Introduction

Tree species produce and disperse seeds to regenerate and expand their distributed area. However, seeds are also a food resource for many granivorous and omnivorous animals. Many seeds produced by trees do not establish as seedlings, because they are subject to mortality from a range of sources, namely, predation by mammals and insects on the tree (e.g., Tanaka et al. 1989; Maeto 1995; Sone et al. 2002), predation by mammals after dropping to the ground (e.g., Shaw 1968; Watts 1968; Zemanek 1972; Sato 2000; Shimada 2001; Sone et al. 2002), and attacks on germinated acorns by mammals, insects, and microbes (e.g., Ueda et al. 1993; Sone et al. 2002). Therefore, seed predation and other forms of seed mortality during the early stages of the population process (i.e., the period from seed production to seedling establishment) may have a serious impact on the regeneration of tree species.

Some rodents are known to hoard surplus foods when they are temporarily supplied with abundant food, and are thought to contribute to seed dispersion and the regeneration of trees through hoarding activities (e.g., West 1968; Smith and Reichman 1984; Jensen and Nielsen 1986; Miyaki and Kikuzawa 1988; Vander Wall 1990; Tamura 2001; Takahashi et al. 2006). Because many rodents are nocturnal, it is very difficult to observe their hoarding behavior directly in the field. To date, many studies have been conducted on census methods for the hoarding behavior of rodents (e.g., Yasuda et al. 1991; Tamura 1994; Isaji and Sugita 1997; Sone and Kohno 1996, Xiao et al. 2006; Yi et al. 2008), and the processes causing the disappearance of dropped or artificially located/buried acorns have been traced using various methods to elucidate the role of these animals on seed dispersion and seedling establishment (e.g., Sone and Kohno 1996, 1999; Akashi 1997; Iida 2006; Takahashi et al. 2006; Xiao et al. 2006; Hirata et al. 2007; Yamagawa et al. 2010). However, most of these studies were conducted over a short period of 1–3 years.

The seed production of tree species belonging to Fagucea varies greatly from year to year, with cyclic mast cropping occurring at 3- to 7-year intervals (e.g., Kanazawa 1982; Kohyama 1982; Hashizume 1987; Imada et al. 1990; Takeda 1992; Harmer 1994). Therefore, long-term studies covering at least 2 or more cycles (i.e., 10–15 years) of fluctuations in seed production are required to identify the conditions needed for the successful regeneration of acorn-producing tree species, and to evaluate the role of the hoarding animals in the process of regeneration.

Pasania edulis (Makino) Makino (Fagucea) is a dominant tree species in many evergreen broad-leaved forests of southern Japan, in which two species of granivorous mice, namely, Apodemus speciosus Temminck and A. argenteus Temminck (Rodentia: Muridae), are distributed widely (Imaizumi 1960; Doi and Iwamoto 1982). P. edulis trees show major year-to-year variation in acorn production (Sone et al. 2002). Because of their large size (2–4 g: 5–10 cal acorn−1) and low tannin content (0.5 %), P. edulis acorns are the primary food resource of Apodemus mice during autumn and winter. In addition to Apodemus mice, the giant flying squirrel (Petaurista leucogenys) and Curculio weevils attack acorns on trees. Furthermore, dropped acorns are fed on by large or medium-sized omnivorous mammals such as wild boar (Sus scrofa leucomystax), badger (Meles meles anakuma), and raccoon dog (Nyctereutes procyonoides viverrinus), as well as the acorn borer (Coccotrypes graniceps) until seedling establishment (Sone et al. 2002). Among these predators, the two species of Apodemus mice not only feed on but also hoard acorns of P. edulis less than 5 cm in the soil (Sone and Kohno 1996, 1999; Sone et al. 2002), and their hoarding activity facilitates germination if not recovered or pilfered as Vander Wall (1990) pointed out. Consequently, it is possible that these mice not only interfere with but also facilitate the regeneration of P. edulis trees.

Sone et al. (2002) speculated on the conditions required for the regeneration of P. edulis and the role of Apodemus mice during the early stages of its population process. This work was accomplished through the life table analysis, using the data of P. edulis acorn crops from five cohorts, in parallel to a survey of the hoarding behavior of Apodemus mice. However, the five cohorts represent just one cycle in the fluctuation of acorn production. Hence, we continued the field studies, and calculated the life tables for 14 cohorts of acorns produced from 1995 to 2008, covering three cycles in the fluctuation of acorn production. In this paper, we describe the results of the life table analysis for the 14 cohorts of P. edulis acorns, and discuss the conditions and determinant factors required for successful regeneration of P. edulis and the role of Apodemus mice in this process.

