OUP user menu

Burrow sharing in the desert-adapted torch-tail spiny rat, Trinomys yonenagae

J. W. A. Santos, E. A. Lacey
DOI: http://dx.doi.org/10.1644/09-MAMM-S-389.1 3-11 First published online: 16 February 2011


Among fossorial rodents burrow sharing is an important behavioral attribute that provides the foundation for multiple aspects of social structure. Within the family Echimyidae the torch-tail spiny rat (Trinomys yonenagae) is distinguished from closely related taxa by its tendency to live in burrows in desert habitats. Preliminary field studies have suggested that burrow systems of this species are shared by multiple adults. To test this hypothesis we used livetrapping and radiotelemetry to quantify patterns of burrow use in a population of torch-tail spiny rats located near Ibiraba, Bahia State, Brazil. Examination of our data indicates that 76.2% of 67 burrow systems monitored were occupied by >;1 adult, including same-sex pairs, male-female pairs, and multiple adults of both sexes. Spatial overlap among adults captured in the same cluster of burrow entrances was extensive (72.0% ± 27.0% based on 95% minimum convex polygons), with 66.7% of animals resident in the same burrow system using the same putative nest site. Collectively, examination of these data indicates that adult T. yonenagae share burrows and thus may be social. To place our findings in a comparative context and identify potential ecological correlates of burrow sharing in T. yonenagae we contrast our findings with data on space use by other fossorial, desert-dwelling rodents.

Key words
  • burrow sharing
  • caviomorph
  • desert habitats
  • rodents
  • social behavior
  • spiny rat
  • Trinomys yonenagae

Whether animals live alone or in groups is a fundamental distinction that has profound implications for multiple aspects of social behavior, including the occurrence of complex forms of cooperation (e.g., alloparental care—Emlen 1991; Solomon and French 1997) and conflict (reproductive skew—Clutton-Brock 1989; Vehrencamp 1983). Although animal groups vary widely with respect to attributes such as size, kin structure, breeding composition, and adaptive value (Alexander 1974; Bennett and Faulkes 2000; Busher 2007; Hare and Murie 2007; Koenig and Dickinson 2004; Lacey and Ebensperger 2007), all tend to be characterized by extensive spatial overlap among group members. Among fossorial rodents such overlap typically includes burrow sharing, in which multiple adults of ≥ 1 sex use the same system of subterranean tunnels and share a common subterranean nest site (Armitage 2007; Ebensperger et al. 2006; Hayes 2000; Lacey 2000; Lacey and Ebensperger 2007; Lacey and Wieczorek 2003; Nevo 1979; Schradin et al. 2006). As a result burrow sharing—generally thought to be rare among fossorial rodents (Michener 1983; Nevo 1979; Randall 2007)—offers important clues regarding social structure, including the probability that a species is group living.

Studies of evolutionary convergence provide important opportunities to test the general applicability of ecological or other hypotheses proposed to explain phenotypic variation, including variation in social behavior (Ebensperger 2001; Ebensperger and Cofré 2001). The spiny rats of the genus Trinomys (Rodentia: Echimyidae) represent a little-known example of phenotypic convergence among burrow-dwelling rodents from desert habitats. Within Trinomys the torch-tail spiny rat (T. yonenagae) is distinguished by its geographic distribution, ecology, and morphology. Although other members of this genus occur in the coastal Atlantic forest and mesic woodlands of Brazil (Lara and Patton 2000), T. yonenagae inhabits semiarid sand dunes in the interior of northeastern Brazil (Rocha 1995). Unlike its generally surface-dwelling congeners, T. yonenagae is fossorial and displays a morphology that is remarkably convergent with other desert-adapted, burrow-dwelling rodents such as kangaroo rats (Dipodomys) and red vizcacha rats (Tympanoctomys barrerae). In particular, T. yonenagae is characterized by elongated hind feet and tail, shortened forelimbs, and enlarged auditory bullae characteristic of saltatorial desert taxa (Rocha 1995; Rocha et al. 2007).

