Oecologia (1996) 105:313-319

9 Springer-Verlag 1996

Z. A b r a m s k y 9 E. S t r a u s s 9 A . S u b a c h 9 B.P. K o t l e r A. Riechman

The effect of barn owls (Tyto alba) on the activity and microhabitat selection of Gerbillus allenbyi and G. pyramidum

Received: 3 May 1995 / Accepted: 20 September 1995

Predation plays an important role in ecological communities by affecting prey behavior such as foraging and by physical removal of individual prey. In regard to foraging, animals such as desert rodents often balance conflicting demands for food and safety. This has been studied in the field by indirectly manipulating predatory risk through the alteration of cues associated with increased risk such as cover or illumination. It has also been studied by directly manipulating the presence of predators in aviaries. Here, we report on experiments in which we directly manipulated actual predatory risk to desert rodents in the field. We conducted a series of experiments in the field using a trained barn owl (Tyto alba) to investigate how two species of coexisting gerbils (Gerbillus allenbyi and G. pyramidum) respond to various cues of predatory risk in their natural environment. The gerbils responded to risk of predation, in the form of owl flights and owl hunger calls, by reducing their activity in the risky plot relative to the control plot. The strongest response was to owl flights and the weakest to recorded hunger calls of owls. Furthermore, when risk of predation was relatively high, as in the case with barn owl flights, both gerbil species mostly limited their activity to the safer bush microhabitat. The response of the gerbils to risk of predation disappeared very quickly following removal of the treatment, and the gerbils returned to normal levels of activity within the same night. The gerbils did not respond to experimental cues (alarm clock), the presence of the investigators, the presence of a quiet owl, and recorded "white noise". Using trained barn owls, we were able to effectively manipulate actual risk of predation to gerbils in natural habitats and to quantify how gerbils alter their behavior in order to balAbstract

z. Abramsky (~) 9E. Strauss 9A. Subach - A. Riechman Department of Life Science, Ben-GurionUniversityof the Negev, Beer Sheva 84105, Israel B.R Kotler Mitrani Center for Desert Ecology, Blaustein Institute for Desert Research, Ben-Gurion Universityof the Negev, Sede Boqer 84993, Israel

ance conflicting demands of food and safety. The method allows assessment of aspects of behavior, population interactions, and community characteristics involving predation in natural habitats. K e y w o r d s Gerbils 9Foraging 9Predation risk. Field experiments 9Trained barn owls

Introduction Animals can normally increase reproductive success by increasing their rate of energy intake. However, other factors with a strong effect on fitness may preclude this as an optimal strategy. Especially important is the need to avoid predators (e.g., Shih 1982; Lima 1985). Many experimental studies have shown that predation is an additional cost of foraging (Edwards 1983; Werner et al. 1983; Kotler 1984a, b; Gilliam and Fraser 1987; Brown et al. 1988; Abrahams and Dill 1989; Kotler et al. 1991, 1993a, b), and as a cost, it affects when, where, and how long an animal forages (Kotler et al. 1991). Therefore, foraging decisions that are made under the risk of predation may differ from those based on energetic considerations alone (Lima and Dill 1990). The behavioral options open to foragers range between maximizing feeding efficiency and minimizing the risk of predation (Sih 1980), Thus, feeding behavior will often be an adaptive compromise between these two conflicting demands (Sih 1980; Lima and Dill 1990). Many observational and experimental studies have tested the effect of predatory risk on the foraging behavior of nocturnal desert rodents. Thompson (1982) found that reduction in the distance between shrubs decreased the risk of predation, increased the number of strategies of spatial utilization available to desert rodents, and increased species diversity. Doucet and Bider (1969), Lockard and Owings (1974), Clarke (1983), Price et al. 1984), and Brown and Alkon (1990) have found that moonlight affects both rodent behavior and owl efficiency. In bright moonlight, prey species are more vulnerable

