Spatial patterns of food storage by Merriam's kangaroo rats

We investigated scatter-hoarding behavior by Merriam's kangaroorats (Dipodomys merriami) in a complex arena in the laboratory.Animals cached seeds in four sand-filled boxes connected byrunways, with an artificial burrow at one end of the systemand a food source at the other. Animals initially cached moreseeds in boxes close to the food source, but amounts cachedbecame more evenly distributed among boxes as sessions progressed.The results were consistent with the hypothesis that scatterhoarding by kangaroo rats may represent a compromise betweenrapidly sequestering seeds, hence making them unavailable tonondigging competitors, and spacing-out caches, making themless vulnerable to pilferage by other rodents.     

森林鼠类对秦岭南坡3种壳斗科植物种子扩散的影响

种子作为森林鼠类的主要食物来源,它们的扩散和更新很大程度上依赖于鼠类的传播。在鼠类扩散种子的过程中,种子特征和食物相对丰富度是影响鼠类对种子进行何种选择策略的重要因素。2011—2012年的8—12月,采用塑料片标记法在秦岭南坡的佛坪国家级自然保护区内调查了森林鼠类对同域分布的3种壳斗科植物(锐齿槲栎Quercus aliena var.acuteserrata、栓皮栎Q.variabilis和短柄枹栎Q.serrata var.brevipetiolata)种子的扩散差异。结果表明:(1)鼠类倾向于贮藏营养价值较大的栓皮栎种子,并且其贮藏距离也最远(2011:1.52 m,2012:4.03 m),3种种子在食物相对丰富度较低年份被贮藏的距离均较远。(2)在食物相对丰富度较高的年份(2011),种子的消耗速率较慢,在种子释放10 d后种子释放点仍有67.33%的种子,贮藏量较高,至实验结束仍有29.67%的种子被贮藏。在食物相对丰富度较低的年份(2012),种子消耗速率较快,在种子释放后10 d内所有种子均被取食或搬离种子释放点,贮藏量较低,至实验结束仅有3.83%的种子仍被贮藏。(3)虽然栓皮栎种子的贮藏量最大,被贮藏后的存留量也最大,但其在实验地的分布却较小,说明种子扩散仅是植物分布与存活的第一步。以上结果表明,鼠类倾向于贮藏营养价值高的种子。在食物相对丰富度较高的年份会更多的贮藏种子,但种子被贮藏的距离较近,在食物相对丰富度较低的年份会更多的取食种子。     

Seed predation and dispersal of glabrous filbert (Corylus Heterophylla) and pilose filbert (Corylus Mandshurica) by small mammals in a temperate forest, northeast China

We investigated the seed dispersal of glabrous filbert (Corylus heterophylla) and pilose filbert (Corylus mandshurica), two large-seeded shrub species in a temperate forest, northeast China, September 2006. Small mammals such as Apodemus speciosus, Clethrlonomys rufocanus, and Eutamias sibiricus, were regarded as the main dispersal agents. More seeds were harvested by small mammals in pilose filbert (98%) than in glabrous filbert (87.5%) till our last survey. Seed removal rates differed between the two species. Fewer seeds of glabrous filbert (17.5%) were eaten in situ than pilose filbert (57.5%). More seeds of glabrous filbert were removed (70%), stay intact after removal (25.5%), eaten after removal (16%) than pilose filbert. However, more seeds were cached after removal in pilose filbert than in glabrous filbert (10.5 and 4%, respectively). Fewer tagged seeds of pilose filberts (14%) were missed than glabrous filberts (24.5%). About 8 and 12 primary caches were found in glabrous filbert and pilose filbert seeds respectively, indicating scatter hoarding. All of the removed seeds were distributed within 10 m of seed stations for both filberts. The average dispersal distances for glabrous filbert did not differ from pilose filbert. Only a small proportion of the caches remained till our last survey (2 and 1%, respectively). Based on the results, we found a difference in dispersal patterns of glabrous filbert and pilose filbert seeds. Evidences showed that glabrous filberts might be a less preferred seed species for small seed-eating mammals compared with pilose filbert, probably due to its harder and thicker husk and low seed profitability.     

