Behavioral Ecology VoL 8 No. 4: 404-413 Why do kangaroo rats {Dipodomys spectabilis) footdrum at snakes? Jan A. Randall and Marjorie D. Matocq Department of Biology, San Francisco State University, San Francisco, CA 94132, USA We examined alternative hypotheses for the benefits of footdrumming in the presence of snakes by the banner-tailed kangaroo rat, Dipodomys spectabilis, by testing whether the target of the signal includes conspecifics, the predator or both. Footdrumming recorded in the field revealed that rats altered their footdrumming signatures when drumming at snakes. In playback tests, however, neighbors failed to show any measurable change in behavior to broadcasts of die snake drumming pattern, but mothers footdrummed significantly more than nonmothers in the presence of a tethered snake. Gopher snakes, Pituophis •mdanoUucus affinis, responded to footdrumming vibrations created by a mechanical thumper. Nonhungry snakes avoided footdrumming, while hungry snakes approached the seismic footdrumming. Snakes decreased stalking rates as footdrummmg increased, but they spent more time stalking drumming than nondnimming rats. We conclude that D. spectabilis footdrums in individual defense and in parental care, rather than to warn adult conspedfics. Footdrumming deters pursuit by informing die snake that die rat is alert and the chances of predation are low. We find little evidence that footdrumming startles, confuses, or harasses die snake. Hungry gopher snakes, however, may locate prey by eavesdropping on territorial footdrumming. Key words: antipredator behavior, communication, deterrence, Dipodomys spectabilis, footdrumming, kangaroo rat, pursuit, snakes. [Behav Ecol 8: 404-413 (1997)] B anner-tailed kangaroo rats, Dipodomys spectabiUs, are solitary, desert rodents that confront snakes in their territories (Randall and Stevens, 1987; Randall et al., 1995). Both males and females approach snakes, jump back to avoid strikes, and usually begin to footdrum. Snakes can probably sense die footdrumming vibrations through low-frequency receptors in their skin, muscle, and ears (Hardine, 1971; Proske, 1969), and hearing sensitivity in gopher snakes (P. m. wtdanoleucus), which ranges from 25 to 700 Hz (Wever and Vernon, 1960), matches die frequency of die footdrumming signal (Randall, 1984). Kangaroo rats also have ears well-adapted to detect low-frequency vibrations (Webster and Webster, 1984), and D, spectabilis responds to airborne sounds transmitted between territories and seismic vibrations transmitted into the burrow (Randall 1984, 1994; Randall and Lewis, in press). Because alarm signals are costly in time and energy and can attract the attention of predators (Caro et aL, 1995; Hump and Shaker, 1984), they must be adaptive and provide fitness benefit to die prey such as warning genetic relatives of danger (Hamilton, 1964). In D. spectabilis, related neighbors can occupy territories close together to form a loosely organized and spatially dispersed aggregation of kin (Jones, 1984; Jones et aL, 1988; Randall, 1984). Territories consist of large dirt mounds, which are a scarce resource that contain burrows essential for shelter, food storage, and reproduction (Randall, 1984, 1991). A single rat occupies a mound, unless it is a female with young, and defends die territory by chasing away intruders. The rats communicate territorial ownership to conspecific neighbors widi individually distinct footdrumming signatures, which they can discriminate as coming from a familiar neighbor or unfamiliar stranger (Randall, 1989, 1994, 1995). Footdrumming in die presence of snakes could also protect vulnerable offspring. Because of dieir smaller size and inex- M. Matocq U currently at the Museum of Vertebrate Zoology, University of California, Berkeley, CA. 94720, USA. Received 17 June 1996; reviied 18 November 1996; accepted 19 November 1996. 1045-2249/97/$5.00 O 1997 Intem»oon»] Sodety for Behavioral Ecology perience (Owings and Coss, 1977; Owings and Loughry, 1985; Tamura 1989), young rats may be more vulnerable to snakes dian adults and benefit from protection by their mothen (Hennessy and Owings, 1988). Female D. spectabiUs have young rats in the mound during die summer when snakes are abundant (Randall, 1991). Footdrumming could inform die offspring directly of danger, or die drumming might assemble adults for snake-directed mobbing (Klump and Shalter, 1984; Owings et aL, 1986). Alarm signals can also benefit die prey by reducing die alarmist's vulnerability (Triven, 1971). Conspecifics could be die target, while die drummer benefits (Hersek and Owings, 1993; Sherman, 1985), or die target of die drumming could be die predator. Alarm calls function in individual defense when they are directed toward die predator to cause it to abandon the attack on die individual exhibiting die behavior (Buder and Roper, 1994; Caro, 1995, 1986a,b; Caro et aL, 1995; Klump and Shalter, 1984). We examined alternative hypotheses about die benefit of footdrumming in die presence of «nalr<»T by die banner-tailed kangaroo rat by testing whether die target of die signal includes conspedfics, die predator, or both (Caro, 1986a,b; Caro et aL, 1995; Hasson, 1991) (Table 1). We hypothesized dial footdrumming might function as an alarm signal to communicate a warning to relatives about the type of predator (Seyfarth et aL, 1980; Slobodchikoff et aL, 1991; Tamura and Yong, 1993) or die extent of threat and degree of risk (Blumstein and Arnold, 1995; Owings and Hennessy, 1984; Weary and Kramer, 1995). We also hypothesized dial footdrumming in the presence of f"a*»t functions in self-preservation of D. spectabiUs (Buder and Roper, 1994; Caro, 1986a,b; Caro et aL, 1995; Klump and Shalter, 1984). First, because snakes rely on crypsis to ambush their prey (Raddiffe et al., 1986; Sweet, 1985), we predicted that footdrumming informs die snake of detection and diat die chances of ambush are thwarted (Woodland et aL, 1980). Second, we predicted that die rats footdrum to advertise their alertness and awareness of die snake and to communicate that continued pursuit is costly (Caro, 1995). Third, footdrumming could inform die snake that die rat is healthy and can out-maneuver it (Caro, 1995; FitzGibbon and Fanshaw, 1988). Finally, footdrumming may Randall and Matocq • Footdnunming at antipredator behavior 405 imUel Hypotheses for to ttdr Hypotheses Behavior directed to conspccifics 1. Signal to other adulti 2. Mother signals young Behavior directed to snake 1. General 2. Signal to predator a. Detection b. Coat of continued pursuit c Healthy condition S. No signal a. Harasses, startles or confuses predator Prediction Outcome of this study Neighbors alert to snake fbotroU Rats without dose neighbors are less likely to drum snake footrofl Only