Effect of Cage Space on Behavior and Reproduction in Crl:CD(SD) and BN/Crl Laboratory Rats

Abstract

The 2011 Guide for the Care and Use of Laboratory Animals contains recommendations regarding the amount of cage space for mothers with litters. Literature on cage-space use in breeding rats is sparse. We hypothesized that, if present, differences in behavior and reproduction would be detected between the smallest and largest cages tested. BN/Crl and Crl:CD(SD) rats were assigned to a cage treatment (580 cm², 758 cm², 903 cm², or 1355 cm²) and breeding configuration (single: male removed after birth of pups; pair: 1 male, 1 female) in a factorial design for 12 wk. All cages received 20 to 25 g of nesting material, and nests were scored weekly. Pups were weaned, sexed, and weighed between postnatal days 18 and 26. Adult behavior and location in the cage were videorecorded by scan-sampling on the litter's postnatal days 0 through 8 and 14 through 21. Press posture in adults and play behavior in pups were recorded according to a 1–0 sampling method. Differences in reproductive parameters were limited to expected differences related to rat genetic background and weaning weight in pups, which was lowest in the pair-bred CD rats in the smallest cages. Press posture in adults in the smaller cages increased as the pups became mobile. Pair-housed outbred rats in the smallest commercially available cage we tested showed behavioral changes and a lower pup weaning weight. Both laboratory animal scientists and caging manufacturers should address the challenge of providing more biologically relevant cage complexity rather than merely increasing floor space.

The Guide for the Care and Use of Laboratory Animals (the Guide) has defined the standard of care of laboratory animals in the United States since its first publication in 1963. The Guide has been revised 7 times since 1963 as new information or changes in accepted best practices made revision necessary. The 1996 Guide and 2011 Guide present cage-space recommendations for rats that are virtually unchanged from those in the 1972 Guide (Table 1), but no edition of the Guide before 2011 has ever directly addressed the space requirements for dams with litters. In the latest version of the 2011 Guide, dams with litters were specifically assigned a space requirements of 800 cm².³⁶ These space recommendations apparently mirror the European regulations, except that the European regulations include the stipulation of an additional 400 cm² for each adult permanently added to the enclosure, resulting in a requirement of 1200 cm² for a breeding pair.²¹ Neither the Guide nor the European regulations cite literature to support the space requirements suggested for mothers and litters. Changes to the Guide were likely intended to improve the welfare of breeding rodents and probably were based on the perception that increased density causes crowding stress, but few studies have been conducted to investigate the cage-space needs of breeding rats,² and none evaluate the behavior of various configurations of breeding rats in various commercially available cages. Cage-space recommendations that are unsupported by scientific study seem suboptimal for this widely used document.

Table 1.

Evolution of the cage-space requirements for rats through versions of the Guide

Version of the Guide	No./weight of rats	Area per rat
1963	1–3/250 g	185.8–650.3 cm2
	4–10/250 g	185.8–464.5 cm2
1972 and 1974	Up to 100 g	110 cm2
	100-200 g	148 cm2
	201-300 g	187 cm2
	Over 300 g	258 cm2
1978 and 1980	<100 g	110 cm2
	100–200 g	148 cm2
	201–300 g	187 cm2
	>300 g	258 cm2
1985 and 1996	<100 g	109.68 cm2
	100–200 g	148.40 cm2
	200–300 g	187.11 cm2
	300–400 g	258.08 cm2
	400–500 g	387.12 cm2
	>500 g	451.64 cm2
2011	<100 g	109.6 cm2
	100–200 g	148.35 cm2
	200–300 g	187.05 cm2
	300–400 g	258.0 cm2
	400–500 g	387.0 cm2
	>500 g	>451.5 cm2
	Mother and litter	800 cm2

Although the use of space by rats has been a subject of investigation almost since scientists began to keep them in laboratories, much of this work has been studies of territory and behavior in wild rats.²⁴ Wild rats spend most of their lives in subterranean burrows consisting of tunnels approximately 7.5 cm high and 8.3 cm wide and in nesting chambers that are approximately 14.5 cm high, 22.1 cm long, 18.4 cm wide.¹⁶ Wild rats are social animals, and their population densities can vary widely over small geographic distances.³⁵ The home range of wild rats varies from 44 m² in agricultural settings with harborage near buildings and abundant food to 2.4 km² for rats living in hedgerows adjacent to agricultural fields.^44,55 Laboratory rats were the primary subjects of classic crowding experiments in the 1950s, which showed that rats will tolerate extremely crowded conditions and generally continue to reproduce, although behavior and reproduction will both vary significantly from those of rats kept in smaller groups.^17,50 The behavioral plasticity and genetic diversity of laboratory rats makes it difficult to define the minimum or maximum of cage space required.

