Davis 195 lc), suggesting that much of the variation is due to phenotypic plasticity. Dif- ferences in annual population dynamics, sex ratios, reproduction, and recruit- ment patterns are also apparent when various studies are compared (Da- vis 1949; Leslie et al. 1952; Farhang-Azad and Southwick 1979; Stroud 1982). Although Norway rats are widespread in urban centers of the world, most recent studies of rat ecology have focused on rural populations. The few urban studies have been in settings resembling rural habitats, e. g. parks or suburbs (see Recht 1982), rather than in densely populated residential centers. Simultaneous comparative studies of rats from widely varying habitats within the same circumscribed region are especially rare. In mis paper we compare patterns of population dynamics, growth, survival, body size distributions, reproduction, movement, and social interactions of Norway rats from several types of habitats in Baltimore to document the extent and potential causes of differences in the population ecology of this species. MATERIALS AND METHODS Trapping. R. norvegicus individuals were trapped from nine sites throughout Baltimore City (Fig. 1) from April 1980 through June 1986. Efforts were concentrated in two periods; April through December 1980 (11.9% of captures), and January 1984 until June 1986 (82.1% of captures). Two areas, Cherry Hill (CH) and Western Run (WR), are parklands (Plate I). CH is primarily grassland dominated by reed (Phragmites communis), aster {Aster multifloria) and goldenrod (Solidago sp.). WR is riparian habitat dominated by maple (Acer rubra) and elm, (Ulmus americana). Abundant, free-flowing water is available at both sites. Human habitation is located across a mowed lawn and street approximately 0.3 km from the trap locations URBAN ECOLOGY OF NORWAY RATS WR WG CV RH PP WA Fig. 1. Locations of the nine study sites within Baltimore City. Abbreviations for the sites are given in the text. CH and WR were parkland sites. PP (PP1, 2, and 3) and WA were low SES areas, RH and CV transitional areas and WG a moderate SES area. at WR and is separated by an intervening, heavily traveled highway at least 0.8 km from trapping sites at CH. These distances are sufficient to minimize the potential contact with commensal rats (Schroder and Hulse 1979). The seven remaining areas are urban residential sites. Rats were trapped in the central alleys and backyards of attached row houses that run the length of each block (Plate I). Residential sites differ in the economic status of the inhabitants (Olson 1976; Taylor et al. 1979). Winston-Govane (WG) is a moderate socio-economic status (SES) area with a 100% occupancy rate in predominantly single family dwellings. Home ownership is common (>70%). Little refuse was available at WG for forage, and rats found harborage under shrubs and hedges. Four sites, Washington (WA), Bradford, Port and Chase (three adjacent blocks designated PP 1 , PP2, and PP3), are low SES areas. The turnover of the human population was frequent at all low SES sites; 90% of the population rented space in dwellings (Taylor et al. 1979). Refuse was discarded in the alleys and rats relied on it for forage (Plate I). At WA only 28% of the houses were occupied and refuse was deposited at only a single corner 4 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY of the alley. Occupancy at all PP sites was high (>90%) but at PP3 half of the perimeter was occupied by an open, grass covered, schoolyard. Charles Village (CV) and Reservoir Hill (RH) are transitional areas, with a combination of single family and multi-residential housing. Areas behind dwellings in low SES and transitional neighborhoods were typically covered by concrete, although some yards contained small gardens and earth. Rats found harborage under the concrete slabs at these sites (Plate I) and occasion- ally infested the basements of houses. Rats were trapped using 12.5 X 12.5 X 40 cm wire mesh single-capture live traps baited with peanut butter. In residential areas 30 traps were placed along runways or adjacent to burrow entrances. Traps were spaced at 7-10 m and most yards were sampled. Traps were placed in the same locations during each trapping period. Traps at CH were placed in a 0.5 ha, 49 station (7 X 7) grid with traps spaced at 10 m. At WR 30 traps were scattered in sites near burrows or runways along the water course. Traps were baited and set prior to sunset. Initially, traps were collected at 0 1 00 hrs, following the major peak of rat activity (Calhoun 1962; Takahashi and Lore 1980). After November 1984 traps were checked at 0100 hrs but left open throughout the night and were collected shortly after sunrise. Capture locality, to household address, or grid station was recorded for each rat A mark-release study was undertaken at CH, WG, WA and PP3 to provide information on growth rates, survival, and movements. Tagging began in July 1984 at WG, in October 1984 at CH and PP3, and in June 1985 at WA. Captured animals were marked with serially numbered ear tags and held for several hours to ensure recovery from anesthesia before release. Recaptures of marked animals were relatively uncommon during portions of the mark- release study so that statistical estimates of rat population sizes were unsat- isfactory. Instead, indices of population size (rats/trap night) were obtained (Emlen et al. 1949). Data Collection and Analysis. Animals were transported to the labora- tory and anesthetized with a 1:10 solution of xylazine and ketamine HC1, injected intramuscularly. While anesthetized, animals were examined for wounds, which were scored on a 5 point qualitative scale (Glass and Slade 1980). Animals with no wounds were scored as 0, minor injuries such as tail wounds were scored as 1. Small bites on the flanks, near the base of the tail or shoulders were 2. Larger wounds (>0.5 cm) were scored as 3 or 4 depending on the degree of damage to underlying tissue. The scoring scheme was similar to Farhang-Azad and Southwick's (1979) except that their sixth category (fatal wounds) was not used. Differences in wounding rates between habitats, seasons, sexes and weight classes (see below) were tested using hierarchical log-linear models (Everitt 1977). Analyses were begun using main effects and statistical significance of the parameters were used as the criterion to infer their importance. The addition of higher order terms was used to test for URBAN ECOLOGY OF NORWAY RATS 5 interactions among main effects. Body weight was recorded to the nearest gram, and after July 1984, standard measurements were recorded to the nearest mm. Associations between body weightand length for males and females from each habitat were examined using Principal Components Analysis (PC A). Differences in these relationships were tested by Analysis of Covariance. External reproductive condition (testes scrotal or nonscrotal; vaginal orifice perforate or imperforate, and lactating or not) was noted. Internal reproductive data were recorded for sacrificed animals after giving anesthesia to excess. Testicular length and width and the presence of epididymial convolutions were noted for males. Epididymial convolutions were consid- ered evidence of spermatogenesis (Davis and Hall 1948; Jameson 1950; Batzli and Pitelka 1971). Testis volume was calculated assuming each testis was a prolate spheroid (V = 4/3 k (Testis width/2)2 (Testis length/2)). Litter sizes were recorded for females. Data on behavioral interactions, foraging and movements by individuals were gathered by direct observation at CV, RH, PP2 and WA (which were lit by street lamps). Observations were made by two individuals using an automobile as a blind. Animal movements and interactions could be observed to a distance of 20-25 m (Childs 1986). Two types of records were made of interactions. The first involved counts of the number of rats present in the alley during a five minute interval and their spatial grouping. Rats within 1 m of each other were considered to be 'together.' Secondly, the types of behaviors made by focal individuals were recorded. Behaviors were classi- fied as locomotory, agonistic (chasing, fighting and jumping), following (including mating), and foraging. Ten minutes were allowed to elapse between intervals to reduce the dependence of observations. Following snowfalls, active burrows and runways were identified by tracks in the snow. Runways and movements were recorded on census bureau plats of each site. During direct observations, each animal's location was noted at 5 second intervals and the path marked as the