Defense of a limited resource, such as space or food, has recently been discovered in snakes and has been widely documented in lizards. Garter snakes (Thamnophis spp.) are historically considered generalist predators such that food is not a limiting resource. However, in this study we show that the common garter snake (Thamnophis sirtalis) and the aquatic garter snake (Thamnophis atratus) show a strong preference for amphibians as their primary food source at the Santa Lucia Preserve (SLP), Monterey County, California. This food preference forces these snake species at SLP to exploit aquatic habitats. Our principle goal was to investigate the aggressive behavior of T. sirtalis and the potential that this aggression displaces T. atratus from its preferred habitat. We found that when individuals from either species are alone, a 100% preference for aquatic or near aquatic habitat is observed. In contrast, when these species are together, T. sirtalis occupy the aquatic habitat and T. atratus occupy an area far removed from water. Thamnophis sirtalis often physically force T. atratus from the aquatic habitat through repeated biting and other displays of aggression.
Citation: Edgehouse M, Latta LC IV, Brodie ED III, Brodie ED Jr (2014) Interspecific Aggression and Habitat Partitioning in Garter Snakes. PLoS ONE 9(1): e86208. doi:10.1371/journal.pone.0086208
Editor: Mark Briffa, University of Plymouth, United Kingdom
Received: May 20, 2013; Accepted: December 9, 2013; Published: January 22, 2014
Copyright: © 2014 Edgehouse et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This work was supported by National Science Foundation grants to EDB Jr and EDB III (DEB-0315172, DEB-0922251). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Aggressive interactions within or between species that arise from competition for a limited resource are an important ecological mechanism structuring communities . Numerous communities of vertebrates, such as desert rodents, salamanders, and lizards, are structured by both intra- and interspecific aggressive interactions –. Conspicuously absent from those vertebrates in which aggression is an important determinant of community structure are snakes. There is evidence that intraspecific aggression, such as competition for a limited resource  and male-to-male combat in response to limited female availability –, can shape the population structure of some snake species. However, in snakes, food partitioning typically shapes the community structure of sympatric snake populations –, while there is no evidence that interspecific aggression is found among communities of snakes.
Snake communities can segregate or coexist on the basis of intra- or intergeneric interactions. A community of Thamnophis sirtalis and Coluber constrictor coexisted based on food partitioning  (T. sirtalis preying mainly on birds and C. constrictor preying mainly on rodents). Information on the diet of a species often provides information on the preferred habitat of an organism when direct observation is not possible –. Additionally, diet data are informative for other aspects of an organism such as growth, resource utilization, and potential competition between closely related species . Another mechanism that shapes snake communities is pre-emptive competition. For example, sympatric Thamnophis spp. may have increased competition based on similarities in body size, diet, and general habits . Often this pre-emptive competition for space or food can lead to segregation or local extinction , , , , .
This study focuses on snake communities at the Santa Lucia Preserve (SLP), Monterey Co. CA. Santa Lucia Preserve is a locality where T. sirtalis, T. atratus and T. elegans coexist with an abundant, toxic prey item, the California newt (Taricha torosa). Thamnophis sirtalis and T. atratus from SLP are aquatic and demonstrate independently evolved elevated levels of resistance to tetrodotoxin (TTX) , a potent neurotoxin, found in the skin of newts that binds to voltage gated sodium channels (reviewed by –). Additionally, during collections at SLP it was noted that T. sirtalis were found near water at certain locales while T. atratus were found far removed from water at these same locales. In contrast, T. atratus were found at the water's edge only at ponds without T. sirtalis, suggesting that the habitat utilization of T. atratus depends on the presence of T. sirtalis.
The similarities in the physiology (resistance to TTX) of these species suggests that they utilize similar prey items, however the drastic change in habitat utilization by T. atratus in the presence of T. sirtalis suggests that T. sirtalis is actively displacing T. atratus.
