Effects of predation risk on rodent foraging behavior in the Colorado Desert

Kobrina E. Boslough1, Jessica Du2, Anass Malabeh3, Caitlyn N. Rich4

1University of California, Berkeley; 2University of California, Riverside; 3University of California, Irvine; 4University of California, Santa Cruz

ABSTRACT

The survival of animals often depends on a tenuous balance between the rewards of food acquisition and the risks associated with predation. In desert environments where resources are incredibly limited, it is imperative for organisms to optimize physiological and behavioral traits in order to maximize nutritional reward while minimizing risks of predation. In this study, we examined the effects of simulated predation risk on the foraging behavior of granivorous desert rodents. We predicted that rodents would become more vigilant and forage less with increased levels of risk. While predator scent did not alter rodent foraging behavior, increased exposure via moonlight and lack of cover reduced rodent foraging activity. This study provides evidence that rodent foraging behavior may depend more on factors that increase exposure and visibility as opposed to predator presence, however further studies are needed to determine the full effects of other predator cues.

Keywords: predation avoidance, predation risk, Dipodomys merriami, rodent foraging, rodent behavior

INTRODUCTION primary driving force in morphological and behavioral adaptations in prey (Evans and Two of the most important factors Schmidt 1990). These non-lethal effects of contributing to species survival are food predators on prey may impose challenges in acquisition and predator avoidance (Lima resource limited environments, such as and Dill 1990). Through predator avoidance, deserts. prey species may lose valuable resource Desert habitats exaggerate prey responses opportunities, indirectly affecting their due to minimal vegetative cover and survival and reproductive success (Bowers stressful abiotic elements, making them 1990). Alterations in anti-predatory ideal systems for understanding the effects strategies due to predator presence and of predator presence (Upham and Hafner non-lethal predation risks may result in 2013). High daytime temperatures and low higher vigilance levels, which may negatively precipitation pose high risk of water loss in affect energy intake rates (Brown et. al all desert species. Many desert organisms 1999). Interactions with predators are also a

CEC Research | https://doi.org/10.21973/N3SS9S Fall 2018 1/6

have adapted to become nocturnal burrow- Borrego Desert Research Center (N dwellers and forage at night (Sjoberg et al. 33.23824° W -116.39278°) from October 31- 1984). The activity of animals, such as small November 7, 2018 (Figure 1). Vegetative mammals, is affected when moonlight cover in Anza-Borrego is sparse and visually exposes prey to predators, causing dominated by creosote and mesquite. The small nocturnal mammals to primarily be reserve is approximately 0.32 km2 and active before moonrise (Upham and Hafner located in San Diego County, California, 2013). The presence of moonlight also bordering Anza-Borrego State Park. Average changes microhabitat use, driving organisms temperature is 24–29 °C. Average away from open areas in order to seek precipitation is 12.7-17.8 cm per year. refuge under brush cover (Upham and During the duration of our study, we Hafner 2013). In addition to abiotic identified three rodent species: kangaroo rat elements, biotic factors such as predation (Dipodomys merriami), cactus mouse risk can promote variance in behavioral (Peromyscus eremicus), and desert pocket responses that may lead to niche mouse (Chaetodipus penicillatus). partitioning and change in risk-taking behaviors among species (Kotler 1984, Kotler 1985, Kotler et al. 1994, and Brown et al. 1988). Other biotic factors, such as competition with other granivores, as well as predator presence, impact rodent habitat use and behavioral patterns (Lima and Dill 1990). In this study, we investigated how predation risk affects rodent foraging behavior. We Figure 1. Location of study site. Research took place baited rodents in areas with high and low at Steele Burnand Anza-Borrego Desert Research predation risk. We predicted that overall Center (N 33.23824° W -116.39278°). Wash area rodent activity would be lower when the where bait and cameras were deployed is represented by the outlined oval. moon was present. In addition, we predicted that higher levels of risk would correspond 2.2 Sampling to more time spent vigilant, lower visitation rates, and less time spent foraging and We tested the effects of predation risk in feeding. The results of this study will provide desert rodents by setting out seed bait evidence of prey behavioral response to treatments and observing behavioral predator presence (Brown et. al 1999). patterns. Each night, we sampled eight creosote bushes (Larrea tridentata) with METHODS visibly active rodent burrows. Treatments included Forti-Diet parakeet feed (20 g per 2.1 Study System dish), water, and commercially purchased Research was conducted in a desert scrub Bare Ground 100% meat-fed coyote urine to habitat located at Steele Burnand Anza- simulate predator presence. Research on deer mice (Peromyscus maculatus) has

