Altitudinal and seasonal variation of frost resistance of Fagus crenata and Betula ermanii along the Pacific slope of Mt. Fuji, Japan


Dirk Gansert, Institut fuer Oekologische Pflanzenphysiologie und Geobotanik, Universitaetsstrasse 1, D-40225 Duesseldorf, Germany (fax ++49 211 81 13335; e-mail


1 Frost resistance of Fagus crenata (Siebold’s beech) and Betula ermanii (Japanese mountain birch) was investigated with respect to the species’ altitudinal distribution on the Pacific slope of Mt. Fuji from 1996 to 1997. Flint’s Index of Injury, which is based on electrolyte leakage from freeze-injured tissue, was used to assess frost hardiness of shoots produced in the previous growing season.

2Fagus crenata is found on the lower slopes (700–1600 m a.s.l.). Mid- to late November hardening of shoots was enhanced, midwinter damage below −30 °C reduced and dehardening delayed nearer the upper limit. To here temperatures began to rise at least 3 weeks before dehardening began. Shade crown shoots were more susceptible to deep-freeze damage than light crown shoots. If the ultimate upper distribution limit was determined by frost hardiness, F. crenata would be expected to occur up to 1800 m altitude.

3Betula ermanii is found between 1600 m and 2800 m, and intensive hardening occurred at all altitudes during the second half of October. Frost hardiness increased considerably with altitude up to the forest limit, where frost acclimation preceded the temperature decline by 2 weeks. Once maximum frost resistance had been attained freezing to −47 °C failed to cause tissue injury. Dehardening began slightly later at the tree line, but the time–course was the same at all altitudes. Main and lateral shoots did not differ in frost hardiness.

4 Comparison of monthly air temperature minima over the past 66 years with the course of frost resistance showed that F. crenata and B. ermanii found on the Pacific slope of Mt. Fuji were unlikely to suffer damage by frost.

5 The observed uppermost distribution limit for B. ermanii at 2800 m altitude on Mt. Fuji is considered both with our observations and with previous hypotheses.


In Japan, Siebold’s beech (Fagus crenata Blume) is the most dominant tree species of the cool-temperate deciduous broad-leaved forest zone. These Fagetea crenatae forests spread naturally over the northern half of Honshu and the southernmost part of Hokkaido (Ellenberg 1980; Miyawaki 1984, 1985; Sakai & Larcher 1987) between 700 m and 1600 m a.s.l. (Miyawaki 1984).

Japanese mountain birch (Betula ermanii Cham.) is a dominant deciduous tree in the mixed coniferous forest of subalpine regions, where it prevails at the tree line, commonly in co-existence with species of the genera Larix, Alnus, Salix and Rhododendron (Ohsawa 1984, 1990; Nakamura 1992). It can withstand severe cold (Sakai 1978; Sakai & Larcher 1987) and therefore also forms tree lines in Hokkaido, Sakhalin, the Kuril Islands and Kamchatka (Ohsawa 1990).

Mt. Fuji rises 3776 m from the Pacific coast of central Honshu and the altitudinal zonation encompasses patterns found in other Japanese forests. Deciduous broad-leaved forest is replaced above 1600 m by mixed subalpine coniferous forest, whose upper regions are characterized by Abies veitchii Lindl., Larix leptolepis Gord. and B. ermanii (Ohsawa 1984; Nakamura 1992). Subalpine deciduous forest, where B. ermanii is the dominant broad-leaved tree species, extends from 2300 m a.s.l. to the forest limit at 2500 m.

Clear skies in winter cause large daily temperature fluctuations and, except on the coastal plains, night frosts are severe here. Thus, frost hardiness determines the altitudinal distribution of F. crenata and B. ermanii in the montane and subalpine zones. Maximum frost hardiness of buds and twigs of B. ermanii was investigated by Sakai (1978), but its response to ambient low temperature and the seasonal variation of frost hardiness has not been investigated. Moreover, there have been no quantitative investigations of F. crenata for changes in its susceptibility to low temperature, a property that might determine the altitudinal distribution of this climax tree species.