Re-caching of acorns by rodents: Cache management in eastern deciduous forests of North America

Scatter-hoarding rodents such as tree squirrels selectively cache seeds for subsequent use in widely-spaced caches placed below the ground surface. This behavior has important implications for seed dispersal, seedling establishment, and tree regeneration. Hoarders manage these caches by recovering and eating some seeds, and moving and re-caching others. This process of re-caching, however, is poorly understood. Here, we use radio-telemetry to evaluate re-caching behavior for the management of acorn caches by rodents in eastern deciduous forests. We also test the hypothesis that as seeds are re-cached, the distance from the source increases. Radio transmitters were implanted in Northern red oak (Quercus rubra) acorns and presented to rodents in a natural setting over 3 seasons. We used radio-telemetry to track and document evidence of recovery and re-caching. We tracked a total of 102 acorns. Of the 39 radio-tagged acorns initially cached, 19 (49%) were cached on two or more occasions; one acorn was cached four times. The hypothesis that rodents move seeds to progressively greater distances from the source is not well-supported, suggesting that acorns are being moved within an individual's home range. Given the species of rodents in the study area, gray squirrels (Sciurus carolinensis) are the most likely to be responsible for the caching and re-caching events. Gray squirrels appear to engage in extensive re-caching during periods of long-term food storage, which has important implications for understanding how caching behavior influences acorn dispersal and oak regeneration.     

Effects of seed distribution and competitors on seed harvesting efficiency in heteromyid rodents

Summary The foraging strategies of four naturally co-existing heteromyid rodent species were investigated: Dipodomys deserti (100 g), D. merriami (38 g), Microdipodops pallidus (13 g), and Perognathus longimembris (7 g). In 208 over-night laboratory foraging trials animals were provided with millet seed distributed in clumped and scattered patterns. Net removal of seeds from the foraging arena and amounts of seeds in cheek pouches and in caches were determined. When alone in an areana none of these species specialized extensively on either clumped or scattered seeds, although each tended to take more clumped than scattered seeds. When placed together with other individuals, animals once again tended to cache more clumped than scattered seeds in all but one paired combination of species: P. longimembris cached more scattered than clumped seeds when opposed by D. deserti. This suggests that the smaller species obtained a less preferred distribution of seeds in the face of competition. The two smaller species showed a great reduction in general foraging success in the presence of either of the two larger species. In general, a species cached less seeds when faced by larger opponent species.     

Rodent handling of Araucaria araucana seeds

In an intermast year of very low seed production, we studied seed handling in an Araucaria araucana (Araucariaceae) forest in Neuquén Province, Argentina. Rodents identified in 844 photographs from automatic cameras removed 589 seeds marked with small embedded magnets. Within 12 days of removal, 460 were recovered using a magnetic field locator; 79% of recoveries were from burrows and caches in litter or soil. Rodents exhibited a diversity of seed‐handling strategies and differed in their potential as dispersal agents. Seed removal was most likely between 15.00 hours and 21.00 hours, but diurnal and nocturnal visits were recorded for all species. Oligoryzomys longicaudatus (Cricetidae) was a seed predator that took 8% of marked seeds and left none whole. Chelemys macronyx (Cricetidae) and Rattus norvegicus (Muridae) removed 25% and 19% of marked seeds respectively. These two species deposited the majority of seeds in groups of 10 or more in burrow larders that were unfavourable seedling establishment sites far from daylight. Abrothrix longipilis (Cricetidae) removed 43% of seeds, scatter‐hoarded the largest percentage of whole seeds (37%), moved some seeds farther than 40 m, and left them in favourable seedling establishment sites near daylight. For all species, the number and proportion of seeds cached whole increased as more seeds were removed. Rodents, especially A. longipilis, may play an important role in regeneration of A. araucana. Seed‐handling strategies and potentially effective dispersal are discussed in terms of masting seed production.     

Seed-caching responses to substrate and rock cover by two <Emphasis Type="Italic">Peromyscus</Emphasis> species: implications for pinyon pine establishment

We examined whether pinyon mice (Peromyscus truei) and brush mice (P. boylii) could act as directed dispersal agents of pinyon pine (Pinus edulis) through differential responses to soil particle size and rock cover. In field experiments, we allowed mice to either cache pinyon seeds or recover artificially cached seeds (pilfer) from quadrats containing large- or small-particle soils. Both species placed most (70%) seed caches in small-particle soil. Pilfering was the same from both particle sizes in the first year, while more seeds were pilfered from large-particle soils in the second year. In separate experiments, rock cover interacted with soil particle size, with both species placing over 50% of their caches in small-particle soil with rock cover. Overall, we found greater seed-caching in small-particle soils near rocks, with equal or lower pilfering from small-particle soils, suggesting more seeds would survive in small-particle soils near rock cover. Three lines of evidence supported the hypothesis that mice could act as directed dispersers by moving pinyon seeds to beneficial microsites for germination and establishment. First, in greenhouse experiments, pinyon seed germination was 4 times greater in small-particle soil cores than in large-particle soil cores. Second, soils near rocks had significantly higher water content than areas of open soil at the driest time of the year, a critical factor for seedling survival in the arid southwestern USA. Third, 75% of juvenile pinyon trees were growing in small-particle soils, and 45% were growing near rock nurses.     