mothers drum snake footroO Mothers drum more than nonmotbers %ung respond to mothers' drumming Rejected Rejected Rejected Accepted Untested Rats drum die snake pattern in absence of con specifics Rats drum in the presence of snakes Accepted Accepted Rat drums at first sighting of •nnfr^ Snake moves away on hearing drumming Rat drums after approaching y^pk^ Rat drums until snake stops pursuit Rats in poor condition do not drum Rejected Rejected Accepted Accepted Untested Snake appears disoriented or exhibits defensive behavior Rejected confuse, startle, or harass the snake so the rat can escape. Footdnunming might be «ixnilar to mobbing, in which an individual rat fbotdrums to harass a snake to cause it to leave (Loughry, 1987; Owings and Loughry, 1985; Tamura, 1989). METHODS Footdrmmning directed to cosspedfica Study sites and animals We tested kangaroo rats on two study sites in southeastern Arizona, USA. Data collected from 1986 to 1992 were from a population of D. sptctabi&s on a 3.6-ha site established near Portal, Arizona, in 1980 (Randall, 1984). Data collected in 1993 were from a site in an arid grassland about 50 km southeast of the Portal site (Jones, 1984). Populations on the Portal site ranged from 23 rats in 1986 to 35 rats in 1989 with mounds that averaged 20-40 m apart (see Randall, 1995, for details). The 54 animals (21 adult females, 16 adult males, and 17 juveniles) on the second site averaged 35 m apart in 1993. We tethered 430-731 g Arizona gopher snakes, P. m. affinis, at the mounds of kangaroo rats to elicit footdnimming. A tether consisted of a strip of nylon underwrap wound around the snake about 20 cm behind the head to prevent damage to the snake's skin while providing a slip-free surface on which to wrap a strip of adhesive tape and to tie monofilament nylon lines (see Randall and Stevens, 1987, for more details). We r"3"|"|1y manipulated the snake's position with the monofilament line until 1990, when we tied the snake between two stakes about 0.6 m apart so it could move and strike. Observations of behavior We used the same procedures in all tests. An observer in a chair 10-15 m from the mound watched the test subject with binoculars or, after 1991, a night-vision scope (Noctron V) mounted on a tripod. The mound was illuminated by moonlight or by a dim light from a lantern on a 2-m high tripod set at least 10 m from the mound. Behavior was described by speaking into a hand-held tape recorder. All rats were habit- uated to the observer presence and the dim lights for a minimum of 4 h on a night before a test. Fooidruwtming signals We compared footdnunming in the territorial (social) and snake (antipredator) contexts of 15 solitary rats (10 males and 5 females) and 6 females with young for changes in the signal elements that compose the footdnimming pattern. We recorded footdrumming during spontaneous territorial drumming and in response to a snake in 1986, 1988, 1991, 1992, and 1993. Three recordings occurred during natural interactions between a rat and snake; the other recordings were obtained after tethering a gopher snake. Drumming of seven rats was recorded in both contexts on the same night; drumming of the remaining rats was recorded on different nights a Tna-riT"'"" of 2 weeks apart We used the same recording procedures and data analysis as reported in Randall (1989, 1994, 1995). We recorded footdrumming with Uher tape recorders at a tape speed of 9.5 cm/s with a 25-dB preamplifier via geophones placed on the mound near a burrow entrance. We digitized multiple sequences of footdrumming for each individual and counted and measured signal elements in the footdrumming pattern that are known to account for individual differences in the territorial footdrumming signatures (Randall, 1989, 1995). The /Statistic associated with Wilts' lambda was used to test for differences in the footdnimming elements of an individual in the two contexts. If a significant result was found, the univariate /kests determined which signal element! differed (SYSTAT MANOVA; Wilkinson, 1990). PtaybaA Usts We conducted two playback tests with slightly different methods to determine whether rats differentiated territorial footdrumming from antipredator drumming. In July 1989, we compared responses of 15 adult D. sptctabilxs (5 males and 10 females) to three playback stimuli: (1) footdnimming in the presence of snakes, (2) the territorial signature of the same rat, and (3) a control of short bouts of single thumps of a hammer hitting the ground at a rate of about 8/s. We used 406 original recordings of three rats, recorded in both the territorial and snake contexts in 1987, in five playbacks each. Footdrumming patterns differed in one or more «ignni elements from the footdrumming pattern of the nearest neighbor (Randall, 1994). We broadcast the three drumming stimuli in a counterbalanced order on the same night into an apparatus that generated both airborne and seismic sound (Randall, 1994). The output of a Uher recorder passed through a car radio amplifier into a voice coil analyzer that transmitted the vibrations through a fiberglass horn with trimmed edges buried face down into the ground 2 m from the base of the mound. The rat was out of the mound and engaged in normal activity for at least 5 min before the playback began. Each playback lasted 10 min, with a 10-min post-test observation and 20-min rest period in between so that a total of 30 min occurred between each stimulus presentation. We set signal amplitude to mimic the sound of natural footdrumming in the mound and equated the length of drumming during each playback. We recorded all footdrumming responses from the test subject and from neighbors at adjacent mounds and counted the number of footrolls from the tapes. We also tabulated the number of approaches to within 1 m of the speaker and the time spent out of die burrow. Data were analyzed, after log transformations, with a repeated-measures ANOVA (SYSTAT) in a 3X2 design: three playback stimuli (hammer, snake, and territorial footdrumming patterns) and two test periods (playback and post-test). We report multivariate /Statistics when possible because assumptions of homogenous variances and compound symmetry are not required (Wilkinson, 1990). Critical levels of post-hoc t tests were corrected by the Bonferroni pairwise procedure. If data could not be normalized, we combined data from the test and post-test periods and analyzed for different responses to the three playback stimuli with nonparametric statistics. In July 1992, we tested four females and six males for their response to playbacks of the snake and territorial footdrumming of neighbors at adjacent mounds. Playback recordings of 4 neighbors were used once and 3 were used twice in the 10 playback tests. Procedures