Behavioral Ecology Advance Access originally published online on June 11, 2004
Behavioral Ecology 2004 15(5):729-734; doi:10.1093/beheco/arh070
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Effects of food value, predation risk, and pilferage on the caching decisions of Dipodomys merriami
School of Psychology, University of Exeter, Exeter EX4 4QG, UK
Address correspondence to L. A. Leaver. E-mail: l.a.leaver{at}exeter.ac.uk.
Received 23 January 2003; revised 20 October 2003; accepted 28 October 2003.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Animals that scatter cache their food face a trade-off between the benefits of protecting caches from pilferers and the costs associated with caching. Placing food into a large number of widely spaced caches helps to protect it from pilferage but also involves costs such as greater exposure to predators. I predicted that animals would disperse food into a larger number of more widely spaced caches when caching (1) a preferred food versus a less preferred food and (2) under conditions of low predation risk versus high predation risk. To test these predictions, I examined the scatter-caching decisions of Merriam's kangaroo rats (Dipodomys merriami). D. merriami distributed caches in clumped patterns, regardless of food preference, but they showed a tendency to invest more in a preferred food by distributing caches more widely. Under the relative safety of the new moon, they did not disperse caches more widely, rather they partitioned the same amount of food into a larger number of caches than they did under the full moon, when predation risk is higher. To examine whether their cache spacing decisions had a significant impact on the success of cache pilferers, I measured discovery by pilferers of artificial caches of two food types at different caching distances. Results indicate that the cache spacing behavior of D. merriami functions to protect caches from pilferers, because increased spacing of artificial caches decreased the probability of pilferage for both types of food.
Key words: caching, Dipodomys merriami, food preference, kangaroo rats, pilferage, predation risk.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
For animals that store food, decisions related to the protection of caches from pilferers can be crucial for surviving periods of scarcity. Many animals store their food in numerous small caches, a method of food storage called scatter caching, as opposed to larder caching, which entails storing repeated loads of food in a single place (Vander Wall, 1990
). Scatter caching presumably provides protection against major loss to pilferers, and wide dispersion of scattered caches further protects caches from pilferage by increasing the area a pilferer must search (Clarkson et al., 1986; Daly et al., 1992b; Sherry et al., 1982; Stapanian and Smith, 1978, 1984).
There are also costs involved in increased cache spacing. Memory demands are higher when an animal must retrieve a larger number of caches, and this cost is reflected in the relative increase in hippocampal size seen in animals that scatter cache compared with animals that do not scatter their caches (Jacobs, 1992; Jacobs and Spencer, 1994; Sherry et al., 1989, 1992). Predation risk also increases as a function of exposure during travel time (Daly et al., 1990), so there is a greater risk of detection by predators associated with greater cache dispersion. Animals that rely on food caching should have evolved to respond adaptively to the changing costs and benefits associated with food caching by modulating their caching decisions in accordance with such factors as the value of the food to be cached, the predation risk associated with caching and recovery, and the risk of cache pilferage.
Foraging theory predicts that animals will conduct a more intensive search (investing more effort in searching over the same area) in the surrounding area after encountering an item of food when foraging (Benhamou, 1992; Tinbergen et al., 1967), and obviously this also applies to animals that find an unfamiliar food cache when foraging. Furthermore, models of cache spacing predict that a "pilferer" will conduct a more intensive search for additional caches after encountering a cache containing a highly valuable food than if it encountered a less valuable food. As a result, animals must compensate by caching more valuable foods at lower densities (Stapanian and Smith, 1978, 1984). Food "value" is measured in various ways, for example, caloric value, nutritional content, or perishability, depending on the species in question. A number of studies show that scatter-caching rodents and birds disperse caches of more valuable food more widely than caches of less valuable foods (grey squirrels, Sciurus carolinensis, Hadj-Chikh et al., 1996; Steele et al., 1996; red squirrels, Tamiasciurus hudsonicus, Hurly and Robertson, 1987; willow and crested tits, Parus montanus and P. cristatus, Jokinen and Suhonen, 1995; heteromyid rodents, Leaver and Daly, 1998; Longland and Clements, 1995; fox squirrels, S. niger, Stapanian and Smith, 1984; Japanese squirrels, S. lis, Tamura et al., 1999; yellow pine chipmunks, Tamias amoenus, Vander Wall, 1995). However, there are very few direct studies of the effect this differential spacing has on pilferage success. Fox squirrels (S. niger) conduct more extensive (covering a greater area) searches in patches containing artificial caches of more valuable food (in terms of calories; Stapanian and Smith, 1984), whereas heteromyid rodents do not search more extensively after finding a large sized cache than they do after finding a smaller one containing the same type of food (Daly et al., 1992b). Heteromyids have cheek pouches that allow them to transport multiple seeds, so when gathering seeds from a single source they can readily vary cache size but not seed value. As a result, the type of seeds in a cache, rather than the size of the cache itself, is likely to be a better predictor of the value of neighboring caches. Although there is conflicting and limited evidence for search strategies used in relation to cache value, there is evidence from studies of a number of different species that increased spacing of artificial caches buried by researchers better protects them from both conspecific and heterospecific pilferage (magpies, Pica pica, Clarkson et al., 1986; kangaroo rats, D. merriami, Daly et al., 1992b; marsh tits, P. palustris, Sherry et al., 1982; fox squirrels, S. niger, Stapanian and Smith, 1978, 1984; Japanese squirrels, S. lis, Tamura et al., 1999).
