- Top of page
Seeds of about 85 of the 110 species of pines are dispersed by wind. These pines have small- to medium-sized seeds with a relatively large wing, and cones that open at maturity or after a fire and shed seeds. The wing causes the seed to autorotate as it falls, slowing its descent and providing opportunities for the seed to be moved laterally by wind, away from the parent plant (Greene & Johnson 1989). A host of animals consume pine seeds on the ground, but some seeds become buried in the soil or plant litter by physical processes.
The relative importance of corvids (e.g. jays and nutcrackers) vs. rodents (e.g. chipmunks and mice) as dispersers of pine seeds is somewhat controversial. When it was first realized that some wingless-seeded pines were dispersed by animals, they were called ‘bird-dispersed’ pines (e.g. Tomback & Linhart 1990; Benkman 1995). The adaptive morphology of the cones of these pines seemed to encourage seed harvesting by birds (Vander Wall & Balda 1977). Rodents were assumed to act exclusively as seed predators (e.g. Hutchins & Lanner 1982; Christensen & Whitham 1993). More recently, we have found that rodents can be effective scatter-hoarders of these pines and in some cases may be more effective dispersers than the corvids (Vander Wall 1992, 1997; Borchert et al. 2003; Hollander & Vander Wall 2004). However, the relative importance of corvids and rodents in dispersing pine seeds is still uncertain because no studies of the effects of a rodent and a corvid have been conducted simultaneously on the same pine.
Pilfering of caches can have both direct and indirect effects. When rodents pilfer from jays (or vice versa), they can have a potentially significant direct effect on the food stores of the jays and, like other forms of interspecific competition, pilferage could reduce the jay's reproductive fitness. If the caching behaviour of the two species is different, then pilfering stored seeds and recaching them elsewhere, which is a common phenomenon (Vander Wall & Joyner 1998; Vander Wall 2002; Vander Wall & Jenkins 2003), could effect pine fitness. Cache characteristics (e.g. depth, size, substrate, microhabitat) vary among species (e.g. Leaver & Daly 2001; Hollander & Vander Wall 2004), and these differences are likely to influence the probability of seed germination and seedling establishment (Shainsky & Radosevich 1986; Drivas & Everett 1988; Callaway et al. 1996). If a species that stores seeds in situations that are generally favourable for seedling establishment (i.e. a good disperser) pilfers from a species that stores seeds in situations that are generally unfavourable for seedling establishment (i.e. a poor disperser), then the act of pilfering and recaching seeds could have a positive indirect effect on the fitness of the pine. These types of species interactions and the potential indirect effects that they might have have received very little attention (Vander Wall 2000; Leaver & Daly 2001).
Here we report on a study that examined interactions among a jay and a rodent that disperse the seeds of a pine that produces seeds morphologically adapted for dispersal by wind. The pine is sugar pine (P. lambertiana Douglas), a fairly common and economically valuable tree in the Sierra Nevada of California and western Nevada. Sugar pine produces relatively large seeds (228 ± 39·4 mg; mean ± 1 SD), borne on a large but flimsy wing. Despite the wing, these heavy seeds are seldom dispersed far by wind. Most wind-dispersed seeds are gathered and cached by yellow pine chipmunks (Tamias amoenus Allen), an abundant and important disperser of pine seeds in the Sierra Nevada (Vander Wall 1992, 1993, 2003a). One of the main competitors of the chipmunk is Steller's jay (Cyanocitta stelleri Gmelin), which gather seeds directly from cones or from the ground. These jays are thought to be important dispersers of pine seeds (Tomback 1978; Vander Wall & Balda 1981), but this aspect of their behaviour has not been studied. Further, both species are diurnal so that they can interact directly, and because of fundamental differences in morphology and behaviour the quality of seed dispersal provided to sugar pine by each species is likely to differ.
Our objectives were: (1) to assess how well sugar pine seeds are dispersed by the wind, (2) to compare the effectiveness of yellow pine chipmunks and Steller's jays as secondary dispersers of sugar pine seeds (direct mutualistic interaction between two animals and sugar pine), (3) to estimate the impacts of chipmunks and jays as pilferers of each other's caches (direct competitive interaction between two dispersers) and (4) to determine whether this interspecific pilferage has any potential indirect effects on sugar pine seedling establishment.
- Top of page
We conducted this study in and around the Whittell Forest and Wildlife Area of the University of Nevada-Reno, located 30 km South of Reno, Nevada (39°15′10″ N, 119°52′35″ W; elevation 1975 m), an open yellow pine forests typical of the eastern Sierra Nevada. Soils consist of decomposed granite and the shrub community includes antelope bitterbrush (Purshia tridentata Pursh), tobacco brush (Ceanothus velutinus Dougl.), Sierra bush chinquapin (Castanopsis sempervirens Kellogg) and greenleaf manzanita (Arctostaphylos patula E. Greene).