were similar to those used in playbacks in other studies (Randall, 1994). We broadcast the territorial and snake footdrumming of neighbors in a counterbalanced order via separate 60-s tape loops from a cassettetape recorder through a battery-operated Realistic-brand speaker (RadioShack) placed on the edge of the neighbor's mound facing the mound of the test subject Drumming on the tape loops occurred 10-15 s apart with approximately 30 s o f drumming each 60-« revolution of the loop. Each playback began with a 10-min pretest for observations of baseline behavior, followed by a 10-min playback and a 10-min post-test A 10-min rest period preceded the next baseline observation. We recorded footdrumming and counted footrolls from the tapes and tabulated the number of alert postures and the time spent out of the burrow. We analyzed data in a 2 x 3 (two playbacks and three test periods) design with the same statistical procedures used in the earlier playback experiment Footdrumming to warn offspring We tested 10 mothers and 7 nonmothers for their responses to tethered gopher snakes in their territories from 3 to 14 June 1993 (Table 2). We trapped at mounds of these females before and after the tests and agaia from 14 to 25 July to determine the presence of pups. The trapping evidence, observations, and vanning for the presence of umall, unmarked rats revealed that the 10 mothers had young in the mound at the time of testing ranging in age from about 1 week to 4 month*. The older juveniles of five females weighed 75-120 Behavioral Ecology VoL 8 No. 4 TaMel Description of belmloi of D. tpretnbiht in retponw to a gopher Behavior Description Alert Rat (tops activity. When quadruped, rat arches la neck about 45° and scant the area. When bipedal the body is held at a 45-90* angle. Rat stands motionless in alert posture and watches the Orient Approach Withdraw Footdrum Latency Distance Rat moves toward the snake in elongated posture with slow, jerky motion. Rat moves away from an approaching snake or jumps back after a strike. Hind feet are lifted and their proximal end repeatedly struck against the ground. Each time the feet hit the ground is a footdrum; these are grouped into footrolls that make up a sequence. Footdrumming is quantified by counting the number of footrolls. Time after the beginning of the test for the first footdrum to occur. Estimated distance from snake when rat approaches or footdrums. g and were estimated to be 2 5 - 4 months old. Because rats of these ages are known to footdrum at snakes (Randall and Stevens, 1987), we removed them from the mound during a test to assure that the animal drumming jp»nH^ the mound was the mother. Four of these five females had a second litter of small pups in the mound. The other five females also had one or two pups in die burrow about 4-5 weeks old, which is the age of first emergence. Pups of this age do not appear able to footdrum with any consistency. We tethered a gopher snake at the base of the mound 2-3 m from the female's main burrow entrance as determined by observed activity of the rat The 10-min test began when the rat emerged from the burrow and approached the snake. We recorded the frequencies of all behavior, tabulated time with a stop watch, and compared the behavior of mothers and nonmothers with nonparametric tests. Footdrumming was recorded continuously during the tests, and footrolls were counted from the tapes. We later analyzed for differences in footdrumming patterns in the territorial and snake contexts for six mothers widi the same procedures described above. Frtrtf oriiiiHiilinT o 4? lrTCtfn tO 8BJUKC8 Study animals We tested 42 P. m, affinis in the laboratory and field from 1990 to 1993. All snakes were captured by hand from a 40-km radius near Portal, Arizona, USA. In the laboratory, each snake occupied a wood and glass terrarrum measuring 70 X 55 X 40 cm with 4-5 cm of fine wood substrate in a windowless room with an average temperature of 24°C and a 12:12 h light:dark cycle. When not in an experiment, we offered snakes live mice weekly, which was important to maintain natural hunting behavior. We housed gopher snakes for field tests on sand in aquaria in the animal quarters at the Southwestern Research Station, which was an open screened area under natural photoperiod and temperatures. The snakes were offered mice every 7-10 days. We transported snakes by automobile to and from the field in large, cotton bags and placed them in their home cages immediately after returning to the research station each night Rats used in the laboratory study were live-trapped in south- Randall and Matocq • Footdrumming as antipredator behavior eastern Arizona in 1988, 1991, and 1992. We transported the rats to San Francisco State University via automobile and maintained them in a windowless room on a 14L:10 h light: dark cycle in individual 42 X 22 X 20 cm plastic cages with wire lids on 4 cm of sand with a tin can burrow. We provided wild bird seed and iceberg lettuce ad libitum. Snakt responses to stismie thuntptr We tested gopher snakes in the field for their responses to the footdrumming patterns of D. sptttabiHt. A "thumper" designed by EJR. Lewis, University of California-Berkeley, created seismic footdrumming from cassette recordings of the territorial and snake drumming patterns by hitting a metal rod onto a pad in the same rhythm as the recordings. We used 60-s tape loops of continuous drumming to run the thumper during the tests. We tested snakes in a large enclosure (18 m long X 3.25 m wide X 1 m high) in an open, desert area near Portal, Arizona. Sides of the enclosure consisted of black, plastic window screening stapled to stakes pounded into the ground every 2 3 m with 10 cm of the bottom of the screening covered with a layer of soiL We observed f"aVT with a night vision scope mounted on a tripod outside the center of the arena. Lanterns on 2 m poles on each end of the enclosure provided dim light for the scope. Soil in the area was raked thoroughly between tests to control for olfactory cues. Snakes could move from die center of the arena toward an adult D. specUMUs that moved freely in 52.7 X 15.5 X 155 cm open wire-mesh cages. We positioned one rat at the back of the seismic thumper and die other rat without the thumper approximately 1 m from die ends of the arena. Footdrumming by die rats was dampened with a 2.3 cm layer of foam rubber on the bottom of die cage. We buried die diumper 1.3 m from die end of die enclosure in a 30-cm deep hole and covered it with a wooden box and soil so there was no visual evidence of its presence. Location of die diumper was shifted to control for position effects. Therefore, half die snakes in a sample heard footdrumming coming from one direction and die other half heard it from die opposite