I conducted a series of field studies to examine the determinants of food-caching decisions in greater detail. In the first study, I tested two predictions that had been supported in a previous laboratory study (Leaver and Daly, 1998) in an attempt to replicate those results in a more ecologically valid setting. These predictions were (1) a preferred food will be partitioned into a larger number of caches, and (2) these caches will be more widely spaced to better protect preferred food from pilferage. In the second study, I tested the prediction that area-restricted search by pilferers will be more extensive if they discover caches containing more highly preferred seeds. Such behavior would be beneficial to a pilferer if seed storers scatter cache a preferred food more widely. In the third study, I examined the relationship between food caching and predation risk by comparing caches made under the relatively darker new moon versus under the full moon. I predicted that caches made under the relative safety of the new moon would be more numerous and more widely spaced.
Merriam's kangaroo rats (D. merriami) are an ideal seed-caching species for testing these predictions. They are primarily granivorous and, similar to many other desert-dwelling heteromyid rodents, they prefer seeds that are relatively high in carbohydrate and low in protein, rather than showing a preference for high-calorie seeds (Frank, 1988; Kelrick et al., 1986; Lockard and Lockard, 1971; Price, 1983). For these desert rodents, food value seems to be closely associated with water conservation. Carbohydrates produce a small amount of water during metabolism (Schmidt-Nielsen, 1964), whereas protein ingestion results in water loss (Frank, 1988). In the laboratory, D. merriami preferentially consume (Price, 1983) and scatter cache a high-carbohydrate, low-protein food, placing seeds with these properties into a larger number of smaller-sized, more widely spaced caches (Leaver and Daly, 1998). Presumably they do this to better protect more valuable caches from pilferage.To test the prediction that a preferred food will be placed into a larger number of more widely spaced caches in the field, I provisioned them with two foods that differed in their ratio of carbohydrate: protein.
Kangaroo rats are sensitive to moonlight as a cue of predation risk and they adjust their activity patterns in response to moonlight while active above ground (Daly et al., 1992a; Price et al., 1984) and during foraging (Bouskila, 1995; Bowers, 1988, 1990; Kotler, 1984; Price, 1978). Predation risk from visually hunting predators is higher under bright moonlight and, as artificial or natural night-time illumination increases, heteromyid activities shift to the relative safety of a microhabitat under the cover of shrubs presumably in order to minimize risk from predators (for reviews, see Munger et al., 1983; Price and Brown, 1983). Accordingly, I studied the effects of moon phase on the caching behavior of Merriam's kangaroo rats to investigate the effects of changes in predation risk on caching decisions.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Field site
The study took place at the University of California's Phillip L. Boyd Deep Canyon Desert Research Center in Palm Desert California, USA. The scrub brush vegetation of the alluvial plain has been described in detail by Zabriskie (1979)
.
Study 1: cache partitioning and spacing in the field
I provisioned D. merriami with seeds dusted in fluorescent powder and used portable ultraviolet lights to locate their caches (protocol from Longland and Clements, 1995). For 17 nights from 4 November 4–5 December 1997, I placed a total of 44 provisioning trays at separate locations in the field. In this study I avoided new and full moon nights, because the extreme lighting conditions in these moon phases cause the animals to alter their foraging and activity patterns from that of the other nights (Daly et al., 1992a; Price et al., 1984).
Each provisioning tray consisted of a 30.6-cm round aluminium tray, the bottom of which was covered with sandpaper and dusted with fluorescent powder (Radiant Color). In the center of each tray, I fastened a 7-cm round Petri dish containing 40 g of whole oats (22 trays) or 40 g of whole lentils (22 trays) that had been mixed with 3 g of green, blue, or pink fluorescent powder. D. merriami prefer oats over lentils (Leaver, 2000). Oats are higher in carbohydrate and lower in protein than are lentils (oats: 68.3% carbohydrate, 13.3% protein, Arrowhead Mills nutrition information; lentils: 61.7% carbohydrate, 24.1% protein, Price, 1983). All seeds in this, and subsequent, studies were sterilized by microwaves to prevent germination.