314

OECOLOGIA 105 (1996) 9 Springer-Verlag

tO visual detection by nocturnal predators (Kotler et al. 1988, 1991; L o n g l a n d and Price 1991). Thus, as light level increased, rodent activity was significantly reduced and shifted from the open to under vegetation cover (Kotler 1984a, b; B r o w n et al. 1988, Kotler et al. 1991). Kotler et al. (1991) tested the effect o f predation by owls on Gerbillus allenbyi and G. pyramidum, which differ in b o d y size, inside an aviary (18x23 m). They found that both illumination and the presence o f owls affected the foraging behavior o f gerbils. Both species foraged fewer patches and shifted their foraging activity towards the bush microhabitat in response to owl predators. Kotler et al. (1988) and B r o w n et al. (1988) perf o r m e d similar aviary experiments with heteromyid rodents and barn owls and found similar results. To date, all experimental studies that have tested the effect o f predation on rodent behavior have been conducted inside aviaries. Although this experimental system is very convenient, it is relatively small (even the largest aviaries are m a n y times smaller than a typical gerbil or owl h o m e range), densities o f prey and predators are unnaturally high (5-40-fold), and the frequency o f encounters between predator and prey m a y abnormally influence prey behavior (Lima and Dill 1990). In addition, this system cannot exactly imitate the c o m p l e x structure o f natural habitats. Rather than being limited by the artificial nature o f cage experiments, we chose to examine the effect of risk o f predation directly in the field by using a trained predator. Caraco et al. (1980) have previously used this method to examine flocking behavior in junkos in response to a trained hawk. We used trained barn owls (barn owls are one o f the major predators of gerbils in the study site) to examine the behavioral responses of two gerbil species, G. allenbyi and G. pyramidum, to predation risk in their natural habitat. This enabled us to directly manipulate risk o f predation in natural habitats. We hypothesized that gerbils would respond to risk o f predation by reducing their activity and by limiting their activity to the bush microhabitat. We also hypothesized that gerbils would not respond to "unnatural" cues not typically associated with predatory risk such as "white noise", the sound o f an alarm clock (to which we trained barn owls), and to the presence o f a quiet owl. We expected that gerbils would respond to natural cues such as owl flights and hunger calls.

During the last 10 years, investigators have studied the population biology (Abramsky 1984), diet (Bar et al. 1984), habitat selection (Rosenzweig and Abramsky 1985, 1986; Abramsky et at. 1990), competitive interactions (Abramsky and Sellah 1982; Rosenzweig et al. 1984; Abramsky et al. 1991, 1994), the relationship between productivity and diversity (Abramsky and Rosenzweig 1984; Abramsky 1988), foraging strategies (Abramsky and Pinshow 1989; Mitchell et al. 1990), mechanisms of coexistence (Kotler and Brown 1988; Ziv et al. 1993; Kotler et al. 1993c), and the effect of predatory risk (in aviaries) on foraging behavior of these two gerbils (Kotler 1992; Kotler et al. 1991, 1993a, b). Both gerbil species prefer the same habitat type, the semi-stabilized sand (Abramsky and Pinshow 1989; Abramsky etal. 1990). However, their secondary habitat preferences differ, with GA preferring the stabilized sand and GP preferring the shifting dunes (Rosenzweig and Abramsky 1986). At a smaller spatial scale, they both prefer to forage in the shrub microhabitat, but GP is relatively more active in the open microhabitat than GA (Brown et al., in press). There are also differences between the two gerbil species in ability to avoid predators, with GP being superior to GA in detecting and avoiding predators away from cover (Kotler et al. 1991). This may lead GP to be competitively superior in open, risky habitats and microhabitats (Abramsky 1988; Kotler et al. 1991). Study site The study was conducted during the summers of 1992 and 1993 at the Holot Mashabim Nature Reserve, situated in the Haluza region, 35 km south of Beer-Sheva, Israel. Average precipitation at the site is i08 ram, with winter rainfall and dew forming on approximately 250 nights per year. Sandy areas in the study region can be divided into two habitat types based on sand mobility, dominant perennial plant species, shrub cover, and the existence of soil crust (Danin 1978). Artemesia monosperma and dead remains of Stipagrostis scoparia dominate semi-stabilized dunes (dunes in the process of being stabilized). In this habitat type, perennial vegetation cover is relatively sparse, open patches of sand are relatively common, and some portions of the dunes are still mobile. A. monosperma and Retama raetam dominate stabilized dunes. In this habitat type, shrub cover is relatively dense, open patches are smaller and none of the sand is mobile because of extensive soil crust formation. Shrubs in both habitat types provide two microhabitat types which differ in the magnitude of predatory risk: the open, risky microhabitat and the safer bush microhabitat (Kotler 1984b; Brown et al. 1988; Kotler et al. 1991). At Holot Mashabim, we used four 1-ha enclosures with rodentproof fences (Abramsky et al. 1990). Fenced enclosures were placed in pairs, which each pair sharing a common fence. To follow natural densities of gerbils, we also used two similar 1-ha unfenced control plots. Methods

Barn owls

Materials and methods Study species

G. allenbyi (GA) and G. pyramidum (GP) coexist in most sandy habitats in the western Negev Desert (Abramsky et al. 1985). Both species have a diet composed mostly of seeds (Bar et al. 1984), are nocturnal, live in burrows, and compete with each other for food and space (Abramsky et al. 1985; Abramsky and Pinshow 1989; Mitchell et al. 1990; Kotler et al. 1993c; Ziv et al. 1993). They are similar morphologically, but differ in body size: the average mass of GA is 26 g and of GP is 40 g (Abramsky et al. 1985).