straight line distance between points. Length of travel was estimated as the straight line distance along the long axis of movement, corresponding to the method of calculating movements from the trapping data. All data were analyzed using an IBM 370 system and 1984 S AS software. All values reported in the text are mean ± one standard deviation, except as noted. RESULTS AND DISCUSSION Population Dynamics and Size Structure. A total of 760 rats were captured 913 times during 8440 trap nights. Of these, 388 were necropsied and 372 were marked and released. Six hundred and twenty-three were 6 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY captured in residential sites while 1 37 were caught in parklands. Rat captures peaked in late autumn 1984, declined over winter, and reached minimal numbers in early spring 1985 (Fig. 2). At most sites populations increased in late spring 1985 and rose towards peak levels in autumn before declining again in the winter of 1985-86. The 1985-86 decline was not as precipitous as in 1984-85. The populations reached minimal numbers in spring 1986. in >0.10, interac- tion; Table 1). Rats in the 200-299 g weight class predominated in parklands (Fig. 3), and the absence of animals in the largest weight categories was the primary reason for the smaller size of parkland rats. Trends in body length mirrored those observed with body weight (Table 1). Because of the homo- geneity of body length/body weight relationships across groups, weight was used to assign each animal to one of six 100 g weight classes up to 500 g and only patterns based on body weight are discussed. 50 i 40 30 - 20 ■ ~ 10 H > o z Ui o UJ & URBAN (n-352) ?? URBAN ( n-331) 50 - a PARKLAND 40 - (n«99) 30 - 20 - 10 ■ ?? PARKLAND (n.68) <1 3 4 ^5 <1 1 ( MASS CLASS (g/x100) 4 £5 Fig. 3. The proportion of male and female R. norvegicus in 100 g weight classes from residential and parkland sites. Males from both sites were significantly larger than females and residential rats larger than parkland rats. Note the near complete absence of parkland rats >400 g. 8 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY Table 1. Summary statistics (£ ± sd; n) for body weight (g) and body length (mm) of 853 captures for male and female Rattus norvegicus from urban residen- tial and parkland sites. Males are significantly larger than females, in both areas. Residential rats are significantly larger than parkland rats (2 way ANOVA). Habitat Body Weight Body Length Residential male female 325.3 ± 150.0; 353 294.1 ±148.1; 332 217.3 ±37.0; 247 206.3 ±34.8; 21 8 Parkland male female 232.4 ± 90.2; 99 211.3 + 88.9; 69 207.0 ±30.5; 99 202.2 ±28.5; 69 Overall, the proportions of each weight class contributing to the popula- tion varied monthly (Fig. 4). Rats in weight classes >400 g were most abundant, ranging from a low of 18% in March to a high of 47% in June. The other classes tended to make up more variable portions of the population. Animals in the smallest weight class (<100 g) were present, in low numbers, throughout the year, making up 5-15% of the population. They had abimodal abundance, peaking in the winter-early spring (February-April) and again in the autumn (October) the months surrounding peak reproduction (see below). June was the only month in which they were not captured. The intermediate weight classes (1 00-299 g) were most abundant in late summer and early autumn . The gradual increase in the 1 00- 199 g weight class during late autumn probably reflects the growth of weanlings from the secondary reproductive peak in early autumn. Rats in the 300-399 g class usually represented 15-23% of the population but reached maximal abun- dance of 37-38% in March and June prior to the onset of the two reproductive peaks (see below). Seasonal patterns in the size structures of residential and parkland popu- lations differed. Residential populations maintained stable size structures between seasons, while parkland populations varied widely (Fig. 5). The major source of variation was in the abundance of sexually immature animals. The smallest weight classes (<100 g and 100-199 g) in parklands, reached maximal abundance in the autumn and winter, when they represented >50% of the populations. In the summer their prevalence was minimal (<10%) while in the spring, their high frequency (37%) was caused by the abundance of rats in the smallest weight class (< 100 g). The largest weight classes (300-499 g) were the only groups with a nearly constant abundance throughout the year (15-18%). Norway rat populations, in a variety of habitats, undergo annual fluctua- tions with a major peak in the late spring or late autumn, followed by a trough u c URBAN ECOLOGY OF NORWAY RATS dd R. norvegicus 99 R. norvegicus >500 g 400-499 300-399 200-299 100-199 <100 g M M J J A S Time (Months) N Fig 4. Monthly distribution of weight classes for male and female R. norvegicus through the year. Animals less than 200 g were considered sexually immature. during the winter or early spring (Davis 1951a; Bishop and Hartley 1976; Hartley and Bishop 1979; Farhang-Azad and Southwick 1979; Stroud 1982). Two studies reporting major peaks in late spring also observed secondary peaks in late autumn (Farhang-Azad and Southwick 1979; Stroud 1982). Populations with spring peaks derived most of their numbers from juveniles 10 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY H d Residential □ 9 Residential | d Parkland I 9 Parkland c a> o i- a> a. Spring Summer tL y = i \uti 1 mn 5 Mass Class (g x 100) Fig. 5. Seasonal comparison of weight classes for male and female R. norvegicus from residential and parkland sites. Weight classes are <1 = 0-99 g, 1 = 100-199 g, 2 = 200-299 g, 3 = 300-399 g, 4 = 400-499 g, and 5 = >500 g. entering the population. Stroud (1982) noted two spring peaks due entirely to the entry of juveniles into the trappable population. Farhang-Azad and Southwick (1979) also found that a large portion of their population (70% of their sample) were juveniles (<200 g) during the spring peak, and that these rats recolonized burrow systems abandoned during the winter. These popu- lations differed from those described by other workers in that much of the habitat was unused when population numbers were low. It may be the availability of resources at some sites, such as parklands, permits extensive recruitment from burrows where rats overwintered. In other, more stable populations, such as those in residential areas, much of the available area remains occupied through the winter and recruitment is reduced. Differences in the annual changes in weight structure of the parkland and residential populations support this interpretation (Fig. 5). At residential sites the size structure remained constant throughout the year and the populations, URBAN ECOLOGY OF NORWAY RATS 11 though fluctuating, remained relatively more stable in numbers. In parklands only the proportion of large adults (>300 g) remained relatively constant, possibly because they could occupy the few available overwintering sites. The remainder of the population was highly influenced by seasonal factors. Growth Rates and Survival. If differences in population dynamics and size structure reflected responses to differences in habitats, this should be evident in the growth and survival of individuals. Growth rates for rats were obtained from the changes in body weight of 96 animals caught at least one month after being tagged. Overall, rats less than 200 g showed extremely rapid growth (Fig. 6), often exceeding 60 g/month. The rate of increase N = 9 32 20 22 o E DC 5 o o 70 1 60 45- 30- 25- 20- 15- 10- 5- 0 -5- -10- -15- -20- -25- <1 i 2 >5 Mass Class (x100g) Fig. 6. Mean growth rates and 95 % Confidence Interval (g/month) for each weight class calculated from recapture data. Sample sizes are shown above each class. Weight classes are defined in Fig. 5. 