The investigation of the use of cage space by rats involves 2 separate factors, one being the number of rats in the cage (group size) and the other being the amount of physical floor space available to each rat (cm²/animal). Much of the peer-reviewed literature on cage space confounds space per rat with group size, primarily by using only one size of cage and then adding or subtracting animals to reach the desired floor space per rat.^29,30,52 When individual rats were given a choice between 2 cages, they chose more space (1620 cm²) over less space (540 cm²), but their cage choices did not change when they were forced to share the larger cage with 4 other rats, which resulted in less physical space than the smaller, solitary cage.⁴⁸ The type of space offered may be important as well. In another study,³ rats showed the same preference as in the previously cited work,⁴⁸ also choosing less space over being alone, but rats also chose cages with more horizontal complexity, which allowed for expression of their normal thigmotactic behavior, rather than vertical complexity. When provided with a taller cage, rats did not choose cages with more vertical space when offered²⁸ but did choose cages with more height when vertical complexity was added to the cage.⁴ In a study that used commercially available cages, male CD rats housed in pairs in different-sized cages with enrichment showed a difference in heart rate variability when compared with rats housed without enrichment, but cage size alone did not alter the measured parameter.¹⁰ One further caveat of many cage-space studies is that investigators often use custom-built caging to manipulate the space parameters and provide different forms of enrichment and structural complexity, often not reported in detail, which makes duplication of their work difficult. This variability also makes it challenging to broadly apply conclusions reached by many cage-space researchers.

Research in which both group size and space per rat are controlled usually has been performed with single-sex groups,¹¹ not breeding animals, because investigators typically are interested in certain behaviors or in altered physiologic parameters that would affect experimental results. In those cases, pregnancy and maternal behavior might be considered a confounder. In one study,¹¹ density and group size were investigated by using both commercially available and custom-built cages. The investigators examined corticosterone levels as a marker of stress in individually housed and group-housed Wistar rats and then manipulated both the density and group size of these single-sex groups. Rats are territorial animals that breed communally; this characteristic reflected in the findings, in which spatial crowding affected the male rats, but group size had a greater effect on female rats.¹¹

Although important to consider, knowing the preferred housing density of single-sex groups is not particularly helpful when housing breeding rats. In commercial breeding applications, breeding rats inhabit most of the cages maintained. In comparison, far fewer rats are used as breeders in research settings, although this proportion is likely to increase as the genetic modification of rats becomes more common.^13,32,58 The cage-space needs of breeding rats rarely are addressed in the literature. In some early work, large groups of laboratory rats were confined to custom-built cages, and behavior, fighting, and reproductive output were measured in these seminatural situations.^7,15 Similarly, in other studies, data on space usage and reproduction were collected from wild-rat populations.^6,8,16,57 Other colleagues recently published a study on the space requirements of laboratory rats breeding in commercially available cages, in which the breeding performance of a rat common at the authors’ institution, the Dahl Salt-Sensitive rat, was examined.² Dahl rats are an inbred strain and are smaller than most outbred rats, with adult females averaging 172 g at 8 wk of age and adult males averaging 222 g. However, the study² confounded density with group size by comparing pair housing and solo rearing of pups by female rats in only one cage size. Interestingly, the reproductive output of these rats did not vary between treatments and, in fact, the weaning weight for pups from the continuously housed pairs was increased and interpreted as a net welfare increase.²

In the present study, we examined the behavior and reproduction of rats in a range of 4 different-sized commercially available cages. We hypothesized that if a difference in behavior and reproduction exists within the limits of commercially available caging, this difference would be detected between the smallest and largest cages. To test this hypothesis, we looked at various measures of reproduction and behavior by studying 2 commonly used types of rats: the small, less fecund, inbred BN/Crl and the large, extremely fecund, outbred Crl:CD(SD).