Given its apparent convergence with other desert-dwelling rodents, T. yonenagae is an important target for studies of interactions between ecology and phenotype, including ecological determinants of social structure. Preliminary field studies indicate that multiple adult torch-tail spiny rats can be captured in the same burrow system (Rocha 1995), suggesting that T. yonenagae may be group living. To date, however, no quantitative studies of burrow or nest sharing by free-living adults have been conducted for this species. As a 1st step toward identifying the ecological factors influencing social behavior in these animals we characterized patterns of space use by adult T. yonenagae. Specifically, we tested the hypothesis that adults share burrow systems and putative nest sites. In addition to providing the 1st quantitative data regarding space use by free-living members of this species, our analyses facilitate efforts to place torch-tail spiny rats within the comparative behavioral framework offered by semifossorial desert species, thereby guiding future efforts to identify the ecological correlates of space use and social structure in T. yonenagae.

Materials and Methods

Study area.—The 5.6-ha study site encompassed 3 parallel sand dunes located along the left bank of the Rio São Francisco, 0.5 km northeast of the village of Ibiraba in Bahia State, Brazil (10°47′S, 42°49′W; Fig. 1). The majority (>75%) of burrow systems used by the study animals were located on the valley floors between dunes (Santos 2004). Vegetation in the valleys was sparse (∼50% bare ground— Rocha et al. 2004), resulting in high soil temperatures (∼60°C) during the day. The herbaceous vegetation consisted primarily of the spiny bromeliad Bromelia antiacantha and the small cactus Tacinga inamoena. Members of the genus Eugenia (Myrtaceae) represented 34.1% of all tree and shrub species (Rocha et al. 2004); their seeds provided most of the food and water consumed by T. yonenagae (Santos 2004).

Fig. 1

Location of the study site for Trinomys yonenagae near Ibiraba, along the Rio São Francisco, Brazil. A) Precipitation gradient map of South America showing the contrast between high-precipitation (dark) and low-precipitation (light) areas; the semiarid Caatinga habitat in northeastern Brazil is partially framed by the rectangle. B) Detailed map of the study region (area included in rectangle in A), with location of the study site for T. yonenagae indicated by the white circle.

The study site was characterized by a semiarid climate, with highly seasonal and unpredictable precipitation (annual range = 400–800 mm—Bahia-SEPLANTEC 1978). The rainy season occurred from October to March, with the dry season extending from April to September (Nimer 1979). Data for this study were collected from June to August 2005–2008. Specifically, fieldwork was conducted for 40 days in 2005, 45 days in 2006, 35 days in 2007, and 25 days in 2008. During data collection ambient temperatures ranged from 15°C at night to 43°C during the day.

Animal capture.—Active burrow systems were identified by the presence of freshly excavated soil at burrow entrances, footprints of spiny rats in fresh mounds of soil, and remains of recently eaten Eugenia seeds around burrow entrances. Preliminary trapping efforts revealed that the mean distance between recaptures of the same individual was 5.5 m ± 3.5 (range = 0.7–14 m, n = 36 recaptures—J. W. A. Santos, pers. obs.). As a result we initially assigned burrow entrances located ≤15 m apart to the same burrow system; burrow entrances >15 m apart (i.e., more than twice the mean distance between recaptures) were assigned to different systems. As part of the current study we used capture locations and telemetry to confirm that burrow entrances assigned to the same cluster were part of the same system. All capture locations were recorded by determining the compass direction and distance of the burrow entrance from a fixed, georeferenced stake placed in the center of each burrow system. We converted each capture locality to x- and y-values that were plotted on a Cartesian coordinate system, allowing localities for different animals to be mapped relative to each other.