Specific nest box designs can improve habitat restoration for cavity‐dependent arboreal mammals

Tree‐cavity‐dependent wildlife faces future shortages of cavities due to a decline in the abundance of large, old trees in many parts of the world. Nest boxes are proposed as a tool to restore habitat value but evidence of their effectiveness for arboreal mammals remains equivocal. This may arise from a poor understanding of design preferences. We conducted investigations in two landscapes in eastern Australia to determine whether species show a preference for specific designs. We observed a preference by some mammal species for particular designs (33–78% occupied/used), suggesting that design refinement can improve the frequency with which nest boxes are used. Although feral species may out‐compete target species for nest boxes, we did not observe this. We recorded feral honeybees (Apis mellifera) in 6–9% of nest boxes but they did not remain, and many occupied boxes were later used by mammals. The introduced common myna bird (Acridotheres tristis) was prevalent in one landscape, but competition for nest boxes was localized. For nest boxes to be an effective habitat restoration tool, they must be able to be occupied over long periods of time. We investigated this for the squirrel glider (Petaurus norfolcensis), an arboreal marsupial threatened through part of its geographic range. Squirrel gliders occupied and bred within nest boxes (100% used) at two locations continuously over a 10‐year period with minimal nest box maintenance. Individuals occupied boxes for up to 7 years. This confirms that targeted nest box programs can be an effective restoration tool for cavity‐dependent arboreal mammals.     

Effects of the risk of competition and predation on large secondary cavity breeders

The effects of competition and risk of predation on secondary cavity breeders were examined between the 2008 and 2009 breeding seasons using an experimental design manipulating two nest entrance sizes (large entrances allowed Barn Owls (Tyto alba) to enter, while the small entrances excluded them). During the 2009 breeding season, the entrance sizes of nest boxes were exchanged, so that if during one year a nest box in a particular location had a small entrance, the second year it had a large entrance and vice versa. Barn Owls and Eurasian Kestrels (Falco tinnunculus) occupied 67.3 and 17.3%, respectively, of large entrance nest boxes. Significantly more Jackdaws (Corvus monedula), House Sparrows (Passer domesticus) and Scops Owls (Otus scops) bred in nest boxes with small than with large entrances. After nest box entrance sizes were exchanged, Barn Owls and smaller species did not breed in the same nest boxes with the new entrance size. Jackdaws probably did not breed in large entrance nest boxes due both to exploitation competition (Barn Owls and Eurasian Kestrels occupied the majority of large entrance nest boxes), and may also have avoided empty nest boxes because of the risk of interference competition; whereas smaller species may have also avoided large entrance nest boxes due to risk of predation.     

Nest box contentions: Are nest boxes used by the species they target?

Nest boxes have grown in popularity as a habitat management tool in Australia during the last decade. This management use remains contentious because some studies suggest nest boxes are ineffective. There are three recent contentions: (i) nest boxes mostly benefit common species, (ii) exotic species may be dominant users of nest boxes, and (iii) species of conservation concern use nest boxes infrequently. We address these contentions using data from 1865 nest boxes involving eight nest box designs. These nest boxes were installed predominantly <200 m from a road in association with highway duplication and re‐alignment across 16 projects in New South Wales. The Common Brushtail Possum (Trichosurus vulpecula) is the species of most relevance to contention 1. It used 9% of boxes overall including 26% of ‘possum’ designated boxes. The most frequent nest box users were small petaurid gliders (mostly Sugar Gliders, Petaurus breviceps) which used 63% of ‘small glider’ designated boxes. This nest box and another suited to the Sugar Glider comprised 40% of all boxes installed, so it is not surprising that this species might be a common user. Exotic species were uncommon users of the nest boxes enabling contention 2 to be rejected. Active hives of Feral Honeybees (Apis mellifera) occupied just 1% of boxes, and another 1% of boxes were used by introduced rodents and birds. The Squirrel Glider (Petaurus norfolcensis) is the species most relevant to contention 3. It was seen in 80 boxes across 11 projects, representing 7% of the three types most frequently used. These observations are not consistent with the third contention. Nest boxes can provide many important insights about the requirements and interactions of hollow‐dependent fauna. However, they are not intended as an alternative to retaining hollow‐bearing trees.     