At sunset (approximately 1700 h) on each of the 17 nights, I placed two to six trays at novel locations on the alluvial plain of Deep Canyon, spaced at least 30 m apart. I left the trays until 0200–0300 h, at which time I returned to each tray and checked whether or not it had been discovered by rodents. At each of the trays that had been discovered by rodents, I used a portable ultraviolet light to follow any trails left by the animal. I also conducted a systematic search around each tray radiating outward in concentric circles at 5- m intervals, covering approximately 70 m in diameter around each tray. By doing so, I was able to find caches that had no trail leading to them. It was not possible to search a wider area thoroughly before dawn. Scatter caches were identifiable by conspicuous disturbance and signs of digging by the caching animal, which left a distinctive sweep of fluorescent powder that was probably not visible to nocturnal pilferers.
I marked trails that led to burrow entrances, caches, and stray seeds with signs of powder around them with numbered pin flags. I removed the contents of any caches (and replenished them with a comparable amount of rolled oats) and placed them in labeled zip-lock bags for later weighing. I also collected and weighed any stray seeds. Caches were defined as consisting of two or more buried seeds. Single seeds were often scattered along trails as if accidentally dropped, and these were not counted as caches.
On the following day, I measured the straight-line nearest-neighbor distances for each cache, with the exception of two cache locations at an oat tray and one cache location at a lentil tray because the pin flags had been blown away in strong winds.
I identified the genus of visitor(s) to the tray by the footprints left around the tray by animals that had walked across fluorescent powder. To confirm that the caches at each tray were made by a single individual, and also to verify the species identity of the caching animals in addition to the footprints, I used dyed seeds at some of the provisioning trays and subsequently set traps locally and examined the trapped animal's feces for dye. I dyed oats at the provisioning trays with a solution of either Fast Green or Eosin Y (Fisher Scientific). Neither of these dyes affects the palatability of seeds for heteromyids (Daly et al., 1992b). It was not possible to dye the lentils because they did not "take" the dye.
At dusk on the night after discovery of an oat tray by a rodent, I set up a five-by-six grid of 30 Sherman live traps baited with rolled oats and set at 5-m intervals centered on the provisioning site. If trapping was not successful on the first night, I reset the traps at the same location the following night. I checked traps between 0000 h and 0200 h. I recorded the weight, sex, and reproductive condition of all trapped animals; checked them for traces of fluorescent powder and dye; and collected one to two fecal boluses from each subject before release. Animals that did not immediately produce a fecal bolus were left in clean traps containing rolled oats for no more than 20 min in order to collect their feces. Seven animals did not produce fecal samples, four Chaetodipus formosus and three D. merriami, but two of each species had traces of dye and fluorescent powder around their mouths, inside their cheek pouches, and around their anuses, so they were assumed responsible for depleting the respective dish.
I placed fecal samples inside small clear plastic vials and transported them to the laboratory, where a few drops of tap water were added to each. Subjective blind judgments were made more than 10 h after the addition of water as to the presence or absence of dye. Twenty-seven fecal samples were collected from trapped animals, and these were rated by four independent raters. Raters disagreed in three cases. Trapping records indicated that in two of these three cases, the animal had traces of fluorescent powder in its cheek pouches and thus was likely to have been responsible for depleting the seeds at the provisioning tray. The third animal had no trace of powder on its body and thus was not considered to be responsible for depleting the tray in the vicinity in which it was trapped. Those animals that had traces of fluorescent powder on their bodies and/or who were unanimously rated as having dye present in their feces were considered to be the animals responsible for caching around their respective trays. I also trapped at two lentil tray locations to ensure that the same species were present at both oat and lentil tray locations.
Clark and Evans' (1954) nearest-neighbor distance method was used to calculate dispersion, and Z tests were used to determine dispersion patterns (clumped, random, or uniform). A two-tailed Mann-Whitney U test was used to compare differences between dispersion of oats and lentils. Two-tailed Mann-Whitney U tests were used to compare the mean number of scatter and larder caches, the mean weight and nearest-neighbor distance of scattered caches, and amount of seeds recovered at trays of oats versus lentils.