To characterize the animals occupying each burrow system we attempted to capture all members of the study population each year using live traps (15 × 15 × 30 cm) constructed of wire mesh (Aramefício Contreras, Cafelândia, São Paulo, Brazil) and baited with small slices of squash. Because T. yonenagae is nocturnal, traps were set in the afternoon (1600 h) and closed the following morning (0600 h). To ensure that only animals using a given burrow system were captured trap entrances were fitted with a 30 × 50-cm canvas sleeve, the other end of which was attached to a 20-cm-long piece of polyvinyl chloride plumbing pipe (12-cm diameter). The open end of the polyvinyl chloride tube was placed in an active burrow entrance. Because the diameter of the tube was slightly greater than most burrow entrances, it was necessary to insert the tube into the soil, providing a tight seal around the focal burrow entrance. As a result, only spiny rats that exited the system via the focal burrow opening could enter the trap, thereby preventing individuals from other systems (i.e., animals traveling aboveground) from being captured. Traps were set simultaneously at all burrow entrances thought to belong to the same system, as determined based on proximity and evidence of recent activity (see above). Individuals were considered residents when captured repeatedly (more than twice) within the same cluster of burrow entrances.

To ensure that all animals resident in a burrow system were trapped each individual captured was placed in a standard polycarbonate rodent cage (dimensions: 15 × 40 × 40 cm), with only individuals trapped in the same cluster of burrow entrances housed in the same cage (≤3 adults per cage). Cage bottoms were lined with a 2-cm layer of dry sand; wet or soiled sand was replaced daily. Captured animals were maintained in a separate room within the building used to house researchers. While in captivity the animals were provided with fresh water and were fed squash and fruits ad libitum. Trapping of a given burrow system continued until no animals had been captured and no activity had been detected at burrow entrances for 48 h (Lacey et al. 1997). Once trapping was complete, all animals held in cages were released at the point of capture.

Marking and tissue sample collection.—For all individuals captured we recorded body mass, sex, and apparent reproductive status. For females reproductive condition was determined by visual inspection of the external genitalia (e.g., perforate vagina) and mammae (e.g., enlarged teats characteristic of lactation) and by palpation of the abdomen (for presence of embryos). Because the testes of males of this species never descend and because T. yonenagae displays no sexual dimorphism in body size, male reproductive status was determined based on body mass. Specifically, because all reproductively active females weighed ≥90 g, we assumed that males weighing ≥90 g also were reproductively mature adults (Santos 2004).

Just before their release newly captured animals were anesthetized lightly by inhalation with isoflurane (Halocarbon Industries, Eagle River, New Jersey), after which they were marked with a uniquely numbered metal ear tag (Monel 1005-1; National Band and Tag Company, Newport, Kentucky) applied to 1 ear. A small tissue sample (≤2 mm of the distal end of 1 pinna) was collected as part of ongoing studies of kinship and parentage in the study population, the results of which are reported elsewhere. Following recovery from the anesthesia, each individual was released into the burrow entrance at which it had been captured.

Radiotelemetry.—During the 2006–2008 field seasons we used radiotelemetry to quantify space use and nest sharing by adults in the study population. Radiotransmitters (model BD-2C; Holohil Systems Ltd., Carp, Ontario, Canada) were attached to the animals using a wire collar that was sheathed in silicon tubing to minimize risk of injury. Collar weight (2.0 g) did not exceed 2.0% of mean adult body mass (130.5 g ± 19.1, n = 238 animals) and had no apparent impact on the behavior of the study animals. Because the number of transmitters available for use was limited, no more than 3 adults from each of the putative burrow systems were fitted with radiocollars. Although T. yonenagae is nocturnal, we were particularly interested in determining whether adults shared burrow systems and nest sites and thus we recorded fixes (i.e., location of the strongest signal for a radiocollared individual) during the daytime (0600–1200 h and 1500–1700 h) when the animals were most likely to be underground.

