589 J.L. Penner, et al. 22001133 SOUSToHuEthAeSaTstEeRrnN N NaAtuTrUaRlisAtLIST 12V(o3l). :1528,9 N–5o9. 83 Are High Pilferage Rates Influenced by Experimental Design? The Effects of Food Provisioning on Foraging Behavior Jennifer L. Penner1,*, Kelly Zalocusky1, Lucy Holifield1, Jordan Abernathy1, Brenley McGuff1, Sarah Schichtl1, Whitney Weaver1, and Matthew D. Moran2 Abstract - We performed a field experiment to determine the effect of supplemental food provisioning on the cache-pilferage rates of a natural squirrel population in Central Arkansas. Provisioning, which is sometimes used to attract animals to a study site and/ or to induce foraging behavior, may increase the local food availability above what is normally found and may in itself have profound effects on behavior. To test the hypothesis that artificial provisioning inflates cache-pilferage rates, we tracked the pilferage rates of experimenter-made caches at field sites that had been either provisioned or not provisioned with supplemental food. We found that squirrels pilfered caches at a significantly higher rate when the site was provisioned (med = 34.8) compared to similar caches placed in non-provisioned plots (med = 27.9), a difference that is likely the consequence of increased squirrel density and foraging intensity at provisioned sites. Based on these results, we suggest that researchers include appropriate non-provisioned controls and be conservative in the interpretation of pilfering rates obtained from study sites with supplemental food. Introduction Scatter-hoarding animals survive periods of food scarcity by storing harvested food throughout their home range. The stored food is buried among multiple sites (caches) and then later retrieved by cache owners who typically rely on memory for individual cache sites (reviewed in Brodin 2010, Smulders et al. 2010). For scatter hoarding to be sustained as an evolutionarily stable strategy that is resistant to “cheaters”, cache owners must be more likely than naïve competitors, or pilferers, to recover their own caches (Andersson and Krebs, 1978, Clarkson et al. 1986, Smith and Reichman 1984). Scatter-hoarding birds and mammals are equipped with behavioral and psychological adaptations aimed at cache concealment and protection (reviewed by Dally et al. 2006) that provide this advantage. Despite their efforts, animals frequently lose 2–30% of their caches daily (reviewed in Vander Wall and Jenkins 2003), with some reports of animals losing even more (e.g., Sciurus [tree squirrels], >95% loss within 24 h; Kraus 1983) and with losses resulting from both conspecific (e.g., Tamias [chipmunks]; Clarke and Kramer 1994) and heterospecific pilfering (e.g., Dipodomys [kangaroo rats]; Leaver and Daly 2001). Such high pilfering rates indicate that the fitness of “cheater” phenotypes may in some cases exceed that of “honest cacher” phenotypes, leading some researchers to propose alternative mechanisms by which caching could evolve as 1Department of Psychology,2Department of Biology, Hendrix College, 1600 Washington Avenue, Conway, AR 72032. *Corresponding author - penner@hendrix.edu. J.L. Penner, et al. 2013 Southeastern Naturalist Vol. 12, No. 3 590 an evolutionarily stable foraging strategy. For instance, Vander Wall and Jenkins (2003) proposed that at least some of the costs of cache loss to an individual owner may be offset by “reciprocal” pilfering. Additionally, the costs associated with pilfering may be outweighed by the value of stored food during seasonal periods of food scarcity (Smith and Reichman 1984). A common protocol for measuring pilfering rates in the field is the provisioning of research sites with nutritious food (sometimes native and sometimes non-native), typically to attract animals and/or to induce scatter hoarding. Such artificial food supplements are meant to mimic the natural, localized abundance that caching animals encounter in the wild (e.g., from masting trees or seed banks). The provisioning of research sites is done in a variety of ways and for a variety of durations, and is usually aimed at inducing foraging behavior so that natural caching, pilfering, and pilferage avoidance behavior can be observed. For instance, researchers may target individual animals during single provisioning trials (e.g., Petroica australis Sparrman [New Zealand Robin; Van Horik and Burns 2007], [kangaroo