Study 2: pilferage from artificial caches
To quantify the rate of discovery of caches of oats versus lentils, and to see whether animals conducted more extensive or effective area-localized search after discovering a cache of oats than lentils, I buried 603 triads of artificial caches and checked them for discovery after 23–25 h (protocol from Daly et al. 1992b). Caches were buried just outside of Boyd Deep Canyon Desert Research Center during three field periods, 23 January 23–2March 1997 (208 triads), 11 October–16 November 1997 (175 triads), and 14 October–30 November 1998 (220 triads). Triads of caches were spaced at least 70 m apart to reduce the likelihood that two triads would be discovered by the same rodent. Each triad consisted of either three caches of oats or three caches of lentils buried at each point of an equilateral triangle, measuring either 0.5 m or 2 m on a side. These are cache distances that have been found to differentially affect discovery by heteromyid pilferers (Daly et al., 1992b). Each cache contained eight seeds buried 1 cm deep in fine sand, extracted from the surrounding substrate, in plastic cups measuring 4.5 cm in diameter and 4 cm deep. The cups were themselves buried in the substrate so that their rims were flush with the surface. Locations were marked by yellow or orange tape tied to nearby shrubs or trees (1–3 m away), and exact cache locations were noted with reference to shrubs and compass directions.
After 23–25 h, I checked artificial caches for discovery. At each triad, I recorded any sign of disturbance, the number of seeds left in each cache, and the number of caches disturbed in each triad. Any triad with a cache that had been noticeably disturbed and/or in which there were seeds missing was considered to be "discovered" by a pilferer. I removed triads in which one or more caches had been discovered, and left triads when none of the three caches had been discovered for one more period of 23–25 h. After removal of artificial caches, the same area was not reused for a subsequent group of triads.
I censused the nocturnal population by trapping for five nights in 1997 at five random sites over the entire area where artificial caches had been made in order to determine the species of potential pilferers. Over five nonconsecutive nights, 420 traps were set and left until dawn. Of the 59 animals caught, the majority were heteromyid rodents; 31 were D. merriami (53%);22 were pocket mice (37%), either Chaetodipus formosus or C. penicillatus; and just six were cricetid rodents, Peromyscus eremicus (10%). Potential diurnal pilferers included ground squirrels, birds, and insects.
To compare proportions of triads when more than one cache was pilfered, given that one cache was pilfered, Zar's (1999) test for comparing two proportions was used to compare pilferage at the two spacings (0.5 and 2 m) and for the two seed types (oats and lentils). Zar's (1999) test for comparing multiple proportions was used to compare pilferage rates between the three field seasons.
Study 3: the effects of moon phase on caching by kangaroo rats
The study took place from 30 April–13 June 1999. Four species of nocturnal rodents were present on the field site during the course of the study, four heteromyids—Dipodomys merriami, Perognathus fallax, P. formosus, and P. penicillatus—and one cricetid, Peromyscus eremicus.
Each of five radio-collared kangaroo rats (MD-2C Holohil Systems) was provisioned with 7 g of millet, a highly preferred food (Price, 1983), once under a new moon and once under a full moon, on cloudless nights. The order of presentation was random, and animals were provisioned opportunistically. For each individual animal, provisioning trays were placed at the same location in each moon phase. Any other rodent attempting to enter the provisioning tray was prevented from doing so by the researcher.
Seed trays were the same as those used in experiment 1. Seven grams of millet seeds were mixed with fluorescent powder and placed in a dish in the center of the tray. Immediately after the seed dish was emptied by the focal kangaroo rat, or 20 min had passed since its last visit, the tray was removed and I searched a 30-m radius around the dish for caches, using portable fluorescent lights. All cache sites were marked in the same way as in experiment 1, with the exception that the cache itself was not disturbed to prevent the rodents from changing their caching behavior over the course of the experiment owing to experimenter pilferage. I recorded the number of caches made by each animal, the length of time that the animal spent emptying the seed tray, and the dispersion of the caches.
To measure the distribution pattern of caches (clumped, random, or uniform), Clark and Evans (1954) nearest-neighbor distance method was used to calculate R for each rodent under both the new and the full moon. For each moon phase, the weighted mean of R (R values weighted by the number of observations contributing to each) was calculated, and pooled Z tests were used to determine dispersion patterns (clumped, random or uniform). A two-tailed Wilcoxon signed-rank test was used to compare differences between distribution under the new and full moon.
Because this study was carried out in an area with grid markers every 10 m at Cartesian coordinates, it was possible to map cache locations and use a more sensitive measure of cache dispersion than nearest-neighbor distance. Dispersion was analyzed by using a dispersion index (DI), which was calculated by measuring the average straight-line distance from any individual cache to any other individual caches made by one animal in one trial. The resulting DI is an average of the distance between any two pairs of caches for a given trial. Dispersion data were analyzed using two-tailed Wilcoxon signed-rank tests.