Directed seed dispersal towards areas with low conspecific tree density by a scatter‐hoarding rodent

Scatter‐hoarding animals spread out cached seeds to reduce density‐dependent theft of their food reserves. This behaviour could lead to directed dispersal into areas with lower densities of conspecific trees, where seed and seedling survival are higher, and could profoundly affect the spatial structure of plant communities. We tested this hypothesis with Central American agoutis and Astrocaryum standleyanum palm seeds on Barro Colorado Island, Panama. We radio‐tracked seeds as they were cached and re‐cached by agoutis, calculated the density of adult Astrocaryum trees surrounding each cache, and tested whether the observed number of trees around seed caches declined more than expected under random dispersal. Seedling establishment success was negatively dependent on seed density, and agoutis carried seeds towards locations with lower conspecific tree densities, thus facilitating the escape of seeds from natural enemies. This behaviour may be a widespread mechanism leading to highly effective seed dispersal by scatter‐hoarding animals.     

Tracking Seed Fates of Tropical Tree Species: Evidence for Seed Caching in a Tropical Forest in North-East India

Rodents affect the post-dispersal fate of seeds by acting either as on-site seed predators or as secondary dispersers when they scatter-hoard seeds. The tropical forests of north-east India harbour a high diversity of little-studied terrestrial murid and hystricid rodents. We examined the role played by these rodents in determining the seed fates of tropical evergreen tree species in a forest site in north-east India. We selected ten tree species (3 mammal-dispersed and 7 bird-dispersed) that varied in seed size and followed the fates of 10,777 tagged seeds. We used camera traps to determine the identity of rodent visitors, visitation rates and their seed-handling behavior. Seeds of all tree species were handled by at least one rodent taxon. Overall rates of seed removal (44.5%) were much higher than direct on-site seed predation (9.9%), but seed-handling behavior differed between the terrestrial rodent groups: two species of murid rodents removed and cached seeds, and two species of porcupines were on-site seed predators. In addition, a true cricket, Brachytrupes sp., cached seeds of three species underground. We found 309 caches formed by the rodents and the cricket; most were single-seeded (79%) and seeds were moved up to 19 m. Over 40% of seeds were re-cached from primary cache locations, while about 12% germinated in the primary caches. Seed removal rates varied widely amongst tree species, from 3% in Beilschmiedia assamica to 97% in Actinodaphne obovata. Seed predation was observed in nine species. Chisocheton cumingianus (57%) and Prunus ceylanica (25%) had moderate levels of seed predation while the remaining species had less than 10% seed predation. We hypothesized that seed traits that provide information on resource quantity would influence rodent choice of a seed, while traits that determine resource accessibility would influence whether seeds are removed or eaten. Removal rates significantly decreased (p < 0.001) while predation rates increased (p = 0.06) with seed size. Removal rates were significantly lower for soft seeds (p = 0.002), whereas predation rates were significantly higher on soft seeds (p = 0.01). Our results show that murid rodents play a very important role in affecting the seed fates of tropical trees in the Eastern Himalayas. We also found that the different rodent groups differed in their seed handling behavior and responses to changes in seed characteristics.     

Winter food caches of beavers<Emphasis Type="Italic">Castor fiber</Emphasis> in NE Poland

Winter food caches of beaversCastor fiber Linnaeus, 1758 were examined in 16 locations within the Suwałki region (north-eastern Poland). Four caches were selected to represent each of four habitat types: river, lake, farmland, and oligotrophic small reservoir called ‘suchar’. Altogether, 92 509 twigs and branches of 15 woody species were identified and measured. Individual caches, with one exception, included branches of only 4 or 5 species. WillowsSalix sp., birchesBetula sp., alderAlnus glutinosa, trembling aspenPopulus tremula, and mountain ashSorbus aucuparia occurred in caches from all four habitats studied. Willows alone constituted 62.5% of the total number of branches cached. Willows, together with birch, hazelCorylus avellana, alder, and trembling aspen, amounted to 97% of all branches in caches. The remaining 10 species provided only 3% of the branches cached. Most branches were 0.5–1 m long and <3 cm thick at their base. Contents of food caches provided a good indication of winter diet and possibly food preference of beavers.     