12 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY slowed in the 200-299 g class (30.6 g/month) and nearly ceased in the >300 g weight classes (9.5 g/month). Variation in growth rates increased in the largest weight class. Overall growth rates (Fig. 6) were similar to those reported by Calhoun (1962) and Bishop and Hartley (1976). Sample sizes were too small to examine monthly differences in growth rates between habitats. Instead, data were grouped by habitat and season for the combined sexes. Seasons were categorized as either warm (May-October; 12°-23°Cmean monthly temperature; NOAA 30 year average for Baltimore) or cool (November-April; 1°-8°C). Weight classes with similar growth rates were grouped into four categories (<200 g, 200-299 g, 300-399 g, >400 g). There were highly significant (/5<0.01) two-way interactions for habitat x season and habitat x weight class, indicating that growth patterns differed between residential and parkland populations (Table 2). Rats from residential sites grew faster than parkland rats regardless of weight class, but particularly among <200 g animals, where they outgrew parkland rats by an average of 29.4 g/month during the warm season and 57.1 g/month in the cool season. Parkland rats in the 300-399 g class, on average lost weight, regardless of season, while residential rats continued to grow (19.8 g/month). During the cool season residential rats grew more rapidly than parkland animals, but differences were not statistically significant during the warm season. Overall, parkland rats grew more slowly during the cool months than during the warm season (F=4.26, />=0.05), while residential rats by contrast grew significantly faster during the cool season (f=6.43, F<0.025). Survivorship, measured between first and last captures of marked animals, for rats recaptured at least once (including within the same trap period), showed rapid exponential declines in both habitats. (In P=- 0.40t + 0.10, ^=0.98, residential; In P=- 0.30t - 0.20, r2 =0.97, parkland; Fig. 7). Survival of tagged animals was somewhat, but not significantly, better in parklands Table 2. Growth rates (g/month) oiRattus norvegicus (X ± se; n), during warm (May to October) and cool (November to April) seasons, in residential and parkland sites. Mass Class (g) Season <200 200-299 300-399 >400 Warm Residential 66.6±5.8;18 43.5 ± 7.4; 7 2.3 ±12.0; 7 4.4 ±8.7; 9 Parkland 37.2 ±7.4; 7 34.1 ±13.8; 4 -12.0 ± 8.0;2 — Cool Residential 81.2±8.0;11 57.0 ±7.0; 2 37.3 ±17.7; 7 22.0 ±10.4; 4 Parkland 24.1 ±11.6; 5 8.2 ±4.5; 7 -8.3 ±23.0; 6 — URBAN ECOLOGY OF NORWAY RATS 13 O.OCHU • Residential x Parkland -0.50 -1.00- > £ -1.50 CO o >» = -2.00- A CO n o -2.50- -3.00- -3.50 6 7 8 Time (Months) 11 12 Fig. 7. Survival time of marked animals on the study sites. Survival followed exponential decays in both populations. The half-life and 90 % turnover time are indicated by dashed lines. (Smirnov test T, = 0.098; P>0.\0). Median life expectancy after first capture was 7 weeks in parkland and 8 weeks in residential populations. There was a 90% population turnover by six months in residential areas and by 7.25 months in parklands. Despite the rapid loss of marked animals, 1 3 individuals lived in excess of six months, three survived at least nine months, and one animal was alive when the study ended, a year after marking. Six of the eight 14 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY residential and 2 out of 5 parkland rats exceeded 250 g when first captured. Given the above growth rates, these animals were probably three to five months old at first capture and at least a year old at the time of last capture. Our median survival time of two months is similar to that reported by other workers. Bishop and Hartley (1976) and Stroud (1982), who studied popula- tions in agricultural areas similar to our parkland, found median survival rates of 1-3 months and maximal survival of 10-12 months, while Davis (1948, 1951 b, c) examined a highly provisioned population with the same results. Thus, survivorship of rats seems to remain fairly constant under a wide range of conditions. Differences in the size structure of local rat populations are not due to differential mortality but to differences in growth rates of individuals as well as possible underlying differences in migration, and reproduction (see be- low). Assuming average growth rates (Table 2), and an initial size at weaning of 40 g (Frazer 1977), residential rats reached 200 g (approximate sexual maturity) within three months, whereas it took parkland rats more than twice as long to reach the same size. As a consequence, for a given size, parkland rats tended to be chronologically older than residential rats. Evidence indi- cates that population dynamics are causally related to individual growth rates by the timing of reproductive maturity (see Reproduction). If so, growth rate variation may explain spatial differences in Norway rat population dynamics. Differences in growth rates may be due to the availability of high energy resources (i. e. refuse) in residential sites. Other workers have also noted that rats from areas with high resource abundance are significantly larger than animals from low resource areas (Leslie et al. 1952). Schein and Orgain (1953) estimated that in low SES neighborhoods there was approximately 2.5 times the amount of refuse necessary to maintain resident rat populations and that removal of refuse significantly reduced the population size (Orgain and Schein 1953). Davis (1949) raised animals from rural and urban areas in the laboratory and found that, with equal resources, there were no differences in growth rates or body sizes despite the observed differences among field populations. This suggests that much of the body size variation among local populations is due to phenotypic plasticity in growth rates. Seasonal differences in growth rates between habitats may be associated with the timing of resource abundance. Grasses, forbs and arthropods were the major food items for parkland rats and were particularly abundant during the warm season. Refuse was available to residential rats the entire year but did not spoil as quickly during the winter. Rotten foods are generally avoided by rats (Schein and Orgain 1953) so the amount of available forage during the warm season may be less than during the cool season. In addition, population densities were higher during the warm season and access to food may have been restricted by dominant population members (see below) (Orgain and Schein 1953). URBAN ECOLOGY OF NORWAY RATS 15 Sex Ratios. Overall, 53.1% of the rats were males, which does not differ significantly from unity (%^3.23; \df, P>0.05). However, there were differ- ences in sex ratios across habitats. In residential sites males made up 51.5% (n=683) of the population, while they comprised 59.3% (n=167) in the parklands (tf=5.15\ ldf, /><0.025). Seasonal trends were similar in both areas with an excess of males in late winter and early spring (Fig. 8), followed by an increase in the proportion of females, with a peak in June. During the remainder of the year, the sex ratio showed a consistent, but nonsignificant, excess of males. 80- 70- 60- (0 50-- 40- 30- 20 T J F M AMJ J A SOND Time (Months) Fig. 8. Monthly changes in the proportion of trapped male R. norvegicus. Numbers in brackets are sample sizes and vertical bars are standard errors. 16 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY Females predominated in smaller weight classes (<300 g) in residential sites, whereas there was an excess of males in the larger classes (x^B.87, /><0.025) (Table 3). In parklands, females also appeared to dominate the smallest weight class while males were at greater frequencies in the remaining groups, but the differences were not significant (x^S.98, P>0.10). Table 3. Proportion of male Rattus norvegicus by weight class, in populations from residential and parkland study sites. Numbers in parentheses are sample sizes. Habitat <100 Body Weight (g) 100-199 200-299 300-399 400-499 >500 Residential Parkland 0.44(70) 0.48(144) 0.39(95) 0.57(160) 0.59(153) 0.56 (66) 0.38(16) 0.57(52) 0.64(67) 0.57(30) 1.00 (3) — The excess of males in parkland populations supports the view that they represent 'colonizing' populations. Male rats tend to disperse farther and more frequently than females (Calhoun 1962). The seasonal trends are consistent with the overall pattern of a minor excess of males, except during late winter and spring (Fig. 8), when a large excess of males was followed by an excess of females. This pattern was also observed by Kendall (1984) and may be a trapping artifact; pregnant females reduce activity outside the burrow late in gestation, and males become more active while searching for mates (Calhoun 1962). The peak in male abundance corresponded with the peak in spring reproduction (Fig. 9) and more females were trapped as reproduction declined during early summer. Most studies (e. g. Davis 1951; Davis and Hall 1951; Bishop and Hartley 1976; Stroud 1982) have observed an excess of males although Leslie et al. (1952), and Farhang-Azad and Southwick (1979) reported a higher propor- tion of females. The latters' trapping methods were more effective at sam- pling animals remaining near burrows and may account for