CEC Research | https://doi.org/10.21973/N3SS9S Fall 2018 2/6

shown that rodents are sensitive to coyote 2.3 Statistical analyses urine (Nolte 1994). Each bush was pseudo- randomly assigned one of four treatments, All statistical analyses were conducted in (1) seed bait under brush drip line JMP statistical software v14. T-tests were (near+seeds), (2) seed bait 2 meters away run to determine how moonrise affects from drip line (far+seeds), (3) seed bait rodent activity, frequency of visitation, and surrounded with coyote urine (5-20 drops) species presence. Logistic regressions were under drip line (near+seeds+urine), and (4) used to examine how distance from shelter water bait 2 meters away from drip line and urine presence affect visitation rates of (control). We reclassified water treatments rodents. A t-test was used to assessed the as a control for behavioral observations effect on rodent behavior by coyote urine because rodents rarely interacted with the and distance from shrub with near seed water dish (one data point excluded). Each treatments. Finally, a Wilcoxon test was treatment had ten replicates, except used to see how behavior budgeting varied far+seeds, which had nine. Bait dishes were across different rodent species. deployed at sunset, left overnight (approximately 12 hours), and collected RESULTS after dawn. Bait was filmed with Bushnell We observed behavior three species: D. game cameras positioned approximately merriami, C. penicillatus, P. eremicus in our one meter away, pointed towards the bait 39 total camera traps. Eighty-five percent of and targeted bush. Motion sensing mode cameras had D. merriami visitation, 15% had was used to record footage that lasted for C. penicillatus visitation, and 5% of cameras either 15 or 60 seconds, with 10 second had P. eremicus visitation. Dipodomys intervals between recordings. merriami activity was higher prior to After cameras were collected, footage was moonrise (N= 69, t=4.1, P< 0.0001; Figure 2). reviewed and behavior budgeting strategies The total amount of C. penicillatus and P. were classified into three categories: eremicus activity marginally increased after foraging, feeding, and vigilance. Foraging moonrise (Nother rodent= 19, tother rodent= 1.8, was defined as rodents actively moving Pother rodent= 0.07; Figure 3). During behavioral around in search for food. Feeding consisted observations, there were three occasions in of rodents remaining stationary while which interspecies competition was inserting food into cheek pouches or actively observed between D. merriami and C. chewing. Vigilance was categorized as penicillatus. In all three instances, D. individuals displaying a lifted head, either merriami excluded C. penicillatus. scanning surroundings or remaining motionless. Behavior time was tracked in each video clip with start and end timestamps. In addition to setting game cameras, we deployed Sherman traps on the third night of the study to accurately identify rodent species in the videos.

CEC Research | https://doi.org/10.21973/N3SS9S Fall 2018 3/6

Figure 2. Effect of treatment type on percent of time spent vigilant. Bars represent percent of time spent Figure 4. Effects of distance from shelter on total vigilant per camera per night. Predator presence, as rodent visitation. Bars represent total rodent simulated by coyote urine spread around the bait visitations per camera per night. There was a slight dish, did not alter behavior in rodents as we had pattern that indicated rodents may avoid foraging out predicted. Error bars represent ± 1 S.E.M. in the open and instead prefer lower-risk, sheltered areas. Error bars represent ± 1 S.E.M.

Figure 5. Observational time budget of each species of rodents. Time spent on each activity (feeding, foraging, or vigilant) did not differ between D. merriami, P. eremicus, and C. penicillatus.

There were marginally fewer visitations at Figure 3. Effect of moon presence on rodent activity far+seeds treatments than at near+seeds by taxa. Bars represent mean number of rodent visitations per camera trap per night. Dipodomys treatments (N= 19, t= 5.9, P = 0.08; Figure 4). merriami was more active when the moon was not However, there was no effect on rodent present, while other rodents (C. penicillatus and P. visitation between near + seeds and near + eremicus) were more active when the moon was seeds+ urine treatments (N= 20, t=5.21, P = present. Error bars represent ± 1 S.E.M. 0.686; Figure 3). In addition, time spent on