Common ravens raid arctic fox food caches

Cache recovery is critical for evolution of hoarding behaviour, because the energy invested in caching may be lost if consumers other than the hoarders benefit from the cached food. By raiding food caches, animals may exploit the caching habits of others, that should respond by actively defending their caches. The arctic fox (Alopex lagopus) is the main predator of lemmings and goose eggs in the Canadian High Arctic and stores much of its prey in the ground. Common ravens (Corvus corax) are not as successful as foxes in taking eggs from goose nests. This generalist avian predator regularly uses innovation and opportunism to survive in many environments. Here, we provide the first report that ravens can successfully raid food cached by foxes, and that foxes may defend their caches from ravens.     

Ectoparasites,nest site choice and breeding success in the pied flycatcher

It has recently been suggested that nest box studies might bias the measurement of behavioural and life-history traits, because the removal of old nests may reduce the load of ectoparasites. This experimental artefact may have notable effects on nest site choice and breeding success in cavity-breeding birds. We tested (i) if pied flycatchers Ficedula hypoleuca prefer clean nest boxes and (ii) if old nest material affects the number of parasites and the breeding success of pied flycatchers. In the first experiment we offered birds one cleaned nest box and one nest box with old nest material from the previous year. The two nest boxes were placed in very similar sites near each other. In this experiment all pied flycatchers clearly preferred dirty nest boxes. In the second part of the study we distributed clean nest boxes and dirty ones on the same study area. After breeding was over we counted the number of fleas Ceratophyllus gallinae in the nest material. This flea species was the most abundant and probably the most hazardous parasite in the nests. Surprisingly, we found that there were significantly more fleas in the nest boxes with nests of the current year only than in the boxes with nests of both current and previous year. This might explain the preference for the dirty boxes. However, our results do not indicate that the number of fleas affects breeding success in the pied flycatcher.     

Substrate type affects caching and pilferage of pine seeds by chipmunks

The abiotic environment often influences the ways in which animalsinteract. By affecting the cues associated with buried seeds,the type of substrate used by seed-caching rodents may alterthe relative probabilities of cache pilferage and cache retrieval.We predicted that, after a wildfire, the presence of ash wouldimpair rodents' ability to smell pine seeds on the forest floor.In a laboratory experiment, we compared the foraging success,caching frequency, and cache recovery of chipmunks (six Tamiasamoenus and six T. quadrimaculatus) in ash versus sand substrates.Initial results supported our hypothesis: chipmunks found only2.3% of 108 caches of Jeffrey pine (Pinus jeffreyi) seeds thatwe buried in ash but found 98% of caches in sand. However, chipmunksmade as many or more of their own caches in ash compared withsand (48% for T. amoenus, 73% for T. quadrimaculatus.) Whenforaging for seeds cached in ash by themselves and by otherindividuals, they found significantly higher proportions oftheir own caches (62%) than of caches made by others (25%).However, when foraging in sand, they found high proportionsboth of their own caches and those of others (86 versus 81%).These results suggest that olfaction is less effective in ashthan in sand, that spatial memory enables chipmunks to recovertheir own caches in ash, and that caching in ash may allow animalsto avoid pilferage of stored food. As chipmunks are importantdispersers of seeds, changes in their foraging patterns or competitiveinteractions after fire could significantly affect pine regeneration.     

Tannin concentration enhances seed caching by scatter-hoarding rodents: An experiment using artificial ‘seeds’

Tannins are very common among plant seeds but their effects on the fate of seeds, for example, via mediation of the feeding preferences of scatter-hoarding rodents, are poorly understood. In this study, we created a series of artificial ‘seeds’ that only differed in tannin concentration and the type of tannin, and placed them in a pine forest in the Shangri-La Alpine Botanical Garden, Yunnan Province of China. Two rodent species (Apodemus latronum and A. chevrieri) showed significant preferences for ‘seeds’ with different tannin concentrations. A significantly higher proportion of seeds with low tannin concentration were consumed in situ compared with seeds with a higher tannin concentration. Meanwhile, the tannin concentration was significantly positively correlated with the proportion of seeds cached. The different types of tannin (hydrolysable tannin vs condensed tannin) did not differ significantly in their effect on the proportion of seeds eaten in situ vs seeds cached. Tannin concentrations had no significant effect on the distance that cached seeds were carried, which suggests that rodents may respond to different seed traits in deciding whether or not to cache seeds and how far they will transport seeds.     