Methods

Study site

This study was conducted at the Takakuma Experimental Forest of Kagoshima University (31°31′N, 130°46′E, and 550 m in altitude) in Kagoshima prefecture, southern Japan. Annual mean temperature and precipitation during the last decade were 14.3 °C and 3,460 mm, respectively. In May 1994, we established a study site of approximately 2 ha in size that was situated on a wide ridge, adjoining a north-facing slope in a secondary evergreen broad-leaved forest. The crown layer of the stand was closed, and P. edulis was the most dominant species and occupied about 50 % of canopy trees. Other dominant tree species were Castanopsis cuspidata, Machilus thunbergii, Neolitsea sericea, Distylium racemosum, Quercus acuta, and Quercus salicina. The understory vegetation was composed of Cleyera japonica, Symplocos lucida, Camellia japonica, and Illicium anisatum. A plantation of the Japanese cedar, Cryptomeria japonica, was situated to the south of the study site.

Aside from P. edulis acorns, some acorns of C. cuspidata and the seeds of Acer rufinerve and Cornus kousa dropped into seed traps almost every year. Although the acorns and seeds of these tree species are available as foods for granivorous mammals, they are much smaller and less abundant than P. edulis acorns (K.S., unpublished data). These findings suggest that at the study site, acorn crops of P. edulis are the most important food resource and primarily determine food conditions for granivorous mammals during autumn and winter.

At the study site, besides the two species of Apodemus mice, handling of acorns by M. meles, S. scrofa leucomystax, and N. procyonoides albus was photo-trapped frequently (K.S., personal observations). The feeding scar of P. leucogenys was recorded in a few occasions (Sone et al. 2002). The Japanese monkey (Macaca fuscata), the Japanese squirrel (Sciurus lis), and Sika deer (Cervus nippon Nippon) also eats acorns, but we did not observe them during the study period. The Eurasian Jay (Garrulus glandarius) and the large-billed crow (Corvus macrorhynchos) can handle acorns, but their acorn handling on the ground was photo-trapped on only few occasions. Therefore, the two species of Apodemus mice and large or medium-sized mammals were thought to be the most important predatory vertebrates for P. edulis acorns.

Mice populations

We set up a study plot measuring 65 × 55 m in the central part of the study site, where we established 56 trap stations at approximately 7-m intervals in a grid (8 rows × 7 columns). We conducted four or five successive night censuses in November 1995, and every month from July 1996 to May 2009. However, we stopped the census from mid-December to mid-March because 10–30 % of captured mice died due to a spell of cold weather on preliminary census occasions. We set a cage trap (9 × 10 × 17 cm) to catch mice with raw peanuts and sunflower seeds as bait and leaf litter and pieces of corrugated cardboard for protection of the captured mice against the cold at each trap station. Each trap was covered with a nylon sheet to prevent the captured mouse from being drenched. We checked the traps every morning. We identified each captured mouse by toe clipping at its first capture, and recorded the trap station where the mouse was captured, species, individual number, and various other characteristics. We then released the mouse at the trap station of capture. The captured mice were divided into two groups: (1) mice actively foraged on the study plot during the census period (the resident mice), and (2) mice captured occasionally. In this study, we defined resident mice in autumn and winter as those that were (1) captured regularly on more than half of census occasions during the acorn-dropping season and (2) captured both in the November/December and March/April censuses. During the 1995–1996 season, we could conduct the census only in November 1995 and July 1996. Therefore, we defined the mice captured on both census occasions as residents.

Acorn crop

We set 24 seed traps (50 × 50 cm) at 1 m height above the ground at 10–20 m intervals throughout the study plot in mid-September 1995. Each trap was supported with three wooden stakes that allowed the mice to access the trap. We counted the number of acorns dropped into each trap every week from early September to late November. On each census occasion, we marked newly dropped acorns with ink, and counted the number of acorns that had been marked to check for the foraging of acorns in traps or the carriage of acorns from traps by mice. Ink marking did not interfere with either consumption or transportation of the acorns by Apodemus mice (Sone and Kohno 1996, 1999). At the end of October of each year, when most acorns had dropped, we randomly collected 1,000–8,000 acorns from the ground. We categorized the collected acorns into 4 groups as follows; sound, abortion (failure to mature), foraged by mammals (mainly P. leucogenys) on the trees, and ingested by insects (mainly Curculio weevils), and counted the number of acorns within each group to estimate the mortality rate of acorns on trees due to each factor.