To locate radiocollared animals we used a digital telemetry receiver (model R-1000; Communication Specialists, Inc., Orange, California) and a 3-element, handheld yagi antenna. Once the signal for an individual had been detected we quietly walked toward the signal until its amplitude indicated that we were standing directly over the animal. Based on fixes recorded for transmitters buried at known locations, we found this procedure to be accurate to 0.5 m. For each animal located via telemetry we determined 3 locations for its position before placing a flag labeled with the animal identification number and date and time of the fix at the point where the strongest signal was detected. At least 30 min were allowed between successive fixes for the same animal; preliminary telemetry data indicated that individuals were rarely in the same location after 30 min, suggesting that this interval was sufficient to minimize autocorrelation of successive data points (Swihart and Slade 1997). Radiotracking continued for 4–5 days, after which the locations of fixes were determined by measuring the compass angle and distance (m) of each flag from the same fixed reference point used to map capture sites (Lacey et al. 1997). At the end of each field season radiocollared adults were recaptured and their collars were removed.

All field procedures followed institutional guidelines and the guidelines of the American Society of Mammalogists (Gannon et al. 2007). The study was conducted under permits issued by the Instituto Brasileiro do Meio Ambiente (0123475 BR).

Analyses of spatial data.—To characterize space use by members of the study population the location of each radio fix was transformed into Cartesian coordinates (Lacey et al. 1998). To characterize space use a 95% minimum convex polygon (MCP) was generated for each radiocollared individual using the Animal Movement extension of ArcView GIS 3.2 (ESRI Technology Inc., Redlands, California). MCPs were used to facilitate comparisons between our data and results obtained from other rodent species (Ribble et al. 2002; Seamon and Adler 1999); 95% rather than 100% MCPs were used because the former are somewhat more conservative with respect to estimates of spatial overlap among individuals. Pairwise calculations of the percent overlap between individuals were completed using ArcGIS 9.0 software package (ESRI Technology Inc., Redlands, California). To determine whether adults shared nests we identified the putative nest for each radiocollared animal as the location most frequently used by that individual. Given the accuracy of our telemetry data, all fixes located within a radius of 0.7 m of one another were treated as the same location.

To characterize the individuals sharing a putative burrow system we identified all adults that were captured in the same cluster of burrow entrances during the same field season. These data were compared to patterns of overlap for 95% MCPs to determine whether individuals captured in the same putative burrow system were spatially distinct from animals captured in other clusters of burrow entrances. Using these data we then determined the number of adults of each sex associated with the same putative burrow system.

Statistical analyses.—Parametric statistical tests were used for data that met the associated assumptions. To determine whether the number of fixes per individual influenced our spatial analyses we used Spearman correlation (rs) analysis to examine the relationship between number of fixes and size of area used (95% MCP) for 6 randomly chosen radiocollared adults (n = 2 per year); significance levels were adjusted using sequential Bonferroni correction when assessing multiple tests. We used a Student's t-test to verify differences in sizes of 95% MCPs obtained with 30- versus 60-min intervals between fixes. We used a Wilcoxon signed-rank test to compare the proportions of fixes associated to the 1st most frequently recorded location versus the 2nd most frequently recorded location. Analysis of variance was used to examine home-range size and percent overlap for 95% MCPs as a function of sex and year, with percentages arcsin-transformed prior to analysis. A Mann–Whitney U-test was used to compare the degree of spatial overlap among adults captured in the same cluster of burrow entrances (i.e., burrow mates) to the degree of overlap among animals captured in distinct clusters of entrances (i.e., non–burrow mates). Statistical analyses were performed using JMP 7.0.2 (SAS Institute Inc. 2009). Means are reported ± 1 SD, with statistical significance set at α = 0.05. All statistical tests were 2-tailed unless otherwise noted.


A total of 180 adult (95 male and 85 female) and 66 juvenile (29 male and 37 female) T. yonenagae were captured during this study. Only 31 (17.2%) of these adults were recaptured in the same burrow system in successive years; no more than 1 adult per burrow system was recaptured in different years. As a result we considered data from each field season to be independent for the purposes of characterizing burrow use.

