Ecological Monographs, 72(4), 2002, pp. 561–577 ᭧ 2002 by the Ecological Society of America SMALL-MAMMAL FORAGING BEHAVIOR: MECHANISMS FOR COEXISTENCE AND IMPLICATION FOR POPULATION DYNAMICS JOHN A. YUNGER,1,3 PETER L. MESERVE,1 AND JULIO R. GUTIE´ RREZ2 1Department of Biological Sciences, Northern Illinois University, DeKalb, Illinois 60115 USA 2Departamento de Biologia, Universidad de La Serena, Casilla 599, La Serena, Chile Abstract. We investigated predation risk and competition as they affected small-mam- mal foraging behavior in semiarid north-central Chile. Giving-up densities (GUD) of seeds were used to measure the foraging activity of the three most common small mammals at the site: degu (Octodon degus), Darwin’s leaf-eared mouse (Phyllotis darwini), and the olivaceous field mouse (Akodon olivaceus), under shrubs (cover) and in the open on pred- ator-excluded and competitor-excluded (Octodon) plots. Experiments were conducted dur- ing both new and full moons. Monthly small-mammal censuses using standard mark– recapture techniques provided data on movement, reproduction, and long-term fluctuations in density between 1989 and 1994. Diurnal Octodon foraged more (had lower GUD) in the absence of predators (although this was confounded by a numerical increase resulting from predator exclusion), and foraged more under shrubs than in the open. However, the lack of a significant cover ϫ predator exclusion interaction and thermoregulation studies suggest that physiological constraints play a greater role than predation risk in determining micro- habitat selection by Octodon. Predation risk, as influenced by lunar light levels and predator exclusion, had only weak effects on microhabitat selection by Phyllotis. The strong tendency of Octodon to forage under cover could depress food availability, forcing Phyllotis to feed more in the open. Concomitantly, Phyllotis exhibits several morphological characters that would favor detection and avoidance of predators. There is extensive evidence that inter- specific competition is the primary constraint on Akodon foraging; predation risk appears to be relatively unimportant. Akodon is also a less efficient forager than the two other species, having significantly higher GUD, on average. This is partially offset by differences in reproductive biology; Akodon exhibits extremely rapid demographic responses to fa- vorable changes in the environment. A fourth species, Abrothrix longipilis, may coexist in the system because of its opportunistic nature, but no data are available on its foraging efficiency. These behavioral and biotic interactions occur within a background of periodic El Nin˜o–Southern Oscillations (ENSO), which may ultimately contribute to species co- existence. Key words: Akodon olivaceus; Chile; competition; foraging behavior; giving-up density; micro- habitat; Octodon degus; Phyllotis darwini; population dynamics; predation risk; semiarid; small mammals. INTRODUCTION lating biotic interactions, it may be possible to elucidate mechanisms of community composition and dynamics. General background Brown (1989) studied foraging behavior and coex- To understand how species coexist within a com- istence of four rodent species in the Sonoran Desert. munity and the temporal dynamics of a system, it is Seasonal rotation in foraging efficiency was the mech- necessary to integrate across ecological levels and anism permitting coexistence among three of the spe- scales (Bissonette 1997). For example, individuals for- cies (Brown 1989). Spatial variation in resource abun- age to procure energy for growth, maintenance, and dance was presumed to be responsible for coexistence reproduction. However, interference from competitors, of a fourth species in the small-mammal assemblage. risk of predation, and physiological constraints all can Selection of open vs. bush microhabitat and temporal influence the resource level at which an individual will variation in resource abundance appeared to be of rel- abandon, or choose not to forage in, a patch. By in- atively little importance. In the Israeli Negev Desert, corporating environmental heterogeneity, while simul- live-trapping at large spatial scales (macrohabitat) taneously measuring foraging behavior and manipu- showed that two species of gerbils occurred most fre- quently in semi-stabilized dune habitats, although un- Manuscript received 9 February 2001; revised 16 January der high densities, secondary habitat preferences di- 2002; accepted 28 January 2002. verged between stabilized and unstabilized dunes (Ro- 3 Present address: Environmental Biology Program, Gov- ernors State University, University Park, Illinois 60466 USA. senzweig and Abramsky 1986, Abramsky et al. 1990). E-mail: [email protected] Both gerbils foraged more under low levels of