We trained barn owls to associate the sound of an alarm clock with food (A. Subach, unpublished work). When the alarm clock was sounded, we trained the owls to fly from one person to another located up to 120 m away. Because of our training methods, we faced two problems: 1. The trained barn owl responded to the sounds of the alarm clock. Therefore, we had to test for the potential effect of the sound to the alarm itself on gerbil activity. 2. The trained adult barn owls gave hunger calls. In natural situations this behavior is typical only of fledgling owls during a brief 4-6 week period. Adult owls do not normally emit these calls. Therefore we had to test for the effect of the hunger calls on gerbil activity.

O E C O L O G I A 105 (1996) 9 Springer-Verlag Plots Prior to each experiment, we trapped gerbils from four enclosed plots for several nights using Sherman traps. Traps baited with millet seeds were set before sunset and checked early in the morning. Each captured gerbil was removed to the laboratory where it was sexed, weighed, and marked with a species-specific toe clip. Individuals of GA were marked by cutting their outer toe on their right hind leg and individuals of GP were marked by cutting their outer toe on their left hind leg. This process continued until we captured all rodent individuals. We released to each of the four enclosed plots 24 GA and 5 GP individuals (the average natural densities at the time of the experiments). We used sand tracking to quantify gerbil foraging activity without disturbing the natural activity of the animals (Kotler 1985b; Mitchell et al. 1990; Abramsky et al. 1990, 1991, 1992, 1994; Kotler et al. 1993c). Gerbil tracks left in the sand were used to measure activity of each species in different experimental plots and microhabitats. Activity in an area is related to the number of tracks left on sand-tracking plots there; animals leave more tracks in areas where they forage more (Kotler et al. 1985b; Abramsky et al. 1990; Mitchell et al. 1990). On each 1-ha enclosed plot, we established 40 sand tracking stations, 20 in the semistabilized dune and 20 in the stabilized sand. At each station, we smoothed the sand in two pairs of 40x40 cm subplots. Each pair consisted of a subplot under a shrub (bush microhabitat) and one in the open microhabitat. Since the gerbils had species-specific toe clips, we were easily able to identify tracks to species and score each subplot for tracks coverage by each species, from 0 (no tracks) to 4 (100% track coverage). The tracks in one pair of subplots at each station were read immediately after the experiment, smoothed and re-read early in the morning. In this way, we were able to test (1) the immediate response of the two gerbils species to predatory risk as well as (2) how their activity changed following the experiment. The tracks in the second pair of subplots were read only in the morning and showed total gerbil activity throughout the night. The activity density of gerbils on each grid was calculated by summing the sand tracking score from every station on the same grid.

Description of the experiments For each pair of fenced enclosures, one enclosure served as the experimental grid and the other as the control on a rotating basis. All experiments began at 2100 h and lasted 120 min. Then the treatment was removed from the experimental grids, data from one pair of tracking subplots from each station were collected, and those tracking subplots were re-smoothed. All tracking subplots were then read in the morning. Each type of experiment was conducted for at least 6 consecutive nights under similar moon phase. The experimental treatments were human presence and the sound of alarm clock, hunger calls with or without the presence of a barn owl, recorded white noise, the presence of a quiet barn owl, and barn owl flights over the experimental plots. We conducted experiments in the summers of 1992, 1993, and 1994. Human presence and the sound of an alarm clock. Barn owls were entrained to the sound of the alarm from an alarm clock. At the sound of the alarm, the owl wound fly from one handler to the other. Thus, to control predator activity using trained owls required the presence of two human handlers and the use of the alarm clock. To be able to assess the reaction of gerbils to the owls, we first had to know how gerbils responded to the sound of the alarm and the presence of human handlers in the absence of the owls. To do so, we sounded the alarm clock at 5-min intervals for 15-20 s during 120 min from a distance of 10 m from either side of the experimental grid in the absence of the trained owls. Hunger calls The trained barn owls emit hunger calls every 5-10 s. We performed two related experiments to assess the effect of the hunger calls. In one experiment, we placed a trained barn owl in a