We trapped rodents in three plant communities: (1) sugar pine–Jeffrey pine–white fir association, (2) Jeffrey pine forests and (3) bitterbrush shrublands to assess the abundance of yellow pine chipmunks and other seed-caching rodents. We established two 40-station trapping grids with 12-m between stations in each of the three habitats. We placed one Sherman live trap baited with sunflower seeds at each station. We checked each grid twice each day for 5 days in early September 1997–99 and early June 1998–2000 for a total of 7200 trap-nights. For each capture, we marked individuals with numbered ear tags, and recorded species, mass, gender, age and reproductive condition. Steller's jays are fairly common throughout the study area, but we did not attempt to estimate population size.
We assessed how rapidly animals removed sugar pine seeds on the ground by placing individual, winged seeds at 5-m intervals along transects, 50 or 100 seeds transect−1. We marked seed stations along transects inconspicuously with unique combinations of cones, twigs, or pebbles (Vander Wall 1994) because chipmunks (and probably other animals) are able to learn conspicuous markers such as pin flags, which cause artificially high removal rates. To ensure that seeds were not blown from stations by the wind, we tethered each seed to a twig with a brown thread Superglued® to the seed. We walked each transect daily until fewer than 10 seeds remained on each transect, recording if a seed was present, missing or eaten at each station. We repeated this method over four field seasons (1997–2000) replicated one to three times per year to produce eight independent measures of rodent foraging pressure on sugar pine seeds. Differences between years and sites was analysed using survival analysis (Allison 1995) with interval censoring and a Weibull distribution.
To assess how far sugar pine seeds are likely to be dispersed by wind, we first determined the mass and wing area of 19 sugar pine seeds. Using these data, we calculated wing loading for each seed. Next we dropped each seed from the ceiling of Lawlor Event Center at the University of Nevada, Reno (a height of 23·7 m). We recorded the time required for the seed to reach the floor and the horizontal distance traveled in the nearly still air of the event centre. We used these data to calculate mean descent velocity of each seed. Because of the long fall distance and because seeds began autorotating within ∼ 1 m after beginning to fall, we consider descent velocity to be a close approximation of terminal velocity.
We used these data to estimate wind dispersal distribution of sugar pine seeds. To calculate a dispersal distance, we used the following formula:
where za = release height (m), u = horizontal wind speed (m s−1), f = the terminal velocity of the seed (m s−1) and x is dispersal distance (m) (Greene & Johnson 1992). We repeated this calculation 100 times using one of 19 randomly selected terminal velocities, 86 cone heights and 139 wind speeds. We gathered data on cone heights from 15 sugar pine trees using a clinometer. We used wind speeds from the online database of the National Climatic Data Center. For this simulation, we used average hourly wind speeds (= 0·5 m s−1) at the Blue Canyon Airport weather station, Placer County, California (39°17′ N, 120°43′ W; elevation 1609 m), for 15 September and 15 October for the years 1997–2000.
We labelled sugar pine seeds with scandium-46 (Sc-46) to determine how Steller's jays and yellow pine chipmunks treated sugar pine seeds. Sc-46 is a biologically inert, gamma-emitting radionuclide with a half-life of 84·5 days (Abbott & Quink 1970; Vander Wall 1992). We also numbered seeds so that we could follow their histories. We attracted jays to a raised feeder by offering peanuts and sugar pine seeds for several days before starting the study. Chipmunks were presented seeds on a nearby boulder or stump. We placed 20 labelled and numbered sugar pine seeds at the seed source and observed individuals removing them from ∼20 m away. Once the subject left the feeding station with a load of seeds, we removed any remaining seeds and by subtraction we determined the number and identity of the seeds taken. We then placed a new set of 20 seeds at the seed source. Assisted by between one and six observers, we tracked each jay visually from the seed source to determine the approximate location of the caches it might make. After approximately 100 seeds had been taken from the seed source, we searched for caches using Geiger counters. For Steller's jays this process was repeated on 4 days in 1998 and 4 days in 1999, and for chipmunks 2 days in 1998 and 2 days in 1999. All located caches were excavated to record depth, number of seeds, substrate, shrub cover, distance from the feeder and numbers on seeds. We handled seeds with forceps to avoid contamination and eliminate the possibility of human odours influencing experimental results. The shrub cover for each cache was divided into open ground (> 10 cm outward from edge of shrub), edge of shrub (within 10 cm from shrub edge) and under shrub (under a shrub > 10 cm from edge). We analysed data on cache microhabitat (shrub cover and substrate) using multiway contingency tables and the fitting of log-linear models. To compare minimum dispersal distances, cache sizes and minimum cache depths between jays and chipmunks, we used a mixed-model analysis of variance (anova), with species and year as fixed factors and trips from the feeder (cluster of caches) as experimental units (jays and chipmunks typically cached all the seeds in a load in a relatively small area). We used an anova model with dispersal agent as a fixed factor to compare the dispersal distance for jays and chipmunks to the dispersal distances generated from the wind-dispersal model.