Materials and Methods

Work was conducted at Charles River's AAALAC-accredited Wilmington, MA (location 1), and Raleigh, NC (location 2), facilities and was approved by Charles River's IACUC (no. P01112012). The study duration was 12 wk. Rats were free of a list of common rat pathogens at the start of the study; additional details can be found at http://www.criver.com/files/pdfs/rms/hmsummary.aspx. No further health monitoring was performed on animals on study. Rats used in both study locations were obtained from Charles River colonies in Raleigh, NC (CD), and Portage, MI (BN). All rats entered the study at 56 d (± 3 d) of age. Rats were housed in solid-bottomed cages of various sizes (Table 2; Figure 1). Cages were purchased from Ancare (A RM1: 39.7 × 23.5 × 18.7 cm, floor area of 758 cm²; Bellmore, NY) or Lab Products (LP 80701: 35.2 × 23.0 × 19.7 cm, floor area of 580 cm²; LP 1400: 40 × 33.7 × 18.7 cm, floor area of 903 cm²; and LP 82110: 35.2 × 48.4 × 19.7 cm, floor area of 1355 cm²; Seaford, DE). Three of the 4 cage sizes used can be purchased directly from the manufacturer; the Ancare cage is produced solely for Charles River and is not typically used for pair breeding of outbred rats. Cages were placed on rolling racks (Metro, Wilkes-Barre, PA) for the duration of the experiment.

Table 2.

Differences between the cage area (cm²) provided in this study and the amount of space recommended in the 2011 Guide

	Cage area provided	Difference between provided and recommended areas
		Female + litter (800 cm2 recommended)	Breeding pair (1252 cm2 recommended)
LP 80701	580	−220	−672
A RM1	758	−42	−494
LP 1400	903	103	−349
LP 82110	1355	555	103

Figure 1.

The 4 different commercially available cages used to test whether cage size altered reproduction or behavior in breeding rats. Cage size increases from left to right.

All cages were open to the environment (no IVC were used). When a filtered lid was an integral part of the cage, the filter was removed to minimize ventilation differences between cages. All cages contained heat-treated chipped hardwood bedding (Beta Chips, NEPCO, Warrensburg, NY), and all cages were provided with 20 to 25 g of a long-fiber paper nesting material (EnviroDri, Shepherd Specialty Papers, Watertown, TN) at the weekly cage change. Food (no. 5L79, LabDiet, St. Louis, MO) and ultrafiltered, hyperchlorinated water (via water bottle) were provided ad libitum. The light cycle was 12:12 light:dark (lights on, 0630; lights off, 1830), humidity was maintained between 30% to 70%, and temperature ranged between 19 to 22 °C. In both study locations, rats and mice³¹ on cage-density studies were housed in dedicated areas that did not contain other, nonstudy animals. Because the rats and mice were of similar microbial status, they were housed concurrently in the same study rooms but on different racks, based on previous work showing no effects on either species.⁴⁹ Rats were evaluated for health daily. At the end of the study, all animals were euthanized with inhaled CO₂, administered to effect.

The experiment was split into 2 arms. The arm of the study conducted in location 1 focused on behavior but also followed reproduction. Rats were assigned randomly to cage and reproductive treatments by using the random integer set generator found at random.org. For half of the rats, the male remained with the female in a stable pair mating. For the other half, the male was removed 24 to 48 h after a litter was born, to model a breeding situation in which the female rat rears the pups by herself. For solo female rats, the male was reintroduced after the pups were weaned only when the female had not become pregnant at the postpartum estrus.

Rats at location 1 were filmed for two 24-h periods, once during postnatal days 0 to 8 of the litter and once during days 14 to 21. The videorecording was evaluated by a single investigator (BNG), who was not blinded as to treatment, due to the nature of the study. Adult behavior was instantaneously scan-sampled every 30 min during both recording periods. Figures 2 and 3 contain the detailed ethograms. To score the adult press posture (Figure 4) and pup play behavior, we applied 1–0 (yes or no) sampling during a 2-min interval every 30 min. Pup play behavior was not scored during recordings from days 0 to 8, because neonates are not mobile enough for recognizable play. The adult press posture was scored in both videos. Reproductive data collected included litter size at birth, litter size at weaning, and weaning weight. From these data, we calculated the interlitter interval between the first and second litters, pup mortality, percentage of female rats pregnant at the end of the study, and the number of pups weaned per female rat per week (production index). We also collected weekly nest scores (scale of 1 to 5):³⁴ 1, manipulated material was present but not a central nest site; 2, flat nest; 3, cup nest; 4, incomplete dome; 5, complete and enclosed dome.

Figure 2.

A general behavioral budget ethogram for breeding rats. For the behavioral analysis according to this ethogram, the videorecording was instantaneously scan-sampled every 30 min over 24 h.

Figure 3.