rats; Leaver and Daly 2001]), repeatedly provision research sites throughout the duration of data collection to observe foraging behavior (e.g., Sciurus lis Temminick [Japanese Squirrel; Tamura et al. 1999]), bait animal live traps (kangaroo rats; Murray et al. 2006), or use populations of animals at parks, college campuses, or research stations that are accustomed to human presence and/or provisioning (e.g., Sciurus carolinensis Gmelin [Eastern Gray Squirrel; Leaver et al. 2007] and Tamias striatus [Eastern Chipmunk; Clarke and Kramer 1994]). Importantly, even when the provisioned food is native to the field sites, it is often presented to animals in an unnatural way (concentrated at a single location) that eliminates or alters the handling or travel time that would otherwise be required for successful harvesting. It is not surprising then that supplemental food has been shown to increase the population density of hoarding animals to unnatural levels. For instance, Klenner and Krebs (1991) provisioned eight study sites ad libitum with sunflower seeds over the course of two summers and found that populations of Tamiasciurus hudsonicus Erxleben (American Red Squirrel) doubled or quadrupled, depending on habitat, and remained inflated for at least six months. More recently, Donald and Boutlin (2011) reported that when American Red Squirrels’ individual middens (larders) were supplemented with peanut butter during the winters over six consecutive years, population densities more than doubled compared to the density recorded before provisioning began. Likewise, Eutamias townsendii Bachman (Townsend’s Chipmunk) populations provisioned with sunflower seeds and whole oats increased to one and a half times that of a control population within months (Sullivan et al. 1983). Increased juvenile recruitment and/or decreased home range appeared to cause the localized increase in density in these studies. These findings indicate that provisioning has the potential to attract enough additional animals to influence population density. In turn, provisioning would be expected to affect pilfering rates. Indeed, a field study conducted with scatter-hoarding birds (Parus montanus von Baldenstein 591 J.L. Penner, et al. 2013 Southeastern Naturalist Vol. 12, No. 3 [Willow Tits]), showed that pilfering rates of experimentally placed sunflower seed caches were negatively correlated with their distance from a sunflower seed feeder, but that there was no relationship between pilfering rates and cache location for the same field sites during trials where the feeder was absent (Brodin 1993). Thus, the presence of an artificial food source can influence the motivation, behavior, and success with which naïve birds find scatter-cached food. To our knowledge, there has been one study to date investigating the relationship between supplemental feeding and pilfering behavior in small mammals. Donald and Boutin (2011) reported that supplemental food (in the form of peanut butter) had no influence on the conspecific pilfering of cones from larder-hoarding American Red Squirrel middens; instead the biggest predictor of pilfering behavior was the absence of the midden owner, a finding that is not surprising since territorial red squirrels physically defend their middens from would-be robbers. What is not yet known is how provisioning might alter the competitive environment and therefore the foraging behavior of scatter-hoarding mammals who, unlike larder hoarders who manage a single larder, are unable to physically defend their multiple, widely distributed caches from pilferers. The goal of our study was to determine whether or not supplemental food would affect the foraging behavior of natural populations of Eastern Gray Squirrels and Sciurus niger L. (Fox Squirrels), and we addressed this question by using methods similar to that of Brodin (1993). Specifically, we predicted that squirrels would discover more experimentally placed caches buried at sites with supplemental food compared to control sites that were not provisioned. Study Site Our study site was located on 24 ha of mature upland forest about 3 km south of Center Ridge, AR in the north-central portion of the state. The forest is dominated by Quercus spp. (oaks), Carya spp. (hickories), and Pinus echinata Mull (Shortleaf Pine). It is bounded on the west, north, and south sides by additional upland forest (total contiguous area is approximately 500 ha) and on the east side by cattle pasture. A forest stand analysis indicated that the forest had not been harvested for about 80 years (M.D. Moran, unpubl. data). Eastern Gray Squirrels and Fox Squirrels were present and commonly seen foraging in study areas. Neither species had been hunted on the property for >10 years (property owner, pers. comm.). Glaucomys volans L. (Southern Flying Squirrel) was also present, although seldom sighted because of its nocturnal behavior. Materials and Methods From 1–31 March 2008, we performed a preliminary, unreplicated, provisioning and pilfering experiment on two 400-m2 plots that were approximately 500 m apart. Based on mean range size assumptions (Bendel and Gates 1987, Koprowski 1994a, b) this distance is great enough that individual squirrel ranges were unlikely to include more than one experimental plot. In the experimental plot, squirrels J.L. Penner, et al. 2013 Southeastern Naturalist Vol. 12, No. 3 592 received black-oil sunflower seeds and raw, in-shell Pawnee pecans (Sun Valley Pecan Co., Fabens, TX) ad libitum during the course of the experiment. The control plot was not provisioned. At both experimental and control sites, pecans used for experimental caches were labeled with small, numbered aluminum tags attached with nylon-coated wire. We buried 40 pecans per plot by first removing a plug of soil with a metal soil sampler, and then placing a tagged pecan in the bottom of a 3.5-cm-deep hole, yielding a cache that was 1 cm deep (the depth at which we had observed most squirrels to bury their own caches; J. Pen ner and L. Holifield, unpubl. data). Individual pecans were placed an average of 4 m apart. We then replaced the extracted plug of soil, pressed the soil down, rearranged the leaf litter over the area, and recorded the cache location on a paper representation of the grid. Gloves were worn throughout the procedure to minimize the introduction of human odor, and pecans never made contact with our hands. Using a metal detector, we tracked cache survivorship for 31 days by monitoring caches daily for the first few days, then every 2–3 days until the termination of the experiment. We counted caches as pilfered when we could no longer detect a tag at the cache site. Usually a pilfered cache was obvious to us (i.e., a conspicuous hole was present). Thus, we could determine cache disappearance with a high degree of certainty. We searched experimental and control sites at the same times and with the same method to standardize the level of human disturbance. From 4 October–8 November 2008, we performed a replicated experiment by establishing six 22-m × 22-m plots at least 100 m apart on the same field site. The distance between plots was smaller for the replicated experiment because of geographic and space limitations. For the replicated experiment, we placed all the study plots on flat terrain, which was limited, and we distributed our six plots as evenly as possible throughout the 24-ha study site. The smaller inter-plot distances likely increased the probability that a few squirrels had ranges overlapping multiple plots. To assist researchers in relocating cache sites during data collection, colored flags were placed in the ground every 2 m around the perimeter of each plot, as well as perpendicularly through the center of the plot. The experiment was performed in October to coincide with the heavy production of fruit (acorns and hickory nuts) and maximal caching activity. In our 3 experimental plots, we provisioned squirrels with sunflower seeds and pecans ad libitum for 10 days prior to burying experimental caches. Provisioned food was presented in hanging feeders near the center of each plot. The other 3 plots served as controls and were not provisioned. We created 16 evenly distributed experimental caches per plot using the methods described for the preliminary experiment. Caches were buried following the procedure described for the unreplicated experiment (see above). However, in the replicated experiment, we buried caches 6 m apart, which represents typical inter-cache distances for squirrels in the wild (Stapanian and Smith 1978). Using the procedure described for the unreplicated experiment, we monitored caches daily for the first 10 days, then every 2–3 days until the termination of the experiment (35 days). 