We monitored jay and chipmunk caches weekly during the autumn, to determine the number of seeds taken from the caches of each species, as well as the timing of seed removal. To help distinguish between cache retrieval and cache pilferage, we paired each animal cache with an artificial cache approximately 30 cm away. Artificial caches contained the same number of seeds, buried at the same depth, and in the same microhabitat. Animals can remember the locations of their own caches, but the only way to locate artificial caches is by using smell or by digging at random. By comparing the rate of removal of real and artificial caches, we estimated the proportion of an animal's caches that it recovered and the proportion pilfered (Vander Wall et al. unpublished data). Differences between rates of removal of real and artificial caches was analysed using survival analysis (Allison 1995) with interval censoring and a Weibull distribution.
We examined the caching and cache recovery behaviour of jays and chipmunks in an aviary (6 × 10 × 2 m). The aviary contained natural vegetation (mostly bitterbrush with three small pines), half of the ground area with decomposed granite substrate and half with pine litter substrate. Initially we began trials by presenting each jay (n = 16) with 100 sugar pine seeds labelled with Sc-46 and allowing the subject 24 h to cache seeds within the aviary. Generally, jays ate seeds but were reluctant to cache seeds. To entice jays to cache, we made several changes to this protocol. These changes included depriving the subjects of seeds for 24 h before trials (subjects were able to eat insects in the aviary) and placing a model of a jay or live jay in a cage near the feeder to simulate competition for seeds in the aviary. Once a jay had either removed most seeds from the feeder or spent 72 h in the aviary, we removed the jay and located caches with a Geiger counter. We excavated seeds, recording number of seeds, depth and microhabitat data, and mapped cache locations. Next, we replaced the seeds at the precise location, and placed an artificial cache of identical size, depth and microhabitat approximately 30 cm from the real cache. It was easy to remake caches without leaving any trace of our digging in the pine litter or friable decomposed granite. We then reintroduced the same jay to the aviary to search for caches for 24–48 h, following which we recorded the number of real and artificial caches found.
We tested whether chipmunks could find jay caches in the aviary. Using the same cache sites that four jays had used (see preceding paragraph), during the summer of 2002 we re-established three sets of jay caches (11–27 caches per trial; for one set of trials, caches of two jays were combined to obtained a sufficiently large sample). Caches were at the same depth with the same number of seeds as the jay caches. Then we allowed 12 yellow pine chipmunks (four for each set) to search individually for these caches. Six of the chipmunks searched for seeds during the summer drought, when soil conditions are very dry (< 0·5% water) and six after a rain (3–10% soil water content) (seed and soil water content are known to affect the strength of olfactory signals released from seeds). After preparing the caches, we placed a chipmunk in the aviary with water but no food (except for natural insects and seeds). The next day, 24 h later, we removed the chipmunks and surveyed the aviary to determine how many seeds had been discovered.
We also assessed the ability of Steller's jays to find caches of yellow pine chipmunks in the aviary. In each of 11 trials we placed a chipmunk in the aviary with 100 labelled sugar pine seeds, and gave it 24 h to make caches. After removing the chipmunk we found caches with a Geiger counter, described the contents as with jay caches, and replaced all seeds. We also placed five sugar pine seeds at each of five locations on the soil surface to motivate the naive jays to search for seeds on and under the surface. Initially we gave naive jays 48 h to search for caches but, due to the unexpected death of the first two jays tested, we reduced the search time to 24 h. We conducted non-observation experiments under both dry (n = 6) and moist conditions (n = 5) to test whether jays use olfaction to find buried seeds. We compared the foraging success of jays under dry and wet conditions using a Mann–Whitney U-test.
In five additional trials, we placed a jay in a wire mesh cage suspended from the top of the aviary, allowing the jay an unrestricted view of the aviary floor. During trials, we gave jays water and 30 sugar pine seeds as food. We placed a chipmunk in the aviary along with 100 labelled sugar pine seeds for 24 h. We then removed both the chipmunk and the jay and located and recorded data on all caches. After mapping and replacing caches, we placed five seeds in each of five piles on the soil surface, reintroduced the jay into the main aviary and allowed it 24 h to search for caches made by the chipmunk. Three of the five trials were conducted under dry soil conditions and two under wet soil conditions. Dry conditions occurred during summer droughts; to create moist conditions, we sprayed the aviary with 100 L of water.