A 1–0 ethogram for stereotypy, play behavior, and press posture in breeding rats and their offspring. For the behavioral analysis, video was observed for occurrence of behaviors described in this ethogram for 2 min every 30 min over 24 h. For grooming, sniffing, or pausing, an event was scored as a bout of stereotypy even when brief interruptions (5 s or less) occurred.

Figure 4.

Image illustrating a female exhibiting the behavior termed ‘press posture’. This behavior is characterized by a lack of movement and the placement of the ventral surface of the body in the corner of the cage. Rats often fell asleep in this position.

Location 2 modeled a barrier-room setting, and observations focused mainly on reproductive performance. Rats were assigned randomly to cage and breeding treatments by using the random integer set generator found at random.org. In half of the cages, the male rat remained with the female, whereas in the remaining half of the cages, the male rat was removed 24 to 48 h after a litter was born. As with the behavior portion of the study, the reproductive data collected included litter size at birth, litter size at weaning, and weaning weight. From these data, we calculated the interlitter interval between the first and second litters, pup mortality, percentage of females pregnant at the end of the study, and the number of pups weaned per female rat per week (production index). Simple behavioral data were collected on the day of weekly cage change. The data collected were limited to noting hair loss (yes or no) and fight wounds (yes or no) when rats were handled at cage change and to observing rats for stereotypies during a 5-s scan of each cage performed 30 min prior to the end of the day by using 1–0 sampling. Rats were considered stereotypic when they displayed a behavior defined as a stereotypy at least 3 times during the scan time period. In addition, nest scores based on the 5-point scale described earlier³³ were collected weekly.

All statistical analyses were conducted in JMP version 9 (SAS Institute, Cary, NC). The likelihood of pregnancy at the end of the experiment was analyzed as a generalized linear model (GLIM). The rest of the analyses were run as ANOVA using a general linear model (GLM); the assumptions of GLM (normality of error, homogeneity of variance, and linearity) were confirmed posthoc.³³ Significant effects were analyzed according to posthoc Tukey tests or Bonferroni-corrected planned contrasts by using custom contrasts in JMP. Cage treatment in all analyses was treated as a categorical variable, given that many aspects of these commercially available cages differed (for example, different dimensions, shape, and depth of feed hoppers) and might have influenced results.

For reproductive results, all data from both locations were included in the analysis, including nonproductive cages but were blocked by location in the model (Table 3 includes the number of cages in each treatment). Data from only one cage were removed from the final analysis, due to an animal found dead within the first week of the study. A full factorial model between strain, breeding treatment, and cage treatment was tested initially. Due to nonorthogonality of the dataset, insignificant interactions were removed from some models.³³ Only data from productive cages were included in the analysis for average weaning weight. Average weaning weight per litter was log transformed for normality and used a similar model as described earlier but was blocked by cage and nested within strain, breeding treatment, and cage treatment. The numbers of litters and pups weaned were tested as covariates, but only the number of pups was significant and thus kept in the model. Pups were weaned between 18 to 26 d of age; age at weaning was tested also but did not significantly explain variation in the data and therefore was not included in the final model. Interlitter interval was calculated for each female rat by calculating the number of days between litters 1 and 2 for cages with multiple litters. Only main effects were used in this model. All values are given as least-squares means and standard error. Log-transformed pup mortality was calculated by cage as 1 – (number of pups weaned ÷ number of pups born [dead or alive]) and contained all main effects and the location blocking factor. All values are given as least-squares means and standard error.

Table 3.

The number of cages from which reproductive data were collected and analyzed

	Location 1		Location 2
	BN	CD	BN	CD
LP 80701	7	7	21	19
A RM1	9	7	19	21
LP 1400	13	7	20	19
LP 82110	13	7	19	21

Time budgets for the adult population were calculated for only location 1 rats that produced litters (32 cages; n = 2 for each treatment combination). Because cage-space requirements differ when a litter is present and maternal behaviors were observed, 2 reproductively successful cages were selected randomly from each treatment combination for behavior observations and analysis. The total number of times each category of behavior or location within the cage (that is, under feeder or other) for each day was divided by the total number of observations for that group. After this calculation, data from the ethogram category ‘unknown behaviors’ were excluded from the analysis (Figure 2). Thus the behavioral time budget does not total 100%, and the independent variables are not colinear. In essence, a change in one behavioral category does not directly influence the level of another behavioral category. A model that included all 3rd-order interactions between strain, breeding treatment, cage treatment, age of the litter, and behavior was tested initially. For datasets that were not orthogonal, nonsignificant high-order interactions were removed from the model.³³ Data were blocked by cage and nested within strain, breeding treatment, and cage treatment for repeated-measures analyses. Play behavior, number of pups in the cage, and age were tested as covariates for both 1–0-scored behaviors, but number of pups was not significant for either analysis and was removed from the model. Nest score was tested similarly by cage nested within strain, cage treatment, and breeding treatment with all 2nd-order interactions as well as for the strain, cage treatment, and breeding treatment interaction. Litter in the nest and litter in the nest by breeding treatment were tested also. Hair loss and wounding were run as a binary logistic regression with Firth bias adjustment also testing main effects.