593 J.L. Penner, et al. 2013 Southeastern Naturalist Vol. 12, No. 3 Mann-Whitney U-tests were used to compare the number of days that caches survived at the 3 experimental plots and the 3 control plots. We also performed a survival analysis by combining the data across plots. All analyses were performed on IBM SPSS Version 20. Results In the unreplicated preliminary experiment, the provisioned site lost 80% (32 of 40) of the experimenter-made caches within 24 hours, whereas the unprovisioned site lost 0% (0 of 40) of the caches during the same time period. After one month, 15% of the caches remained in the provisioned site, whereas all the caches in the unprovisioned site were still present. Foraging by squirrels was observed regularly in both plots, although we have no quantitative measurements of squirrel activity or foraging. In the replicated experiment, a two-tailed, independent samples Mann- Whitney U test revealed that the median number of days a cache survived was significantly lower in the provisioned sites (med = 26.5) compared to unprovisioned sites (med = 34.8, U = 0.0, P = 0.050). Likewise, the rate of cache disappearance (number of caches lost per day) was significantly higher in provisioned sites (med = 0.2) compared to unprovisioned sites (med = 0.04, U = 0.0, P = 0.050). We performed a survival analysis by combining the data across plots to compare the survivorship of caches in provisioned versus unprovisioned plots; we found that caches declined at a significantly higher rate in the provisioned plots (c2 = 9.57, P = 0.002; Fig. 1). Discussion We found that caches disappeared more quickly in sites that had been consistently provisioned with pecans and sunflower seeds, compared to caches in unprovisioned sites, even though we commonly observed squirrels foraging at both types of sites. This study is the first to directly compare pilfering rates in provisioned versus unprovisioned sites for scatter-hoarding rodents. However, our findings are similar to the pilfering rates reported from a similar experiment with scatter-hoarding Willow Tits, who discovered caches more quickly (up to 15% cache loss per day) in the presence than in the absence of feeders (Brodin 1993). From our replicated experiment, we found that, on average, 33% (range: 25–43%) of the caches at our provisioned sites disappeared over the course of our experiment (35 days), whereas only 8.3% caches at unprovisioned sites experienced the same fate (range: 0–19%). The pilfering rate we report for our unprovisioned sites is similar to the “background” pilfering rate reported for Willow Tits at sites without feeders, where only 11% (11 of 100) of artificial caches disappeared (Brodin 1993). Apparently, the presence of an artificially enhanced food supply increases the foraging effort of squirrels in the area surrounding the feeder (resulting in a 30% cache loss), while the pilfering rate under more natural conditions is relatively low (about 10%). J.L. Penner, et al. 2013 Southeastern Naturalist Vol. 12, No. 3 594 The pilferage rates we report from our experimental (provisioned) sites are comparable to, and sometimes lower than, those reported for other smallmammal communities. For instance, Eastern Chipmunks discovered nearly half (45.7%) of experimental caches within hours after burial, a finding that the authors suggest could be at least partially attributable to the f act that the study site was located in a popular park where animals were reliably fed by people, which probably elevated population density (Clarke and Kramer 1994). Likewise, Dasyprocta leporina L. (Red-rumped Agoutis) found 80% of experimenter caches after 4 hours (Guimarães et al. 2005), a finding that may, at least in part, be attributable to the fact that the study site was located in a popular urban park with tame agoutis, although the agoutis were not provisioned directly by the researchers. Figure 1. Survival curves showing proportion of caches surviving over the course of the experiment between 3 provisioned and 3 unprovisioned plots in the replicated field experiment. 