lunar 561 562 JOHN A. YUNGER ET AL. Ecological Monographs Vol. 72, No. 4 light intensity (Kotler 1984a) and beneath shrubs as degus, 140.9 Ϯ 20.9 g, X¯ Ϯ 1 SE), which is diurnal opposed to open areas (microhabitat; Kotler et al. (Meserve and Le Boulenge´ 1987, Wilson and Reeder 1991). Such foraging behaviors are putative modes of 1993). All three species feed largely or exclusively on reducing predation risk. In the Namib Desert, Hughes plant material, including seeds. Seed utilization of the et al. (1994) studied the foraging behavior of the noc- three species in decreasing order is Phyllotis, Akodon, turnal dune hairy-footed gerbil (Gerbillurus tytonis)in and Octodon (Meserve 1981a, b). three microhabitats: open, grass, and brush. This gerbil foraged the most in brush, followed by grass and then Hypotheses open areas. The authors attributed the pattern to re- Predation risk.—Analysis of vertebrate predator di- duced predation risk in brush and grass microhabitats. ets showed that Octodon were consumed more than In all three microhabitats, G. tytonis foraged more un- would be predicted based on total prey availability, der a new moon than a full moon. Effects of a putative followed by Phyllotis, with Akodon representing a com- competitor were unclear, given no patterns in micro- paratively low proportion of the predator diets (Jaksic habitat shifts as a result of the removal. et al. 1993, 1997). From these results, two alternative Because predators were never directly manipulated, predation risk hypotheses can be suggested. First, Oc- these studies provide limited insight into the influence todon should show a clear pattern of foraging more of predation. Although Hughes et al. (1994:1397) pur- under shrubs, followed by Phyllotis, with Akodon ex- ported to ‘‘provide direct experimental evidence on the hibiting the least microhabitat preference. Alternative- role of predation risk...,’’lunar light levels were used ly, Akodon may be under moderate or high predation as a surrogate. This approach has commonly been em- intensity, but are least observed in predator diets be- ployed when investigating small-mammal foraging cause they are good at avoiding depredation, e.g., by (e.g., Lockard and Owings 1974, Kotler 1984b, c, Price staying under shrubs. At the same time, the high dep- et al. 1984, Bowers 1988). Although it provides useful redation rate of Octodon could be the result of indi- information, alternative explanations, such as ther- viduals spending a proportionately greater amount of moregulatory constraints (Goodfriend et al. 1991, La- time in the open. Two lines of evidence support the gos et al. 1995a) or intraspecific competition, may at former hypothesis. First, preliminary field observations least partially explain the foraging regimes observed suggest that Akodon spent more time in the open than under different lunar light levels (Reichman and Price did Octodon. Because the vast majority of the predator 1993). Under low levels of lunar light intensity, ver- assemblage includes species that have been observed tebrate predators may still be present and may forage to feed more in the open than in or under shrubs, Ako- successfully (Dice 1945). In addition, lunar light in- don may not be under high predation intensity. Second, tensity is irrelevant for diurnal prey species. Octodon has exhibited a significant numerical response In this study, predator and competitor exclosures to predator exclusion, whereas Akodon has not (Mes- were used to test four spatial and three temporal hy- erve et al. 1999). potheses on small-mammal coexistence. These hy- Acceptance of the first alternative hypothesis leads potheses, which are not mutually exclusive, were tested to three predictions. First, Octodon GUD should be using natural variation in cover and light intensity. In significantly lower under shrubs than in the open, the context of these manipulations and variation, the whereas there should be no significant difference in foraging patterns of small mammals were examined Akodon GUD between the two microhabitats. Second, using giving-up densities (GUD), the level at which an Octodon should show significantly lower GUD in the individual ceases to use a particular resource (Brown absence of predators than in their presence, followed 1988, Kotler et al. 1991, Brown et al. 1992); similar by Phyllotis and then Akodon. Third, Octodon should methods have been used elsewhere (e.g., Bowers et al. show the greatest shift toward foraging in the open as 1993, Bowers and Breland 1996). Most often the re- a result of predator exclusion, followed by Phyllotis, source is some type of food mixed with a matrix ma- with little or no significant response by Akodon. terial (e.g., sand), resulting in a diminishing rate of Physiological constraints.—Diurnal