Results

Reproduction.

From the total of 230 cages assessed for reproductive parameters between the 2 locations, all CD cages had at least 1 litter, whereas 6 BN cages did not produce any litters. One BN female rat was found dead of unknown causes 1 wk after the start of the study. Of the 230 cages originally, only 152 cages (59 BN cages and 93 CD cages) had 2 or more litters.

The total number of pups born per cage was affected by a 2nd-order interaction between the strain of rat and breeding treatment (GLM: F_3,221 = 15.8; P < 0.001). Breeding configuration in BN rats resulted in no difference in the number of pups born over the 12 wk (BN Pair, 8.55 ± 0.71 pups per cage; BN Single, 7.64 ± 0.64 pups per cage; P > 0.05), but CD rat pairs had more pups than did single CD female rats (CD Pair, 29.3 ± 0.73 pups per cage; CD Single, 22.8 ± 0.73 pups per cage; P < 0.05). The number of pups weaned was also affected by breeding treatment, but the effect was dependent on the strain of rat (GLM: F_1,221 = 13.4; P < 0.001). Breeding configuration in BN rats showed no difference (BN Pair, 8.4 ± 0.72 pups per cage; BN Single: 7.3 ± 0.66 pups per cage; P > 0.05), but CD pairs weaned significantly more pups (CD Pair, 29.0 ± 0.74 pups per cage; CD Single, 22.7 ± 0.74 pups per cage; P < 0.05). Pup weaning weight was significantly altered by a 3rd-order interaction between the 3 experimental variables (GLM: F_{3, 199} = 62.6; P < 0.001; Figure 5).

Figure 5.

Average weaning weight per litter in the 4 different cage sizes by strain and breeding treatment. Least-square means and standard errors are plotted against a log-transformed y axis. Asterisks indicate a significant (P < 0.05) difference between groups, according to posthoc Tukey comparisons.

A significant negative correlation was found between number of pups in the litter and the average pup weaning weight (GLM: F_1,199 = 35.1; P < 0.001). Thus, pups from larger litters weighed less at weaning than did those from smaller litters. None of the cage sizes affected the production index (GLM: F_3,199 = 2.7; P = 0.05). Breeding treatment only affected the production index between CD pair-housed and single-housed rats (CD Pair, 2.38 ± 0.06 pups weaned per female rat per week; CD Single: 1.9 ± 0.06 pups weaned per female rat per week; P < 0.05). Interlitter interval was only altered by breeding treatment, in that pair-housed rats had a slightly longer interval between litters than did singly housed rats (Pair, 30.0 ± 0.61; Single: 27.0 ± 0.61 d; GLM: F_1,146 = 12.3; P < 0.001). None of the parameters in the analysis affected the pup mortality rate. Finally, CD rats were more likely to be pregnant at the end of the study than were BN rats (no. of rats pregnant: CD, 62; BN, 19; χ² = 44.6; P < 0.001).

Behavior.

Cage size did not alter a general behavior budget for adult rats (GLM: F_9,760 = 0.26; P = 0.98). However, breeding treatment did change the behaviors observed (GLM: F_3,760 = 4.08; P = 0.007). Bonferroni-corrected tests identified that pair-housed rats were less active than were singly housed female rats (α/4; F_1,760 = 8.2; P = 0.004). None of the treatments in this experiment affected play behavior in pups during postnatal days 14 through 21, although play behavior increased as pups approached day 21 (GLM: F_1,18 = 16.4; P < 0.001). Press posture was observed more often in CD rats in smaller cages (GLM: F_3,26 = 9.42; P < 0.001; Figure 6), but no difference in press posture behavior was seen in BN (P > 0.05). The age of pups in the cage also affected press posture in different cage treatments (GLM: F_3,26 = 4.9; P = 0.008; Figure 7). For each cage size, the age of the pups in the cage significantly affected the occurrence of press posture (slopes were different from 0), except for the largest cage, where the age of the pups did not affect the behavior (α/4; LP 80701 (smallest cage): F_1,26 = 31.3; P < 0.001; A RM1: F_1,26 = 22.7; P < 0.001; LP 1400: F_1,26 = 7.43; P = 0.01; LP 82110 (largest cage): F_1,26 = 0.92; P = 0.35). Rat strain also affected press-posture behavior at different pup ages (GLM: F_1,26 = 6.2; P = 0.02; Figure 8). The slopes of regression lines for each strain were significantly different from 0 (α/2; BN: F_1,26 = 9.14; P = 0.006; CD: F_1,26 = 52.9; P < 0.001). The amount of time rats spent under the feeder varied among different-size cages (GLM: F_3,83 = 134.8; P < 0.001). In all of the Lab Products cages (LP 80701, LP 1400, and LP 82110), rats spent less time under the feeder than in all other areas of the cage (P < 0.05). In contrast, rats allocated to the A RM1 cages spent equal amounts of time in both areas (P > 0.05).