The extent of and seasonal variation in frost hardiness were investigated for F. crenata and B. ermanii along an altitudinal transect on Mt. Fuji to enable a better understanding of the adaptive potential of a climax and a pioneer tree species to decreasing temperature. In this cold-determined Pacific winter climate, we assessed the upper distribution limits of F. crenata and B. ermanii in terms of the cold hardiness of the previous year’s shoots and disruption of the phenological cycle by spring and autumn frost.

Materials and methods

Study areas

Four study sites were established at different altitudes on the Pacific-facing south slope of Mt. Fuji (Table 1). The lowest plot, P1, was located in a natural deciduous broad-leaved forest of the upper montane zone where the climatic conditions are suitable for growth of F. crenata. Plot P2 was located at an altitude where the upper distribution limit of F. crenata intersects with the lower limit of B. ermanii. Plot P3 was established in the mixed subalpine coniferous forest where B. ermanii is the most abundant broad-leaved tree species and individuals vary considerably in size. The uppermost plot, P4, was located near the forest limit and had Larix leptolepis and B. ermanii as dominant trees. At this altitude B. ermanii characteristically has a multistem growth habit and the decrease in its average height (from 10 m to about 1 m) between 2380 m and 2500 m is marked.

Table 1.  Location of study sites and characterization of the tree species investigated (d.b.h. = stem diameter at breast height). Age was estimated by tree-ring counting
Plot no.Altitude (m a.s.l.)Co-ordinatesSpecies investigatedMax. height (m)Max. d.b.h. (cm)Estimated age (Years)Population code
138°41′32.7″ E
35°17′53.1″ NFagus crenata2175F1 
138°44′21.3″ E
35°18′46.6″ N
Betula ermanii
Fagus crenata
138°44′01.0″ E
35°19′30.8″ NBetula ermanii163885B3
138°44′30.3″ E
35°20′06.5″ NBetula ermanii516B4

The geological parent material of all the study sites is scoriaceous basaltic ejecta and volcanic rock (Tsuya et al. 1971). The pH(H2O) values of the soil range between 4.5 and 6. An intensively rooted organic layer and A-horizon are prevalent in both the mountain soil at higher altitude and the brown forest soil at the lower sites.

At each study site, five representatives of each of the study species present were selected as sample trees (Table 1). Where necessary stepladders were used to gain access to the upper crown.

Sampling and pretreatment of shoots

At monthly intervals (from October 1996 to June 1997) 20 shoots formed in the previous growing season (1996) were collected from each tree using a crossbow to sample from the outermost portion of the crown. Further samples formed in 1997 were taken on 30 October and 21 November 1997. Fagus crenata shoots were divided into those from light and shaded portions of the crown, and B. ermanii shoots into main and lateral shoots taken randomly from the upper crown. The samples were transferred to the laboratory in a cold box and stored at 3 °C overnight. The shoots of each species and type were combined after rinsing three times with distilled water and drying with soft tissue. The internodes of 2–3 mm in diameter were then trimmed into segments of 11 mm in length and mixed to give a composite sample.

Determination of frost hardiness

Frost hardiness was determined using the method of electrolyte leakage from freeze-injured tissues (Green & Warrington 1978; Hallam & Tibbits 1988; Silim & Lavender 1994). Based on preliminary tests, six internodal segments were placed in a screw cap glass vial and represented a single specimen. The vials were immersed in a pre-cooled bath of methanol–ethanol (1 : 1) at 0 °C and frozen to −10, −20, −30, −40 or −47 °C at a cooling rate of 5 K h−1: the final temperature was maintained for 30 min before removal. For autumn and spring samples (i.e. those taken before November or after March) a −5 °C treatment was included in place of −47 °C. Control (unfrozen) specimens were stored at 3 °C and frozen samples were returned to this temperature overnight. There were four replicate samples for each freezing treatment and the control at each sampling date.