the observed differences. Reasons for the weight associated changes in sex ratios, particu- larly in the residential populations, are unclear. Some workers have shown that at or shortly after birth the sex ratio is slightly, but consistently, skewed towards males. Leslie et al. (1952) reported 51.9% of 509 offspring taken from nests were males. Farhang-Azad and Southwick (1979) also noted that 51% of animals < 50 g were males. But Bishop and Hartley (1976) and Davis (1951c) reported shifts towards an excess of males in larger weight classes, while females predominated in weight classes less than 200 g. Males reportedly grow more quickly than females (Calhoun 1962), so such a pattern is expected. In our samples, males weighed 10% more than females (Table 1) but we could not show sexual differences in the growth rates with our relatively small samples. URBAN ECOLOGY OF NORWAY RATS 17 0-....0 9 9 A- ---A Pregnant 0) o ■a o 100- 90- 80- 70 60- 50- 40- 30- 20- 10- V / J ' F ' M A M J S O N D Time (Months) Fig. 9. Monthly changes in the proportion of reproductive animals trapped. Open circles (females) and closed circles (males) are percentages of perforate and/or lactating females and scrotal males observed. Triangles represent the proportion of pregnant females determined from necropsy. Reproduction. External examination indicated reproduction occurred in males from all weight classes and during every month of the year. Overall, 84.0% (n=424) of males had scrotal testes when captured. There was a significant difference in the proportion of scrotal males among weight classes (subdivided into <100 g, 100-199 g, > 200 g; %2 = 183.56; 2df, F<0.0001). Males less than 200 g had descended testes less often than larger males. Males less than 100 g were occasionally scrotal (9/32) but six of these were 92-99 g, and the three others had unconvoluted epididymides when necropsied. There was probably no active spermatogenesis in these individuals (see below). A greater proportion of the 100-199 g class males from residential sites were scrotal (61%) than were their parkland counterparts (33%) (%2=6.20; Idf, P<0.025). There were no significant differences between residential and parkland rats in any other weight category. 18 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY The frequency of scrotal males varied seasonally (Fig. 9), decreasing in late autumn and early winter and peaking in late winter and spring. This was due to annual changes in the proportion of smaller individuals and in their reproductive status in the population. Nearly all males > 200 g (99.0%) were scrotal, while smaller males, which represented an increasing fraction of the population in autumn and early winter (Fig. 4), were frequently nonscrotal during these seasons. As small individuals disappeared or matured in late winter and early spring, there was a concommittant rise in the proportion of scrotal males. Smaller males were the only group to show seasonality in reproductive condition. Although 81% (n=26) captured between June and August had descended testes, only 50% (n=24) were scrotal in the spring, and only 30.4% (n=69) were scrotal during the autumn and winter. Necropsies of 97 (23 parkland, 74 residential) males complemented the results obtained from external examination. Among males greater than 200 g, convoluted epididymides were present in all scrotal individuals and both nonscrotal males examined (Fig. 10). Only 3/n of the males <200 g had convoluted epididymides and all had scrotal testes. Of the remaining 14 nonreproducuve males, six were scrotal. Overall, males had a mean testicular volume of 1.53 ± 0.71 cc. Testis volume increased as a function of body weight up to approximately 300 g and then leveled off at approximately 1.7 cc. Males less than 200 g had testicular volumes less than 0.6 cc. The relationship between body weight and sperma- togenesis held for both parkland and residential rats. Despite the greater age of 200 g parkland rats, they failed to become sexually mature until they reached this weight (Fig. 10). External examination suggested reproduction among females of every weight class during every month of the year (Fig. 9). Eighty-two per cent («=382) of females had perforate vaginal orifices or were lactating. Females less than 100 g were rarely perforate (u/46) and never lactating. Of 95 females between 100-199 g, 66.3% had perforate vaginal orifices but none were lactating or pregnant. All but three females greater than 200 g were perforate (n=241) and 37.1% were lactating. Lactating females were rarely observed during the late autumn and winter. Lactation peaked in April and May and again in September and October (Fig. 9). Females at residential sites showed external evidence of reproduction at a larger average body weight than parkland females. Only 16% of residential females < 100 g were perforate while 56% of parkland females from this class were perforate (%2=6.17; Idf, P<0.025). The proportion of residential females between 100-199 g with perforate orifices was also lower than in parklands (63% vs 77%), but the difference was not significant (x2=2.62; Idf, P>0.25). The three nonperforate females >200 g were from residential sites. Necropsy data from 193 females supported observations based on external examination. No female less than 200 g (n=60) was pregnant, regardless of URBAN ECOLOGY OF NORWAY RATS 19 3.0 2.4 o I 1.8 O > = 1.2 to a) 0.6- 0- *. •• .%:. A* I * ••• i 000° °A O 100 200 300 Body Mass (g) 400 500 600 Fig. 10. Relationship between testis volume and body weight for 97 R. norvegicus. Closed circles represent residential males with convoluted epididymides and open circles represent residential males with unconvoluted epididymides. Closed triangles are parkland males with convoluted epididymides and open triangles are parkland males with unconvoluted epidi- dymides. vaginal condition. Overall, 42.1% (aj=133) of females >200 g were pregnant (Fig. 11). The proportion of pregnant females between 200-499 g was relatively constant (200-299 g,41.7%; 300-399 g, 37.8%; 400-499 g 33.3%) X2=0.5 1; 2df, P=0.98). Among females >500 g 70.8% (n=24) were pregnant. The proportion of pregnant animals among 200-399 g parkland rats was significantly higher than among residential rats. Eighty percent (ai=10) of parkland females were pregnant compared to 30.8% (n=52) of residential females (%2=9.03; Idf, P<0.005). This accounts for the slightly higher frequency of pregnancy among 200-399 g rats, inasmuch as no parkland female exceeded 400 g. The prevalence of pregnancy in residential rats did not approach that of parkland rats unul residential females exceeded 500 g. Pregnant females were found throughout the year, although reproduction decreased during the summer and winter (Fig. 9). Two breeding peaks were noted, one in April and the other in September. Between these times breeding was minimal in June. The peaks in pregnancies were due to seasonal breeding in <400 g females. Rats in larger weight classes tended to be pregnant throughout the year at a relatively constant rate while females <400 g were pregnant primarily between May and September. Only 3 of 14 of females less than 400 g were pregnant between October and April while 21 of 47 were pregnant between May and September (%2=3.81; Idf, P=0.05). In larger weight classes the 20 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY O c 0) 3 200-99g (N = 24) 300-99g (N=37) 400-99g (N=48) >:500g (N = 24) T 1 4 6 8 10 12 14 t Litter Size Fig. 1 1 . Distribution of litter sizes among four female weight classes. Females <200g were never pregnant. Arrows indicate the median litter size for each weight class. URBAN ECOLOGY OF NORWAY RATS 21 comparable values were 11 of 25 and 22 of 47 (x2=0.60; idf, P>0.50). Seasonal pregnancy patterns did not appear to differ between habitats, though sample sizes were too small to statistically evaluate the data. Litter sizes also followed a weight-specific pattern. Mean litter size was 10.49 ± 0.45 (Jc± se, n=57; range 1-16). Litter sizes increased with female body weight, up to 500 g, and then declined (Fig. 1 1). Despite this, the largest litters were found among females in the 300-399 g weight class. Females >500 g had a broad range of litter sizes and, consequently, reduced average litter size. The small litters in this class tended to be found in females with all embryos in a single uterine horn. Usually, the other horn was reduced and thread-like. Median litter sizes were the same for 200-399 g females from both residential and parkland populations (median=10). Sexual maturity appears to