Data are reported as means ± 1 SD unless otherwise stated.
- Top of page
Yellow pine chipmunks were the most common rodent species at all sites, and long-eared chipmunks (T. quadrimaculatus Gray) were common in sugar pine forests (Table 1). All the species in Table 1 act as sugar pine seed predators as well as potential seed dispersers. Long-eared chipmunks behave like yellow pine chipmunks but appear to be somewhat less effective as seed dispersers (Briggs & Vander Wall, in preparation). The chipmunks, deer mice (Peromyscus maniculatus LeConte) and golden-mantled ground squirrels (Spermophilus lateralis Merriam) gather wind-dispersed seeds from the ground surface and the ground squirrels can also remove seeds from closed cones.
Table 1. Number of individuals captured during 200 trap days during four field seasons (autumn 1997–spring 2001). Data are means ± 1 SD
|Sugar pine forest*||Jeffrey pine forest||Bitterbrush shrublands|
|Yellow pine chipmunk Tamias amoenus||24·6 ± 8·7||12·1 ± 10·2||27·8 ± 13·4|
|Long-eared chipmunk Tamias quadrimaculatus||22·9 ± 15·7|| 1·6 ± 2·6|| 1·6 ± 2·8|
|Golden-mantled ground squirrel Spermophilus lateralis|| 5·6 ± 5·5|| 9·5 ± 3·4|| 4·1 ± 3·7|
|California ground squirrel Spermophilus beecheyi†|| 0·0|| 0·6 ± 0·7 || 1·0 ± 1·4|
|Deer mouse Peromyscus maniculatus|| 7·5 ± 7·3|| 1·6 ± 1·7|| 7·0 ± 3·5|
We recorded three to seven individual Steller's jays in sugar pine stands during visits. Jays foraged for sugar pine seeds from the ground, from foliage and directly from cones. Jays foraged on cones by flying from a nearby branch, striking the cone with their bill and catching a dislodged seed before it hit the ground (Tevis 1953).
Mass of sugar pine seeds including the wing was 230 ± 39 mg and wing area was 2·59 ± 0·39 cm2, resulting in a wing loading of 91·0 ± 21·0 mg−1 cm2 (n = 19). Mean descent velocity of sugar pine seeds was 2·1 ± 0·9 m s−1 (n = 19). The model of wind dispersal (Greene & Johnson 1992) of sugar pine seeds showed that with a mean wind speed of 2·8 ± 1·6 m s−1 and mean cone height of 18·3 ± 5·3 m, sugar pine seeds would travel 26·7 ± 19·5 m (range = 2·6–117·1 m) (n = 100) from the source tree (Fig. 1).
Figure 1. Distances sugar pine seeds were transport to cache sites by yellow pine chipmunks and Steller's jays and simulated distance that seeds were transported by wind.
Download figure to PowerPoint
Animals removed sugar pine seeds rapidly from seed removal transects (Table 2). The half-life of seeds (times for half of the available seeds to be removed) was highly variable and ranged from 0·31 to 2·55 days, equivalent to rates of removal ranging from 23·8 to 89·0% day−1 (49·2 ± 19·9% day−1). We repeated the transect at SP-1 (Table 2), an open, old-growth stand of sugar pine, in each of 4 years. Five of six between-year comparisons of removal rates from this transect were significantly different (P < 0·001). Three of five within-year comparisons of removal rates in 1997, 1999 and 2000 were also significantly different (P < 0·001). Only 4·8 ± 6·3% of removed seeds were eaten at the stations; the remaining seeds were removed intact and many were probably stored. Our field observations of foraging animals suggest that most of the seeds along these transects were probably removed by chipmunks.