Each thawed or control specimen was infiltrated with 5 ml of 3% propanol in distilled deionized water and evacuated in a desiccator for 4 h at 25 °C, before incubation in a water bath for 20 h at 25 °C in the dark. The electrical conductivity (EC) of the solution was then measured with a temperature-compensated platinum electrode after 1 day of infiltration (to give ECfrozen and ECcontrol). The shoots were killed by autoclaving for 5 min at 110 °C before incubation for a further 24 h and determination of total conductivity (ECtotal). Comparison with more drastic methods of tissue destruction and longer incubation showed that this method gave reliable estimates of ECtotal. EC values of samples (frozen and control) were corrected for conductivity changes of the pure propanolic water before (ECpw frozen) and after autoclaving (ECpw total) and, following Ritchie (1991), relative electrical conductivity (RC) was calculated as:


Index of Injury (It), derived from this according to the method of Flint et al. (1967), was used to quantify the degree of cold injury on a percentage basis as:


It-values ranging from 0% (i.e. no freeze damage) to 100% (i.e. tissue completely killed) were calculated for and averaged over the four replicates per treatment. Frost hardiness was expressed by plotting the It-values as a function of test temperature. For statistical analysis, the Mann–Whitney ranked sum test (U-test) was employed. Unless otherwise stated, significant values have P < 0.01.

Correlation with temperature

Daily records of air temperature at different altitudes were obtained from several stations on Mt. Fuji. Four meteorological stations, located on the west slope at 2350 m 1700 m, 1280 m and 500 m altitude, had data for 1970 onwards. Air temperature was measured 120 cm above ground level in meteorological shelters in the subalpine forest (highest station) or near the forest edge. Air temperature was recorded continuously at a site at 2400 m on the southern slope from April 1982 to March 1989 (Masuzawa et al. 1989, 1990). We recorded air temperature at 30-min intervals in the upper crown of one sample tree of each species in each of plots P1–P4, starting in July 1996. Similar data were available for 1990 onwards for a F. crenata tree in P1 (Y. Kakubari, unpublished data).

Mean and minimum temperature regimes at the altitudes of the study sites were calculated by linear regression analysis from corresponding daily records for the past 17 years obtained from the Mt. Fuji meteorological observatory, which is located at the summit. These regression lines were also used to calculate the site-specific absolute monthly minima from the corresponding minimum temperatures recorded at the summit since 1932.


Cold injury related to time and altitude

Fagus crenata

Until mid-November, F. crenata at 1624 m (population F2; Table 1) was more frost hardy than at the lower elevation (F1): It-values at the higher site were significantly lower for temperatures down to −30 °C (Fig. 1). Frost damage at −40 °C and below was similar at the two altitudes, with shade crown shoots showing stronger frost injury than those from the light areas.

Figure 1.

Seasonal course of the Injury Index (It-value) of previous year’s shoots of F. crenata (F.) and B. ermanii (B.) at their upper and lower distribution limits on the Pacific slope of Mt. Fuji. Fagus crenata data differentiates between shoots of light (l) and shade (s) crowns, and B. ermanii data between main (m) and lateral (l) shoots.

From mid- to late November, F2 underwent further frost hardening, with tissue injury at −30 °C approximately halved in both light and shade crown (P < 0.01). At −20 °C, F2 light crown shoots showed significantly less frost injury (It = 1.34%) than F1, where 10% of the tissue was killed. No damage was induced at either site by freezing down to −10 °C.

Maximum frost hardiness was achieved by the end of January. Although temperatures down to −30 °C had similar effects at the two altitudes, lower temperatures caused significantly (P < 0.05) higher tissue damage in F1 than in F2 shoots.

In early March, a significant increase in cold injury at −30 °C was observed for both F1 and F2. The It-values of light crown shoots increased from 12.3% (F1) and 10.9% (F2) to 29.5% at both altitudes. From then until mid-April, frost deacclimation was observed at both altitudes, but was more extensive at F1. F1 shoots showed significantly higher injuries at −10 °C than F2, where It-values were still less than 3% and considerable damage (41%) at −20 °C was about double that at F2.