direction. We introduced snakes into die center of die test arena, one at a time, in their cotton bags. Appearance of die snake's head from die bag marked die beginning of die test. We recorded quiedy into a hand-held tape recorder die time snakes sampled dieir environment by moving their heads and flicking dieir tongues and die time to travel to die end of die arena. A test ended when die snake struck at die caged rat, reached die end of die arena, or laid in die bag widiout moving for 45 min. We tested snakes diat varied in hunger for dieir responses to die footdrumming patterns. Injury 1990, we tested six large gopher snakes > 1 m that had been kept at die Soudiwestern Research Station since capture and dius had been widiout food for at least 2 months. In July 1991, we tested 14 snakes ranging from 0.82 to 13 m in length and weighing 214-998 g for dieir responses to die territorial and snake footrolls of diree different rats, selected at random, so diat half of die snakes were tested first for dieir response to die territorial footdrumming and half to die snake footdrumming. We only used data from die first test, however, because snakes tended to move in die same direction die second time in die arena as in die first independent of die position of die diumper. We fed die snakes when captured and deprived diem of food for 4 weeks to control for hunger. In July 1992 we conducted a diird experiment. We tested 18 snakes diat differed in die amount of time deprived of food for dieir response to die snake footrolL We captured 12 snakes ranging from 0.69 to 1.26 m and weighing 100-1136 407 g (average 525 g). The remaining six snakes weighed an average of 753 g and had been in die laboratory since 1991 and returned to Arizona by car in 1992. We divided die 6 snakes evenly between 2 food regimes and tested 8 snakes widiin 1 week of eating 1-2 mice and die remaining 10 snakes after being deprived of food for 6 weeks. The snakes deprived of food showed a significant weight loss of 23 g (« = 3.18, df 9, p - .01). Because we predicted die snakes would move away from die diumper, we tested for choice as determined by direction moved widi a one-tailed binomial test. We compared hungry and nonhungry snakes for die time it took to reach die end of the test arena widi a 2 (hungry and nonhungry) X2 (drumming and no drumming) ANOVA after data were normalized widi log-transformations (SYSTAT, Wilkinson 1990). We designated snakes diat had not eaten in 6 weeks or more as hungry (n « 15) and snakes diat had eaten in 4 weeks or less as nonhungry (n = 21). Drumming and nondrumming rats Because only a portion of D. sptctabihs drum consistendy in laboratory tests, we could compare snake responses to drumming and n on drumming rats. We tested 10 gopher snakes (434-1240 g) from December 1992 to February 1993 in encounters widi four rats diat were active footdrummers and four rats diat had "never drummed during a rat-snake interaction. We fed die snakes a regular diet of live mice and water but deprived diem of food for 5 weeks before a test to increase dieir predatory behavior. We were unable to use data from diree snakes because rats failed to footdrum as predicted or die snake captured die rat. We analyzed videotapes of nine additional animal* ranging in size from 200 to 791 g diat were used as stimulus animah widi rats in odier experiments (Randall JA, Matocq MD, Hatch SM, unpublished data). All procedures were die same, except we deprived die snakes of food 2-4 weeks instead of 5 weeks. We used procedures developed by Randall et al(1995) in all tests. We staged 15-min interactions in die laboratory in a large 4X1.2X1 m rectangular arena widi OS m of sand in die bottom. We habituated bodi die snake and rat to die arena. The snake moved freely in die arena for 1 h. We removed die snake, raked die sand dioroughly to distribute odors, and introduced die rat into die arena in its tin can burrow for 2436 h widi food and in a normal lighcdark cycle. We videotaped 15-min encounters between a rat and snake simultaneously under red lights widi two 8-mm video, low-light intensity cameras on tripods from a raised (2 X 2 X 0.675 m high) pladbrm positioned 1.0 m from die end of die experimental arena. We focused die video cameras directly into die arena or on a 1.35X1.2 m mirror suspended at a 20 s angle at die end of die arena opposite die camera. Filming began 3 5 h into die dark cycle immediately after removing die tin can burrow. We reintroduced die snake into die opposite end of die arena from die test subject. An encounter was initiated when eidier die rat or die snake approached to widiin 1 m and an interaction occurred, which we defined as a change in behavior of die animal being approached. Capture of a rat (n =• 3) by die snake caused immediate termination of die test and removal of die rat from die snake. We returned bodi die rescued rat and die snake to dieir respective home cages and fed die snake laboratory mice. Sand in die arena was raked dioroughly after each test. We compared behavior of die 16 snakes in diree 5-min segments for dieir responses to drumming and nondrumming rats in (1) time stalking defined by an approach of a snake to widiin 1 m and continued monitoring of die rat until die 408 snake moved away and (2) number of approaches that resulted in stalking. We also counted the number of footroHs drummed by rats in response to stalking snakes. We analyzed data with multivariate repeated-measures tests, after log transformations, in a 2 (drumming versus nondrumming)XS (times) design. If dgnifiranf Fkesa were found with Wilks' lambda, we did paired / tests with Bonferroni corrections. Snake behavior was examined in more detail in tests with drumming rats. We tabulated the number of approaches, moving away and lying with head oriented toward footdnimming rats. Neutral behavior consisted of moving around the edge of the arena with no noticeable response to the rat Because we found no measurable difference* across time, we combined the data, and, after log transformation, compared the four behaviors with a one-way ANOVA and paired t tests with Bonferroni corrections. Data are presented as means ± SE. RESULTS Comparison sod territorial footdrmmiiing Footdrumming signals All rats drummed different patterns in the territorial and snake contexts as determined by «ign Hi r^pt Wilks' lambda. The rats changed the two signal elements of the footdrumming pattern that account for individual differences in territorial drumming (Randall, 1989), the number of fbotdrums in the first footroll, and the number of footrolls in a sequence (Figure 1). Paired t teats of the means of these two signal elements for each rat in each context showed a «igntfimnt decrease by 7.68 footdrums in the average length of the first