315

cage in the middle of the experimental plot for 120 rain. In a second experiment, we recorded the hunger calls and broadcast them for 120 min using a tape recorder placed in the middle of the experimental plot. Hunger calls were broadcast for 20 s every 5 min. We used hunger calls as a cue to risk of predation since they were easy to obtain and since they have biological meaning. The gerbils may be exposed to natural hunger calls for a 4-6 week period in the spring when the young owls leave the nest and travel with their parents begging to be fed. Our experiments followed this event by a few weeks. White noise. To assess whether gerbils may have been responding to the broadcast equipment rather than to the recorded owl hunger calls themselves, we transmitted recorded white noise [noise which includes all possible frequencies (72 Hz-10 kHz) at equal intensities] for 120 rain. The white noise was transmitted for 20 s every 5 min. Quiet barn owl. We placed a mature barn owl in a cage in the middle of the experimental plots for 120 rain to see if gerbils can use other senses to detect a stationary predator. Barn owlflights. To assess whether gerbils can respond directly to the presence of an owl when both visual and auditory cues are available, we allowed the barn owl to fly over the experimental grid, in 10-min bouts for a total of 120 min. The flying owl behaves in a similar manner to an owl that is truly hunting and should represent the highest level of risk to gerbils for our experiments. Data analysis. To allow easy qualitative comparison between the results obtained from different experiments that had different treatments and activity levels we plotted the ratio between the activity of the gerbils on the experimental and the adjacent control plots. A ratio of 1.0 means no difference between the activities of the gerbils on the experimental and control plots. A ratio smaller than 1.0 means that the activity on the experimental plot was less than that on the control plot, i.e., the gerbils responded to the treatment in a predicted manner. A ratio higher than 1.0 would mean that the rodents preferred the treatment over the control plot. To analyse the results we used paired t-tests to examine if the activity of the gerbils on the experimental plot was significantly different from that on the adjacent control plot. We used ANOVA to test if the ratios of different treatments were significantly different.

Results T h e two e n c l o s e d grids o f e a c h pair are s i m i l a r I n the a b s e n c e o f a n y t r e a t m e n t , n o s i g n i f i c a n t d i f f e r e n c e (t=0.75, df=15, P > 0 . 0 5 ; p a i r e d t-test) was d e t e c t e d b e t w e e n the a c t i v i t y d e n s i t i e s o f the t w o g e r b i l s s p e c i e s o n a d j a c e n t grids d u r i n g the three e x a m i n e d t i m e p e r i o d s . S i m i l a r l y , n o s i g n i f i c a n t d i f f e r e n c e was f o u n d a m o n g the f o u r e n c l o s e d grids ( F = 0 . 7 1 , dr=31, P = 0 . 5 5 4 ) . T h i s result s u p p o r t s e a r l i e r w o r k (Abramsky et al. 1991, 1992, 1994) w h i c h also s h o w e d that w h e n g e r b i l p o p u l a t i o n d e n s i t i e s are similar, so are their a c t i v i t y densities.

T h e effect o f h u m a n p r e s e n c e a n d the s o u n d of an alarm clock on gerbil activity T h e t w o g e r b i l s p e c i e s d i d n o t alter a c t i v i t y levels ( P > 0 . 0 5 ) in r e s p o n s e to the s o u n d o f the a l a r m c l o c k a n d the p r e s e n c e o f h u m a n s i n the area (Fig. 1). S i m i l a r l y , the a c t i v i t y d e n s i t y o f e a c h g e r b i l s p e c i e s i n the e x p e r i m e n t a l a n d c o n t r o l grids w e r e n o t s i g n i f i c a n t l y d i f f e r e n t

316

OECOLOGIA 105 (I996) 9 Springer-Verlag 1.5

1.5

[] [] a

.A O

Z 0

1.0~

.J 0 I-:Z 0

-r

1

IZ 0.5

GA GP

a

Z

Z

lal 0. X Ul

a

GA GP

1.0 Z

I-Z U.I U.I Q. X UJ

0.5

0.0

0.0

u~

g

o

i

tn "----.

E

ac:

.c

o

Fig. 1 Comparison of the effect of the treatments on the activity of the two gerbil species in the experimental relative to the control plots measured immediately following the experimental treatment. A ratio of 1.0 means no difference between the activities of the gerbils on the experimental and control plots. A ratio smaller than 1.0 means that the activity on the experimental plot was less than that on the control plot. Significant difference (* P<0.05, ** P< 0.001, *** P<0.0001; paired t-test) between experiment and control. Treatments with different letters are significantly different, P<0.05). GA GerbilIus allenbyi, GP G. pyramidum (P>0.05) for the second part of the night (Fig. 2) or throughout the night (Fig. 3). The effect of barn owl hunger calls and white noise on gerbil activity Live barn owl hunger calls affected the activity of both gerbil species. Their activity density on the experimental grid immediately following the experiment was significantly lower than on the control (Fig. 1; for GA: t=6.34, df=-6, P=0.001: for GP: /=-11.09, df=6, P<<0.001, df=5; paired t-test). However, by morning we found no difference in gerbil activity between the experimental and contril grids (Figs. 2 and 3). It seems that following the removal of the barn owls the effect quickly disappeared as the gerbils resumed normal activity levels. Recorded hunger calls also significantly affected (Fig. 1) the activity of GA (t=3.34, df=6, P=0.02; paired ttest), but not the activity of GP (P>0.05). GP activity in both experimental and control plots was so low during this experiment that we would not have been able to find a significant difference between experimental and control plots, whether or not it actually existed. In this experiment we collected data only immediately following the experimental treatment. Recorded white noise transmitted in the same frequency and duration as the recorded hunger calls did not significantly (P>0.05) affect the activity of either gerbil