Cache site selection by chipmunks (Tamias spp.) and its influence on the effectiveness of seed dispersal in Jeffrey pine (Pinus jeffreyi)

The effectiveness of Jeffrey pine (Pinus jeffreyi) seed dispersal performed by seed-caching yellow pine chipmunks (Tamias amoenus) and lodgepole chipmunks (Tamias speciosus) was compared to that of wind dispersal in the Sierra Nevada of western Nevada. Wind-dispersed seeds typically fall under or near the parent tree. Chipmunks removed 90 and 97% of 1064 radioactive seeds from each of two simulated wind-dispersed seed shadows in less than 24 h. Wind-dispersed seeds were deployed within 12 m of the two source trees, but chipmunk caches were found from 2–69 m from the trees. Chipmunks carried nearly all seeds away from source trees, greatly reducing the density of seeds under and near source trees. Caches contained from 1–35 seeds and most were buried 7–21 mm deep. Chipmunks cached in open bitterbrush shrubland with mineral soils much more than expected and cached in closed-canopy Jeffrey pine and lodgepole pine forests with thick needle litter much less than expected. Many Jeffrey pine seedlings and saplings grow in the bitterbrush habitat and few grow in the pine forests. Ten and 20% of the original caches survived until April, the time of seed germination, at the two sites. The movement of wind-dispersed seeds is random relative to environmental variables important in seedling survival, and the wind in coniferous forests cannot quickly bury seeds. The quality of seed dispersal rendered by chipmunks was superior to that provided by the wind because the chipmunks quickly harvested seeds on the ground, moved them away from source trees, and buried them in the ground in habitats and microhabitats where they were more likely to establish new seedlings. The increased quality of seed dispersal provided by animals relative to the wind may help explain why over twenty species of pines have evolved seeds and cones that are adapted for dispersal by seed-caching animals.     

PVC nest boxes are less at risk of occupancy by feral honey bees than timber nest boxes and natural hollows

Feral European Honey Bee (Apis mellifera) has been identified as a potential nest competitor for Australian hollow nesting species, but few studies have investigated the impact of feral honey bee competition on Threatened species. Our study used data from Glossy Black‐cockatoo (Calyptorhynchus lathami halmaturinus) nests on Kangaroo Island, monitored and managed over an 11‐year period, and found 12% of nests became occupied by feral honey bees during that period. Our results indicate that feral honey bees were less likely to occupy nest boxes made of PVC (5%) compared with wooden nest boxes (24%) or natural hollows in Eucalyptus trees (14%). The removal of feral honey bee hives from nests is a priority for long‐term conservation of glossy black‐cockatoos on Kangaroo Island. We recommend that PVC nest boxes are chosen for future nesting habitat restoration, due to the more frequent use of wooden nest boxes by feral honey bees.     

Impact of site‐induced mouse caching and transport behaviour on regeneration in Castanea crenata

Abstract. Dispersal and retrieval site selection by mice, transport distance, cache depth, and emergence and survival of seedlings of Castanea crenata (Japanese chestnut) were investigated by a magnet‐locating experiment in two habitat conditions (gap vs. forest understorey). Magnets were inserted into nuts (n= 450) and the nuts placed in the edge of forest gaps. Although wood mice (Apodemus speciosus and A. argenteus) initially buried nuts singly in shallow surface caches near the nut source, by the following spring these cached nuts were retrieved and re‐cached in larger, deeper caches farther from the source, particularly in forest understories, probably to reduce the threat of pilferage. All the nuts cached in the forest understories were consumed, but 4 seedlings emerged in gaps, apparently because of lower foraging activity in the gaps by the mice. Seed size was not correlated with cache depth or cache site selection. With increasing seed size, transport distance increased, particularly in gaps, possibly due to a greater potential energy gain (relative to handling cost to the cacher), or to attempts to prevent density‐ or mass‐dependent loss of caches by other foragers. Variable seed dispersal behaviour based on variation in seed size may influence the chances of colonization and distribution of the light‐demanding Castanea trees in mosaic landscapes and may play an important role in community organization and dynamics.