footroll (t -» 6.47, df - 14, p - .0001) and an average increase by 5.6 footrolls in the number of footrolls in a sequence (/ " 4.29, df - 14, p - 0.001; Figure 1). Mothers altered the same signal elements that were changed by adult rats without offspring. In the presence of f ? t ^ « . they decreased the average length of the first footroll in their territorial footdrumming by 3.6 footdrums (t «• 2.4, df ™ 5, p •» .058) and increased die average number of footrolls in a sequence from 5.0 ± 0.7 in die territorial context to 8.2 ± 0.5 in the snake context (( - 5.1, df • 5, p - .004). Drumming rate significantly increased from 18.0 ± 1.7 fbotdrums/s during territorial drumming compared with 20.5 ± 1.7 footdrums/s during die snake footdrumming ( ! « 2.8, df - 5, p - .058). Footdnimming in die presence of snakes did not attract conspedfics. No neighbors visited territories of drumming rats during the tethered-snake trials. Playback tests The rats altered their drumming with die type of playback (main effect for playback: Fra — 4.76, p - .014; Figure 2a). They fbotdmmmed significantly more to die territorial playback than to die control (t « 2.32, df » 14, p <.05). The amount of drumming did not differ significantly between die snake playback and the control, between the territorial and snake playbacks after die Bonferroni correction (p >.O5), or between die test and post-test (Flta - 0.98, p •» .33). The playback by tune interaction was not significant {F±a •» 2.06, p = .14). The rats spent similar time out of die mound during each playback (F%At «= 0.26, p - .77): 3.2 ± 0.9 min during the hammer thumps, 2.5 ± 0.9 during die snake playback, and 2.81 ± 0,85 d u m g the territerial playback. The average total time out of the mound during die playback test was 2.82 ± 0.9 min, compared with die post-test time of 4.6 ± 1.2 min (*i.« " 3 - 3 2 . P " ° 7 6 ) - The r* 0 approached die speaker an average of two to three times during die 10-min test and posttest periods for all three stimuli. Behavioral Ecology VoL 8 No. 4 Footdrt j directed to Signal to adult neighbors Dipodomys sptctabiBs did not respond to playbacks of antisnake footdrumming of neighbors. In the first test, neighbors did not differ significantly in die number of footrolls drummed in response to playbacks of die territorial, snake, and hammer control (Freidman's - 4.625, df -> 2, p » .099; Figure 2b). In me second test, few rats drummed in response to broadcasts of neighbor footdrumming. Six of 10 rats drummed during die test and post-test of the territorial playback. No rats drummed during die snake playback, and only diree drummed during die post-test of the snake playback. Two rats drummed in die pretest The neighbor rats seemed to hear die playback broadcasts; mey stopped activity and stood in alert postures when die playbacks began. The number of alert postures did not differ significantly, however, in die two playback contexts(.Fj,ls » 0.93, p - .45). The rats exhibited 6.8 ± l& alert postures during die playback of die territorial footdnimming compared with 4.4 ± 1.7 during the snake playback. Pretest and post-test responses were similar (p > .05). The rats stood alert 8.5 ± 2.85 times in die pretest and 5.1 ± 2.4 during die posttest of the territorial playback and 4.9 ± 0.95 in die pretest and 6.8 ± 2.8 in the post-test after die snake playback. They spent about die same percentage of time out of die mound, widi 54.4% out during die territorial playback and 56.9% out during the snake playback. Signal to offspring Females with pups footdrummed at higher rates than those with no pups and actively came closer to die snake (Table 3). Nine of 10 mothers drummed compared widi 4 of 7 nonmothers (Fisher's Exact test, p < .05). Mothers drummed 62% of dieir fbotroDs on the mound at an estimated distance of 0.3-13 m from die snake. Only two of seven nonmothers drummed out of die mound. Eight mothers approached die head of die snake, compared widi only two nonmothen. AH females were very active when a snake was present, and both mothers and nonmothers spent a majority of time out of die mound. Other behaviors did not differ significantly (Table 3). Most mothers (80%) had vulnerable pups in tile burrow that were quiet and did not exit the mound or footdrum inside the burrow when die mother drummed during a test Only one rat at a time drummed inside die mound and was presumed to be die mother. Co Response to seismic thumping Snakes detected die seismic vibrations but did not differentiate the snake and territorial drumming (Table 4). Snakes that had not eaten in 6 weeks or more took 11.7 ± 3.0 min (n " 6 snakes from 1990) to reach the end of die arena widi die territorial footroll compared widi 9-5 ± 1.5 min to reach the end of die arena widi die snake footroll (n — 4 snakes in 1992; t - 036, df = 8, p >.O5). Hunger influenced tile snakes' response to die seismic thumping. Snakes that had recently eaten tended to move to the end of the area without die thumper (16 of 21; Table 4). In contrast, 10 hungry snakes approached die seismic drumming, while five hungry snakes moved away (x1 with date's concoction • 4.98, df • 1, p •= .026). We found a significant interaction between hunger and direction; hungry snakes moved more slowly than nonhungry snakes when approaching the thumper (.FI-M «• 8.607, p - .006; Figure 3). Nonhungry and hungry snakes did not differ significantly in total time traveling (main effect for hunger F liM •* 0.43, p m 32) or in 409 Randall and Matocq • Footdrumming as anrJpredator behavior 1 i i • i i i i I i i i 6 8 10 12 14 16 18 20 22 24 26 28 Footrolls in Sequence (X±SE) Figure 1 Footdrumming patterns of 15 D. sptrtnNHt during territorial drumming (shaded ellipses) compared with drumming patterns (open ellipses) of the same rats in the presences of snakes. Ellipses are plots of the average (± 1 SE) number of footdrums in the first footroD and average (± 1 SE) number of footrolls In a sequence. time traveling toward or away from the thumper (main effect for direction: FlM = 3.01, p = .088). The time sampling with head out of the bag before moving from the center of the arena also was not significant {flM » 0.34, p » .82). Response to drumming rats Gopher snakes spent significantly more time stalking drumming ran (mean ± SE, 4.09 ± 1.0 min) than nondnimming rats (2.22 ± 0.46 min) during the 15-min tests {F^ - 4.47, p = .045). Total time stalking decreased significantly during the test (Fug " 17.6, p " 0.0001), with significantly more time stalking in the first 5 min interval (1.8 ± 0.36 min) than during the second (0.81 ± 0.22 min; I = 439, df - 31, p < .01) or third (033 ± 0.25 min; t = 6.09, df = 32, p < .01) 5 min intervals.'The snakes stalked drumming rats for 2.29 ± 0.41 min in the first 5 min of the test, compared with 1.1 ± 0.2 min for nondrumming rats (interaction: /"us ~ 3.498, p " .044). The snakes sharply decreased the time they stalked drum- ming rats during the first 5 min of the test (Figure 4). Time stalking and the number of footrolls were negatively correlated (r • - . 