be size — rather than age-dependent. In our populations 200 g seems to represent a physiological threshold necessary for reproduction. Males have convoluted epididymides only after reaching 200 g. Although residential males are scrotal at lower weights they apparently cannot breed. These results are consistent with Davis and Hall's (1948) observations. They noted that males became scrotal (at 105 g) well in advance of producing spermatozoa (200 g). They also reported there was no seasonal testicular regression so males were reproductive throughout the year, but smaller males tended to be scrotal more often during the summer than the winter. Similarly, females do not become pregnant until they reach at least 200 g, despite having perforate vaginal orifices. These results support Davis' (1949, 1951c) conclusions that rats in rural areas mature more slowly than those in urban environments. Recent studies examining the effects of exercise and diet on the onset of puberty in laboratory rodents support the hypothesis that body weight rather than age is the critical factor in reproductive maturity (Perrigo and Bronson 1983; Bronson 1987). Comparisons of the age and size at first ovulation for ad lib fed rats and slowly growing, exercising rats showed a delay of nearly 2.5 months in onset among exercising rats until they reached the same weight as control rats (106 vs 1 12 g). This size threshold for maturity profoundly affects population dynamics, because the slower growth rates of parkland rats mean fewer animals survive to reproduce (Slade et al. 1984; Sauer and Slade 1987). As a result, migration may become a relatively more significant factor in maintaining these populations. The size-dependent relationship between litter size and body weight observed in this study is typical of rats (Davis 1951c; Leslie et al. 1952; Farhang-Azad and Southwick 1979; Hartley and Bishop 1979). In most populations litter size is an increasing function of female body size. In a few studies a small proportion of females < 200 g were pregnant (Leslie et al. 1952; Farhang-Azad and Southwick 1979) but none were observed in this study. Although females < 200 g have been considered reproductively mature 22 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY by some workers (e. g. Leslie et al. 1952; Farhang-Azad and Southwick 1979) because of perforate vaginal orifices or the presence of corpora lutea, they contributed little to these populations' reproductive outputs. Determining the size at which females reach sexual maturity has a significant impact on estimated fecundities. For example, Davis (1951c) considered females < 200 g to be sexually immature and calculated a mean litter size of 10.1 young/ female, close to our value of 10.5. Farhang-Azad and Southwick (1979) used all females > 139 g (based on corpora luteal counts) in their estimates of average litter size (8.7). However, when their data are stratified by the female's weight class (their Table III), their medians and quartiles do not differ from ours. Similarly, Leslie et al. (1952) noted that altered criteria for sexual maturity produced large differences in productivity estimates within the same population. The bimodal, seasonal pattern of reproduction (Fig. 9) is characteristic of most rat populations, with both midsummer and winter declines in fecundity (Davis 1951c; Davis and Hall 1951; Leslie etal. 1952; Calhoun 1962;Bishop and Hartley 1976; Pye and Bonner 1980; Stroud 1982). Leslie et al. (1952) noted peaks in late spring and autumn. Davis (1951c) observed a decrease from a peak pregnancy rate of 45% in March-April to a low of 21 % in early summer followed by a secondary peak in autumn. Farhang-Azad and South- wick (1979) described bimodal reproductive peaks in April, and September or October. Our data indicate that seasonal reproduction occurred primarily in smaller rats (<400 g) while larger animals bred throughout the year. Davis and Hall (1951), and Leslie et al. (1952) also noted seasonal breeding by smaller animals, while larger females bred throughout the year. The midsummer breeding depression may be due to the large influx of nulliparous young females born in the spring (Fig. 4) that have reached sexual maturity but have not become pregnant, while the second autumn reproductive peak occurs as they are recruited into the breeding population. The winter decline in breeding may represent a delay in the maturation, because of environmental condi- tions, that often characterizes the autumnal cohort of young rodents (Calhoun 1962). The proportion of pregnant females remained relatively constant with body size, ranging from about 35^0%, except among females >500 g, which were nearly always pregnant. These results are similar to those of Leslie et al. (1952) and Davis (195 lc) who reported pregnancy rates of 37.9% and 30.4%, respectively. Median litter size increased with female weight up to the largest weight class (>500 g) and then declined, suggesting that, despite the high pregnancy rate, these females may be approaching reproductive senescence. Davis (1951c) also found a decline in litter sizes among very large females, although Leslie et al. (1952) observed the decrease only among their 'non- rick' females. Farhang-Azad and Southwick (1979) reported that litter size URBAN ECOLOGY OF NORWAY RATS 23 continued to increase even in their largest animals. However, in their study the proportion of pregnant females in this class declined dramatically compared to the rest of the population. Movements. We recaptured 107 of 372 (29%) marked rats 153 times. Distances between adjacentcaptures were significantly greater in parkland (3c = 24.7 ± 4.55 m; 95% Confidence Interval, n=64) than in residential areas (3c = 13.5 ± 5.17 m, n=77). Most distances between recaptures were relatively short, particularly in residential areas (Fig. 12). Residential rats were recap- tured 27.3% of the time at the same location. In contrast, in the parkland only 7.8% of ratrecaptures were at the same trap sites. There were several instances of relatively long-distance movements. Two rats moved more than 85 m in residential sites, spanning the trapping areas. Movements based on tracking after snowfall, in alleys, were significantly longer than those based on capture data (21.6 ± 3.61 m, ai=39). These movements were direct, going from a burrow or brush pile, along fence rows and other cover, to a food source and returning. There was little evidence of exploratory behavior or other movements. Data from direct observations of individuals were not significantly different in the mean distance traveled, compared with tracking, but were slightly longer (22.4 ± 7.33 m; /i=73) and there was more exploratory behavior (see below). These data probably under- estimate the real distance travelled because of difficulties in observation. Some rat movements were much farther than indicated by the mean distance travelled. Four individuals traversed the length of the alleys during a single foray — a distance of approximately 90 meters. One individual travelled 165 m, during 15 minutes, moving the length of the block, crossing a side alley, the main street, and moving part way down the next block before it was lost going under a fence. Rats crossed major thoroughfares, although levels of movement were lower than within alleys. During 21 hours of censusing at one site, a mean of 0.5 rats/hour crossed the street, while an average of 41.5 rats/hour traversed the alley. However, these rates varied depending on the levels of rat and human activity. The movement data shed some light on differences noted by other workers. Distances traveled in residential areas (based on trapping (Fig. 12) were short, typically less than 15 m. This matches the results of Davis et al. (1948) who, using various methods, found 96.3% of movements were less than 25 m. Visual observations and tracking in snow indicate that trapping results underestimate the linear distance rats move, and fail to reveal the tendency for rats to traverse entire alley systems. As such, they support Taylor's ( 1 978) and Recht's (1982) criticisms that trapping seriously underestimates rat move- ments. However, observations and tracking also indicate most movements in residential areas are short and remain within alleys while crossing major streets is relatively rare. As such, residential rats are probably restricted to much smaller areas than rats in parkland environments. 