CEC Research | https://doi.org/10.21973/N3SS9S Fall 2018 4/6

vigilance did not alter between the three We found that distance from creosote treatment types (NFar vs Near= 19, TFar vs bushes did not affect vigilance levels among Near=3.16 , PFar vs Near= 0.17; NUrine vs rodents, but did marginally affected overall Near= 20, TUrine vs Near= 3.4, PUrine vs visitation rates by rodents. Even though far Near= 0.50; Figure 4). In behavioral treatments posed greater predation risk, D. observations, the three rodent species merriami’s sensitive auditory bullae and foraged, fed, and were vigilant in similar bipedalism may have provided them with a proportions. (2Feeding= 1.83, PFeeding= predator evasion strategy in both near and 0.40; 2Foraging= 1.55, PForaging= 0.46; far treatments (Kotler 1985). Despite 2Vigilance= 0.12, PVigilance= 0.94; Figure 5). olfactory stimuli, predator presence had no observable effects on D. merriami behavior DISCUSSION or visitation rates between treatments of bait with and without coyote urine. Absence Dipodomys merriami activity was greater of fear-inducing auditory cues from potential before moonrise than after, therefore nearby predators allowed little variance in D. moonlight may be a determining abiotic merriami behavior and visitation rates. factor for nocturnal rodent foraging Therefore, possible future studies can mimic behavior (Kotler 1984, Longland and Price varying predator types that may induce 1991). In Upham and Hafner’s (2013) study stronger fear based behavioral patterns. In conducted between March and October addition to coyotes, foxes, snakes, and owls (1999-2006), a similar pattern of decreased are common rodent predators. Utilizing owl D. merriami foraging activity after the onset pellets or snake musk may signaling predator of moonrise was observed. With a decrease presence to rodents differently, and it is in D. merriami activity after moonrise, C. possible that the rodents will react penicillatus and P. eremicus activity may differently to differing predators. Because D. have increased due to reduced competition merriami has enlarged auditory bullae, a and exclusion. Dipodomys merriami study focusing on them can situate speakers aggression was documented in three next to the bait and emit sounds of instances, showing D. merriami preventing predators, testing the rodent’s reaction to P. eremicus from feeding on seed bait auditory stimuli (Kotler 1985). through direct competition. These Behavioral patterns in rodents are rarely observations of dominance by D. merriami quantified in ecological research, making it may explain why sample population size of difficult to gather data on responses to D. merriami was greater than P. eremicus predation risk. Our study may fill a and C. penicillatus during our behavioral knowledge gap by combining quantitative observation periods. However, patterns in and observational data of predation behavioral budgets were similar between D. avoidance strategies in rodents. This merriami, P. eremicus, and C. penicillatus, information may help us predict how allowing us to evaluate rodent behavior for predator-prey interactions function in other each treatment as a whole rather than for ecosystems. With further understanding of each species individually. the direct and indirect negative effects on fitness caused by fear responses, we may be

CEC Research | https://doi.org/10.21973/N3SS9S Fall 2018 5/6

able to begin to understand the broad-scale combinations of predation risk and energy cost. implications of interspecies interactions. Animal Behavior 30:637–640. Jorgensen, E. E. and S. Demarais. 1999. Spatial scale ACKNOWLEDGMENTS dependence of rodent habitat use. Journal of Mammalogy 80:421-429. This work was performed at the University of California’s Steele Burnand Anza- Kotler, B. P. 1984. Risk of predation and the structure Borrego Desert Research Center, of desert rodent communities. Ecology 65:689– doi:10.21973/N3Q94F. We would like to 701. thank Krikor Andonian, Tim Miller, and Kate Kotler, B. P. 1985. Owl predation on desert rodents Melanson for their help, support and which differ in morphology and behavior. American guidance. Society of Mammalogists 66:824–828.

REFERENCES Kotler, B. P., Brown, J. S., & Mitchell, W. A. (1994). The role of predation in shaping the behavior, Banta, M. L. 2003. Merriam’s kangaroo rats morphology and community organization of desert (Dipodomys merriami) voluntarily select rodents. Australian Journal of Zoology 42:449–466. temperatures that conserve energy rather than water. Physiol. Biochem. Zool. 76:522–532. Laundre, J. W., W. J. Ripple, and L. Hernandez. 2010. The landscape of fear: ecological implications of Bean, M. L. 1983. Adaptive physiology of Heteromyid being afraid. The Open Ecology Journal 3:1–7. rodents. Great Basin Naturalist Memoirs 7:65–76. Lima, S. L. and L. M. Dill 1990. Behavioral decisions Bowers, M. A. 1990. Exploitation of seed aggregates made under the risk of predation: a review and by Merriam's kangaroo rat: harvesting rates and prospectus. Can. J. Zool. 68:619–640. predatory risk. Ecology 71:2334–2344. Nolte, D. L., J. R. Mason, G. Epple, E. Aronov, and D. L. Brown, J. S., J. W. Laundre, and M. Gurung. 1999. The Campbell. 1994. Why are predator urines aversive ecology of fear: optimal foraging, game theory, and to prey? Journal of Chemical Ecology, 20:1505– trophic interactions. Journal of Mammalogy 1516. 80:385–399. Sjoberg, D. E., J. A. Young, K. McAdoo, and R. A. Evans. Evans, D. L. and J. O. Schmidt. 1990. Insect Defenses: 1984. Kangaroo rats. Rangelands 6:11–13. Adaptive Mechanisms and Strategies of Prey and Predators. SUNY Press. New York, New York, USA. Upham, N. S. and J. C. Hafner. 2013. Do nocturnal rodents in the Great Basin avoid moonlight? Grubb, T. C., Jr., and L. Greenwald. 1982. Sparrows Journal of Mammalogy 15:59–72. and a brushpile: foraging responses to different

CEC Research | https://doi.org/10.21973/N3SS9S Fall 2018 6/6