Snakes were collected by hand at the following locations throughout the SLP; 36°26.24.32 N 121°47.33.43 W, 36°26.001 N 121°44.664 W, 36°27.741 N 121°46.876 W, 36°27.179 N 121°46.684 W, 36°26.21.85 N 121°46.49.68 W, 36°27.27.19 N 121°47.58.07 W. All of the collection localities had the same principle conditions of a water source surrounded by vegetation. Upon collection, animals were forced to regurgitate by stomach palpation , measured for snout-vent length (SVL), weighed and uniquely scale clipped for identification. Animals were then housed individually in bags in a cooler at 20 C and watered once a day for five days prior to the start of each behavioral trial. Housing snakes individually for five days insured that all animals used in the trials were exposed to the same conditions prior to the start of each behavioral trial. We do not believe that the collection of natural history data nor the ventral clipping affected the behavior trials.
Ten 4 m×1 m×0.45 m enclosures made with black, half centimeter weatherproof nylon mesh were utilized to determine microhabitat preference. Enclosures were marked each meter and set 1 m in water and 3 m out of water. One cover object (0.3 m×0.3 m×12.5 mm Styrofoam®) was provided on the ground in the meter farthest from the water (zone 1; Fig. 1). Each trial consisted of a 24-hour period with one animal in the enclosure. Snakes were placed in the enclosure at 0700 hours, six observations of position, movement, and behavior were recorded for each snake during each trial. Data were recorded at 0800, 1000, 1200, 1400, 1600 and 1800 hours, from a distance of approximately 20 m to avoid observer influence; observations were repeated as above for trials with two snakes in the chamber. All observations during trials were made using binoculars by the same individual (ME). At each observation, the snake's position was recorded as 1 (zone farthest from water) 2, 3, or 4 (zone in water; Fig. 1), depending upon the zone in which the majority of the snake's body was located. If an animal was on the border between two zones, the head of the snake was used to determine the zone. Instances of aggressive behavior directed toward the heterospecific (during trials with more than one snake), such as biting, chasing, head flattening, and hissing were recorded. Observations of initial hissing were removed from analysis because of possible observer influence, subsequent observations of hissing were made by observing behavior similar to that noted during initial hissing observations; this behavior has been classified as head flattening for analysis. Biting often occurred as 2–3 directed strikes occurring rapidly in a row. Each sequence of strikes was counted as 1 bite. More than one aggressive behavior was recorded as observed. For example, an obvious head flattening followed by 3 strikes was counted as 1 head flatten and 1 bite. Trials were conducted on the southeast corner of a managed wetland at SLP (36°26.25.39 N, 121°47.35.19 W). This site was chosen to conduct behavior trials because shallow water enabled the observer to position himself at multiple angles to the enclosure 20 meters away.
Figure 1. Behavior test chamber diagram.
Enclosures were 4- meter delineations, by 1 meter wide, by .45 meters high. Each 1-meter zone was assigned numbers 1, 2, 3, or 4. Zone 1 was farthest from water and contained one 30 cm×30 cm×1 cm Styrofoam cover object (indicated by an asterisk), zone 4 was in the water. Snakes were introduced into the enclosures in zone 1.doi:10.1371/journal.pone.0086208.g001
Animals were assigned to trials matching SVL as close as possible (Table 1). Twenty T. sirtalis vs. T. atratus trials were conducted starting on 06 May 2005. These trials consisted of placing one T. sirtalis in each enclosure for 24 hours and recording observations as described. Upon completion of this 24-hour period a T. atratus was added and six observations were again recorded for 11 hours as described above. Because we could not collect enough snakes, and because we did not use any individual in more than one trial, 13 T. atratus vs. T. sirtalis trials matching sex (all male) and SVL as closely as possible were conducted starting on 26 May 2006. These trials consisted of placing one T. atratus in each enclosure for 24 hours and recording observations as described. Upon completion of this 24-hour period a T. sirtalis was added and observations were again recorded for 11 hours. Subjects were used in only one set of trials. Two sets of 10 control trials with either T. sirtalis or T. atratus used as both focal and introduced species were conducted following the same design.
Table 1. Descriptive statistics for snakes used in behavior trials.doi:10.1371/journal.pone.0086208.t001
Snakes were collected by hand throughout SLP between 05 May 2004 and 27 September 2006. Upon capture, each animal was SVL measured, weighed, given a unique ventral scale clip for future identification, and forced to regurgitate by stomach palpation ; each food item was identified. Subjects were then released at the site of capture.
The study was observational and subjects were held a minimum length of time for the purpose of this study; each individual was used only once. No individual was injured. Research was approved by the Utah State University Institutional Animal Care and Use Committee, Protocol #1008. We thank we thank California Department of Fish and Game for providing scientific collecting permits (801071–05).