Between April and May, the It-values of F1 shade crown shoots increased significantly at −10 °C and at −20 °C, but there was no further deacclimation of light crown shoots. At 1624 m, frost susceptibility of both light and shade crown shoots at these temperatures increased significantly, but the injury indices were less than at F1. Thus, higher frost resistance at the higher altitude was maintained until mid-May.

Frost susceptibility at up to freezing to −20 °C was similar for sun and shade crowns, except in May when shade crown shoots were more affected by freezing at −10 °C. However, from November onwards, freezing to −30 °C and below often caused significantly (P < 0.05) stronger tissue damage in shade crown shoots. This enhanced damage of shade crowns at lower temperatures is seen as a deviation from the correlation between shade and light It-values (open symbols; Fig. 2a).

Figure 2.

Correlation of tissue injury in the previous year’s shoots of the shade vs. the light crown of F. crenata (a) and lateral vs. main shoots of B. ermanii (b). It-values for F. crenata were grouped in two temperature ranges.

On 21 November 1997, frost hardiness was reassessed and compared with samples from similar altitudes on the north slope of Mt. Fuji. An additional −80 °C treatment was included in which samples frozen to −50 °C were transferred to a deep-freeze for a further 2 h.

It-values of F1 and F2 at the south slope (Fig. 3) were consistent with those in late November of the previous year (Fig. 1), indicating pronounced tissue hardening during mid-November. At 1600 m a.s.l. shoots from the north slope were slightly more resistant down to −20 °C than at the south slope at 1624 m. Freezing to −80 °C did not increase further damage, suggesting that maximum injury of about 60% had been induced at −40 to −50 °C.

Figure 3.

Comparison of frost hardiness of F. crenata (F.l = light crown, F.s = shade crown) and B. ermanii (B.) at corresponding altitudes on the north and south slopes of Mt. Fuji on 21 November 1997.

Betula ermanii

On 10 October, frost injury down to −20 °C was similar for B. ermanii (B2) and F. crenata (F2) at the site (1624 m) at which they both occurred. Progressively less damage at −20 °C was observed at higher altitudes (B3–B4) and a similar pattern was observed for the effect of −30 °C and −40 °C (Fig. 1). Frost acclimation in B. ermanii was therefore correlated with altitude, even in early October, with lowest It-values at the tree line.

By mid-November, B. ermanii at all altitudes was more or less completely frost hardy to −20 °C. The It-values at lower temperatures showed a pronounced frost acclimation of B. ermanii at its upper limit, where tissue destruction of 1.6–8.5% was half that at 2000 m, with the lower population again being the most injured.

By early December, little frost injury could be induced at any altitude by freezing to −30 °C. The It-values of B3 and B4 were less than 1%, although tissue damage varied considerably between replicates. Lower temperatures caused some damage that was more marked at lower altitudes, but even B2 shoots showed only 2.8% injury at −30 °C, increasing to 10% at −47 °C. Two months later, when frost resistance was at its maximum, freezing to nearly −50 °C failed to cause tissue damage, except to a limited extent in B2 (It = 5.9%).

Dehardening was observed at 1624 m at the end of April and at higher sites in samples taken a week later. Nevertheless, tissue destruction was still low (all sites <7% at −30 °C) and only showed >10% damage at −40 °C, indicating that frost hardiness is maintained more strongly at higher altitudes.

Considerable deacclimation had occurred by mid-May and the pattern was remarkably consistent along the transect. Only at −20 °C were there any significant differences in tissue injury (B2 higher). As in early October, B. ermanii at 1624 m showed similar frost resistance to F. crenata at the same site.

Between mid-May and early June, continued deacclimation was reflected in a doubling of injury at −10 °C at all altitudes. Damage at −20 °C to −40 °C (It = 41.8–45.3%) was greater at the tree line, suggesting that dehardening was more advanced here than at the lower altitudes.

No differences were detected at any time between main and lateral shoots (Fig. 2b).

When frost hardiness of B. ermanii was reassessed on 30 October 1997, dwarf individuals from above the forest limit (up to 2600 m) were also sampled. Damage at −5 °C to −40 °C was similar at all altitudes from 2000 m upwards, indicating a similar state of frost resistance over 600 m difference in elevation: no frost damage could be induced by freezing to −20 °C and limited damage at −30 °C was discernible only at 2000 m. Betula ermanii at 1624 m showed considerably higher tissue destruction. It-values at all sites were close to those in mid-November 1996, suggesting that intensive frost hardening took place at all altitudes during the second half of October.