Based on the locations of active burrow entrances, the study site contained a mean of 54.8 ± 8.8 (range = 42–60) occupied burrow systems per year (Table 1). The increase in number of burrow systems per year over the course of the study was due primarily to the presence of newly occupied burrows on the study site in each successive year. When all clusters of burrow entrances (active and inactive) were considered, the mean percentage of burrow systems occupied per year was 55.5% ± 14.2% for the 4 years; mean annual density of occupied burrow systems was 9.9 ±1.6 systems/ha (Table 1). Based on the locations of burrow entrances assigned to the same system, the area encompassed by individual burrow systems ranged from 4 to 320 m2; however, because subterranean portions of these systems were not quantified, these values do not provide robust estimates of actual sizes of individual burrow systems. When data from all years were considered, mean distance between the spatial centers of adjacent burrow systems (as determined from the locations of burrow entrances) was 22.6 ± 7.3 m (range = 8.4–41.7 m, n = 100 pairwise comparisons of adjacent systems). The number of burrow entrances per system ranged from 1 to 13 (= 5.4 ± 4.2 entrances/system, n = 219 systems).

View this table:
Table 1

Occurrence of burrow systems of Trinomys yonenagae on the study site in Brazil. For each year of the study number and density of burrow systems (occupied and unoccupied) are indicated, with number and percentage of all burrow systems occupied by T. yonenagae.

YearTotal number of burrow systems (n)Density of burrow systems mapped (total/ha)Occupied burrow systems
n% of total
All 4 years (X̄ ± SD)54.8 ± 8.813.6 ± 2.229.5 ± 3.455.5 ± 14.2

Analyses of burrow use.—Of 34 adults fitted with radio-collars, 3 disappeared from the study population and 2 moved to a different burrow system during the course of data collection. An additional 2 radios failed before data collection was completed. As a result, analyses of space use were based on radiofixes from 27 individuals (2006: 6 males and 5 females; 2007: 5 males and 5 females; 2008: 4 males and 2 females). Telemetry data collected over 4–5 consecutive days yielded a minimum of 33 fixes per individual (2006: 48.1 ± 4.2 fixes/individual, n = 11 adults; 2007: 50.2 ± 3.9 fixes/individual, n = 10 adults; 2008: 59.7 ± 5.4 fixes/individual, n = 6 adults).

The number of fixes per individual was correlated with 95% MCP size for each of the 6 randomly selected individuals examined (all rs values ≥0.60, all P < 0.0001, Bonferroni corrected α = 0.008). In all cases, however, the relationship between number of fixes and cumulative 95% MCP size reached a plateau at ≤35 fixes per individual (2006: ≤27 fixes; 2007: ≤29 fixes; 2008: ≤35 fixes), suggesting that the mean number of fixes per animal obtained during this study provided a reliable estimate of the burrow area used by an individual. Changing the interval between fixes appeared to have no effect on estimates of the area used by an individual; resampling of the data set revealed no significant differences in size between 95% MCPs calculated for the same individual using 30-min compared to 60-min intervals between fixes (n = 19 individuals; t36 = −0.259, P = 0.79). Although these analyses do not preclude temporal autocorrelation among telemetry fixes for the same individual, they suggest that our analyses of space use were not influenced substantially by the interval between fixes.

When all radiocollared animals were considered, the sizes of 95% MCPs did not differ with sex (males: 5.63 ± 4.40 m2, n = 15, females: 6.68 ± 4.90 m2, n = 12) or year (2006: 4.42 ± 4.09 m2, n = 11; 2007: 8.49 ± 4.98 m2, n = 10; 2008: 5.16 ± 3.43 m2, n = 6; F2,27 = 2.487, P = 0.10). When data from all individuals were pooled, the mean size of 95% MCPs was 6.36 ± 4.44 m2 (range = 0.55–15.81 m2, n = 27 adults).

Evidence of burrow sharing.—Few burrow systems were trapped completely (i.e., all animals captured) in 2008, and thus this year was not included in analyses of burrow sharing and occupancy. During the 2005–2007 field seasons all adults (n = 156) resident in 88 burrow systems were captured. Of these animals, 121 (77.6%) were caught at >1 burrow entrance; in all cases, capture localities for the same individual were located within the same cluster of burrow entrances. Based on these data, we assigned animals captured within the same cluster of burrow entrances to the same burrow system.