Transportation and predation of dropped acorns

We established four plots measuring 1 × 1 m in the northeastern, central, and southwestern parts of the study plot, respectively, in September 1995. We numbered newly dropped acorns with ink, and checked the presence or absence of each numbered acorn in each plot at approximately 1-week intervals until all sound acorns had disappeared. When the acorns were eaten on the plot, we recorded the acorn number on the remains of the nutshell and attempted to identify the predator species by using the teeth marks on the nutshells and photo-traps by a camera with an infrared sensor switch (Fieldnote II, Marifu, Iwakuni, Japan). Based on this information, we estimated the ratio of dropped acorns predated by Apodemus mice and large or medium-sized mammals on the plots and those transported by Apodemus mice from the plots.

To estimate the ratio of acorns that were hoarded after being transported by Apodemus mice, we established a food station in the same three areas in September 1995. Each food station was covered with an 8 mm wire mesh cage (30 × 30 × 20 cm) that was lifted 5 cm above the ground, to allow only Apodemus mice access to the acorns in the food station. To trace the fate of the acorns transported by Apodemus mice, we tagged miniature transmitters (9 × 7 × 20 mm; antenna length, 11 cm; weight, 2.1 g; model 393 or F515, ATS, Isanti, MN) on the seed coat of each acorn with a water-proof bond (Konishi, Tokyo, Japan) (T-acorns). The weight of the T-acorns ranged from 5.0 to 7.0 g. Sone and Kohno (1996) reported that the attachment of a transmitter to a P. edulis acorn did not interfere with the hoarding activity of A. speciosus. We placed two or three T-acorns together with 10–50 P. edulis acorns that were marked with ink at each of the 3 food stations. At 1–7 days later, we located the position of each T-acorn using a receiver (FT-690mk II, Vertex Standard, Tokyo, Japan) and recorded the condition of the acorn (eaten or hoarded). We traced the fate of 16–57 T-acorns each year.

Fate of hoarded acorns

Besides the trace of the fate of each T-acorn, we buried acorns artificially in the soil at the end of the acorn-dropping season (i.e., from late-October to late-November) every year from 1996 to 2008. At our study site, mice either hoarded single acorns in the soil just beneath the litter layer (scatter hoarding) or hoarded four or five acorns in the soil at 10–15 cm depth (larder hoarding) (K.S., unpublished data). Therefore, we selected 24 undergrowth trees in 1996, 40 trees in 1997 and 1998, 53 trees in 1999, and 55 trees in 2000–2008 for scatter and larder hoarding plots. We marked selected trees with colored tape at about 1.5 m height above the ground. The trees were distributed uniformly at approximately 7–10 m intervals at and near the study plot. For scatter hoarding, we cached a single acorn ≦5 cm deep in the soil at four hoarding sites located on the circumference of a circle, with a radius of 20 cm at 90° directional intervals around each selected tree. For larder hoarding, we buried five acorns 10–15 cm deep in the soil at a site located 20 cm from the trunk of the selected tree. We checked the survival of each acorn in each cache at 1–4 month intervals. If the cache site was dug up and vertical hole measuring greater than 10 cm in diameter remained, we considered the acorn(s) to be consumed by large or medium-sized mammals. If there was a horizontal or vertical tunnel measuring 3–4 cm in diameter at the hoarding site, we considered the acorn(s) to have been removed by Apodemus mice.

Data analysis

Based on data on acorn crops and acorn mortality due to each factor at each stage estimated by the methods of Sone et al. (2002), we calculated life tables for 14 cohorts of P. edulis acorns that were produced from 1995 to 2008.

To determine the mortality factor that contributed the most to the annual fluctuation in the total mortality from acorn production on the trees to seedling establishment (Total K), the “key factor”, we divided Total K into three stage mortalities as follows:

  • k 1: Mortality on trees. k 1 was further divided into three submortalities; the failure to mature (empty acorns) (k 11), the mortality due to the attack by Curculio weevils (k 12), and that due to the predation by mammals (mainly P. leucogenys) (k 13).

  • k 2: Mortality from dropping to germination. k 2 included mortalities due to the predation by Apodemus mice (k 21), that by large or medium-sized animals (S. scrofa leucomystax, M. meles, and N. procyonoides albus) (k 22), and rotting and/or other factors (k 23).