When north and south aspects were compared in November 1997, no differences were observed up to −40 °C (Fig. 3) and, as in F. crenata, the It-values corresponded to those in late November and early December 1996. At 2000 m altitude frost susceptibility at −40 °C was slightly higher on the north slope (6.9%) than the south (5.6%) and both It-values were higher than in early December of the previous year (3.2%).

Freezing at −80 °C caused much less tissue destruction than in F. crenata. A significant increase compared with −40 °C was measured at 1600 m on both slopes, but at 2000 m and 2450 m only on the south slope. Thus, at high altitudes, B. ermanii on the north slope seemed to be even more resistant to extreme freezing than on the south slope of Mt. Fuji.

Cold injury related to air temperature

Monthly minimum air temperature and the temperature of the mildest freezing treatment (It5-temperature) are compared in Fig. 4. This illustrates that the viability of previous year’s shoots is unlikely to have been affected by the minimum ambient temperatures experienced during this study. Even in October and May, when frost resistance is low, the It5-temperatures were at least 10 °C below the air temperature minima.

Figure 4.

Comparison of the temperature that causes more than 5% tissue damage (It-value < 5%) in F. crenata (F) and B. ermanii (B) with the monthly minimum air temperature from October 1996 to May 1997.

Comparison of the altitude-corrected absolute monthly air temperature minimum for the past 66 years with the recorded minimum from August 1996 to July 1997 (Fig. 5), shows that the minimum of −25 °C observed at the forest limit on 22 January 1997 was an extremely cold event equalling the coldest known record. Moreover, none of the monthly minima from October 1996 to January 1997 at any altitude was more than 5 °C above the absolute minimum since 1932, indicating a colder than average period. In November, the minima were merely 1.3 °C (2000 m) to 3.5 °C (1624 m) higher than the coldest of the many years on record.

Figure 5.

Comparison of monthly minimum air temperature measured at the study sites from August 1996 to July 1997 with the altitude-corrected absolute monthly minimum, derived from the air temperature records from the summit of Mt. Fuji since 1932.

Minimum temperature data also suggest that shoot-damaging frost events are unlikely during September and May. In May, low temperature extremes at the summit were recorded only during the first week, and the lowest (−18.9 °C on 3 May 1934) corresponded to a minimum of about −5 °C at the forest limit. As −10 °C is required to induce <1% tissue injury at this time, natural frost injury is improbable. Severe frost, which generally occurred in late October and early April, was also insufficient to cause tissue damage. There is therefore strong evidence that, even at their present upper distribution limits on the Pacific slope of Mt. Fuji, F. crenata and B. ermanii were unlikely to have suffered significant frost damage to young shoots over the past 60 years or more.


Significance of frost injury for the upper distribution of beech and birch on mt. fuji

Fagus crenata at 1624 m altitude

The enhanced hardening of shoots observed between mid- and late November 1996 took place during a period when there were two sequences, of 5 and 7 days, with minor night frosts of down to −3.5 °C. Sakai & Larcher (1987) reported that natural hardiness of trees increases markedly when the daily minimum temperature falls to subzero for a week, as was the case here. The subsequent 5-day period (30 November 1996 to 4 December 1996) when minimum temperatures were between −6.9 °C and −9.3 °C did not therefore cause frost damage. A sudden drop in air temperature during late November and early December is characteristic of the climate on Mt. Fuji, to which F. crenata at its upper distribution limit is obviously well adapted, as shown on 21 November 1997 when freezing to −10 °C caused less than 1% of tissue damage.