Analyses of 95% MCPs revealed that adults captured in the same burrow system displayed extensive spatial overlap with one another. Percent overlap between 95% MCPs for adults captured in the same burrow system did not differ between years (2006: 66.3% ± 29.2%, n = 20 pairs of burrow mates; 2007: 83.4% ± 20.4%, n = 10; 2008: 72.2% ± 23.3%, n = 6; F2,36 = 1.282, P = 0.19) or with the sex(es) of the individuals compared (male–male: 72.9% ± 23.9%, n = 10 pairs of burrow mates; female–female: 72.9% ± 44.7%, n = 4; male–female: 71.4% ± 25.4%, n = 22; F2,36 = 0.0119, P = 0.87), and thus data from all burrow mates were pooled for subsequent analyses. Based on this pooled data set, the mean overlap between 95% MCPs for burrow mates was 70.0% ± 27.1% (n = 36 pairs of burrow mates). In contrast, no overlap was detected between 95% MCPs for the 36 pairs of non–burrow mates for which telemetry data were available. This difference in percentage overlap was significant (U2,36 = 7.87, P < 0.0001), suggesting that individuals overlapped spatially only with burrow mates.

On average, the single most frequently recorded location for an animal represented 56.4% ± 23.7% of the total number of fixes for that individual (n = 27 adults). In contrast, the 2nd most frequently recorded location represented only 18.6% ± 6.5% of the total number of fixes per individual. This difference was significant (Z = −4.829, P < 0.0001). Because telemetry data were recorded during the day when study animals were largely inactive, we assumed that the most common fix locality for each adult represented that individual's nest site (Fig. 2). For individuals assigned to the same burrow system, 18 (66.7%) of 27 burrow mates shared the same putative nest site. Overall, the mean distance between putative nest sites was 0.84 ± 0.41 m (n = 27 pairs of burrow mates). In comparison, the mean distance between putative nests for adults resident in different burrow systems was 56.0 ± 40.9 m (n = 10 randomly selected pairs of nearest-neighbor nests). Collectively, examination of these data suggests that adults in the study population shared burrow systems and, in some cases, apparent nest sites.

Fig. 2

Spatial overlap for 6 adult Trinomys yonenagae in Brazil monitored via radiotelemetry during 2008. The animals shown were resident in 3 distinct burrow systems; for each system the locations of burrow entrances are denoted by squares. For each individual a 95% minimum convex polygon (MCP) was constructed based on locations determined during daylight hours by radiotelemetry. For each burrow system MCP is denoted by a solid line for 1 spiny rat and a dashed line for the other; the shared putative nest site for each pair of animals is indicated with a star. In all cases additional adults were resident in each burrow system but were not monitored via telemetry.

Characterization of burrow occupancy.—More than 1 adult was captured at 233 (64.4%) of the 362 burrow entrances at which traps were set during this study. Adults in 12 of these systems also were monitored via telemetry. In all cases telemetry data confirmed that the individuals captured in a given cluster of burrow entrances were resident in that burrow system. Across years (2005–2007) the majority of burrow systems (76.1% ± 2.7%, n = 67) were occupied by ≥2 adults (Table 2). The number of individuals captured ranged from 2 to 5 (2.9 ± 1.3) adults and 0 to 4 (1.7 ± 0.48) juveniles per burrow system. Burrow systems occupied by opposite sex pairs (1 male and 1 female) or multiple adult males (≥2 males and no females) were most common, although burrow systems occupied by multiple females (≥2 females and no males) and by mixed sex groups (≥3 adults of both sexes) also were encountered (Table 2). Twenty burrow systems contained multiple adult females; in 7 (35%) of these systems, >1 female was reproductively active, suggesting that multiple adult females can breed while occupying the same burrow system.

View this table:
Table 2

Characterization of adult Trinomys yonenagae residents in 88 burrow systems monitored in Brazil in 2005–2007. In all cases all adults resident in the same burrow system during the same field season were captured. Data from 2008 are not included due to the limited number of burrow systems from which all animals were trapped. Same-sex groups consisted of ≥2 adults of the same sex; mixed-sex groups consisted of ≥3 adults of both sexes.