Figure 6.

Incidence of press posture (% of total observations) in different-sized cages for the 2 types of rats tested. Least-square means and standard errors are plotted against the y axis. NS indicates the lack of significant difference between cages; bars with different letters differed significantly (P < 0.0125), according to Bonferroni-adjusted custom contrasts.

Figure 7.

Incidence of press posture (% of total observations) in adults at different pup ages in the 4 cage sizes. Data points are plotted with the least-squares line.

Figure 8.

Incidence of press posture (% of total observations) in adults at different pup ages in 2 types of rats tested. Data points are plotted with the least-squares line.

Cage treatment significantly altered the nest scores of CD rats (GLM: F_3,218 = 3.9; P = 0.009). Tukey tests revealed a difference only between the A RM1 and LP 1400 cages (P < 0.05). BN rats built significantly more complex nests than did CD rats (BN, 3.19 ± 0.06; CD, 2.02 ± 0.06; GLM: F_1,218 = 198.8; P < 0.001). In addition, an interaction between breeding treatment and whether a litter was present in the nest affected nest scores (GLM: F_1,218 = 4.82; P = 0.03). Nest scores were higher when a litter was present, but solitary female rats had the most complex nests (P < 0.05). When no litter was present, the breeding configuration did not affect nest scores (P > 0.05).

Stereotypies were observed in behavioral observations at location 2 but were too infrequent for analysis. Hair loss was more likely in BN than CD rats (GLIM: χ² = 18.63; P < 0.001), but CD rats were more likely to have minor wounds, presumably from agonistic interactions (GLIM: χ² = 60.8; P < 0.001). Single rats were, unsurprisingly, less likely to be wounded (GLIM: χ² = 9.06; P = 0.003).

Discussion

The smaller, inbred BN rats showed no difference in reproduction, regardless of cage size or breeding configuration. Due to abnormal placentation,³⁹ BN have smaller litters than do other strains or stocks of rats, which was confirmed by the overall significantly lower reproductive performance of BN when compared with CD in the current study. We found that CD rats housed in pairs weaned more pups than did CD dams rearing pups alone, regardless of cage size. In contrast to previous results,² we did not find an overall difference in weaning weight between CD rats bred in pairs compared with single female rats. This difference might be due to the different genetic background of the rats used in in the two studies or the fact that we controlled for larger litter size in CD pairs by our analysis of weaning weight (larger litters typically have lighter pups at weaning). In addition, we found a linear relationship of pup weaning weight with cage size in pair-housed CD rats. Pups from the largest cage were heavier than those from the smallest cage. This result cannot be explained by the number of pups weaned in each cage type and perhaps may be the only reproductive indication of space-related stress in this study. Production index was not affected by any cage size, but pair-housed animals did have a higher production index per cage than did female rats separated from male rats. Pair housing did slightly increase the interlitter interval, perhaps due to embryonic diapause in the lactating female⁴⁵ or to mating events taking place later than the immediate postpartum estrus. Pup mortality was not affected by cage size, which seems to indicate that either the rats were not stressed enough to perform infanticide or that the mating-induced block on infanticide in male rats⁴⁶ was not affected by the cage size.