9 3 , p = .025). A majority of rats (81%) began to footdrum during the first 5 min of the test (Figure 4). Snakes spent almost the same amount of time stalking drumming rats during the second (0.92 ± 0.27 min) and third (0.89 ± 0.32 min) 5-min periods of the test, and rats continued to footdrum at an average rate of 3-5 footdrumming bouts/min for the remainder of the test Although time stalking differed in tests with drumming and nondrumming rats, the number of predatory approaches was similar <flx ™ 2.43, p «= .13). The snakes averaged 1.27 ± 0 5 predatory approaches to drumming rats, compared with 1.93 ± 0 3 to nondrumming rats in the first 5 min of the test. In the second 5-min interval, snakes approached drumming rats 1.0 ± 0.03 times and nondrumming rats 0.94 ± 0 3 times. The number of predatory approaches continued to decrease, and snakes approached drumming rats only 0.75 ± 035 times 410 Behavioral Ecology VoL 8 No. 4 Table 4 Response of gopher snakes, P. m. aflmb, at different leveb of hunger to playbacks of D* tpn titbuu territorial and soaks footdramming crested by a seismic dumipei Weeks since Type of footroll Experiment 1 Experiments Experiments Territorial Territorial Snake Snake Snake Approach Avoid Strike thumper thumper (%)• >6 4 4 1 6 0* 5* 6 5 5 100 14 0 14 14 CO ' p < .03, binomial test for choice of direction; data are combined in experiments. * Strike tnAirmttt percentage of snakes striking at a rat protected in a wire cage. 1 3- (b) T 1 r -. 21 - T- It __ Hammer Territorial Snaka Figure 2 Rate of footrolls by (a) residents (n ~ 15) and (b) neighbon (n 10) to playbacks of three different drumming patterns: hammer thumping as a control, snake footroll, and territorial footdnimming signatures of the same rat. and nondrununing rats 0.13 ± 0.1 times in the last 5 min of the test. The snakes exhibited four different responses to the footdrumming rats (main effect for response: Fi3> — 6.7, p — .006). They moved both toward and away from drumming rats, oriented to the drummer, and engaged in neutral behavior of exploring the arena (Figure 5). The snakes explored the test arena at a consistently higher rate than they engaged in any other behavior. They averaged 8.3 ± 1.5 explorations Table 5 OTiiUxrlnTii of responses of fc ale* with young and whh no young m the natal mound to a gopher snake tethered at the base of the nimiiwl dm lug a 10-mm test (•"—"« ± SEj Mann-Whitney V teat) Mothers (n - 10) during the 15-min test compared with approaching the rat 3.0 ± 0.73 times (t - 3.38, df - 15, p - < .01), moving away 1.25 ± 0.60 times(< - 4.46, df - 15, p - < .001), and orienting to the rat an average of 4.1 ± 1.42 times (t •* 2.9, df « 15; p > .05 after Bonferroni correction for this and all other comparisons). DISCUSSION We are able to answer the question of why D. sprctabilis footdrums in the presence of snakes from tests of alternative hypotheses summarized in Table 1. In general, we reject the hypothesis that D. sptttabilis footdrums to warn conspedfics, except in the case of mothers with young, and accept the hypothesis of footdnimming directed to the snake. The kangaroo rats footdrum after an initial interaction with a snake to deter its continued pursuit, rather than as a signal of detection or to harasses, startle, or confuse the snake. ; directed to adult conspedfics It is unlikely that D. sptctabiHi footdrums to warn neighbors of danger (Table 1). First, D. spectabihs is a solitary specie*, and both males and nonreproductive females drum in the presence of snakes in the absence of conspecifics in both the field and laboratory (Randall and Stevens, 1987; Randall et aL, 1995). In contrast, rodents that emit alarm calls to warn E 10 • LU 53 Nonmothers (n-7) IX - * Frequency Footrolls Approach snake Alert to snake Time (min) Oriented to snake Out of mound Latency to drum Closest distance (m) LU 190.2 ± 43.4 7.6 ± 1.4 5.1 * 0.96 1 1 ^ ~ 60.9 £ 36.7 6.9 ± 23 4.4 ± 1.1 .035 .402 .768 ZS 7.5 5.6 0.56 .242 .432 .128 2 f= • FED • HUNGRY • •• IH 11H 5 • TO DRUMMING AWAY FROM DRUMMING 1 DIRECTION 4J i 0.9 5.0 * 3.77 2^*0.94 0.12 + 0.03 i • £ * 1.1 3.1 1.71 0.18 .047 Figure 5 Time taken for hungry gopher snakes to move toward (n - 10) or away (n ~ 5) from seismic drumming compared with movement of nonhungry (fed) snakes toward (n — 5) or away (n - 16) from the drumminir. Randall and Matocq • Footdrumming as antipredator behavior 2 3 4 5 Figured Average time (s) spent by gopher snake* (filled circles) stalking footdrumming rats during each minute of 10-min of a 15-min encounter compared with the mean frequency of fbotroDs drummed by the rats (open drdes) and the number of rats that began drumming (open bars) each 6 MINUTES conspedfics of danger only call when relatives are nearby (Hoogland, 1983, 1996; Sherman, 1977). Second, neighbors did not respond to the snake footdrumming. Although the territorial and snake drumming patterns differed, the rats failed to respond to playbacks of the snake footrolL In contrast, the rats discriminated playbacks of territorial footdrumming from the general thumping of a hammer and drummed more to the territorial drumming than to the snake drumming. These results are consistent with behavior observed in other experiments in which rats recognized differences in territorial footdrumming patterns (Randall, 1994). If the snake footroll communicates predation risk, the rats should be able to discriminate the signal and respond \n • 0 • accordingly (Blumstein and Arnold, 1995; Owings and Hennessy, 1984; Seyferth et aL, 1980; Slobodchikoff et al., 1991; Tamura and Yong, 1993; Weary and Kramer, 1995). Third, the drummer may gain little benefit from warning neighbors. Neighbors can be unrelated as well as related, and the distribution of territories makes it impossible for D. spectabiUs to direct its footdrum warning to benefit only neighboring kin. Furthermore, the drumming fails to recruit conspecifics for mobbing or to cause confusion (Caro, 1986a; Hersek and Owings, 1993; Owings and COM, 1977). Therefore, the risk of a continued interaction with the snake and energy cost of drumming seem to outweigh any benefits the drummer might receive from warning neighbors, even at high AWAY TOWARD ORIENT NEUTRAL 0-5 5-10 MINUTES 10-15 Figure 5 Frequency of moving away, toward, and orientation of gopher snake*, P. mdancUucus, (n - 16) to footdrumming D.sptctabilu during 5-min intervals of a 15-min encounter. Neutral behavior was mainly exploratory without regard to the presence of the rat. Behavioral Ecology