Same-sex groups
Lone adultsOpposite-sex pairsMalesFemalesMixed-sex groupsTotal number of systems
YearMalesFemales% of totaln% of totaln% of totaln% of totaln% of total


Examination of our data indicates that free-living adult T. yonenagae share burrow systems and, in some cases, putative nest sites. In our study population multiple adults were captured in >75% of the burrow systems monitored. Adults captured within the same cluster of burrow entrances exhibited substantial spatial overlap, and individuals captured in different clusters never overlapped spatially with one another. Individuals occupying the same cluster of burrow entrances also typically shared the same putative nest site. Collectively, these findings suggest that T. yonenagae meets both spatial criteria—burrow sharing and nest sharing—typically used to diagnose sociality in fossorial rodents (Lacey 2000; Urrejola et al. 2007). Although additional research is needed to characterize the social structure (e.g., patterns of kinship among burrow mates) of T. yonenagae in detail, our data are consistent with previous studies that have reported high rates of affiliative behavior among captive adult torch-tail spiny rats (Freitas et al. 2008, 2009) and provide the 1st quantitative evidence of group living among free-living members of this species.

The number and sex(es) of the animals occupying burrow systems varied markedly within the study population. Adults sharing the same burrow system included same-sex pairs, male–female pairs, and groups of ≥3 adults of both sexes. This variation in group composition resembles that reported for other species of desert-dwelling rodents such as Meriones unguiculatus (Ågren et al. 1989), Rhabdomys pumilio (Schradin and Pillay 2004), and Rhombomys opimus (Randall et al. 2005). Randall et al. (2005) have argued that this type of flexible social structure is an adaptive response by members of desert species that must cope with harsh and unpredictable environments. More specifically, as ecological conditions in desert habitats change, animals may shift between living alone and living in groups to maintain the social setting that is most adaptive for a given set of environmental conditions. The dune habitat in which T. yonenagae lives varies temporally with respect to factors such as rainfall and food availability (Rocha et al. 2004; Rocha and Rodrigues 2005), suggesting that variation in burrow occupancy in this species also might reflect adaptive responses to variable environments.

Three ecological factors that are frequently identified as promoting sociality are food resources, predation, and thermoregulatory requirements (Alexander 1974; Ebensperger 2001). Although the ecology of T. yonenagae has not been well characterized, both the patchy spatial distribution and variable temporal production of the Eugenia seeds that this species consumes suggest that availability of food resources and, in particular, the formation of food caches (Santos 2004) might be an important component of burrow sharing in this species. No quantitative studies of predation have been conducted for T. yonenagae, but these animals are the only small mammals found in the dunes near Ibiraba (Rocha 1995) and are thus likely to be the primary prey item for multiple predators. T. yonenagae appears to evade most predators by darting into burrows, suggesting that these structures are an important source of predator protection (Ebensperger and Blumstein 2006; Ebensperger and Wallem 2002; Hayes et al. 2007). Burrow systems also could function as refugia from the extreme heat and aridity of the environment in which T. yonenagae occurs (Rocha 1991). Construction of subterranean burrows is thought to be energetically expensive (Ebensperger and Bozinovic 2000; White et al. 2006) and burrow sharing could allow individuals to reduce the costs associated with access to this critical resource. Future studies will evaluate these potential ecological influences on burrow sharing by T. yonenagae in greater detail.