Even at maximum frost hardiness, F. crenata showed a strong increase in It-values if the shoots were frozen below −30 °C. The method of differential thermal analysis (Quamme et al. 1972; Burke et al. 1976; Malone & Ashworth 1991) applied to artificially hardened 1-year-old twigs of F. crenata revealed a low temperature exotherm (LTE) at −40 °C that initiated at –32 °C (Sakai 1978; Sakai & Larcher 1987). This exotherm is attributed to the freezing of deep supercooled symplasmic water of the xylem ray parenchyma cells (Hong et al. 1980; Hong et al. 1980; George et al. 1982; Ashworth et al. 1983). This causes cell death, and the strong increase in the injury index is therefore most probably due to electrolyte leakage from destroyed xylem ray parenchyma cells and from the pith, which also exhibits LTE (Sakai & Larcher 1987). In Fagus, the pith is a major storage compartment and represents a large percentage of the parenchymatous cells in young shoots. The reduced frost damage observed within the LTE range suggests that these least hardy tissues in the inner shoot axis have developed a higher cold resistance at the species’ upper distribution limit.

The significant increase in cold injury to about 30% at −30 °C in early March indicates the onset of dehardening processes. However, it is not associated with an increased risk of injury, since at that time minimum ambient temperatures do not drop below −16 °C, a temperature at which complete hardiness is maintained. Kaku & Iwaya (1978) reported that in spring the initiation temperature of LTE in twigs of very hardy deciduous tree species ranged between −15 °C and −25 °C and intracellular freezing, with its potential risk of injury, can be precluded here. Shoots are not susceptible to damage by temperatures in the range of −12 °C to −16 °C until towards the end of April, suggesting that the course of dehardening is at least 3 weeks behind the gradual temperature increase.

In 1997, budburst was observed on 5 May and frondescence started on 13 May. As absolute air temperature minima in May over the past 66 years indicate that even mild frosts (to −2.5 °C) have been confined to the first week, F. crenata at its present upper distribution limit does not incur the risk of late spring frost damage. In 1997, frondescence was delayed by 1.7 days per 100 m ascent, which is in the range of 1–3.4 days 100 m−1 reported for F. sylvatica (Lausi & Pignatti 1973; Tranquillini & Plank 1989). Along the Pacific slope, over the past 17 years, the lapse rate of minimum temperature in May has been 1.05 °C 100 m−1, indicating that F. crenata will approach the risk of frost damage to unfolding leaves at 1800 m altitude. Here, temperature minima to −3 °C are possible during the most sensitive state of the phenological cycle.

Kira (1949) described the distribution limits of Fagetea crenatae forests by introducing a ‘Warmth Index’[WI = Σi=1, . . . , n (ti−5), month-degrees °C], where t is the monthly mean air temperature and n the number of months with mean temperatures above 5 °C. He postulated an index of 45 for the upper distribution limit of F. crenata. Altitude-corrected monthly mean temperatures throughout the past 25 years indicate that this value corresponds to the same limit (1800 m) as predicted by our frost susceptibility data, and suggest an alternative explanation for the distribution.

Betula ermanii at the present forest limit

The state of frost hardiness in early October 1996 and at the end of the month in 1997 indicate that frost acclimation at the forest limit precedes the temperature decline by at least 2 weeks. During October, the injurious temperature range for the previous year’s shoots of B. ermanii up to 2600 m altitude is lowered from at least 7 to about 22 degrees below the natural minimum temperature regime. Thus, young shoots are well hardened to withstand the frequent frost (to −6.5 °C) that has been recorded between 16 and 27 October over the past 17 years.

In contrast to F. crenata, no strong increase in tissue damage at a given freezing level was observed during this study. This finding is consistent with the lack of a low temperature exotherm in dormant 1-year-old twigs (Sakai 1978; Sakai & Larcher 1987), from which it was concluded that B. ermanii hardens by extracellular freezing and tolerates freeze-dehydration (Li & Sakai 1982). In stems of B. alleghaniensis seedlings, Calméet al. (1995) reported an inverse ratio of increasing frost tolerance to decreasing water content during the frost-hardening process. Although the water content of the shoots of B. ermanii was not measured, we observed that the shoots shrivelled conspicuously and became soft and pliable, pointing to water deprivation during the hardening period.