To assess how the presence of a conspecific or heterospecific snake influences habitat preference in the focal snake species we analyzed data at each of six sampling intervals for three groups that describe the position of the focal subject, conspecific, or heterospecific snake. The groups were: 1) the position of the focal snake alone in the enclosure, a measure of the habitat preference of the focal species; 2) the position of the focal snake in the enclosure in the presence of the conspecific or heterospecific species, a measure of the habitat preference of the focal species in the presence of a conspecific or heterospecific; and 3) the position of the conspecific or heterospecific subject in the enclosure in the presence of the focal, a measure of the habitat preference of the conspecific or heterospecific.
In order to account for the non-normality and heteroscedasticity in our data, and the fact that we have repeated measures on the same subjects, we used the Friedman test to assess statistical differences between the three groups we defined previously . The Friedman test is a non-parametric test for differences among levels of a grouping factor when there is also a blocking factor with multiple levels that may contribute to variation in the levels of a group. For our analysis, the grouping factor is snake species/combination (three levels) and the blocking factor is the sampling interval (six levels).
If the position of the focal species alone does not differ from the position of the focal species in the presence of the conspecific or heterospecific then the conspecific or heterospecific does not alter the habitat preference of the focal species. Alternatively, if the position of the focal species alone does differ from the position of the focal species in the presence of a conspecific or heterospecific then the conspecific or heterospecific does alter the habitat preference of the focal species. All habitat preference analyses were conducted in Program R using the agricolae package .
One concern is that the patterns we observe may be driven by differences in body size between the focal and added snakes. In order to test whether size is an important determinant of movement between positions in the experimental chambers we first calculated two metrics that describe differences in body size and position. The first metric, size differential, was calculated by subtracting the SVL of the added species from the SVL of the focal species. The second metric, movement differential, was calculated for each time-step by subtracting the position of the focal species in the experimental chamber after the addition of the second species from the position of the focal species before the addition of the second species. We calculated these metrics for the dataset in which T. sirtalis was the focal species and T. atratus was the added species, and also for the dataset in which T. atratus was the focal species and T. sirtalis was the added species. For each dataset, we then performed regression analyses using size differential as the independent variable and movement differential as the dependent variable.
To examine the differences in diet among the snake species at SLP we used a generalized linear model assuming a Poisson distribution for the prey counts with species treated as the main effect. We obtained p-values by using a Chi-squared test to compare the reduction in deviance of the main effect to the residuals. Each of the two prey types recovered from the snake species (i.e., anurans and salamanders) were analyzed separately.
Both T. sirtalis and T. atratus occupied the aquatic portion of the test chamber when in the chamber alone (Fig. 2). There was a significant difference in the spatial occupation of T. atratus (focal species) and T. sirtalis (χ2 = 10.33; p = 0.006) when in the test chamber together. There was also a significant difference in spatial occupation of T. atratus when solo in the test chamber and the position of T. atratus when the chamber was co-occupied by T. sirtalis (χ2 = 6.00; p = 0.014). In every series of trials (13 total) with a single T. atratus in the test chamber, T. atratus preferred positions close to, or in the water (positions 3 and 4; Fig. 2 and 3). However, in the presence of T. sirtalis, the position of T. atratus shifted significantly to terrestrial habitats (positions 1 and 2; Fig. 3A). Most of the T. sirtalis used in these trials were observed to exhibit aggression (10 of 13). Fifteen aggressive displays (11 head flatten, 4 strikes) were observed from 10 different T. sirtalis; it is likely we failed to observe other aggressive interactions from the distance of 20 m. Fourteen of 15 aggressive displays were observed between 800 and 1200 hours. We saw no aggression from T. atratus in any trial. Additional trials alternating focal species (T. sirtalis solo, T. atratus introduced) were conducted and the results indicate no significant differences in spatial occupation by T. sirtalis were present in any trial (χ2 = 2.33; p = 0.311); when T. sirtalis were in the test chamber first they occupied the water, or near the water and T. atratus did not approach the water (or the T. sirtalis). There were no observations of aggression from T. sirtalis, as the focal snake. In the presence of conspecific snakes, neither T. sirtalis (χ2 = 1.09; p = 0.580) nor T. atratus (χ2 = 1.45; p = 0.483) altered their position in the enclosure.