VoL 8 No. 4 411 population densities when territories are close together and neighbors have a greater chance of being related (Jones et aL, 1988). uzrcctco to The data support the hypothesis that mothers footdrum in the presence of snakes to protect vulnerable offspring (Table 1). Kangaroo rat mothers respond more intensely than nonmothers to the presence of a snake by footdrumming at higher rates and coming closer to the snake. What is unclear is whether females footdrum to warn their pups of a snake on the mound or if the behavior is directed toward the snake. The mother's footdrumming could convey the extent of threat and risk to her pups (Bhimstein and Arnold, 1995; Nikolskii et aL, 1994; Owing* and Hennessy, 1984; Weary and Kramer, 1995). Because we removed older juveniles from the mound and the young pups remained inside, we have no information about responses of young rats to the drumming. They seemed to remain quiet in the burrow. The temporal differences in the territorial and snake footdrumming sequences could convey different messages, and it may be important for young rats to discriminate difference* in the two patterns drummed by their mothers, especially if an inappropriate response results in their death. Footdrmnming directed to the winkf Our results support the hypothesis that D. sptttabiht footdrums in the presence of snakes to reduce its own vulnerability (Hump and Shaker, 1984; Trivers, 1971) (Table 1). The rats drummed the snake footroll in the absence of conspedfics, and a decrease in stalking behavior of the snakes was highly correlated with an increase in footdrumming by die kangaroo rats. Snakes have been observed leaving the mound during natural interactions in the field (Randall and Stevens, 1987). Hence, footdrumming seems to decrease predatory behavior of the snake and possibly inhibits further pursuit Footdrumming, therefore, seems to communicate to the snake that the rat is alert and aware and thus that the snake's chances of successful capture are low. The continuous series of footroDs may function as a tonic signal to maintain the snake's attention (Hersek and Owings, 1993). The long, repeated rhythmic bouts of footdrumming may be the most efficient way for a rat to communicate directly to the snake. We suggest that the change in footdrums from the territorial to the snake footroll is a scaling change. The rats increase footroll repetition rate as they grow more excited, which could communicate increased arousal and enhanced awareness of the snake by the rat. We reject the hypothesis that footdrumming informs the snake of detection, because antipredator drumming always occurs after an encounter with the snake, not at first sighting. We have never seen a rat begin to footdrum in the presence of a snake before it has approached and interacted with it (Randall et aL, 1995, this study). More likely, the physical approach of the rat to the head of the snake informs it of detection and that the chances of ambush are lowered. Antipredator behavior continues when the rat footdrums an intense series of short fbotrous at a safe distance to inform the snake its presence is being monitored. Rats footdrummed when they interacted with more predatory snakes, which suggests that predatory beHaVfor of (he fnake increases the probability that a rat will footdrum. We are unable to rule out that footdrumming communicates a healthy condition and the ability to avoid predation (Caro, 1995). We have no evidence that rats in poor condition fail to footdrum or are more vulnerable to snake strikes. We do have evidence that very old rats footdrum in the presence of snakes and can avoid snake strikes as well as younger rats (Randall JA, Matocq MD, Hatch SM, unpublished data). Finally, there is little evidence that footdrumming alone confuses, harasses, or startles snakes. The snakes that moved toward the end of me arena with the seismic thumper did not avoid the vibrations and moved over the top of the buried thumper to reach the caged rat. The f?fc*^ exhibited no obvious sign of discomfort or confusion when responding to the artificial thumper or when a rat footdrummed. Sometimes snakes pulled their head back and hissed defensively at the approach of a rat, but this behavior was never observed in response to footdrumming alone. We never observed rats biting or physically harming a snake. We conclude, therefore, that D. spedabiUs footdrums in the presence of snakes to communicate continued awareness of the location and presence of the snake. Footdrumming functions as a less dangerous defense than contact with the snake to deter pursuit after the rat has initially approached the snake and interacted with it Do snake* locate prey from footdrammmg? Our results suggest that hungry ima^r* may locate kangaroo rats from their territorial footdrumming. Hungry snakes moved toward the seismic thumper during drumming of both the territorial and snake footrolls, while nonhungry snakes tended to move away from the drumming. By locating the territorial footdrumming of kangaroo rats, hungry snakes could save search time. Some predators are able to use their prey's intraspecific interactions to locate their prey (Ryan et al., 1981; Tuttle et aL, 1982). A third party can eavesdrop on a dyadic interaction because any attempt to reduce eavesdropping may decrease effectiveness of die signal (MarU, 1985). Snakes, therefore, could locate kangaroo rats by their territorial advertisement Hungry snakes often travel long distances in search of rodents with patchy distributions (King and Duvall, 1990). Occupied D. spettabiHs mounds can vary considerably from year to year in some areas, so it would be to a snake's advantage to locate an area of occupied mounds from the territorial footdrumming and then wait in ambush of individual rats at specific locations. Our deep appreciation goe* to Maureen Sullivan for her tireless help with the field portion of this study and to Ted Lewis, Department of Electrical Engineering and Computer Science, University of California, Berkeley, for use of the thumper*. We also appreciate assistance from Susan Hatch, Evon Hekkala, Patricia Kennedy, Jennifer Nymark, Allie Rich, Jay Shore, Mary Beth Stone, and the staff at the Southwestern Research Station. We thank Peter Waser for allowing us to work on his Rucker Canyon site in 1993, Wade Sherbrooke for the use of his gopher snakes in 1990, and Larry Wolf and anonymous reviewer! for their reviews and editing of the manuscript. We are grateful for support to J.A.R. by the National Science Foundation, BNS 87-16860. BNS 89-08827, BNS 91-09850, IBN 95-06688 and the National Geographic Society (3273-86). Research approved by University Animal Care and Use Committee. REFERENCES Bhimstein DT. Arnold W, 1995. Situational specificity in