24 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY Residential N = 77 Parkland N = 64 6 >7 Distance (x 10m) 6 >7 Fig. 12. Distributions of distances between recaptures for residential and parkland rats. Median distances are indicated by arrows. Parkland rats moved significantly farther than residential rats. Even if trapping underestimates movements, comparisons of parkland and residential sites show that rats move farther in rural than in urban areas (Fig. 12). Other workers have noted that individual movements are labile and frequently a function of the distance between the burrow site and food. If food is readily available rats move very little while other individuals in the same population may move considerable distances to forage (Davis et al. 1948; Taylor 1978; Hardey and Bishop 1979; Pye and Bonner 1980). If food availability is a major determinant of the distance rats travel, movements in urban areas may be shorter than in rural settings due to the presence of refuse. URBAN ECOLOGY OF NORWAY RATS 25 Most rats in our residential sites had to move less than 20 m from their burrows to reach food. Parkland rats relied on grasses and arthropods for food. Grasses, in particular, are nutritionally inferior to the refuse available in residential areas (Schein and Orgain 1953) and may require parkland rats to move further or more often to obtain resources. Wounding. Wounding status was scored for 702 rats. Patterns of wound- ing showed significant sexual, seasonal and weight based differences (Fig. 13). Fifty-two percent of residential males, 42.3% of residential females, and 25.0% and 9.4% of parkland males and females, respectively, were wounded. There was no statistically significant difference in wounding patterns be- tween residential and parkland rats (x2=6.01; Adf, P=0.20) after sexual, seasonal and size factors were considered. The apparently lower prevalence of wounding in parklands appears to be due to differences in weight structure. Therefore, both habitats were combined for further analysis. The effects of the remaining variables on wounding were mutually independent, as indicated by the lack of statistically significant pairwise and higher order terms. Males had significantly more severe wounds than females (x2^ 12.98; Adf, /^O.Ol). Severe wounds (grades 3^4) were found on 15.9% of males, but only 10.5% of females were seriously wounded. Wounding prevalence increased > o z UJ o LU DC oV 6 29 38 21 CSSSSSg PARKLAND ?? 22 20 13 m mm < 1 1 2 >3 <1 RESIDENTIAL 27 58 35 71 70 24 > 3 ?? GRADE 4 GRADE 3 GRADE 2 GRADE 1 GRADE 0 32 69 46 48 41 23 £ 5 MASS CLASS (X 100 9> Fig. 13. Frequencies of wounding for male and female rats from residential and parkland sites stratified by weight class. There were no significant differences in wounding rates between habitats within sexes or weight classes. Weight classes are defined in Fig. 5. 26 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY steadily with weight in both sexes. Prereproductive (<200 g) individuals had lower wounding rates than sexually mature rats. Two males < 200 g («=121) had body wounds and only six had minor tail wounds. Nine females < 200 g (n= 133) had body wounds. Both sexes had approximately equal wounding in the 200-299 g weight class (Fig. 13). Wounding increased dramatically among males in the larger classes so that 84.2% of males >500 had body wounds and more than half had grade 4 wounds. In contrast, less than half of all females in the largest weight classes had body wounds and only two had grade 4 wounds. Instead, most (60.7%) were tail or minor rump wounds. Seasonally, serious wounding occurred at significantly higher rates during the summer and autumn (37.2% of animals) than during the winter and spring (10.1%) (x^ 10.85; Adf, T^O.03). Particularly among females, wounding was nearly absent during the spring, and was elevated during the summer among animals in the 300-499 g weight classes (1 of 30 versus 12 of 16 animals). The choice of 200 g as a criterion for characterizing mature animals is consistent with the data on wounding patterns. Adult rats are rarely aggressive towards sexually immature individuals. However, as they approach sexual maturity, aggressive encounters with socially dominant animals increase (Calhoun 1962; Blanchard et al. 1985). In our study, animals < 200 g were rarely wounded and were never observed involved in prolonged fights (see Behavioral Interactions). Their wounds were rarely severe. At about 200-299 g wounding increased in both frequency and severity. Wounding increased dramatically in larger males, presumably due to inter- male conflict for access to breeding females (Calhoun 1962). Although none of the attacks could be directly ascribed to this, only a few fights involved access to food or other identifiable resources. Severity of wounding became so widespread in large males (Fig. 13) that it represented a significant risk to individuals (Calhoun 1962; Farhang-Azad and South wick 1979). Unlike Blanchard et al. (1985), we noted a strong sexual difference in wounding severity consistent with description from enclosed populations (Calhoun 1962). The decrease in wounding among males during the winter may represent decreased intermale conflict as female reproduction and population size declines overwinter. Causes of wounding in females have received less attention in the litera- ture. However, despite its lower severity than in males, wounding occurred in a significant proportion of the females (Fig. 13). Male rats are, generally, unaggressive towards females (Calhoun 1962; Blanchard et al. 1984). Dominant females reportedly attack unknown males and may be bitten in return, but this is rare (Blanchard et al. 1984), implying that most wounds are due to interactions with other females. Increased wounding rates during the summer may represent increased conflict among reproductive females as populations increased from winter and spring lows. This is supported by the URBAN ECOLOGY OF NORWAY RATS 27 restriction of increased wounding rates to the 300-499 female rats, the major reproductive classes. Behavioral Interactions. Observations were made at four residential sites, and involved 964 observations of 1 130 rats in 92 hours. Rats were seen as lone individuals in 87.0% (n=839) of all observations, in pairs in 10.6% (n=105) and in trios or larger groups in 2.1% (n=20) of the cases (Table 4). Table 4. Behaviors and group size of Ran us norvegicus viewed in alleys. Number of Rats Observed* Behavior 1 2 3 >4 Total Locomotor 817 10 9 3 839 (87.0) Foragingb 21 4 1 2 28 (2.9) Agonisticc — 78 3 — 81 (8.4) Folio wingd — 13 2 1 16 (1.7) Total 838(87.0) 105(10.9) 15(1.4) 6 (0.6) 964 Observations of >1 rat includes only animals seen together as a group. bIncludes feeding, drinking, and food carrying. Includes chasing, fighting, leaping, and displacing, deludes passive following and mating. Overall, 87.0% (n=838) of all behaviors were classified as locomotory (Table 4). Locomotory behaviors usually involved single rats moving across an alley (a distance of 3>-A m), and were typically of short duration (<3 seconds). Movements across alleys were rapid, involving a 'galloping' gait (Ewer 1971; Recht 1982). However, on occasion, animals would amble and pause for bouts of grooming which typically lasted 0.5-3 minutes. Move- ments, along the entire alley (above), took 5-20 minutes, and periods of rapid movement were interspersed with pauses, grooming, and exploration of the immediate surroundings. Agonistic encounters were the most commonly observed interactions (Table 4). Chasing was the most common agonistic behavior (ai=28 pairs) and involved rapid movements, without prolonged contact, accompanied by frequent squealing that could be heard from a distance of 20 m. Chases lasted from a few seconds to more than a minute. Short duration chases usually terminated with the chased animal escaping into a burrow. Longer chases usually involved access to a resource, such as food, that neither animal would abandon. The pursuing animal was in all cases the larger of the pair. One 254 g female rat maintained sole access to a food pile for more than 90 minutes by repeatedly chasing four small juveniles that approached her while she fed. Fighting («=21 observations) and displacing (n=20) occurred at similar 28 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY frequencies during agonistic encounters. Fighting involved actual physical contact, squealing, tumbling, and apparent biting; it frequently terminated in chases. Fights normally lasted less than 10 seconds, although one lasted more than a minute with periods of rest interspersed between fighting. Submissive behaviors, other than escapes, were never observed. Displacing