  • k 3: Mortality from germination to seedling establishment. k 3 included mortalities due to the predation by Apodemus mice (k 31), that by large or medium-sized mammals (k 32), the attack by the acorn borer (k 33), and unknown factors (k 34).

K-, k i-, and k ij-values were calculated as follows (Varley and Gradwell 1960):

  • K = log (no. of produced acorns on trees) − log (no. of seedlings established in the first season)

  • k i  = log (no. of acorns at the beginning of stage i) − log (no. of acorns at the beginning of stage i + 1),

  • k ij  = log (no. of acorns before the mortality factor j acted during the stage i) − log (no. of acorns after the mortality factor j acted during the stage i) = −log(1 − q ii)

where q ij is the mortality rate due to the mortality factor j during the stage i, K = k 1 + k 2 + k 3, k 1 = k 11 + k 12 + k 13, k 2 = k 21 + k 22 + k 23, and k 3 = k 31 + k 32 + k 33 + k 34.

During each stage of the population process, two or more mortality factors acted on acorns. In this study, we estimated the mortality rate due to each factor assuming that they acted simultaneously on the acorns.

In life table data analysis, we identified a key factor using the regression method presented by Podoler and Rogers (1975). In this method, k i- and k ij-values are plotted on the y-axis against K- and k i-values on the x-axis, respectively, and the mortality that provides the highest regression coefficient is designated as the key factor. The density relationship of each mortality was examined by calculating the correlation coefficient between K-, k i-, and k ij-value and the log-transformed acorn density.

Results

Annual changes in acorn crops of Pasania edulis

The number of acorns produced on trees varied greatly from 38,000 to 692,000 acorns ha−1 (Fig. 1). In this study, we tentatively classified acorn production into four classes as follows: mast (>400,000 acorns ha−1 were produced), semi-mast (about 300,000 acorns ha−1), medium (100,000–300,000 acorns ha−1), and poor (<100,000 acorns ha−1). According to this criterion, mast cropping was recorded in 4 years (1996, 1999, 2003, 2006), semi-mast cropping in 2002 and 2007, medium cropping in 4 years (1995, 1997, 2000, 2001), and poor cropping in 4 years (1998, 2004, 2005, 2008).

Fig. 1
figure 1

Annual changes in the numbers of resident Apodemus mice during autumn and winter (left y-axis) and dropped acorns (bars) (right y-axis)

Full size image

Annual changes in the number of resident mice

Two granivorous Apodemus mice, A. speciosus and A. argenteus, were captured on each census occasion. Two species of rodents (Microtus montebelli and Urotrichus talpoides) were caught on a few occasions.

The number of resident A. speciosus and A. argenteus mice during autumn and winter ranged from 1 to 16.5 and from 1 to 13.5 individuals, respectively (Fig. 1). The numbers of resident individuals of both species were highest during the 1996–1997 season, and showed a similar pattern in annual fluctuation. The total number of resident mice ranged from 2 to 30 individuals. If we tentatively classified the abundance of resident mice as follows; abundant (>15 mice), intermediate (6–15 mice), and low (<6 mice); the abundance of resident mice were categorized as abundant in the 1996–1997, 1997–1998, and 2007–2008 seasons, medium in the 1998–1999, 2000–2001, 2001–2002, and 2002–2003 seasons, and low in the 1995–1996, 1999–2000, 2003–2004, 2004–2005, and 2005–2006 seasons. Annual changes in the number of resident mice were not synchronized with the acorn crops (Fig. 1). In the mast years, the number of resident mice was not categorized as abundant, except for the 1996–1997 seasons. In years with a poor acorn crop, the number of resident mice was low, except for the 2008–2009 season. In years with a medium acorn crop, the number of resident mice varied with the years.

Mortalities from acorn production on trees to seedling establishment

Table 1 shows the life tables of the 14 cohorts of P. edulis acorns produced on trees from 1995 to 2008 until they were established as seedlings in the following year. In the subsequent sections, we present a detailed description of the mortality at each stage.

Table 1 Life tables for Pasania edulis acorns of 14 cohorts produced from 1995 to 2008
Full size table

Mortality on trees (k 1)

The mortality rate at this stage varied from 19.1 to 52.4 %. The primary mortality factor was abortion (failure to mature), which varied from 11.3 to 47.6 %. The other important mortality factor was predation by Curculio weevil larvae, and the predation rate varied from 4.8 to 23.8 %. The mortality rate due to other mortality factors such as predation by P. leucogenys was very low (0–2.1 %).