595 J.L. Penner, et al. 2013 Southeastern Naturalist Vol. 12, No. 3 Interestingly, studies from research sites with minimal provisioning, or utilizing more natural populations of animals who are unaccustomed to human provisioning, report lower pilfering rates. For instance, Dipodomys merriami Mearns (Merriam’s Kangaroo Rats), found only 13% of experimental caches within 24 hours of cache burial in unprovisioned sites (Leaver 2004). In a different study, kangaroo rats found 37% of experimental caches within 24 h at sites located several hundred meters away from the “principal” study site, wh ere animals had been provisioned with rolled oats (Daly et al. 1992). Likewise, studying Redrumped Agouti, Galveza et al. (2009) reported that experimental caches lasted between 4 and 35 days (which marked the end of the monitoring period). While it appears that the agoutis in this study were, at least in some sites, offered marked seeds to cache (for another part of the study), the number of seeds provided (25 at a time) was relatively low, and the type of seed was native to the study site. Artificial food sources may affect robbery rates in a variety of ways. For instance, the supplemental food in our study may have drawn higher than normal numbers of squirrels to the feeding sites, attracting more animals to raid our experimenter-made caches. This explanation is supported by reports of increased population densities in study sites where supplemental food is provided (Donald and Boutlin 2011, Klenner and Krebs 1991, Sullivan et al. 1983). Additionally, animals may be responding to the expectation that caches are buried in higher densities around a food source, thereby increasing their foraging activity (Stapanian and Smith 1984). Although our experimenter-made caches were buried at equal densities throughout both control and experimental plots, the supplemental food likely increased the cache density in our experimental plots because animals were observed caching the pecans they gathered from the feeder. Cache density is known to be negatively correlated with cache survivorship (Stapanian and Smith 1984) because smaller nearest-neighbor distances increase the success of area-restricted searching after the discovery of a single cache. Similarly, Tamura et al. (1999) reported that feeder-dispensed walnuts naturally buried close to the feeder were more likely to be robbed by conspecific Japanese Squirrels than were caches buried farther away. Foraging animals may use other visual cues as predictors of buried food. For instance, animals are known to target their foraging efforts near prominent objects such as trees (Red-rumped Agouti; Guimarães et al. 2005), and to associate experimenter-placed markers with buried food (Tamias amoenus Allen [Yellow-pine Chipmunks]; Vander Wall and Peterson 1996), so much so that some animals actively avoid such objects when burying their own food (Devenport et al. 2000, Penner and Devenport 2011). It is therefore reasonable to assume that squirrels in the current study learned that the presence of the supplemental feeder, and/or the heightened number of foraging conspecifics was associated with a valuable food supply, prompting them to increase their foraging activity and their pilfering behavior beyond that found at naturally occurring food bonanzas. Presumably, the 8.3% loss at our unprovisioned sites is representative of small-mammal pilfering rates under natural conditions at our site, yet this value J.L. Penner, et al. 2013 Southeastern Naturalist Vol. 12, No. 3 596 is lower than many pilfering rates reported for birds and mammals in the field under similar conditions and durations (e.g., 78% percent pilferage of Eastern Fox Squirrel caches buried 4.6 m apart; Stapanian and Smith 1978, reviewed in Vander Wall and Jenkins 2003). Thus, while we acknowledge that most researchers take precautions to avoid inflated, unnatural pilfering rates, we suggest here that more could be done to ensure that pilfering rates are representative of those occurring in the wild. Likewise, the discrepancy between ours and other studies may also be due in part to another important distinction—our squirrels were not tame, were unaccustomed to artificial food sources, were not habituated to human activity, and had not previously been used for research purposes. It is common for field researchers to use as subjects animals that are tame and frequently fed by people (e.g., Red-rumped Agoutis [Guimarães et al. 2005], Eastern Chipmunks [Clarke and Kramer 1994]) or animals observed at field sites that are part of ongoing studies (e.g., parks or college campuses); such animals may already associate human activity with unnatural concentrations