Figure 2. Median positions of Thamnophis atratus (TA) and Thamnophis sirtalis (TS) when alone in the enclosures at SLP.doi:10.1371/journal.pone.0086208.g002
Figure 3. Median positions of Thamnophis atratus (TA) and Thamnophis sirtalis (TS) in enclosures for the SLP locality.
The focal species before (filled circles and solid lines) and after (open circles and solid lines) the addition of the second species (squares and dashed lines) are indicated in each panel. A) The response of T. atratus to the addition of T. sirtalis. B) The response of T. sirtalis to the addition of T. atratus. Only T. atratus showed a significant change in position in response to the addition of T. sirtalis.doi:10.1371/journal.pone.0086208.g003
In both sets of trials T. sirtalis was larger than T. atratus (Table 1). Our analyses for the effect of size on movement suggested that differences in SVL do not explain differences in movement in the experimental chamber. Neither the regression when T. sirtalis was the focal species, nor when T. atratus was the focal species, yielded significant results (T. sirtalis: df = 102, F = 1.614, p = 0.2069; T. atratus: df = 115, F = 0.28, p = 0.598).
Diets of T. atratus and T. sirtalis were similar based on generalized linear modeling (anuran: deviance = −0.005, p = 0.94; salamander: deviance = −1.34, p = 0.25; Figure 4). Stomach contents reveal that both T. sirtalis and T. atratus are amphibian specialists at this locality; 100% of the prey recovered from stomachs of both species were amphibians. The most abundant amphibian prey recovered in the stomachs and observed in the field were Taricha torosa and Pseudacris regilla (Northern Pacific Treefrog). We have no food records from T. atratus found away from water and cannot confirm they were feeding when in the terrestrial habitat.
Thamnophis sirtalis potentially displace T. atratus with aggressive behavior under experimental conditions and field observations did not detect these two aquatic species syntopically at any of the ponds at SLP. This is the first time interspecific aggression has been implicated in microhabitat distribution within snake communities. Similar to observations of kukrisnakes that defend a food resource from conspecifics by aggression , this defense of territory by T. sirtalis was often accompanied by direct physical interaction. Ten of 13 T. sirtalis were observed to display aggression. Fifteen aggressive displays (11 head flatten, four strikes) were observed from these 10 different T. sirtalis. However, in contrast to kukrisnakes, which bit conspecifics, this behavior involved physical displays of body and head flattening and typically repeated strikes towards the anterior portion of the heterospecific individual. The majority of aggressive displays were observed between 800 and 1200 hours, possibly indicating that dominance had been established.
The shift in the preferred spatial occupation of T. atratus from aquatic to terrestrial habitats upon addition of T. sirtalis indicates that a unique interaction is occurring at SLP. Both species clearly prefer aquatic habitats when alone, however, the addition of T. sirtalis to a chamber occupied by T. atratus resulted in an abrupt shift in zonal occupation by T. atratus from aquatic habitats to terrestrial habitat. When the experiment was conducted in reverse fashion (T. sirtalis focal species) no habitat displacement of T. sirtalis was seen. During trials in which T. sirtalis was the focal species, T. atratus did not enter into the zone occupied by T. sirtalis for the majority of the trial time frame. Only after T. sirtalis moved from zone 4 to zone 3 did T. atratus occupy zone 4 (Fig. 3B). These results suggest that T. sirtalis is the behaviorally dominant species. This conclusion is corroborated by the fact that aggressive displays were made by T. sirtalis toward T. atratus, but never by T. atratus toward T. sirtalis. We acknowledge that using a single observer limited the time frame in which observations were conducted, possibly resulting in an underreporting of aggression. In all trials, T. sirtalis were larger (SVL and mass) than T. atratus. The difference in size may have influenced the behavior of the snakes.