Fig. 2 Comparison of the effect of the treatments on the activity of the two gerbil species in the experimental relative to the control plots during the second part of the night. In all treatments, the activity on the experimental plots is not significantly different from that on the control plots. GAG. alIenbyi, GP G. pyramidum 1.5 ~A

n" I'Z 0 (,3

1.0

Z IJJ mml Q. X

0.5

0.0

f. o

.c

Fig. 3 Comparison of the effect of the treatments on the activity of the two gerbil species in the experimental relative to the control plots during the entire night. In all treatments, the activity on the experimental plots is not significantly different from that on the control plots. GA G. allenbyi, GP G. pyramidum species (Fig. 1). Similar results were obtained for the second part and throughout the night (Figs. 2 and 3). We can therefore attribute the response of gerbils to recorded hunger calls to the actual calls themselves rather than to the operation of the broadcasting equipment. The effect of a quiet barn owl A quiet caged barn owl in the middle of the experimental grid did not significantly (P>0.05) affect the activity of

O E C O L O G I A 105 (1996) 9 Springer-Verlag

317

mizing risk (Sih 1982; Werner et al. 1983). Often, when risk of predation is high, foragers prefer to feed in relatively poor but safe areas over rich but risky areas (e.g., Kotler 1984a; Brown et al. 1988; Nonacs and Dill 1990). Similar results have been documented for gerbils: in the field, they often respond more strongly to the presence of a predator than to the presence of a competitor (Abramsky et al. 1991, 1993), and in aviary experiments, The effect of barn owl fights on gerbil activity they respond most dramatically to the presence of predaBoth gerbil species significantly reduced their activity tors and to a lesser extent to the indirect cue of added il(for GA: t=12.7, df=6, P<0.000; for GP: t=22.47, df=-6, lumination (Kotler et al. 1991, 1992, 1993a, b). In the P<0.001; paired t-test) during the experiment relative to field, up to 91% of foraging costs can be attributed to their activity in the control plots (Fig. 1). Yet, during the predation (Brown et al., in press). Under these circumsecond part of the night, after the owls were removed, stances, risk of predation can interact with competition activity densities of gerbils in the experimental and con- to determine mechanisms of species coexistence (Brown trol plots were similar (P>0.05; Fig. 2). Moreover, no 1986, 1989; Kotler and Brown 1988) and alter conditions significant effect of owl flights remained in plots from for competitive coexistence (Kotler and Holt 1989; e.g., which data were collected only at dawn (P>0.05; Fig. 3). freshwater insects, Sih 1982; ants, Nonacs and Dill 1990; It seems that the gerbils recovered very quickly from the fish, Werner et al. 1983). To test the importance of risk treatment and may even have increased their activity to of predation on the activity and microhabitat selection of compensate for the time lost when the barn owls were nocturnal gerbils in the field, it would be useful to maflying. nipulate directly risk of predation. The experiments presented here show the efficiency of manipulating predatory risk to gerbils in the field usMicrohabitat selection ing trained barn owls (A. Subach, unpublished work). The gerbils responded to our experimental treatments Under control situations (no treatment), neither species by reducing their activity in the experimental plots relaexhibited a microhabitat preference. The preference of tive to the control plots. Gerbils gave the strongest reGA to the shrub microhabitat was 0.51_0.045 sponses to situations and cues most immediately or di(mean+SD) and that of GP 0.53_+0.026. Neither value rectly related to predatory risk (Fig. 1). Thus, both gerbil differed significantly from random use (0.50) of the two species responded more strongly to actual hunger calls microhabitats. In the presence of owl flights, both gerbil from an owl present on the grid than to recorded hunger species significantly shifted their activity to the shrub calls. The strongest responses were seen in the most danmicrohabitat. For GA the preference for the shrub micro- gerous context: when an owl was flying overhead. Only habitat was 0.85_+0.25 (t=2.87, df=6, P=0.028; paired t- when the response to risk of predation was the strongest test), and for GP it was 0.93_+0.19 (t=3.35, df=6, did the gerbils also shift their microhabitat use to the safP=0.015: t-test). In contrast, hunger calls with the owl ei- er shrub microhabitat. We have previously noted that ther present or absent did not significantly change the gerbils are similarly able to appropriately order risk in microhabitat preference of the gerbils. aviary experiments. There, gerbils foraged least and showed the greatest avoidance of the open microhabitat in the presence of predators and increased illumination Discussion and foraged most on dark nights when predators were absent (Kotler et al. 1991, 1992, 1993a, b). Furthermore, Predation plays an important role in structuring ecologi- gerbils responded more strongly to barn owls than to two cal communities by affecting prey behavior and by the other owl species less specialized on rodents (Kotler et physical removal of prey individuals (Kotler and Holt al. 1991). Our current results show a similar ability of 1989). Perhaps the most important effect of predation gerbils to correctly assess risk in the field. The possible comes from its influence on prey behavior by altering exception was that gerbils did not change their activity foraging decisions (Sih 1980, 1982; Dill and Fraser levels in response to the presence of a quiet owl. Since 1984; Lima 1985; Lima et al. 1985; Anderson 1986), barn owls hunt while flying or from perches, this may inhabitat selection (Milinski and Heller 1978; Cerri and dicate an inability of gerbils to detect this potentially Fraser 1983; Edwards 1983; Ohman et al. 1983; Werner dangerous situation in which reliable cues are largely et al. 1983; Lima et al. 1985; Holbrook and Schmitt weak or lacking. 1988; Nonacs and Dill 1990), activity times (Sih 1980, Although gerbils respond significantly to barn owl 1982), group size (Caraco et al. 1980; Cassini 1991), flights and owl hunger calls by reducing activity, the regrowth rates (Stamp and Bowers 1991) and reproduction sponse disappeared shortly after we removed the cue or strategies (Magnhagen 1991). It has been shown that for- source of risk. Within the same night, the activity of the aging decisions under risk of predation are often a com- gerbils returned to its normal level. This result contrasts promise between maximizing net energy gain and mini- with aviary experiments (Kotler 1992) in which GA did either of the two gerbil species (Fig. 1). Similar results were obtained for the second part and throughout the night (Figs. 2 and 3). Gerbils did not appear to be able to use olfaction or sight to detect the presence of the stationary owl.