The focused behavioral evaluation, which was conducted entirely at location 1, showed no significant differences in the overall behavioral budget between cage size or rat strain. However, behavioral budget was affected by breeding treatment, with pairs spending more of their time inactive than did single female rats. Rat pups showed the same level of play behavior, regardless of parental breeding treatment, cage size, or genetic background, and the play behavior increased as the pups approached weaning. One behavioral measure that differed between cages was the amount of time spent under the feeder. In the Lab Products cages, rats spent significantly less time under the feeder, whereas rats in the A RM1 cages spent equal amounts of time in both locations. Figure 1 illustrates a potential explanation for this difference: the feeder in the A RM1 cage has a much different configuration and overhangs a larger proportion of the cage than does that of the LP cages. Rats are thigmotaxic and find open spaces anxiogenic; although the Lab Product cages were both the largest and smallest used in the study, rats in the largest cage nonetheless appeared to huddle under the shelter provided by the feeder. However, this effect could also be due to other factors inherent to commercial caging that we could not control, such as cage height or configuration of the cage lid. Although not specifically measured, a subjective difference between the cages is that the rats used the V-shaped feeder in the A RM1 cages as an improvised cage divider, with the pups corralled on one side by a wall of bedding and nesting material built by the adults.

In this study, we observed a rarely mentioned behavior described as ‘press posture’^22,51 in both CD and BN rats (Figure 4). This behavior, in which a female rat remains near pups but either presses her ventrum into the cage floor or against a vertical surface, has been interpreted as a behavior that either dissipates heat or denies access to the nipples and thus encourages weaning of pups.²² This behavior was not part of the original ethogram, but during data collection, the vertical press posture was seen repeatedly, so the ethogram was amended and videos were reevaluated to include this behavior. We observed this behavior in both sexes, although we could not find a description of this behavior in male rats elsewhere in the literature. In BN rats, the occurrence of press posture was not related to cage size. In CD rats, the onset of this behavior was related to both the age of the pups in the cage and the size of the cage. Press posture was rarely seen during the first week after the birth of a litter, and when seen, it was more likely in the smallest cage (Figure 6). As pups grew, especially in the smaller cages, the frequency of the behavior increased (Figure 7). Both male and female rats exhibited this vertical press posture, and this behavior appeared to be an attempt to remove themselves from direct interaction with the pups. Because adult rats fell asleep in this position, it may allow them to obtain relatively undisturbed rest in the smaller cages. Inability or disinclination to sleep in normal postures may indicate stress or crowding in these smaller cages. We cannot conclude that this behavior is solely stress-related, since this behavior still occurred in another study when female rats were able to escape direct contact with the pups.²² Perhaps not the behavior itself but rather its frequency is indicative of space-related stress and crowding. Crowding should be distinguished from density:⁵⁴ density is defined solely by the space available, whereas crowding is a motivational state that occurs through the interaction of spatial and social factors, with motivation directed toward ending the perceived spatial restriction. Crowded rats have a motivation to perform various behaviors, such as the vertical press posture we noted, to gain more space or cease interactions with others in the cage.

In both study locations, numerous behaviors were evaluated. Stereotypies occurred at such low frequencies that their statistical analysis was impossible. Rats exhibit stereotypies only rarely in standard laboratory housing, so low levels of stereotypical behavior is not surprising.⁴⁰ Hair loss, which could be due to over-grooming or to hair-plucking behavior, was higher in BN rats than in CD. Rat breeders report anecdotally that BN are more likely than CD to show hair loss, but no etiology of the phenomenon has been investigated. To monitor possible agonistic behavior between paired rats, we scored minor wounds, which were more likely to occur in CD rats, regardless of cage size. This result may indicate that BN rats are less aggressive in defense of a litter from a cagemate or that their overall aggressive tendencies are less than those of the CD rat. Rats in this study were provided with nesting material as an enrichment and to enable us to evaluate nest-building behavior. The use of nesting material by virgin rats may be a learned behavior,^37,56 and these rats had not encountered nesting material before the study, so we were unsure whether or how they would use the material. Pregnant rats, however, are motivated to build nests, and the nest score was higher when a litter was present, regardless of breeding configuration. Singly housed dams and the smaller BN rats had higher nest scores than did pair-housed animals or CD rats. This finding may indicate that the smaller or singly housed rats are under increased thermal stress or that having more rats within a cage caused simple mechanical disruption of the nest. The amount of nesting material provided to each cage was sufficient, because the rats were able to build structurally complex nests.