behavior occurred when a rat approached a second individual and the second rat abandoned his/her location and left the area. The movements of both animals were unhurried and the approaching rat did not chase the displaced animal. Jumping or leaping behavior was seen on five occasions. Typically, one rat would leap with all four legs off the ground as a second animal approached. This behavior is similar to Robitaille and Bovet's (1976) 'startling jump,' except that touching by the second animal did not always occur. Leaping is reportedly an escape from aggressive behavior (Ewer 1971; Robitaille and Bovet 1976; Hart 1982). The jumping animal immediately fled the area, often precipitating a chase by the other rat Following behavior was differentiated from agonistic behavior, especially chases, by the slowness of the gait and the absence of squealing. Seven of the 16 following observations involved pairs of large adults. In each case illumination was sufficient to establish that the following animal was a male and the leading animal probably a female (indicated by large size and the absence of scrotal testes). This form of following behavior may be related to courtship (Calhoun 1962; Ewer 1971; Hart 1982), but mating was only seen once. It involved a single male following at a female's flank and mounting twice. Each mounting was followed by the male licking his genitals (Ewer 1971; Dewsbury 1975). Following was also observed among small rats. Five pairs of juveniles (<100 g) emerged from burrows and jointly explored areas adjacent to the burrows. Groups of three or more rats formed the remaining observations. In all cases the animals were small (<100 g) and engaged in exploratory behavior. This observation was supported by trapping. In three instances we captured three individuals in a single trap and all the animals were <100 g. This could only have occurred if the animals were following very closely. Foraging behavior involved actual feeding, drinking or carrying of objects in the mouth. Most observations of foraging (21 of 28) involved single rats (Plate I, Table 4) but occasionally, large groups of rats (6 or 7) fed from a single refuse pile. Generally, there were no overt interactions among individuals at such times, but five chases resulted from individuals attempting to preempt food. Rats were most frequently observed feeding on and carrying refuse items, although animals also ate fallen mulberry fruits. Foraging on garden plants and grasses was indicated by piles of clipped vegetation at burrow entrances. Rats drank from pools of water that accumulated in the alleys during rainstorms or following human activity (car washing, gardening, etc.). Drinking, or at least the lowered posture associated with it (Ewer 1971), URBAN ECOLOGY OF NORWAY RATS 29 occurred for several minutes without interruption. Observations indicate that away from burrow systems most rats acted as solitary individuals with little direct social interaction (Table 4). Most interactions that did occur were agonistic, and while most did not involve actual physical contact, fighting occurred in a high proportion of cases. Fights were of short duration and typically occurred between nearly equally sized animals. Small rats were never seen fighting, although they were occasionally chased. These data are consistent with Robitaille and Bovet's (1976) obser- vations of rat interactions at a landfill, where rates of agonistic encounters were high. Also consistent with their observations was the absence of submissive behaviors (other than retreating from chases, and displacement), that characterize interactions among laboratory rats (e. g. Hart 1982). This lack of complex submission behaviors may be due to the inability of subordinate rats to escape under laboratory conditions, or because more elaborate submissive behaviors only occur among members of the same burrow, which we could not observe. Escape may be the only submissive response between rats from different burrows. The wounding data are also consistent with the interpretation that complex appeasement behavior may not occur among free-ranging rats. CONCLUSION This study suggests that many of the reported differences in the character- istics of rat populations primarily reflect plastic phenotypic responses to the environments; methodological differences among workers are of secondary importance. Direct comparisons of population composition, movements, and survival and growth rates from different habitats within the same geographic area show the broad range of responses of a very labile species. Individuals in parkland populations grow more slowly, are smaller sized, mature later, live at lower densities, are more vagile, but live as long as rats from residential sites. Many of these factors may be directly related to the growth rates of individuals and as such may reflect variation in resource availability and quality. Future studies should focus on quantifying the availability and distribution of resources in differing habitats and relating them to the dynamics of the populations. ACKNOWLEDGMENTS Financial and logistical support for this study was provided by contract D AMD-894-C-40 1 5 to J . E. Childs. The views of the authors do not purport to reflect the position of the Department of the Army or Department of Defense. We wish to thank the people of Baltimore, especially the residents of our study areas for their understanding and patience. We appreciate the 30 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY assistance of the Department of Parks and Recreation for providing permis- sion to use various city locations for our study. Animals were handled in accordance with NIH Guidelines "Guide for the Care and Use of Laboratory Animals" and were trapped under permits from the Maryland Department of Natural Resources. David E. Davis, Robert D. Holt, and two anonymous reviewers read and greatly improved earlier drafts of this paper. SUMMARY Norway rat (Rattus norvegicus) populations were studied at nine sites in the city of Baltimore by a combination of removal and mark-recapture methods. Trapping locations varied in the density and socioeconomic status of human habitation, ranging from rural, parkland habitat to high density, low income, urban areas. Rat population sizes varied annually with major peaks of abundance in autumn. Parkland numbers were smaller than in residential areas, except in spring when an influx of animals resulted in a temporarily large population. Populations in parklands appeared to be more influenced by local movements of animals than by in situ recruitment. Rats moved nearly twice as far (3c=25 m) between recaptures in parklands as in residential areas (3c=14 m). Direct observation of residential rats showed movements were primarily restricted to a single alley system and generally involved move- ments from a burrow to a food source (i. e. refuse). Animals in parklands were significantly smaller than those in residential sites. Survival rates of marked animals did not differ among sites, but residential rats grew more rapidly and stopped growing at larger body weights than parkland rats. Rats did not become sexually mature until a body weight of 200 g, based on necropsy data, in both habitats. Litter size averaged 10.5 young/pregnant female and 42% of females were pregnant. Pregnant rats were more common in parklands than in residential areas. Mean litter size was maximal for female body weights of about 300 g. Reproduction was seasonally bimodal with peaks in spring and autumn. Bimodality was due to seasonal breeding in smaller animals; individuals >400 g bred at a constant rate year round. Wounding followed seasonal, sexual and weight-specific patterns. Sexually immature animals were rarely wounded. Wounding increased with size for both sexes but was more severe among males than females. Wounding tended to decrease during the late winter when populations and reproductive rates were at their lowest. Females 300-499 g had their highest rates of wounding in late summer as population sizes and reproduction began to peak. Wounding rates were consistent with observations of behavioral interactions. More than 16% of all behaviors were agonistic, and actual Fighting represented more than 5% of recorded observations. Many of the differences reported in rat population biology from throughout the species' range were observed within the various populations in Baltimore and may be due to phenotypic responses to resource availability. URBAN ECOLOGY OF NORWAY RATS 31 LITERATURE CITED Anderson, P. K. 1970. Ecological structure and gene flow in small mammals. Symposia of the Zool. Soc. Lond., 26:299-325. Batzili.G. O. andF. A. Pitekla. 