The similarity in diet of T. sirtalis and T. atratus may help explain the unique interaction we have observed. Stomach contents reveal that both T. sirtalis and T. atratus are amphibian specialists in this area, 100% of the prey recovered from stomachs were amphibians. Interspecific aggression leading to displacement is more likely to occur between biologically similar species . If both, T. sirtalis and T. atratus are vying for a limited resource(s) (such as habitat and/or food); it is more likely that interspecific territoriality will occur . The spatial partitioning we have seen at SLP between T. atratus and T. sirtalis may be a direct result of food availability and aggressive defense of this food source. The most abundant amphibian prey recovered in the stomachs and observed in the field were T. torosa and P. regilla. Feldman et al. 2010 showed that 100% of the T. sirtalis and roughly 75% of the T. atratus from SLP are resistant enough to not be affected by the toxin that the newts possess, making this prey a viable food source for all T. sirtalis and the majority of T. atratus. The result of this defense of a limited resource has manifested itself as the spatial occupation shift that we observe.
Summer months at SLP bring many ecological changes, including drying of grasslands and greatly reduced water levels. The changes that occur throughout the summer reduce the suitable habitat and resources for Thamnophis sp. If two species are ecologically similar enough and persist in a homogeneous environment with limited but definable resources, they should exhibit interspecific territoriality –. This hypothesis is applicable to the interactions observed at SLP. During the summer, resources (water and food) utilized by T. atratus and T. sirtalis are declining at a rapid rate, forcing the animals to occupy a more compact space; resulting in a behavior that has not been seen at other locales. Further trials at SLP are warranted to test this hypothesis and observe the long-term consequences of these behaviors. Our data confirm that T. sirtalis from SLP are using aggression to defend a limited resource, the toxic prey, T. torosa, resulting in an ecological niche shift of the behaviorally inferior T. atratus.
The composition of Thamnophis populations throughout Monterey Co. CA suggest aggressive behavior is structuring snake communities there and possibly throughout the region. Levels of tetrodotoxin resistance of snakes vary throughout this region, however, the majority of snakes in Monterey Co, both T. sirtalis and T. atratus, have the ability to consume T. torosa as a viable food source. The factors that maintain segregated snake populations in Monterey Co. are not entirely defined, but diet and behavior data from SLP suggest that these life history aspects help maintain segregation.
Conceived and designed the experiments: ME LCLIV EDBIII EDBJr. Performed the experiments: ME LCLIV EDBIII EDBJr. Analyzed the data: ME LCLIV EDBIII EDBJr. Contributed reagents/materials/analysis tools: ME LCLIV EDBIII EDBJr. Wrote the paper: ME LCLIV EDBIII EDBJr.
- 1. Maynard-Smith J (1982) Evolution and the Theory of Games. Cambridge: Cambridge University Press.
- 2. Pinter-Wollman N, Dayan T, Eilam D, Kronfeld-Schor N (2006) Can aggression be the force driving temporal separation between competing common and golden spiny mice. J Mammal 87: 48–53. doi: 10.1644/04-mamm-a-194r2.1
- 3. Lancaster DL, Jaeger RG (1995) Rules of engagement for adult salamanders in territorial conflicts with heterospecific juveniles. Behav Ecol Sociobiol 37: 25–29. doi: 10.1007/bf00173895
- 4. Smyers SD, Rubbo MJ, Townsend VR Jr, Stewart CC (2002) Intra- and interspecific characterizations of burrow use and defense by juvenile ambystomatid salamanders. Herpetologica 58: 422–429. doi: 10.1655/0018-0831(2002)058[0422:iaicob]2.0.co;2
- 5. Griffis MR, Jaeger RG (1998) Competition leads to an extinction-prone species of salamander: Interspecific territoriality in a metapopulation. Ecology 79: 2494–2502. doi: 10.1890/0012-9658(1998)079[2494:cltaep]2.0.co;2
- 6. Kleeberger SR (1984) A test of competition in two sympatric populations of desmognathine salamanders. Ecology 65: 1846–1856. doi: 10.2307/1937782
- 7. Hess NE, Losos JB (1991) Interspecific aggression between Anolis cristatellus and A. gundlachi: Comparison of sympatric and allopatric populations. J Herpetol 25: 256–259. doi: 10.2307/1564669
- 8. Huang WS, Greene HW, Chang TJ, Shine R (2011) Territorial behavior in Taiwanese kukrisnakes (Oligodon formosanus). Proc Nat Acad Sci U S A 108: 7455–7459. doi: 10.1073/pnas.1101804108
- 9. Firmage M, Shine R (1996) Battles for mates and food: Intraspecific combat in island tigersnakes (Notechis ater) from southern Australia. Amphib-reptil 17: 55–65. doi: 10.1163/156853896x00298
- 10. Greene HW (1997) Snakes, the evolution of mystery in nature. Berkeley, CA: University of California Press.