Table 2. Seed removal from the ground surface on eight transects in four sugar pine stands in the Whittell Forest and Wildlife area
|Year||Transect||Initial seeds (n)||Final seeds (n)||Time elapsed (days)||Half-life (days)||Removal rate (% day−1)|
|SP-3|| 52|| 4||3·12||0·84||56·5|
|SP-4|| 48|| 8||4·02||1·56||36·1|
|1999||SP-1|| 50|| 5||1·04||0·31||89·0|
|SP-2|| 50|| 6||3·18||1·04||48·6|
|2000||SP-1|| 50|| 9||3·81||1·54||36·2|
|SP-2|| 50|| 6||7·81||2·55||23·8|
The mean (± 1 SD) distance that Steller's jays carried radioactively labelled sugar pine seeds to cache sites was 203·7 ± 104·5 m (max. = 370 m; n = 24 loads), and for yellow pine chipmunks 19·9 ± 14·0 m (max. = 38 m; n = 14 loads). A mixed-model anova for dispersal distance of sugar pine seeds by Steller's jays, yellow pine chipmunks and wind indicated a significant effect of type of dispersal (mixed anova, F2,7 = 22·3, P < 0·001). Jays dispersed seeds significantly further than yellow pine chipmunks and wind (Tukey's honest significant difference (HSD), P = 0·006 and P = 0·003, respectively). However, there was no significant difference between the distances that seeds were carried by yellow pine chipmunks and wind (Tukey's HSD, P = 0·75).
Steller's jays and yellow pine chipmunks differed little in the number of sugar pine seeds they took from a feeding station in a single load. Steller's jays took 11·4 ± 3·9 seeds per trip (n = 23 trips) in 1998, and 5·7 ± 2·4 seeds per trip (n = 22 trips) in 1999. Yellow pine chipmunks carried 9·9 ± 4·7 seeds per trip (n = 11 trips) in 1998 and 12·1 ± 3·3 seeds per trip (n = 7 trips) in 1999. There was a significant interaction between year and species (two-way anova, F1,471 = 95·9, P < 0·001). After we eliminated year as a factor in the analysis, we found no significant difference in load size between species (F1,56 = 2·6, P = 0·11). Jay load size was significantly correlated with transport distance to cache sites (data for 1998 and 1999 combined; r = 0·733, d.f. = 23, P < 0·001).
Caching substrate use by yellow pine chipmunks and Steller's jays varied considerably between years (Fig. 2a). Log-linear analysis of both substrate and shrub cover of caches indicated that a three-way interaction was necessary to create the best-fitting model for the analysis (χ2 = 22·58; d.f. = 3; P < 0·0001), so no further interpretation was attempted. Thick deposits of pine litter were the most commonly used cache substrate for Steller's jays in both 1998 and 1999 (74·5 ± 15·8% of cache sites). These sites were almost invariably under the closed canopy of pine forests. Yellow pine chipmunks, in both 1998 and 1999, made most caches in either mineral soil or light litter (combined substrates: 1998, 50·0 ± 19·2%, n = 44; 1999, 90·0 ± 17·7%, n = 40). The fairly heavy use of pine litter by chipmunks in 1998 was because of one subject.
Figure 2. (a) Percentages of caches made by yellow pine chipmunks and Steller's jays in four substrate types: HL, heavy litter (> 10 mm deep, non-pine); LL, light litter (< 10 mm deep, non-pine); MS, mineral soil; PL, pine litter (> 10 mm deep, closed canopy pine forests). (b) Percentages of caches made by yellow pine chipmunks and Steller's jays in three cover types: under shrub (> 10 mm from shrub edge), edge of shrub (± 10 mm from shrub edge), open (> 10 mm from shrub edge).
Download figure to PowerPoint
Steller's jays made caches much further away from shrubs than yellow pine chipmunks (Fig. 2b); 42·1% of jay caches were > 2 m away from the nearest shrub (n = 95) whereas 95·3% of the chipmunk caches were under or < 1 m from a shrub (n = 84) (none was > 2 m from a shrub). Furthermore, when jays cached under shrubs, they often chose chinquapin growing under Jeffrey pine trees, whereas chipmunks used bitterbrush growing in forest openings.
We found no significant difference in number of seeds per cache between Steller's jays (1·8 ± 0·9; range 1–5, n = 99) and yellow pine chipmunks (1·4 ± 0·8; range 1–3, n = 85) (mixed anova, F1,36 = 1·5, P = 0·224). Steller's jays made caches slightly deeper (11·8 ± 6·8 mm to top of seeds) than did yellow pine chipmunks (6·2 ± 4·7 mm; Fig. 3) (mixed anova, F1,36 = 4·3, P = 0·045).
Figure 3. Caching depth profiles of yellow pine chipmunks (closed circles, n = 79 caches) and Steller's jays (open squares, n = 94).
Download figure to PowerPoint
Our analysis of the disappearance of real and artificial caches in the radioactive seed study revealed that both chipmunks and jays had a recovery advantage when foraging for their own caches; real caches disappeared 3·6 times faster than artificial caches (χ2 = 31·77; P < 0·0001 for both species). However, after rain, artificial chipmunk caches disappeared just as quickly as real chipmunk caches (P > 0·5). We were unable to test the effect of dry vs. wet conditions on jay caches because jays made caches several days after rain. Caches made by chipmunks (generally in the bitterbrush habitat) disappeared 3·4 times faster than caches made by jays (generally in the open Jeffrey pine forest habitat) (χ2 = 38·03, P < 0·0001). We observed a similar pattern when we compared the rate of removal of artificial chipmunk caches to artificial jay caches (artificial chipmunk caches disappeared 3·3 times faster than artificial jay caches; χ2 = 20·01, P < 0·0001).