alpine-marmot alarm communication. Ethology 100:1-13. Butler JM, Roper, JT, 1994. Escape tactics and alarm responses in badgers Mtia mtUn a field experiment. Ethology 99J13-322. Caro TM, 1986a. The function of stoning: a review of the hypotheses. Anim Behav 34:649-662. Caro TM, 1986b. The function of Hotting in Thomson's gazelles: some tests of the predictions. Anim Behav 34:663-664. Randall and Matocq • Footdrummlng as antiprcdator behavior Caro TM, 1995. Pursuit-deterrence revisited. Trend* Ecol Evol 10-.500503. Caro TM, Lombardo L. Goldizen AW, Kelly M, 1995. Tail-flagging and other antiprcdator signals in white-tailed deer new data and synthesis. Behav Ecol 6:442-450. FitzGibbon CD, Fanshawe JH, 1988. Stoning in Thomson's gazelles: an honest signal of condition. Behav Ecol Sociobiol 23:68-74. Hamilton WD, 1964. The genetical evolution of social behaviour. J Theor Biol 7:1-52. Hanon O, 1991. Pursuit-deterrent signals: communication between prey and predator. Trends Ecol Evol 6325-329. HartHne PH, 1971. Physiological basis for detection of sound and vibration in snakes. J Ezp Biol 54:349-371. Hennessy DF, Owing* DH, 1988. Rattlesnakes create a context for localizing their search for potential prey. Ethology 77:317-329. Hersek MJ, Owings DH, 1993. Tailflagging by adult California ground squirrels: a tonic signal that serves different functions for males and females. Anim Behav 46:129-138. Hoogland JL, 1983. Nepotism and alarm calling the black-tailed prairie dog (Cynomys tudoviaanus). Anim Behav 31:472-479. Hoogland JL, 1996. Why do Gunnison's prairie dogs give anti-predator calls? Anim Behav 51:871-880. Jones WT, 1984. Natal philopatry in bannertaOed kangaroo rats. Behav Ecol Sociobiol 15:151-155. Jones WT, Waser PM, Elliott LF, link NL, Bush BB, 1988. Philopatry. dispersal, and habitat saturation in the banner-tailed kangaroo rat, Dipodomyj sptctabilis. Ecology 69:1466-1473. King MB, Duvall D, 1990. Prairie rattlesnake seasonal migrations: episodes of movement, vernal foraging and sex differences. Animal Behav 39324-935. Khunp CM, Shalter MD, 1984. Acoustic behaviour of birds and mammals in the predator context I. Factors affecting structure of alarm signals, n. The functional significance and evolution of alarm signals. Z Tierpsychol 66:189-226. Loughry WJ, 1987. The dynamics of snake harassment by black-tailed prairie dogs. Behaviour 103:27-48. Mark] H, 1985. Manipulation, modulation, information, cognition: some of the riddles of communication. In: Experimental behavioral ecology and sociobiology (HoDdobler B, Lindauer M eds). Sunderland, Massachusetts: Sinauer; 163-193. Nikolskii AA, Nesterova NL, Suchanova MV, 1994. Sltuational variations of spectural structure in Marmota bobacmu]} alarm signal. In: Actual problems of marmots investigations (Riimiantsev V Yu, ed). Moscow: Theriology Society, Russian Academy of Science; 127-168. Owings DH, Coss RG, 1977. Snake mobbing by California ground squirrels: adaptive variation and ontogeny. Behaviour 6230-69. Owings DH, Hennessy DF, 1984. The importance of variation in sdurid visual and vocal communication. In: The biology of grounddwelling squirrels (Murie JO, Michener CR eds). Lincoln: University of Nebraska Press; 169-200. Owings DH, Hennessy DF, Leger DW, Beckett G, 1986. Different functions of "alarm" calling for different time scales: a preliminary report on ground squirrels. Behaviour 99:101-116. Owings DH, Loughry WJ, 1985. variation in snake-elicited jump-yipping by black-tailed prairie doge ontogeny and snake-fpedfidty. Z Tierpsychol 70:177-200. Proske U, 1969. Vibration-sensitive mechano-receptors in snake skin. Exp Neurol 23:187-194. Raddiffe GW, Esteop K, Boyer T, Chizar D, 1986. Stimulus control of predatory behavior in red spitting cobras (Naja mmurmtica paltida) and prairie rpftlwufcy (Crotahu n viriduis). Anim Behav 34:804814. Randall JA, 1984. Territorial defense and advertisement by footdrumming in banncrtail kangaroo rats (Dipodowtys sptctaikhs) at high and low population densities. Behav Ecol Sociobiol 16:11-20. Randall JA, 1989. Individual footdrumming signatures in bannertailed kangaroo rats, Dipodomys sptrtahiKt Anim Behav 38:620-630. Randall JA, 1991. Mating strategies of a nocturnal desert rodent (Dipodonys tptdabiUs). Behav Ecol Sodobiol 28:215-220. Randall JA, 1994. Discrimination of footdrumming signatures by kangaroo rats, Dipodowtys sptaabiUs, Anim Behav 47:45-54. Randall JA, 1995. Modification of footdrumming signatures: changing territories and gaining new neighbours. Anim Behav 49:1227-1237. Randall JA, Hatch SM, Hekkala ER, 1995. Inter-specific variation in 41S anti-predator behavior in sympatric species of kangaroo rat Behar Ecol Sociobiol 36:243-250. Randall JA, Lewis ER, in press. Seismic communication between the burrows of kangaroo rats, Dipodomys sptctabiKt J Com Physiol A. Randall JA, Stevens CM, 1987. Footdrumming and other anti-predator responses in die bannertail kangaroo rat (Dipodomys sprrtaHHt). Behav Ecol Sociobiol 20:187-194. Ryan MJ. Tuttle MD, Taft LK, 1981. The costs and benefits of frog chorusing behavior. Behav Ecol Sodobiol 8:273-278. Seyfarth RM, Cheney DJ, Marler P, 1980. vervet monkey alarm calls: sematic communication in a free-ranging primate. Anim Behav 28: 1070-1094. Sherman PW, 1977. Nepotism and the evolution of alarm calls. Science 197:1246-1253. Sherman PW, 1985. Alarm calls of Belding's ground squirrels to aerial predators: nepotism or self-preservation? Behav Ecol Sodobiol 17: 313-323. Slobodchikoff CM, KiriazisJ, Fisher CCreefE, 1991. Semantic information distinguishing individual predators in the alarm calls of Gunnison's prairie dogs. Anim Behav 42:713-719. Sweet CT, 1985. Geographic variation, convergent crypsb and mimicry in gopher snakes (Pitup/us mtlanoUucus) and western rattlesnakes (Crotatus viridis).] Herpetol 1935-67. Tamura N, 1989. Snake-directed mobbing by the Formosan squirrel Cattouatrut trythixuus Uiaheonensis. Behav Ecol Sodobiol 24:175180. Tamura N, Ybng H, 1993. Vocalizations in response to predators in three spedes of Malaysian Callosdurus (Sduridae). J Mammal 74: 703-714. Trivers, RL, 1971. The evolution of reciprocal altruism. Q Rev Biol 46:35-57. Tuttle MD, Taft LK, Ryan MJ, 1982. Evasive behavior of a frog in response to bat predation. Anim Behav 30393-397. Weary DM, Kramer DI_ 1995. Response of eastern chipmunks to conspecinc alarm calls. Anim Behav 49:81-93. Webster DB, Webster M, 1984. The specialized auditory system of kangaroo rats. Sensory Physiol 8:161-196. Wever EG, Vernon JA, 1960. The problem of hearing in snakes. J Aud Res 1:77-83. Wilkinson L, 1990. SYSTAT. The system for statistics. Evanston, Illinois: SYSTAT. Woodland DJ, Jaafar A, Knight ML, 1980. The "pursuit deterrent" function of alarm signals. Am Nat 115:748-753.
© Copyright 2024 Paperzz