Data are reported as mean plus or minus standard deviation.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Study 1: caching behavior
Of the 44 provisioning trays, 26 (16 oat trays, 10 lentil trays) were discovered by rodents. In the 70 m2 that I searched around each of these 26 trays, I found a total of 171 scattered caches (71 oat, 100 lentil) and 41 larder caches (36 oat, five lentil). In the area searched around three tray locations, there was only one scattered cache (two oats, one lentil). There were two or more scattered caches in the area that I searched around 13 (six oats, 6/16 = 37.5%; seven lentils, 7/10 = 70%; Fisher's Exact test p =.226) of these 26 trays. Signs of larder caching (traces of powder around burrow entrances) were present in the area searched around 14 trays (eight of the 13 trays with two or more scattered caches, two of the three trays with one scattered cache, and four seed trays with no scattered caches). Larder caches were more often found in the 70-m2 area searched around trays containing oats (11/ 16 trays discovered by rodents) than lentils (3/10 trays), but this difference was not significant (Fisher's Exact test p =.105).
Contrary to the prediction that rodents would partition a more valuable food into a greater number of scattered caches, the rodents made significantly more scattered caches of lentils than of oats, and significantly more larder caches of oats than of lentils (Table 1). There was no significant weight difference between caches of lentils and oats (Table 1).
|
I recovered significantly less of the total provisioned 40 g of oats per tray than lentils (Table 1). Kangaroo rats eat approximately 3g of food/night, so consumption cannot account for the unrecovered amount. It is possible that the excess food was larder cached in burrows, because there were more larder caches in the area around trays containing oats than lentils, and less oats were recovered in total.
Caches of both oats (R =.73, Z = 4.18, p <.05) and lentils (R = 0.40, Z = 11.36, p >.05) were significantly clumped, rather than distributed randomly or uniformly. There was no significant difference between the dispersion of oats and lentils (oats N = 6, lentils N = 7, U = 11, p = ns). Although the mean nearest-neighbor distance between caches was greater for oats than for lentils, as predicted, this difference was not significant (Table 1).
Identification of cachers
I set a total of 390 traps around 14 of the 16 oat tray locations that had been discovered by rodents. Inclement weather prohibited trapping at the two remaining tray locations. At one of the 14 trapping locations, no animals entered any of the traps. Out of 35 animals trapped, traces of fluorescent powder and/or fecal dye were present for 9/15 D. merriami, 5/11 C. formosus, 2/8 C. penicillatus, and 0/1 C. fallax.
At six of the seven oat trays at which D. merriami were trapped, only one individual had traces of powder on their bodies and/or traces of dye in their feces. At the remaining tray, two kangaroo rats had traces of powder on their bodies.
I set 30 traps around each of two lentil tray locations and trapped a total of 15 animals. Traces of powder were present on 2/10 D. merriami, 0/4 C. formosus, and 0/1 C. penicillatus. At one tray location, there were footprints made by kangaroo rats, but none of the four kangaroo rats trapped at the site had traces of powder on their bodies. At the other tray location, two of six trapped kangaroo rats had traces of powder on their bodies.
At the 11 trays from which I had both footprint identification and trapping data (nine oat trays, two lentil trays), the species that was identified by its fluorescent tracks was also trapped and, at nine of these locations (eight oat trays, one lentil tray), had traces of dye or powder. At the remaining two tray locations, two species had been identified by footprints, but only one had traces of dye or powder when trapped. In future studies, use of fluorescent powder to identify the caching animal by footprints seems sufficient.
I set traps at 11 (nine oat trays, two lentil trays) of the 13 tray locations where I had found more than two scattered caches, and kangaroo rats were present at all of the tray locations. Therefore, I attributed all scattered caches to kangaroo rats. Previous direct observations of habituated kangaroo rats and pocket mice collecting and hoarding provisioned food at Deep Canyon suggest that kangaroo rats scatter cache, whereas pocket mice seldom do so (Leaver and Daly, 2001).
Study 2: discovery of artificial caches
Eighty of the 603 triads were discovered, so the discovery or pilferage rate was 13.3%. Pilferage rate differed between field seasons: 32/208 (15.4%) in winter 1997, 13/175 (7.4%) in autumn 1997, and 35/220 (15.9%) in autumn 1998 (
2 = 7.30, v = 2, p <.05). Increased spacing reduced the probability that more than one cache in a triad would be pilfered if at least one was, from 74% at 0.5 m to 45% at 2 m (Z2 = 2.35, p <.01, one-tailed) (Table 2). Contrary to prediction, there was no effect of seed type on the probability that further caches in a triad would be discovered, given that one was found (Z2 = 0.19, p = ns, one-tailed) (Table 2).
|
Study 3: caching and moon phase
As predicted, the kangaroo rats made significantly more caches under the relative safety of the new moon (12.2 ± 5.0) than under the light of the full moon (5.2 ± 6.9; Z = –2.03, p <.05). As an indication of the effort involved in making a larger number of caches, it took the rodents significantly more time to empty the seed trays under the new moon (41.2 ± 28.5 min) than under the full moon (16.2 ± 18.4 min; Z = –2.02, p <.05).