Only four of 16 Steller's jays that we tested cached seeds in the aviary. The four subjects that did store seeds made seven caches containing a total of nine seeds, 27 caches (56 seeds), four caches (four seeds) and 21 caches (43 seeds). During the recovery phase of the experiment, these four jays found a significantly greater percentage of their own caches (93·3%) than the artificial caches just 30 cm away (1·3%) (paired t-test, t = 12·99, d.f. = 3, P < 0·001).
Chipmunks were effective in finding jay caches only when the soil was moist. Under dry conditions, chipmunks found 5·3 ± 7·7% (range 0·0–19·0%) of available caches (n = 354) within 24 h; however, when the soil was moist, chipmunks found 67·9 ± 29·7% (range 18·2–96·3%) of available caches (n = 354). This difference in foraging success was significant (one-way anova: F1,10 = 26·22, P < 0·001).
Steller's jays were unable to find chipmunk caches if they did not observe the caches being made. Two of six jays that searched for 15–30 caches in the aviary under dry conditions each found one cache and four jays found no caches. During four of five trials when the substrate was moist, the jays removed only one of 79 caches present. In the fifth trial, the jay recovered 18 of 30 caches made by a chipmunk, but the behaviour of this subject was very different; the jay excavated much of the substrate in the aviary by using its bill to turn over large areas of plant litter and to remove litter from under shrubs. The evidence suggests that this jay used exploratory digging (random search) to find buried seeds. Recovery of chipmunk caches by Steller's jays did not differ between wet or dry conditions (U5,6 = 20, P > 0·1). Jays in all the wet and dry trials found the five surface caches, indicating that they had searched for food in the aviary.
Five Steller's jays had the opportunity to observe a chipmunk forage in the aviary. Only two of the chipmunks in these trials made any caches. One chipmunk made only a single cache, and placed it behind a shrub where the jay did not have a clear view. The other chipmunk made 30 caches many in view of the observing jay. During the recovery phase the jay observer was allowed to search for caches, but it was found dead the next morning, having failed to find any of the chipmunk caches or the five surface caches we had prepared in the aviary. The fact that this jay observer failed to remove even the surface caches indicates that it spent little or no time searching the aviary for food. Following the death of this subject, we suspended this experiment. Chipmunks made significantly fewer caches when a jay was observing (6·2 ± 13·3) than when no jay was observing (23·3 ± 6·8) (U5,7 = 30, P = 0·025).
- Top of page
Most sugar pine seeds are dispersed by the wind when cones open in mid-September However, heavy wing-loading causes them to fall rapidly (nearly 2 m s−1) compared to other winged pine seeds (Siggins 1933; Johnson et al. 2003). This rapid falling speed is caused, in part, by constraints on wing length imposed by cone scale length (Benkman 1995). In comparison, the descent velocities of lodgepole pine (P. contorta Loudon) and ponderosa pine (P. ponderosa), species found in the region that have smaller seeds more typical of wind-dispersed pines, were 0·89 ± 0·14 m s−1 and 0·99 ± 0·19 m s−1, respectively (Johnson et al. 2003). Even under moderate winds sugar pine seeds are not likely to travel far (Fig. 1), although our analysis no doubt underestimates the potential for infrequent, long dispersal distances caused by high winds and updrafts in mountainous terrain (Cain, Milligan & Strand 2000; Nathan, Safriel & Noy-Meir 2001).
Wind-dispersed seeds are gathered rapidly from the ground; rates of removal along transects averaged ∼50% day−1 (Table 2). The variation between sites and years was probably because of differences in background food densities and rodent population sizes. Even at slower rates (~25% day−1), virtually all seeds would be gathered by animals within a few weeks. These rates are much faster than those determined for smaller pine seeds (Vander Wall 1994), and are no doubt caused by the large size (high nutritional value and conspicuousness) of sugar pine seeds. Most seeds (∼95%) were carried away intact, suggesting that they could have been cached for later use. We were unable to determine what proportion of the seeds along transects were taken by Steller's jays vs. yellow pine chipmunks, but numerous yellow pine chipmunks searched the ground under sugar pine trees, finding a seed every minute or so, and placed the seeds into their cheek pouches. Steller's jays occasionally foraged on the ground but focused much of their attention on opening cones. The jays foraged by striking cones with their beak while hovering and catching falling seeds (Tevis 1953) and searched for fallen seeds lodged in the foliage below open cones. During the period of cone opening, the per capita harvest of seeds by chipmunks and jays was probably similar, but after most seeds had been shed from cones, chipmunks probably obtained a greater proportion of the seeds on the ground.