1971. Condition and diet of cycling populations of the California vole Microtus calif ornicus . J. Mammal., 52:141-163. Bishop, J. A. and D. J. Hartley. 1976. The size and age structure of rural populations of Rattus norvegicus containing individuals resistant to the anticoagulant poison warfarin. J. Anim. Ecol., 45:623-646. Blanchard, D. C, C. Fukunaga-Stinson et al. 1984. Dominance and aggression in social groups of male and female rats. Behav. Proc., 9:31-48. Blanchard, R. J., L. Pank et al. 1985. Conspecific wounding in free-ranging R. norvegicus from stable and unstable populations. Psych. Rec., 35:329-335. Bronson, F. H. 1987. Puberty in female rats: relative effect of exercise and food restriction. Am. J. Physiol., 252:R140-R144. Calhoun, J. B. 1962. The Ecology and Sociology of the Norway Rat. U. S. Dept. Health, Education and Welfare, Public Health Serv. Publ., 1008:1-288. Childs, J. E. 1986. Size dependent predation on rats (Rattus norvegicus) by house cats (Felis catus) in an urban setting. J. Mammal., 67:196-199. Davis, D. E. 1948. The survival of wild brown rats on a Maryland farm. Ecology, 29:437-448. Davis, D. E. 1949. A phenotypical difference in growth of wild rats. Growth, 13:1-6. Davis, D. E. 1951a. The relation between level of population and pregnancy of Norway rats. Ecology, 32:459-461. Davis, D. E. 195 lb. The relation between level of population and size and sex of Norway rats. Ecology, 32:462-464. Davis, D. E. 1951c. A comparison of reproductive potential of two rat populations. Ecology, 32:469-475. Davis, D. E. 1955. Social interactions of rats as indicated by trapping procedures. Behaviour, 8:335-343. Davis, D. E., J. T. Emlen, and A. W. Stokes. 1948. Studies on home range in the brown rat. J. Mammal., 29:207-225. Davis, D. E. and O. Hall. 1948. The seasonal reproductive condition of male brown rats in Baltimore, Maryland. Physiol. Zool., 21:272-281. Davis, D. E. and O. Hall. 1951 . The seasonal reproductive condition of female Norway (brown) rats in Baltimore, Maryland. Physiol. Zool., 24:9-20. Dewsbury, D. A. 1975. Diversity and adaptation in rodent copulatory behavior. Science, 190: 947-954. Emlen, J. T, A. W. Stokes, and D. E. Davis. 1949. Methods for estimating populations of brown rats in urban habitats. Ecology, 30:430-441. Everitt, B. S. 1977. The Analysis of Contingency Tables. Halsted Press, London. Ewer, R. F 1971. The biology and behavior of a free-living population of black rats (Rattus rattus). Anim. Behav. Monogr., 4:127-177. Farhang-Azad, A. 1977a. Ecology of Capillaria hepatica (Bancroft 1893) (Nematoda). I. Dynamics of infection among Norway rat populations of the Baltimore Zoo, Baltimore, Maryland. J. Parasitol, 63:117-121. Farhang-Azad, A. 1977b. Ecology of Capillaria hepatica (Bancroft 1893) (Nematoda). II. Egg- releasing mechanism and transmission. J. Parasitol., 63:701-706. 32 OCCASIONAL PAPERS MUSEUM OF NATURAL HISTORY Farhang-Azad, A. and C. H. Southwick. 1979. Population ecology of Norway rats in the Baltimore Zoo and Druid Hill Park, Baltimore, Maryland. Ann. Zool., 15:1-42. Frazer, J. F. D. 1977. Growth of young rats after birth. J. zool. London, 183:556-559. Glass, G. E. and N. A. Slade. 1980. The effect of Sigmodon hispidus on spatial and temporal activity of Microtus ochrogaster: evidence for competition. Ecology, 61:358-370. Hart, M. 1982. Rats. Allison and Binsby, London. Hartley, D. J. and J. A. Bishop. 1979. Home range and movement in populations of Ratlus norvegicus polymorphic for warfarin resistance. Biol. J. Linn. Soc, 12:19-43. Jameson, E. W. 1950. Determining fecundity in male small mammals. J. Mammal., 3 1 :433^436. Kendall, P. B. 1984. Seasonal changes of sex ratio in Norway rat (Rattus norvegicus) populations in Wales. J. Zool. Lond., 203:288-291. King, H. D. 1916. The relation of age to fertility in the rat. Anat. Rec, 11:269-287. King, H. D. 1939. Life processes in the grey Norway rat during fourteen years in captivity. Amer. Anat. Mem., 17:1-72. Leslie, P. H., U. M. Venables, and L S. V. Venables. 1952. The fertility and population structure of the brown rat (Rattus norvegicus) in com-ricks and some other habitats. Proc. Zool. Soc. Lond., 122:187-238. Olson, S. 1976. Baltimore. Harper & Row, Cambridge, MA. Orgain, H. and M. W. Schein. 1953. A preliminary analysis of the physical environment of the Norway rat. Ecology, 34:467-473. Perrigo, O. G. and F. H. Branson. 1983. Foraging effort, food intake, fat depostion and puberty in female mice. Biol. Reprod., 29:455-463. Pye, T. and W. N. Bonner. 1980. Feral brown rats, Rattus norvegicus, in South Georgia (South Atlantic Ocean). J. Zool. Lond., 192:237-255. Recht, M. A. 1982. The fine structure of the home range and activity pattern of free-ranging telemetered urban Norway rats Rattus norvegicus (Berkenhout). Bull. Soc. Vector Ecol., 7:29-35. Robitaille, J. A. and J. Bovet. 1976. Field observations on the social behaviour of the Norway rat, Rattus norvegicus (Berkenhout). Biol. Behav., 1 :289— 308. Sauer, J. R. and N. A. Slade. 1987. Body size as a demographic categorical variable for vertebrates. Ann. Rev. Ecol. Syst., 18:71-90. Schein, M. W. and H. Orgain. 1953. A preliminary analysis of garbage as food for the Norway rat. Amer. J. Trap. Med. Hyg., 2:1117-1130. Schroder, G. D. and M. Hulse. 1979. Survey of rodent populations associated with an urban landfill. Amer. J. Publ. Health, 69:713-715. Slade, N. A., J. R. Sauer, and G. E. Glass. 1984. Seasonal variation in field-determined growth rates of the hispid cotton rat (Sigmodon hispidus). J. Mammal., 65:263-270. Stroud, D. C. 1982. Population dynamics of Rattus rattus and R. norvegicus in riparian habitat. J. Mammal., 63:151-154. Takahashi, L. K. and R. K. Lore. 1980. Foraging and food hoarding of wild Rattus norvegicus in an urban environment. Behav. Neurobiol., 527-531. Taylor, K. K. 1978. Range of movement and activity of common rats (Rattus norvegicus) on agricultural land. J. Appl. Ecol., 15:663-677. Taylor, R. B., S. N. Brower, and W. Drain. 1979. Toward a neighborhood-based data file: Baltimore. Center for Metropolitan Planning and Research, Johns Hopkins Univ. URBAN ECOLOGY OF NORWAY RATS 33 Plate I. Characteristic urban habitats trapped for R. norvegicus within Baltimore included urban parkland at CH (top) and urban residential sites at PP (middle) and RH (bottom). Note large mounds of dirt excavated from under concrete slabs that mark the entrances to rat burrows (arrows) in residential areas. Large quantities of refuse were available for foraging in some low SES residential areas (middle and bottom). Foraging by rats and cats were common at these locations (bottom). Note the smaller rat underneath the garbage near the cats as well as the larger individual to the left (arrow). A kinked tail with several scars are visible on the larger rat. University of Kansas Publications MUSEUM OF NATURAL HISTORY The University of Kansas Publications, Museum of Natural History, beginning with volume 1 in 1946, was discontinued with volume 20 in 1971. Shorter research papers formerly published in the above series are now published as Occasional Papers, Museum of Natural History. The Miscellaneous Publications, Museum of Natural History, began with number 1 in 1946. Longer research papers are published in that series. Monographs of the Museum of Natural History was initiated in 1970. Authors should contact the managing editor regarding style and submission procedures before manuscript submission. All manuscripts are subject to critical review by intra- and extramural specialists; final acceptance is at the discretion of the Director. The Occasional Papers and Miscel- laneous Publications are produced by a Macintosh computer, Las- erWriter, and the software Microsoft® Word and PageMaker®. Conversion to current format by E. O. Wiley. Institutional libraries interested in exchanging publications may obtain the Occasional Papers and Miscellaneous Publications by addressing the Exchange Librarian, University of Kansas Li- brary, Lawrence, Kansas 66045. Individuals may purchase sepa- rate numbers of all series. Prices may be obtained upon request ad- dressed to Publications Secretary, Museum of Natural History, University of Kansas, Lawrence, Kansas 66045-2454 U.S.A. Editor: Robert D. Holt Managing Editor. Joseph T. Collins Design and Typesetting: Kate Shaw and Joseph T. Collins PRINTED BY UNIVERSITY OF KANSAS PRINTING SERVICE LAWRENCE, KANSAS DATE DUE -MAR H 1996 DEMCO. INC. 38-2931 II r 3 2044 093 361 707 ii