Caches were clumped under both the full moon (mean density/m2 = 0.002, weighted mean R =.61, pooled Z = 3.33, p <.05) and the new moon (mean density/ m2 = 0.003, weighted mean R =.47, pooled Z = 5.68, p <.05). There was no difference in distribution patterns made under the full moon (mean R =.41 ± 0.47) and under the new moon (mean R = 0.40 ± 0.25; Z = –0.405, p = ns). Although caches were more dispersed under the new moon (DI = 13.1 ± 3.7 m) than under the full moon (DI = 7.4 ± 7.1 m) the difference was not significant (Z = –1.21, p = ns).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
The results of these studies provide some evidence that kangaroo rats adaptively adjust their caching behavior in a manner that protects more valuable food from pilferage, and that they are sensitive to changing trade-offs between predation and pilferage risk while caching. Contrary to prediction, the kangaroo rats made more larder caches of the preferred food, oats, and more scattered caches of the less preferred food, lentils. Leaver and Daly (1998)
found that in the laboratory D. merriami made more scattered caches of a preferred food, but the rodents in that study were not given the opportunity to larder hoard, which may account for the different results between the two studies. Further investigation of the relative benefits of scatter versus larder caching in D. merriami is warranted to investigate this phenomenon, but the conflicting results highlight the need for ecological validity in artificial laboratory environments.
Every possible effort was made to ensure that only one animal was caching around each food tray. Data collected from intensive trapping, use of fluorescent powder, and dyed seeds provide evidence that caches around the trays were made by a single kangaroo rat. However, it is not possible to be 100% sure that the caches at each tray were made by a single individual, so caution must be taken when interpreting the data. An observational study showing that individual kangaroo rats were responsible for emptying the majority of food trays containing oats and lentils also provides some indirect evidence to support the assumption of independence (Leaver and Daly, 2001). If it is the case that there were multiple harvesters at lentil trays more often than at oat trays, then it can provide an alternative explanation for the shorter nearest-neighbor distances of lentil caches without invoking the hypothesis that a preferred food is cached more widely to protect it from pilferage. Future investigations are needed to test the assumption that a single kangaroo rat made the caches around each tray.
When caching under the new moon, when predation risk was low, the kangaroo rats took more time to distribute seeds into caches than they did under the full moon. They also made a greater number of caches under the relatively safe darkness of the new moon. Because more caches were made under new moon conditions than under full moon conditions, one might expect cache dispersion to be less; the fact that spacing was relatively greater suggests that animals were not only making more, presumably smaller, caches under new moon conditions but also spacing them more widely.
To avoid disrupting the animals' subsequent caching behavior, I did not measure the size of the caches, but because the caches made under the new moon were more numerous, they must also have been smaller in size than the ones made under the full moon because they cached the same amount of provisioned food in each moon phase. They seem to have made a greater investment in protecting caches against pilferage when predation risk was relatively low.
Wider cache dispersion almost surely entails a greater investment of time, energy, memory, and exposure to predators. The results of the pilferage study reinforce the theory that increased spacing functions to protect neighboring caches from pilferage. It is clear that cache dispersion is relevant to the issue of protecting a preferred food from pilferers but that kangaroo rats respond adaptively to changes in the associated costs; they place food into a smaller number of caches when predation risk is high, and they show a tendency to increase cache spacing when pilferage risk is high.
The spacing of artificial caches in the present study influenced their vulnerability to pilferage. Wider dispersion of artificial caches provided better protection from discovery, as has been found in previous studies (see Introduction). However, contrary to the predictions made by various models of caching (Clarkson et al., 1986; Hurly and Robertson, 1987; Stapanian and Smith, 1978, 1984), pilferers showed no evidence of searching more widely for neighboring caches upon discovering a cache containing a preferred seed type. The more widely dispersed caches of oats made by D. merriami suggest that caching strategies eliminate any advantage that would accrue from differentially extensive area-localized search (assuming that the pilferer is "aware" that the seeds they have discovered constitute a cache rather than seed rain). The conditional probability of finding additional caches in a triad after one was discovered was similar for oats (0.61) and lentils (0.56), so it is clear that the type of seed in the cache did not affect probability of discovery. In a study carried out by Daly et al. (1992b), the number of seeds per cache was similarly irrelevant to the probability of discovery. Stapanian and Smith (1984) found that fox squirrels conduct more extensive area-localized search for neighboring caches when they find a cache of higher caloric value than when they find a cache containing a less valuable nut. However, Merriam's kangaroo rats do not seem to adjust the intensity of their searches based on the value, in terms of calories or nutritional characteristics, of the caches that they encounter.