Our observations of caching behaviour using radio-labelled seeds in the field revealed that Steller's jays and yellow pine chipmunks cache seeds in a similar way. Both species made relatively small caches (one or two seeds) and scatter-hoarded seeds in the soil surface (∼5–25 mm deep). The greatest difference in their caching behaviour is that jays placed most caches in pine litter under trees > 2 m away from shrubs, whereas chipmunks cached more in mineral soil under or close to bitterbrush shrubs. Yellow pine chipmunks usually avoid caching in thick deposits of pine litter (Vander Wall 1993). One possible reason for this difference in cache site selection is that chipmunks might prefer to stay closer to shrub cover to reduce risk of predation, whereas jays reduce the risk of predation by staying closer to the trees.
Our aviary experiments demonstrated that Steller's jays are able to recover their own caches, but are much less effective in finding artificial caches or those of chipmunks. These results are consistent with the findings of many studies on cache recovery by corvids and parids (e.g. Sherry, Avery & Stevens 1982; Kamil & Balda 1985; Brodin & Clark 1997) and suggest that jays cannot smell buried seeds even under moist conditions and used spatial memory to find their own caches. One jay was able to find chipmunk caches, but it apparently used exploratory digging to do so. This appears to be an energetically expensive foraging behaviour, and it is unlikely to be an effective means of finding caches outside the aviary where cache density is much lower. Yellow pine chipmunks, on the other hand, are able to find their own caches using spatial memory (Jacobs & Liman 1991; Vander Wall 1991; Jacobs 1992) and olfaction is an effective means of finding buried seeds only under moist conditions (Johnson & Jorgensen 1981; Vander Wall 1998, 2000). These results mean that jays should be ineffective at finding chipmunk caches, but that chipmunks can easily find jay caches except when the soil is very dry.
Steller's jays are able to pilfer the caches of other species when they observe an animal make a cache (Tomback 1978; Burnell & Tomback 1985). Steller's jays watch other animals prepare caches and pilfer the seeds after the cacher departs. Burnell & Tomback (1985) also found that grey jays (Perisoreus canadensis L.) would not cache in the presence of Steller's jays. We were unable to confirm this behaviour with chipmunks because most chipmunks in our experiments failed to cache in the presence of jays. However, it was not clear from our study whether chipmunks seldom cached in the presence of jays because the chipmunks considered the jays to be competitors or to be potential predators.
What do these aviary studies mean for jays and chipmunks interacting in the field? Once seeds are cached, chipmunks can pilfer from jays when the conditions are moist (but less so when conditions are very dry). In the semi-arid pine forests of the east slope of the Sierra Nevada, long summer droughts interrupted by infrequent rain events provide occasional opportunities to pilfer caches in summer and early autumn. On the other hand, jays are ineffective pilferers of chipmunk caches unless they see caches being prepared. The species interaction is asymmetrical. Chipmunks can pilfer from each other so their pilferage can be reciprocal, which gives the interaction some stability (Vander Wall & Jenkins 2003). However, when a jay is pilfered by a rodent the jay cannot reciprocate effectively. The best strategy from the jay's perspective is to avoid being pilfered in the first place.
How do jays avoid being pilfered by rodents? Our paired-cache study provides a partial answer to this question. Real caches of both chipmunks and jays disappeared much faster than artificial caches. The difference between the rates of removal of real and artificial caches is a measure of the recovery advantage of the cacher (Vander Wall et al., unpublished data). This advantage was similar for both chipmunks and jays, suggesting that the jays are not at any great disadvantage when competing with chipmunks for their own cached seeds. If chipmunks had pilfered many of the jay caches, we would have expected them not to discriminate between real and artificial jay caches (as the jays should), resulting in more similar removal rates of the real and artificial jay caches. However, both real and artificial jay caches (i.e. in open Jeffrey pine forest habitat) disappeared much slower than real and artificial chipmunk caches (i.e. in bitterbrush habitat). These results suggest that the open Jeffrey pine forest habitat where chipmunks are far less abundant (Table 1) and spend less time foraging (Vander Wall 1993, 1994; K. M. Kuhn, personal observation) are safer cache sites for jays. Steller's jays appear to avoid competition with yellow pine chipmunks for stored seeds by caching in a habitat that the chipmunks seldom use.