Mortality of dropped acorns until germination (k 2)

No or few acorns that dropped into the seed traps were predated or transported by animals. During the study period, 12,000–443,000 sound acorns ha−1 dropped to the ground. In total, Apodemus mice and large or medium-sized mammals predated 31.0–99.6 % of the acorns on the places where they dropped without being transported and hoarded by Apodemus mice. The predation rates by Apodemus mice and large or medium-sized mammals were subject to large variation among 14 cohorts, 0.8 to 99.6 % and 0 to 67.7 %, respectively.

The mortality rates of hoarded acorns were very high, i.e., 76.5–100 %. Predation by Apodemus mice and large or medium-sized mammals represented almost all of the mortality of hoarded acorns. Rotting and other factors also destroyed 0.04–2.7 % of hoarded acorns.

As a result, 97.6–100 % of the acorns were predated by these mammals before they germinated. Predation rates by Apodemus mice and large or medium-sized mammals during this stage ranged from 20.0 to 100 % and 0 to 78.4 %, respectively. The predation rate by large or medium-sized mammals was zero or very low (<3 %) in 3 of the 14 cohorts (cohort 1995, 2005, 2006).

Mortality of germinated acorns until seedling establishment (k 3)

The number of germinated acorns varied from 0 to 18,000 acorns ha−1. The mortality rate of germinated acorns ranged from 12.5 to 100 %. The primary mortality factor was predation by Apodemus mice, which accounted for 16.7–100 % of mortality. The other important mortality factor was the attack by the acorn borer, which destroyed 0–66.7 % of the germinated acorns. Predation by large or medium-sized mammals and unknown mortality factors contributed little to the mortality of germinated acorns (0 % for 8 and 9 out of 12 cohort, respectively).

As a result, 0 to 3,413 acorns ha−1 were established as seedlings during the study period. Of the 14 cohorts, 5 (cohort 1995, 1997, 1998, 2004, 2005) did not produce any seedlings, and 3 (cohort 2000, 2001, 2008) produced few seedlings (<100 seedlings ha−1) in the first growing season. In contrast, about 1,000 or more seedlings ha−1 were established from the cohorts of mast (cohort 1996, 1999, 2003, 2006) or semi-mast years (cohort 2007) in the first growing season.

Life table analysis

Among the stage mortalities, the mortality of dropped acorns before germination (k 2) showed the highest regression coefficient (0.9006) to Total K (Table 2). Among the submortalities of k 2, predation of dropped acorns by Apodemus mice (k 21) and that by large or medium-sized mammals (k 22) fluctuated compensatory to each other (Fig. 2). The regression coefficient of k 21 on k 2 (0.7535) was much higher than that of k 22 (0.0627), suggesting that k 21 contributed most and k 22 did not contribute to the annual fluctuations of k 2 (Table 3). Among the mortality factors, the regression coefficient on Total K was highest for k 21 (0.6450) (Table 3). These results indicated that the predation of dropped acorns by Apodemus mice before germination (k 21) was the key factor causing the annual fluctuations of Total K.

Table 2 Regression coefficients (r) for various k i- and k ij-values on Total K
Full size table
Fig. 2
figure 2

Annual changes in the mortality of dropped acorns due to predation by Apodemus mice (k 21) and that due to predation by large or medium-sized animals (k 22)

Full size image
Table 3 Regression coefficients for various k 2j -values on k 2
Full size table

Table 4 shows the density relationship of the total mortality (Total K), stage mortalities (k i), and mortalities due to main factors (k ij ). A negative correlation with acorn density was detected by Total K, k 1, k 2, and k 11. k 21 tended to act inversely with acorn density. A significant density relationship was not detected in k 22, k 3 and its submortalities (k 3j).

Table 4 Correlation coefficients (r) for mortality and main mortality factors on the density of P. edulis acorns
Full size table

Discussion

Mast cropping has been recorded by many tree species of Fagaceae (e.g., Kanazawa 1982; Kohyama 1982; Hashizume 1987; Imada et al. 1990; Takeda 1992; Harmer 1994). During the study period, the acorn production of P. edulis also varied greatly, and mast cropping was recorded at 3- or 4-year intervals. Abundance of seed production has great impacts on the regeneration of acorn-producing tree species and some pine species (e.g., Imada et al. 1990; Vander Wall 2002; Jansen et al. 2004; Sun et al. 2004; Garcia et al. 2005). Therefore, mast cropping might have considerable impact on the regeneration of P. edulis at the study site.