of food, an association that could easily influence population density, foraging behavior, or both, during field studies. We provisioned with two food sources that were not naturally found at the field site (sunflower seeds and pecans), so it is possible that the novel and highly nutritious food caused the high rates of cache disappearance. Nonetheless, using non-native or processed foods is not uncommon in frequently cited investigations of caching-related behavior (e.g., Clarke and Kramer 1994), and it is possible that differences in provisioning procedures and type of supplemental food may explain some of the variability reported for caching and pilfering in the field. Our experiment, using feeders that provided food ad libitum, may represent the upper end of provisioning documented in published experiments. However, our experiment effectively demonstrates the potential effects of manipulating the quantity and quality of food available to animals. Thus, we suggest that native and seasonally appropriate food sources should be used when provisioning research sites. In addition, we encourage researchers to report the extent to which they attempt to match naturalistic foraging conditions (regarding both quantity and quality of food) in their experiments. At our research sites, the pilfering behavior of scatter-hoarding squirrels intensified when supplemental food was present. However, David and Boutin (2011) reported that the pilfering behavior of larder-hoarding red squirrels was largely unaffected by provisioning. Instead, American Red Squirrel pilfering behavior was more closely tied to the physical defense of individual lar ders; pilfering occurred more frequently at middens where the owner was absent (i.e., the owner died during the winter or was experimentally removed from the midden site) than where the owner was present. In contrast to larder-hoarding animals, whose food is concentrated in a single larder (or midden, for American Red Squirrels), scatter-hoarding animals cannot physically defend multiple scatter-cache sites that are, by their very definition, widely dispersed. Thus, the difference between our findings and those of David and Boutin (2011) suggest that the competitive 597 J.L. Penner, et al. 2013 Southeastern Naturalist Vol. 12, No. 3 dynamics of scatter-hoarding animals might make them especially vulnerable to the effects of supplemental food that were observed in the current st udy. In conclusion, we suggest that offering supplemental food at research sites can lead to pilfering rates that are not representative of pilfering behavior under more natural conditions. Others have identified methodologies aimed at more closely pinpointing the pilfering rates of natural populations by, for instance, taking into account the recovery advantage of cache owners when assessing cache survivorship (Devenport et al. 2000, Vander Wall et al. 2006). We suggest here that the type and quantity of food offered at research sites should also be taken into account in an effort to more accurately estimate pilferage rates under natural conditions, and we encourage researchers to employ, when possible, appropriate non-provisioned sites as controls in their experimental designs. Although we understand that some provisioning and some exposure to humans is a realistic expectation for many field sites, we simply recommend that researchers, in their dissemination efforts, attempt to include enough detail about site-provisioning, human-animal exposure, and population densities relative to non-research sites so that the ecological validity of their findings can be properly interpreted. We suggest that these factors may make the animals used in such studies qualitatively different than animals in more natural settings, and that these differences and their effects on foraging behavior warrant further investigation. Acknowledgments The Charles Brewer Endowment for Undergraduate Research in Psychology and the Hendrix College Odyssey Program for Experiential Learning, Conway, AR, provided funding for this research. Literature Cited Andersson, M., and J. Krebs. 1978. On the evolution of hoarding behaviour. Animal Behaviour 26:707–711. Bendel, P.R., and J.E. Gates. 