In our study, we observed the behavior of 2 commonly used ‘breeds’ of rats in typical reproductive configurations and measured both observed behavior and reproductive output, instead of biochemical or histopathologic methods, as a way to examine stress. Methods that require complicated analyses or sophisticated equipment will be limited in their applicability. We chose to examine reproduction and behavior in light of their practicality and relevance as well as their potential to directly influence how rats are managed. Biochemical methods, such as serum or fecal corticosterone levels, are subject to normal fluctuations based on circadian rhythms,^38,53 and the data in the literature on corticosterone levels in pregnant rats are from studies in singly housed animals,^5,12,19 which may be an additional source of stress.⁴¹ We were unsure whether the examination of physiologic measures would have differentiated between stress induced by crowding and the normal stresses of pregnancy, parturition, and lactation. In addition, we were wary of the increased handling necessary for collecting serum, feces, or urine in that handling might alter the behavior toward offspring and adult cagemates. Furthermore, we chose to avoid excess handling because rats acclimate to procedures,²⁷ thus perhaps changing behavioral measures throughout the study. We acknowledge that not all scientists consider biologic functioning as indicated by successful reproduction to be a definitive measure of welfare in a highly fecund species that has been domesticated and further selected for this trait under a variety of conditions, but we deemed reproduction appropriate, biologically relevant, reproducible and necessary for this scientific question in breeding rats.

Rat behavior in this study was measured through the observation of both behavior over time and the examination of animals for physical changes associated with unwanted behavior. Other methods of behavioral analysis, such as preference testing, were not used, although they may be relevant. The birth of a litter complicates the analysis of space preference in rats, given that offspring are fully dependent on parents until they are mobile, parental care extends until weaning, and the preferences of parents may differ from those of offspring. Offspring occupy more physical space within the cage near weaning than at birth, thus perhaps altering the interactions of the breeding group. Attempting to ascertain the preferences of breeding rats for cage space has not been done in a systematic manner and would be difficult to accomplish in commercially available caging. In addition, when examining the preferences of animals, consideration should be given to their intended purpose; they may prefer conditions that make their captivity (and freedom from disease, hunger, thirst, and predation) impractical but not show a strong preference when asked to choose between conditions that can be made reasonably available in the laboratory.²⁰ Preference testing, although useful in determining the preferences of animals, should always be placed in context, given that animals (and humans) choose things that they prefer in the short term but that may be detrimental over the long term.²⁵

The present study supports the concept that laboratory rats can breed successfully in a variety of cage sizes. However, we noted several behaviors, such as an increased frequency of press posture by both sexes, suggesting that some of the cages were too small for breeding outbred rats. This study was unable to identify an amount of cage space in which rats would not reproduce (a positive control for the experiment). However, the smallest commercially available cage that we used was not acceptable for breeding rats. The bedding in the smallest cage became soiled within hours of changing and would require frequent cage changes to provide the rats with an environment that is clean and esthetically pleasing to humans.^9,23 Frequent cage changing might diminish welfare in other ways, for example by increasing cannibalism of litters.¹⁵ Although rats do not necessarily avoid feces and in fact use fecal matter to signal feeding sites,^42,43 the Guide suggests that they have areas in which to rest that are free of fecal matter. In addition, small cages may require additional daily management to ensure that the housed rats have appropriate amounts of food and water, given that the food hopper and water bottle both are proportional to the size of the cage. In our current study, the average litter size of CD rats was 12 pups, and we did not cull litters to an arbitrary number. Assuming an average pup weight at 21 d of 50 g, an average weight of a 20-wk female rat of approximately 300 g, and an average weight of a 20-wk male rat of approximately 500 g, the smallest cage housing a pair of CD rats with a litter ready to wean contained 1400 g of rat in 580 cm²—a great deal of rat body mass in a very small space. Similar situations may occur briefly in the wild, perhaps in burrow chambers, but in that case, adult wild rats have access to more space by simply exiting the burrow. The impracticality of giving laboratory rats the same amount of space their wild cousins have should not be taken to mean that less space than is customarily provided is acceptable or that large outbred rats should routinely be bred in the smallest cage we tested.

Cage space should be considered as a performance standard rather than as an engineering standard. Simply increasing space and assuming that animal wellbeing will also increase is an invalid assumption. Animals should be able to express normal postures and have some freedom of movement; both conditions were difficult to achieve in the smallest cages when 2 adult rats and a litter were present. Rather than absolute floor space available as the only consideration, perhaps increasing the complexity of the cage would be more beneficial for breeding rats. Nonbreeding rats prefer cage complexity over simple space,^1,47,48 and the effects of adding complexity to breeding cages should be examined. Other species raised intensively benefit from cage complexity, such as elevated resting platforms that allow the adults to escape from their offspring,^14,26 and breeding rats likely would benefit as well, although the work is still preliminary.¹⁸ Both laboratory animal scientists and caging manufacturers should address this challenge of providing more biologically relevant cage complexity rather than merely increasing floor space.