Acorns are subject to mortality caused by various factors from production on trees to seedling establishment. In this study, the mortality rate of acorns produced on trees before seedling establishment was very high, i.e., 99.7–100 %, for all 14 cohorts. The mortality rate was higher for dropped acorns than acorns on trees. A high mortality rate of dropped acorns was also reported for other Fagaceae tree species (e.g., Kanazawa 1982; Yamashita 1994; Akashi 1997).

The main mortality factors were the attack by Curculio weevil on trees, the predation by Apodemus mice and large or medium-sized mammals from dropping to germination, and the attack by the acorn borer from germination to seedling establishment. Among these four mortality factors, predation by Apodemus mice (k 21) caused the highest rate of mortality and was thought to have the greatest impact on the population process of P. edulis as reported for other acorn-producing tree species by Shaw (1968), Watts (1968), Zemanek (1972), Sato (2000), Shimada (2001), and others. Large or medium-sized mammals (k 22) were also responsible for the mortality of dropped acorns in some cohorts. Predation by these two groups of mammals showed compensatory fluctuation, indicating that these two groups competed for dropped acorns at the study site.

The analysis of life table data showed that the annual fluctuation pattern in the total mortality (Total K) was determined mainly by that of mortality due to the predation by Apodemus mice (k 21) and that the annual fluctuation of the predation of dropped acorns by the Apodemus mice should be a determinant factor of that of seedling establishment for P. edulis as Sone et al. (2002) had speculated. The tendency of k 21 to act in a negative density relationship suggests that more acorns could escape from predation by Apodemus mice and establish as seedlings in mast years. Curculio weevils, large or medium-sized mammals, and the acorn borer were responsible for the high mortality of acorns on trees, dropped acorns, and germinated acorns, respectively, in some years. However, these mortality factors acted on acorns in a density-independent manner. Therefore, although the predation by these animals occasionally led a high mortality of acorns, it was not a determinant factor in the annual fluctuation pattern of seedling establishment. The failure of acorns to mature (k 11) also correlated negatively with the number of acorns produced on trees, as reported by Tanaka et al. (1989). However, the regression coefficient of k 11 to Total K was much lower than that of k 21; k 11 seemed to contribute little to the annual fluctuations in the total mortality and seedling establishment.

Table 5 presents a summary of the acorn crop, the abundance of resident Apodemus mice during autumn and winter, and the probability of seedling establishment in the first growing season for the 14 cohorts of P. edulis acorns. A high probability of regeneration success, seedling establishment, was expected for the cohort 1999, 2003, and 2007. In these cases, acorn production was categorized as mast or semi-mast. Cohorts of poor and medium crop years produced few seedlings. These results show that mast cropping of acorns is necessary to produce a sufficient number of seedlings. A negative density relationship of Total K suggests that the more acorns produced, the more seedlings established. A large number of seedlings established in the season following mast years. This phenomenon has also been reported for other tree species (e.g., Imada et al. 1990; Vander Wall 2002; Jansen et al. 2004; Sun et al. 2004; Garcia et al. 2005).

Table 5 Summary of acorn production, abundance of resident mice, and probability of regeneration for the 14 cohorts of Pasania edulis acorns produced during the period from 1995 to 2008
Full size table

However, mast cropping did not guarantee the production of plenty of seedlings alone. A small population of Apodemus mice might be needed. Although the most abundant acorns (692,000 acorns ha−1) were produced in 1996, only 891 seedlings ha−1 were established in the first growing season and the probability of regeneration success was not expected to be high. In the autumn and winter of 1996/1997, the number of resident Apodemus mice was highest during the study period, and contributed to the high mortality of dropped acorns. For the successful regeneration of P. edulis, the control of predation pressure of Apodemus mice is also needed (Harmer 1994).

From these results, we conclude that, at this study site, mast cropping was a requisite but a small population of Apodemus mice, namely a low predation pressure was also necessary for the successful regeneration of P. edulis. At this study site, Apodemus mice reproduce only once a year from late autumn to early spring (Oishi et al. 2010), and the mice population could not increase immediately in the mast years as reported by Miguchi (1988) and Saitoh et al. (2008). This delayed increase in Apodemus mice population after mast cropping could produce a condition in which Apodemus mice acted as a seed disperser; mast cropping and a small population of Apodemus mice. For the acorn cohorts of years with poor and medium crop, Apodemus mice contributed little to regeneration of P. edulis trees irrespective of their population size, and acted as seed predators.