Spiny rats in the genus Trinomys are primarily forest-dwelling animals that lack conspicuous morphological adaptations for specialized forms of locomotion. T. yonenagae differs from other echimyids—including other species of Trinomys—in numerous ways, including morphology (elongated hind limbs and tail—Rocha 1995), mode of locomotion (e.g., saltatorial movement associated with sandy habitats— Rocha et al. 2007), and reliance on subterranean burrows (fossoriality—Mares 1980). Although other echimyids generally have been assumed to be solitary (Trinomys [Bergallo 1994, 1995], Proechimys [Aguilera 1999; Emmons 1982], and Thrichomys [Streilein 1982]), several recent studies suggest that the social systems of these animals might be more complex than previously realized. For example, home ranges of adult gorgona (or Tome's) spiny rats (Proechimys semispinosus) overlap extensively, with males and females sharing nest sites on a short-term basis (Endries and Adler 2005; Seamon and Adler 1999). The arboreal southern bamboo rat (Kannabateomys amblyonyx) appears to change its mating system from social monogamy to polygyny in response to the distribution of food resources (Silva et al. 2008). Finally, both field studies (Guichón et al. 2003) and molecular analyses (Tunéz et al. 2009) indicate that the semiaquatic coypu (Myocastor coypus) lives in groups composed of multiple adults of both sexes. Collectively, these studies suggest that some degree of sociality could be more common among echimyids than typically has been assumed, with the result that group living in T. yonenagae might not represent a marked contrast to other members of this family.

Because T. yonenagae shares a number of phenotypic attributes with other burrow-dwelling rodents from arid habitats (Rocha 1995; Rocha et al. 2007), comparative studies of these taxa could yield important insights into the ecological bases for group living in desert species. Several other arid-adapted rodents are known to engage in burrow sharing (R. opimus [Randall et al. 2005], M. unguiculatus [Ågren et al. 1989], and Notomys alexis [Watts and Aslin 1981]), providing the opportunity to use convergent patterns of social structure to generate robust tests of ecological hypotheses for group living (Ebensperger 2001; Ebensperger and Cofré 2001). Future studies of torch-tail spiny rats will exploit this framework to explore the effects of food resources, predation, and thermoregulation on the tendency to live in groups. In addition to clarifying the ecological bases for burrow sharing in T. yonenagae, such studies will facilitate understanding of variation in social structure among all burrow-dwelling desert rodents.


We thank L. D. Hayes and L. A. Ebensperger for inviting us to contribute to this special feature on caviomorph behavior. For helpful comments and suggestions on a 1st version of this manuscript we thank J. R. Burger and W. Lidicker. We also thank N. Bennett, B. H. Blake, and an anonymous reviewer whose comments and suggestions greatly improved this manuscript. We are indebted to A. Rio Branco, J. C. de Souza, and I. Bispo de Assis from Ibiraba, Brazil, for their tremendous assistance in the field. For their invaluable support during this work, we thank T. Argolo, T. Barduke, R. Burger, W. Fahning, J. N. Freitas, J. Hamilton, A. Mendonça, I. Moradillo, R. Oliveira, V. Rios, and, especially, E. Sena, T. Praseres, and T. Serravalle de Sá. Comments and suggestions by J. Woodruff, L. Benedict, M. MacManes, and S. Diaz-Muñoz, from the Lacey Lab, greatly improved the manuscript, and we thank them. We thank J. Wieczorek, whose ideas and suggestions greatly improved our field techniques and data analyses. Logistic and intellectual support was provided by P. L. B. da Rocha, M. T. Rodrigues, and J. L. Patton. We are grateful to M. Koo from the Museum of Vertebrate Zoology at the University of California, Berkeley, for her help with the map illustration. During the preparation of this manuscript JWAS was supported by a predoctoral fellowship (200755/2004-8) granted by the Conselho Nacional de Desenvolvi-mento Científico e Tecnológico from the Brazilian Ministry of Science and Technology. For permission to conduct fieldwork we thank the Instituto Brasileiro do Meio Ambiente e dos Recursos Naturais Renováveis. This study was supported financially and logistically by the Museum of Vertebrate Zoology and the Department of Integrative Biology at the University of California, Berkeley. Research grants were provided by the American Society of Mammalogists, the IDEAWILD Foundation, the Center for Latin American Studies at the University of California, Berkeley, and Sigma-Xi, The Scientific Research Society.


  • Special Feature Editor was Barbara H. Blake.

Literature Cited

View Abstract