1987. Home range and habitat partitioning of the Southern Flying Squirrel, Glaucomys volans. Journal of Mammalogy 68:243–255. Brodin, A. 1993. Low rate of loss of Willow Tit caches may increase adaptiveness of long-term hoarding. The Auk 10:642–645. Brodin, A. 2010. The history of scatter-hoarding studies. Philosophical Transactions of the Royal Society, Biology 365:869–881. Clarke, M.F., and D.L. Kramer. 1994. The placement, recovery, and loss of scatter-hoards by Eastern Chipmunks, Tamias striatus. Behavioral Ecology 5:353–361. Clarkson, K., S.F. Eden, W.J. Sutherland, and A.I. Houston. 1986. Density dependence and magpie food hoarding. Journal of Animal Ecology 55:111–121. Dally, J.M., N.S. Clayton, and N.J. Emery.2006. The behaviour and evolution of cache protection and pilferage. Animal Behaviour 72:13–23. Devenport, J.A., L.D. Luna, and L.D. Devenport. 2000. Placement, retrieval, and memory of caches by Thirteen-lined Ground Squirrels. Ethology 106:171–183. Donald, J.L., and S. Boutin. 2011. Intraspecific cache pilferage by larder-hoarding Red Squirrels (Tamiasciurus hudsonicus). Journal of Mammalogy 92:1013–1020. J.L. Penner, et al. 2013 Southeastern Naturalist Vol. 12, No. 3 598 Galveza, D., B. Kranstaubera, R.W. Kaysc, and P.A. Jansena. 2009. Scatter-hoarding by the Central American Agouti: A test of optimal cache spacing theory. Animal Behaviour 78: 1327–1333. Guimarães, P.R., B.Z. Gomes, Y.J. Ahn, and M. Galetti. 2005. Cache pilferage in Redrumped Agoutis (Dasyprocta leporine) (Rodentia). Mammalia 69:3–4. Koprowski, J.L. 1994a. Sciurus niger. Mammalian Species 479:1–9. Koprowski, J.L. 1994b. Sciurus carolinensis. Mammalian Species 480:1–9. Klenner, W., and C.J. Krebs. 1991. Red Squirrel population dynamics. I. The effect of supplemental food on demography. Journal of Animal Ecology 60:961–978. Kraus, B. 1983. A test of the optimal density model for seed scatter-hoarding. Ecology 64:608–610. Leaver, L.A. 2004. Effects of food value, predation risk, and pilferage on the caching decisions of Dipodomys merriami. Behavioral Ecology 15:729–734. Leaver, L.A., and M. Daly. 2001. Food caching and differential cache pilferage: A field study of coexistence of sympatric Kangaroo Rats and Pocket Mice. Oecologia 128:577–584. Leaver, L.A., L. Hopewell, C. Caldwell, and L. Mallarky. 2007. Audience effects on food caching in Grey Squirrels (Sciurus carolinensis): Evidence for pilferage avoidance strategies. Animal Cognition 10:23–27. Murray, A.L., A.M. Barber, S.H. Jenkins, and W.S. Longland. 2006. Competitive environment affects food-hoarding behavior of Merriam’s Kangaroo Rats (Dipodomys merriami). Journal of Mammalogy 87:571–578. Penner, J.L., and L.D. Devenport. 2011. A comparative study of caching and pilfering behavior in two sympatric species, Least Chipmunks (Tamias minimus) and Eastern Chipmunks (Tamias striatus). Journal of Comparative Psychology 125: 375–384. Smith, C.C., and O.J.Reichman. 1984. The evolution of food caching by birds and mammals. Annual Review of Ecology and Systematics 15:329–351. Smulders, T.V., K.L. Gould, and L.A. Leaver. 2010. Using ecology to guide the study of cognitive and neural mechansims of different aspects of spatial memory in food-hoarding animals. Philosophical Transactions of the Royal Society, Biology 365:883–900. Stapanian, M.A., and C.C. Smith. 1978. A model for seed scatter-hoarding: Coevolution of Fox Squirrels and Black Walnuts. Ecology 59:884–896. Stapanian, M.A., and C.C. Smith. 1984. Density-dependent survival of scatter-hoarded nuts: An experimental approach. Ecology 65:1387–1396. Sullivan, T.P., D.S. Sullivan, and C.J. Krebs. 1983. Demographic responses of a Chipmunk (Eutamius townsendii) population with supplemental food. Journal of Animal Ecology 52:743–755. Tamura, N., Y. Hashimoto, and F. Hayashi. 1999. Optimal distances for squirrels to transport and hoard walnuts. Animal Behavior 58:635–642. Vander Wall, S.B., and S.H. Jenkins. 2003. Reciprocal pilferage and the evolution of food-hoarding behavior. Behavioral Ecology 14:656–667. Vander Wall, S.B., and E. Peterson. 1996. Associative learning and the use of markers by Yellow Pine Chipmunks (Tamias amoenus). The Southwestern Naturalist 41:88–90. Vander Wall, S.B., J.S. Briggs, S.H. Jenkins, K.M. Kuhn, T.C. Thayer, and M.J. Beck. 2006. Do food- hoarding animals have a cache recovery advantage? Determining recovery of stored food. Animal Behaviour 72:189–197. van Horik, J., and K.C. Burns. 2007. Cache spacing patterns and reciprocal cache theft in New Zealand Robins. Animal Behavior 73:1043–1049.