318

OECOLOGIA 105 (1996) 9 Springer-Verlag

not fully resume normal levels of activity for at least 5 nights following the removal of predators and GP did not resume normal activity levels for 1 or 2 nights. What may underlie these differences? One explanation may be that aviary experiments represent unnaturally high levels of risk not normally encountered in the field and avoided in our current field experiment with trained owls (barn owls were constantly present in the aviary for 1-3 nights prior to aviary experiments; they were present in the field for only 2 h). Effects of predators lingered far longer in the aviary where levels of risk were so much higher. It is also possible that in aviary experiments with unnaturally high levels of risk, as in many laboratory experiments, animals do not behave naturally, and this results in certain artifacts. In any case, the manipulations of risk using trained owls in the field reported here appear to give a more realistic picture of the magnitude of responses that occur in nature. The ability to alter and manipulate predatory risk for gerbils in the field has great potential for studying how predation affects foraging behavior, population interactions, and higher-order interactions. As we have demonstrated, the behavioral response of gerbils in nature to risk of predation in the guise of owl flights can be quantified using techniques such as sand tracking (Kotler 1985b; Mitchell etal. 1990) or manipulated resource patches (Brown 1988; Brown et al. 1988; Kotler et al. 1988, 1993a, b). The reduction of activity and the avoidance of open areas seen here suggest that gerbils forgo all but the richest foraging opportunities rather than to expose themselves to high risk. Gerbils are thus trading off food against safety. Using trained owls to manipulate risk also allows assessment of how gerbils respond to increasing risk by altering the frequency with which the owls are allowed to overfly the experimental grid. In this manner, it can be shown that gerbil activity is negatively related to number of owl flights (Z. Abramsky, M.L. Rosenzweig, and A. Subach, unpublished work), and the precise relationship in which foraging behavior is altered in response to changes in predatory risk can be quantified. The use of trained owls can be combined with manipulations of prey densities and level of predation to measure the shape of victim isocline under the assumption of ideal free distribution (Fretwell 1972; Z. Abramsky, M.L. Rosenzweig, and A. Subach, unpublished work). In a similar manner, manipulations of resource availability or of competitor density coupled with manipulations of the intensity of predatory risk can also be used to examine how the intensity of per capita competitive interactions change with changes in predator density (higher order interactions). This tells us how the interaction of two species are altered and how the outcome of that interaction is perhaps changed by the presence of predators. Thus, the ability to manipulate actual predatory risk in the field allows for studying aspects of behavior, species interactions, and community properties not otherwise accessible.