Kangaroo rats are capable of adjusting the size of their caches by limiting the number of seeds that they release from their cheek pouches, whereas squirrels tend to carry and cache one nut per caching trip. It is possible that kangaroo rats adjust the value of each cache that they make to be relatively equal to any other cache in relation to a number of variables associated with food value, pilferage risk and predation risk. If this is the case, all caches will be similar in value, and it would not be adaptive for conspecifics to adjust their search intensity according to cache value. This may also help to explain why kangaroo rats, unlike fox squirrels, do not appear to have flexible search strategies.
If kangaroo rats obtain a high proportion of their food through pilferage, then it is possible that they have evolved the ability to recognize caches and adjust their search strategies accordingly. Scrub jays (Aphelocoma coerulescens) adjust their caching strategies depending on whether or not they have experience as a pilferer (Emery and Clayton, 2001), so it is possible that caching animals are capable of recognizing the act of pilferage as something different from primary foraging.
Discovery rates of artificial caches were significantly lower in autumn 1997 than in either winter 1997 or autumn 1998. Similar seasonal differences in the discovery rates of artificial caches were related to food availability in fox squirrels (Stapanian and Smith, 1984), and in the present study there was an abundance of food in autumn 1997 owing to increased rainfall in the summer months. The kangaroo rats were also slightly heavier than average (39 g in autumn 1997 versus 36 g in winter 1997 and 34 g in autumn 1998). Thus, artificial caches were less likely to be discovered when food was relatively abundant, presumably because more food (natural and provided) is left unharvested.
|
ACKNOWLEDGEMENTS |
|---|
I am grateful for the tireless field assistance of Magdalena Sztejnmec, Susan Robinson, and Christopher Evans; the technical assistance of Al Muth, Mark Fisher, and Dennis Hebert at the Phillip L. Boyd Deep Canyon Desert Research Center; and the helpful advice and assistance of Martin Daly, Margo Wilson, and Stephen Lea. This work formed part of an Ontario Graduate Scholarship and Natural Sciences and Engineering Research Council of Canada (NSERCC) funded PhD studentship completed by L.L. and was additionally funded by a NSERCC research grant awarded to Martin Daly. Trapping was conducted under scientific collecting permits issued to L.L. by the State of California Department of Fish and Game.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Benhamou S, 1992. Efficiency of area-concentrated searching behaviour in a continuous patchy environment. J Theor Biol 159:67-81.[CrossRef]
Bouskila A, 1995. Interactions between predation risk and competition: a field study of kangaroo rats and snakes. Ecology 76:165-178.[CrossRef]
Bowers MA, 1988. Seed removal experiments on desert rodents: the microhabitat by moonlight effect. J Mammal 69:201-204.[CrossRef]
Bowers MA, 1990. Exploitation of seed aggregates by Merriam's kangaroo rat: harvesting rates and predatory risk. Ecology 71:2334-2344.[CrossRef]
Clark P, Evans F, 1954. Distance to nearest neighbour as a measure of spatial relationships in populations. Ecology 35:445-453.[CrossRef]
Clarkson K, Eden SF, Sutherland WJ, Houston AI, 1986. Density dependence and magpie food hoarding. J Anim Ecol 55:111-121.
Daly M, Behrends PR, Wilson MI, Jacobs LF, 1992a. Behavioural modulation of predation risk: moonlight avoidance and crepuscular compensation in a nocturnal desert rodent, Dipodomys merriami. Anim Behav 44:1-9.
Daly M, Jacobs LF, Wilson MI, Behrends PR, 1992b. Scatter hoarding by kangaroo rats (Dipodomys merriami) and pilferage from their caches. Behav Ecol 3:102-111.
Daly M, Wilson M, Behrends PR, Jacobs LF, 1990. Characteristics of kangaroo rats, Dipodomys merriami, associated with differential predation risk. Anim Behav 40:380-389.[CrossRef]
Emery NJ, Clayton NS, 2001. Effects of experience and social context on prospective caching strategies by scrub jays. Nature 414:442-446.
Frank CL, 1988. Diet selection by a heteromyid rodent: role of net metabolic water production. Ecology 69:1943-1951.[CrossRef]
Hadj-Chikh LZ, Steele MA, Smallwood PD, 1996. Caching decisions by grey squirrels: a test of the handling time and perishability hypotheses. Anim Behav 52:941-948.[CrossRef]
Hurly AT, Robertson RJ, 1987. Scatterhoarding by territorial red squirrels: a test of the optimal density model. Can J Zool 65:1247-1252.