- 11. Gregory PT, Macartney JM, Larsen KW (1987) Spatial patterns and movements. In: Seigel RA, Collins JT, Novak SS, editors. Snakes: ecology and evolutionary biology. New York: Macmillan Publishing Company. 366–95.
- 12. Duvall D, Schuett GW, Arnold SJ (1993) Ecology and evolution of snake mating systems. In: Seigel RA, Collins JT, editors. Snakes: ecology and behavior. New York: McGraw Hill Inc. 165–200.
- 13. Olson DJ, Warner RE (2001) Grassland snakes diet. Herpetol Rev 32: 186–187.
- 14. Kjoss VA, Litvaitis JA (2001) Community structure of snakes in a human-dominated landscape. Biol Conserv 98: 285–292. doi: 10.1016/s0006-3207(00)00167-1
- 15. Rossi JV, Rossi R (2003) Snakes of the United States and Canada. Malabar, FL: Krieger Publishing Co.
- 16. Luiselli L (2006) Resource partitioning and interspecific competition in snakes: the search for general geographical and guild patterns. Oikos 114: 193–211. doi: 10.1111/j.2006.0030-1299.14064.x
- 17. Reichenbach NG, Dalrymple GH (1980) On the criteria and evidence for interspecific competition in snakes. J Herpetol 14: 409–412. doi: 10.2307/1563698
- 18. Simpfendorfer C (1992) Biology of tiger sharks (Galeocerdo cuvier) caught by the Queensland shark meshing program off Townsville, Australia. Mar Freshw Res 43: 33–43. doi: 10.1071/mf9920033
- 19. Plummer MV (1981) Habitat Utilization, Diet and Movement of a Temperate Arboreal Snake (Opheodrys aestivus). J Herpetol 15: 425–432. doi: 10.2307/1563532
- 20. Voris HK, Voris HH (1983) Feeding Strategies in Marine Snakes: An Analysis of Evolutionary, Morphological, Behavioral and Ecological Relationships. Am Zool 23: 411–425. doi: 10.1093/icb/23.2.411
- 21. Feldman CR, Brodie ED Jr, Brodie ED III, Pfrender ME (2010) Genetic architecture of a feeding adaptation: garter snake (Thamnophis) resistance to tetrodotoxin bearing prey. Proc R Soc, B 277: 3317–3325. doi: 10.1098/rspb.2010.0748
- 22. Hanifin CT (2010) The chemical and evolutionary ecology of tetrodotoxin (TTX) toxicity in terrestrial vertebrates. Mar Drugs 8: 577–593. doi: 10.3390/md8030577
- 23. Kao CY (1996) Tetrodotoxin, saxitoxin, and their significance in the study of excitation phenomena. Pharmacol Rev 18: 997–1049.
- 24. Narahashi T, Moore JW, Poston RN (1967) Tetrodotoxin derivates: chemical structure and blockage of nerve membrane conductance. Science 156: 976–979. doi: 10.1126/science.156.3777.976
- 25. Carpenter CC (1952) Comparative ecology of the common garter snake (Thamnophis s. sirtalis), the ribbon snake (Thamnophis s. sauritus), and Butler's garter snake (Thamnophis butleri) in mixed populations. Ecol Monogr 22 (4) 236–258. doi: 10.2307/1948469
- 26. Conover WJ (1999) Practical nonparametric statistics 3rd edition. New York: John Wiley and Sons.
- 27. deMendiburu F (2010) Agricolae: Statistical Procedures for Agricultural Research. R package version 1.0–9. Available: http://www.inside-r.org/packages/agricolae/versions/1-0-9. Accessed 2013 Dec 31.
- 28. Toft CA (1985) Resource partitioning in amphibians and reptiles. Copeia 1: 1–21. doi: 10.2307/1444785
- 29. Wolff JO, Freeberg MH, Deuser RD (1983) Interspecific territoriality in two sympatric species of Peromyscus (Rodentia: Cricetidae). Behav Ecol Sociobiol 12: 237–242. doi: 10.1007/bf00290776