Acknowledgements We thank Dr. M.L. Rosenzweig for making many useful comments on the data analysis and on an earlier version of this work. We also thank M. Gabii, T. Ninari, Y. Kapitanov, N. Negler, A. Bar, R. Bluch, A. Shuat, Y. Witztom, and I. Zurim for their assistance in feeding and training the owls. We also thank Kibbutz Mishmar Hauegev and Kibbutz Behari for providing space to keep the owls. This project was supported by the Inter-university Ecology Foundation and by the United States-Israel Binational Science Foundation grant number 93-00059. This is publication number 209 of the Mitrani Center for Desert Ecology.

References Abrahams MV, Dill LM (1989) A determination of the energetic equivalence of the risk of predation. Ecology 70:999-1007 Abramsky Z (1984) Population biology of Gerbillus allenbyi. Mammalia 48:197-206 Abramsky Z (1988) The role of habitat and productivity in structuring desert rodent communities. Oikos 52:107-114 Abramsky Z, Sellah C (1982) Competition and the role of habitat selection in Gerbillus allenbyi and Meriones tristrami: a removal experiment. Ecology 63:1242-1247 Abramsky Z, Rosenzweig ML (1984) Tilman's predicted productivity-diversity relationship shown by desert rodents. Nature 309:150-151 Abramsky Z, Brand S, Rosenzweig ML (1985) Geographical ecology of gerbilline rodents in sand dune habitats of Israel. J Biogeogr 12:363-372 Abramsky Z, Pinshow B (1989) Changes in foraging effort in two gerbil species correlate with habitat type and intra- and interspecific activity. Oikos 56:43-53 Abramsky Z, Rosenzweig ML, Pinshow B, Brown JS, Kotler BR Mitchell WA (1990) Habitat selection: an experimental field test with two gerbil species. Ecology 71:2358-2369 Abramsky Z, Rosenzweig ML, Pinshow B (1991) The shape of a gerbil isocline. Ecology 72:329-340 Abramsky Z, Ovadia O, Rosenzweig ML (1994) The shape of a Gerbillus pyramidum (Rodentia: Gerbillinae) isocline: an experimental field study. Oikos 69:318-326 Anderson PK (1986) Foraging range in mice and voles: the role of risk. Can J Zool 64:2645-2653 Bar Y, Abramsky Z, Gutterman Y (1984) Diet of gerbilline rodents in the Israeli desert. J Arid Environ 7:371-376 Brown JS (1986) Coexistence on a resource whose abundance varies: a test with desert rodents. Ph.D dissertation, University of Arizona, Tucson Brown JS (1988) Patch use as an indicator of habitat preference, predation risk and competition. Behav Ecol Sociobiol 22:37-47 Brown JS (1989) Desert rodent community structure: a test of four mechanisms of coexistence. Ecol Monogr 59:1-22 Brown JS, Alkon PU (1990) Testing values of crested porcupine habitats by experimental food patches. Oecologia 83:512518 Brown JS, Kotler BR Smith RJ, Wirtz WO (1988) The effects of owl predation on the foraging behavior of heteromyid rodents. Oecologia 76:408-415 Brown JS, Kotler BE Mitchell WA (in press) Patch use as the basis for the structure of a Negev desert granivore community. Ecology Caraco T, Martindale S, Pulliam HR (1980) Avian flocking in the presence of a predator. Nature 285:400-401 Cassini MH (1991) Foraging under predation risk in the wild guinea pig Cavia aperea. Oikos 62:20-24 Cerri RD, Fraser DF (1983) Predation and risk in foraging minnows: balancing conflicting demands. Am Nat 121:552-561 Clarke JA (1983) Moonlight's influence on predator/prey interactions between short-eared owls (Asioflammeus) and deer mice (Peromyscus maniculatus). Behav Ecol Sociobiol 13:205-209 Danin A (1978) Plant species diversity and plant succession in sandy area in the northern Negev. Flora 167:409-422