Jacobs LF, 1992. Memory for cache locations in Merriam's kangaroo rats. Anim Behav 43:585-593.[CrossRef]
Jacobs LF, Spencer WD, 1994. Natural space-use patterns and hippocampal size in kangaroo rats. Brain Behav Evol 44:125-132.[Web of Science][Medline]
Jokinen S, Suhonen J, 1995. Food catching by willow and crested tits: a test of scatterholding models. Ecology 76:892-898.[CrossRef]
Kelrick MI, Macmahon JA, Parmenter RR, Sisson DV, 1986. Native seed preferences of shrub-steppe rodents, birds and ants: the relationship of seed attributes and seed use. Oecologia 68:327-337.[CrossRef]
Kotler BP, 1984. Risk of predation and the structure of desert rodent communities. Ecology 65:689-701.[CrossRef]
Leaver LA, 2000. Ecological determinants of foraging and caching behaviour in sympatric Heteromyid rodents (PhD dissertation). Hamilton, Canada: McMaster University.
Leaver L, Daly M, 1998. Effects of food preference on scatter-hoarding by kangaroo rats (Dipodomys merriami). Behaviour 135:823-832.
Leaver L, Daly M, 2001. Food caching and differential cache pilferage: a field study of co-existence of sympatric kangaroo rats and pocket mice. Oecologia 128:577-584.[CrossRef]
Lockard RB, Lockard JS, 1971. Seed preference and buried seed retrieval of Dipodomys deserti. J Mammal 52:219-221.[CrossRef][Web of Science][Medline]
Longland WS, Clements C, 1995. Use of fluorescent pigments in studies of seed caching by rodents. J Mammal 76:1260-1266.[CrossRef]
Munger JC, Bowers MA, Jones WT, 1983. Desert rodent populations: factors affecting abundance, distribution, and genetic structure. Great Basin Nat Mem 7:91-116.
Price MV, 1978. The role of microhabitat in structuring desert rodent communities. Ecology 59:910-921.[CrossRef]
Price MV, 1983. Laboratory studies of seed size and seed species selection by heteromyid rodents. Oecologia 60:259-263.[CrossRef]
Price MV, Brown JH, 1983. Patterns of morphology and resource use in North American desert rodent communities. Great Basin Nat Mem 7:117-134.
Price MV, Waser NM, Bass TA, 1984. Effects of moonlight on microhabitat use by desert rodents. J Mammal 65:353-356.[CrossRef]
Schmidt-Nielsen K, 1964. Desert animals: physiological problems of heat and water. Oxford: Clarendon Press.
Sherry D, Avery M, Stevens A, 1982. The spacing of stored food by marsh tits. Z Tierpsychol 58:153-162.
Sherry DF, Jacobs LF, Gaulin SJ, 1992. Spatial memory and adaptive specialization of the hippocampus. Trends Neurosci 15:298-303.[CrossRef][Web of Science][Medline]
Sherry DF, Vaccarino AL, Buckenham K, Herz RS, 1989. The hippocampal complex of food storing birds. Brain Behav Evol 34:308-317.[Web of Science][Medline]
Stapanian MA, Smith CC, 1978. A model for seed scatterhoarding: coevolution of fox squirrels and black walnuts. Ecology 59:884-896.[CrossRef]
Stapanian MA, Smith CC, 1984. Density-dependent survival of scatterhoarded nuts: an experimental approach. Ecology 65:1387-1396.[CrossRef]
Steele MA, Hadj-Chikh LZ, Hazeltine J, 1996. Caching and feeding decisions by Sciurus carolinensis: responses to weevil infested acorns. J Mammal 77:305-314.[CrossRef]
Tamura N, Hashimoto Y, Hayashi F, 1999. Optimal distances for squirrels to transport and hoard walnuts. Anim Behav 58:635-642.[CrossRef][Web of Science][Medline]
Tinbergen N, Impekoven M, Franck D, 1967. An experiment on spacing-out as a defence against predation. Behaviour 28:307-321.
Vander Wall SB, 1990. Food hoarding in animals. Chicago: University of Chicago Press.
Vander Wall SB, 1995. The effects of seed value on the caching behaviour of yellow pine chipmunks. Oikos 74:533-537.[CrossRef]
Zabriskie JG, 1979. Plants of deep canyon. Riverside, California: University of California Press.
Zar JH, 1999. Biostatistical analysis. Englewood Cliffs, New Jersey: Prentice Hall.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  L. J. Hopewell, L. A. Leaver, and S. E.G. Lea Effects of competition and food availability on travel time in scatter-hoarding gray squirrels (Sciurus carolinensis) Behav. Ecol., November 1, 2008; 19(6): 1143 - 1149. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||