An alternative means of avoiding pilferage by chipmunks would be for jays to recovery caches relatively quickly, before chipmunks had much of an opportunity to pilfer them (pilferage avoidance; Vander Wall & Jenkins 2003). However, our analyses of the removal rates of real jay and chipmunk caches do not support this hypothesis. Jay caches disappeared significantly slower than chipmunk caches.
Competition between distantly related taxa has been documented in many ecological systems (e.g. Smith & Balda 1979; Christensen & Whitham 1993). Although this study does not demonstrate competition between Steller's jays and yellow pine chipmunks, there is evidence that the two species interact over sugar pine seeds and that the potential for competition clearly exists. During autumn in years when sugar pines produce many cones, food is not limiting. However, both species store harvested food for the winter and over the longer term (September–March) food often is limiting. It seems likely that jays and chipmunks compete intensely for stored pine seeds in the late autumn.
Our radiolabelled seed studies revealed that Steller's jays dispersed sugar pine seeds about 10 times further than yellow pine chipmunks and nearly eight times further than moderate wind. However, these comparisons should not be taken as measures of the relative potential benefits of dispersal by wind, chipmunks and jays because they ignore the fact the many sugar pine seeds are being dispersed in two phases, combining wind- and seed-caching animals (Vander Wall & Longland 2004). Many of the seeds initially dispersed by wind are moved subsequently to a cache site by a chipmunk. Jays often pre-empt wind dispersal by removing seeds directly from cones and storing them, but they can also serve as agents of secondary dispersal by gathering seeds from the ground or foliage and transporting them to cache sites.
The habitats and microhabitats in which Steller's jays and yellow pine chipmunks cache sugar pine seeds may have important consequences for seed germination, seedling establishment and subsequent seedling growth (Shainsky & Radosevich 1986; Drivas & Everett 1988; Callaway et al. 1996). Shrub cover greatly benefits sugar pine seedling establishment compared to seedlings emerging in openings between shrubs (E. C. H. Hager, unpublished data). The effect of shrubs on pine seedlings can be positive when the shrub acts as a ‘nurse-plant’, providing shade to prevent desiccation, reducing herbaceous competitors and providing protection from herbivory (Callaway et al. 1996; Callaway & Walker 1997; Chambers 2001; Hollander & Vander Wall 2004). However, we do not yet know whether sugar pine seedlings grow and survive well under the canopy of mature trees. For Jeffrey pine seedlings, shaded pine litter appears less suitable for growth and establishment than mineral soil under the shelter of a shrub (Vander Wall 1993).
The interaction between jays and chipmunks might also have indirect effects on sugar pine in at least two ways. First, chipmunks appear to have influenced cache site selection of jays through pilfering jay caches. The forest sites that jays use appear to be low quality for sugar pine seedlings. It seems likely that, if it were not for rodent pilferage, jays might distribute caches more widely with a higher proportion of seeds in open, shrubby areas. Many animals are known to shift caching microhabitat in response to interspecific competition (Moreno, Lundberg & Carlson 1981; Pravosudov 1986; Alatalo & Carlson 1987; Suhonen & Alatalo 1991). These shrubby areas should be more favourable to sugar pine seedling establishment (E. C. H. Hager, unpublished data). Consequently, rodent pilferage may have forced jays to be less effective dispersers of sugar pine seeds. Secondly, pilfering of seeds by chipmunks from jays and then recaching those seeds could significantly influence the probability of germination and successful establishment of a seedling. For example, a chipmunk might pilfer a jay cache in a low-quality site (e.g. in thick pine litter under the canopy of other pines) and recache the seeds in higher-quality sites (e.g. in mineral soil under shrubs in a forest opening). This type of secondary dispersal could be highly beneficial to the pine: jays could provide long-range dispersal and then chipmunks could pilfer jay caches and, by recaching the seeds, ‘upgrade’ the quality of the deposition site.
Sugar pine has been assumed to be a wind-dispersed pine based on seed and cone morphology (e.g. Siggins 1933). However, this study shows that, much like winged ponderosa, Jeffrey and Coulter pine seeds (Vander Wall 1992, 2002; Borchert et al. 2003), scatter-hoarding animals play an important role in sugar pine dispersal. Although chipmunks do not carry sugar pine seeds far, the bitterbrush habitat in which they cached may be favourable to sugar pine recruitment. On the other hand, Steller's jays can carry seeds long distances but often select poor cache sites from the pine's perspective. The effect of this dichotomy in dispersal patterns on sugar pine is still unclear. However, it would seem that yellow pine chipmunks may provide a greater proportion of local seed dispersal and subsequent sugar pine recruitment. Steller's jays may provide sugar pine with long distance dispersal, important for colonizing new habitats.