OECOLOGIA 105 (1996) 9 Springer-Verlag Dill LM, Fraser AHG (1984) Risk of predation and the feeding behavior of juvenile coho salmon (Oncorhynchus kisutch). Behav Ecol Sociobiol 16:65-71 Doucet GJ, Bider JR (1969) Activity of Microtus pennsylvanicus related to moon phase and moonlight revealed by the sand transect technique. Can J Zool 47:1183-1186 Edwards J (1983) Diet shifts in moose due to predator avoidance. Oecologia 60:185-189 Fretwell SD (1972) Populations in a seasonal environment. Princeton University Press, Princeton Gilliam JF, Fraser DF (1987) Habitat selection under predation hazard: test of a model with foraging minnows. Ecology 68:1856-1862 Holbrook SJ, Schmitt RJ (1988) The combined effects of predation risk and food reward on patch selection. Ecology 69:125-134 Kotler BP (1984a) Harvesting rates and predatory risk in desert rodents; a comparison of two communities on different continents. J Mammal 65:91-96 Kotler BP (1984b) Risk of predation and the structure of desert rodents communities. Ecology 65:689-701 Kotler BP (1985b) Microhabitat utilization in desert rodents: a comparison of two methods of measurement. J Mammal 66:374-378 Kotler BE Brown JS (1988) Environmental heterogeneity and the coexistence of desert rodents. Annu Rev Ecol Syst 19:281-308 Kotler BP, Holt RD (1989) Predation and competition: the interaction of two types of species interactions. Oikos 54:256-260 Kotler BE Brown JS, Smith RJ, Wirtz WO (1988) The effects of morphology and body size on rates of owl predation on desert rodents. Oikos 53:145-152 Kotler BE Brown JS, Hasson O (1991) The specter of predation: factors affecting gerbil foraging behavior and rates of owl predation. Ecology 72:2249-2260 Kotler BE Blaustein L, Brown JS (1992) Predator facilitation: the combined effect of snakes and owls on the foraging behavior of gerbils. Ann Zool Fenn 29:199-206 Kotler BE Blaustein L, Dednam H (1993a) The specter of predation: the effect of vipers on the foraging behavior of two species of gerbils. Isr J Zool 39:11-22 Kotler BE Brown JS, Slotow R, Goodfriend W, Strauss M (1993b) the influence of snakes on the foraging behavior of gerbils. Oikos 67:309-316 Kotler BE Brown JS, Subach A (1993c) Mechanisms of species coexistence of optimal foragers: temporal partitioning by two species of sand dune gerbils. Oikos 67:548-556 Lima SL (1985) Maximizing feeding efficiency and minimizing time exposed to predators: a trade-off in the black-capped chickadee. Oecologia 66:60-67 Lima SL, Dill LM (1990) Behavioral decisions made under the risk of predation: a review and prospectus. Can J Zool 68:619-640

319

Lima SL, Valone TJ, Caraco T (1985) Foraging-efficiency-predation-risk trade-off in the grey squirrel. Anita Behav 33:155-165 Lockard RD, Owings DH (1974) Moon-related surface activity of bannertail (Dipodomys spectabilis and Dipodomys nitratoides) kangaroo rats. Anita Behav 22:262-273 Longland WS, Price MV (1991) Direct observations of owls and heteromyid rodents: can predation risk explain microhabitat use? Ecology 72:2261-2273 Magnhagen C (1991) Predation risk as a cost of reproduction. Trends Ecol Evol 6:183-186 Milinski M, Heller R (1978) Influence of a predator on the optimal foraging behavior of sticklebacks (Gastetvsteus acuIeatus L.). Nature 275:642-644 Mitchell WA, Abramsky Z, Kotler BP, Pinshow B, Brown JS (1990) The effect of competition on foraging activity in desert rodents: theory and experiments. Ecology 71:844-854 Nonacs P, Dill LM (1990) Mortality risk vs. food quality tradeoffs in a common currency: ant patch preferences. Ecology 71:1886-1892 Ohman MD, Frost BW, Cohen EB (1983) Reverse dial vertical migration: an escape from invertebrate predators. Science 220:1404-1407 Price MV, Waser NM, Bass TA (1984) Effects of moonlight on microhabitat use by desert rodents. J Mammal 65:353-356 Rosenzweig ML, Abramsky Z (1985) Detecting density-dependent habitat selection. Am Nat 126:405417 Rosenzweig ML, Abramsky Z (1986) Centrifugal community organization. Oikos 46:339-348 Rosenzweig ML, Abramsky Z, Brand S (1984) Estimating species interactions in heterogeneous environments. Oikos 43:329-340 Sih A (1980) Optimal behavior: can foragers balance two conflicting demands? Science 210:1041-1043 Sih A (1982) Foraging strategies and the avoidance of predation by an aquatic insect, Notonecta hoffmanni. Ecology 63:786 796 Stamp NE, Bowers MD (1991) Indirect effect on survivorship of caterpillars due to presence of invertebrate predators. Oecologia 88:325-330 Thompson SD (1982) Structure and species composition of desert Heteromyid rodent species assemblages: effects of a simple habitat manipulation. Ecology 63:1313-1321 Werner EE, Gilliam JF, Hall DL, Mittelbach GG(1983) An experimental test of the effects of predation risk on habitat use in fish. Ecology 64:1540-1548 Ziv Y, Abramsky Z, Kotler BE Subach A (1993) Interference competition and temporal and habitat partitioning in two gerbil species. Oikos 66:237-246