In 1997, bud swelling began on 10 May, during a warm period with daily maxima up to 18.4 °C and minimum temperatures of −2.7 °C; this year was warmer than the many in which a low temperature regime down to −6 °C persisted during the first week of May. The warm period continued until 19 May, when the buds were in various phenological stages, most of them still being closed. Studies of budburst in B. pubescens and B. pendula prove that warm periods in early spring considerably reduce the thermal time requirement (accumulated day degrees >0 °C) for budburst and thus advance budburst (Heide 1993). Swelling and bursting of buds dramatically increase frost sensitivity and therefore severe freezing after a warm period represents the greatest risk of frost damage. Coincidentally with the onset of budburst, a cold period of 9 consecutive frost days, with an absolute minimum of −3.3 °C on 22 May 1997, caused an interruption in budburst for 4 weeks. Data from the past 17 years indicate that frost to −3 °C until late in May is a common climatic feature at the forest limit. Although there is no experimental data as to whether burst buds of B. ermanii might be injured by such temperatures, we did not observe frost damage. Leafing started on 17 June, by which time frost has ceased at 2450 m altitude, and was completed on 30 June. Cold periods in May, such as those recorded in 1997, occur regularly and it is therefore assumed that the observed phenological course was representative for B. ermanii at the forest limit. Spring frost damage to B. ermanii is therefore unlikely unless burst buds are vulnerable.

Betula ermanii at 2800 m altitude

On the Pacific slope of Mt. Fuji, the uppermost dwarf individuals of B. ermanii, less than 50 cm tall, grow at 2700 m altitude. On the north and west slope, however, B. ermanii ascends to 2800 m altitude (Ohsawa 1984; Masuzawa 1985), which corresponds to the upper distribution limit in the Southern Alps of Honshu. The observed budburst at 2000 m and 2450 m altitude in 1997 indicates a delay of about 2 days 100 m−1 ascent, suggesting that budburst at 2800 m altitude will begin towards the end of May. The lapse rate of minimum temperature of 1.05 °C 100 m−1 in May, suggests that frost to −7 °C is likely to occur at 2800 m altitude until late in this month, and frost damage in dehardened buds therefore becomes a serious risk for the survival of B. ermanii at this altitude. Moreover, if the level of susceptibility is similar to that measured at bud-break at 2450 m, up to 6% tissue damage might be caused to the young shoots by frost to −7 °C.

How does the altitudinal distribution limit of B. ermanii at 2800 m on Mt. Fuji compare with two hypotheses relating tree lines to monthly mean temperatures? Kira (1949) postulated that, in Japan, tree lines will form at altitudes with a warmth index of 15 (Ohsawa 1990). On the Pacific slope, the mean warmth index at the present forest limit appears to be 26 (from continuous temperature measurements throughout the years 1981–88; Masuzawa 1985; Masuzawa et al. 1989, 1990; and for 1996–97, present authors, unpublished data). However, an average warmth index of 16.8 calculated for 2800 m from altitude-corrected monthly mean air temperatures recorded at the summit throughout the past 25 years is quite similar to Kira’s index. Alternatively, Brockmann-Jerosch (1919) suggested that a mean air temperature of c. 10 °C in July was a major determining climatic factor for the formation of tree lines. We calculated a mean temperature of 10.4 °C in July at 2800 m altitude, averaged over the past 25 years (assuming a lapse rate of 0.6 °C 100 m−1 ascent) that was in accord with Masuzawa (1985). Thus, both postulations suggest an altitudinal distribution limit for B. ermanii at 2800 m on Mt. Fuji under present climatic conditions, consistent both with its observed occurrence and with our predictions from investigations of frost resistance and phenology.


The authors are gratefully obliged to the German Alexander von Humboldt-Stiftung and the Japan Society for the Promotion of Science, who generously granted a 2-year research scholarship in Japan. We are also indebted to the Fuji Sabo Work Office, Chubu Regional Construction Bureau, Ministry of Construction, Japan, and the Meteorological Service Station of Shizuoka for providing the many years of weather records for Mt. Fuji.

revision accepted 17 September 1998

Received 28 April 1998