Introduction

The growth period of plants decreases with increasing altitude, in association with changes in air temperature and precipitation. Such altitudinal environmental changes affect plant growth along altitudes. Many studies investigated the effects of climate on radial growth of trees by using dendrochronological techniques (Cullen et al. 2001; Nöjd and Hari 2001; Peterson and Peterson 2001; Hopton and Pederson 2005; Fan et al. 2009). In a study of the change in growth–climate relationships along an altitudinal gradient, the tree-ring width of Picea abies (L.) H. Karst negatively and positively correlated with early summer temperature and precipitation, respectively, at a low altitude, but inversely correlated at a high altitude (Wilson and Hopfmueller 2001). Similar altitudinal changes in growth–climate relationships were also observed in other studies (Adams and Kolb 2005; Wang et al. 2005; Massaccesi et al. 2008; Peng et al. 2008), although growth–climate relationships vary at the same altitude according to topographical aspects (Liang et al. 2006). Therefore, the results of previous studies suggest that the factors limiting tree growth are drought stress at low altitudes and low temperatures at high altitudes on a meso scale.

Within a genus, plant species are often distributed at different altitudes. For example, Betula platyphylla var. japonica Hara is distributed in the upper part of the montane zone (~1,600 m a.s.l.) in central Japan, but this species is replaced by Betula ermanii Cham. in the subalpine zone above 1,600 m a.s.l. In this case, the altitude of 1,600 m a.s.l. is the altitudinal ecotone between the two species. The climatic conditions of an altitudinal ecotone are cold and moist for species of the upper distribution limit within its altitudinal distribution range, but in the same ecotone are warm and dry for species of the lower distribution limit within its altitudinal distribution range. Thus, species at the upper distribution limit may respond to climatic conditions differently from species at the lower distribution limit in an altitudinal ecotone. For example, growth of B. ermanii at the lower distribution limit is more reduced by drought stress compared with B. platyphylla var. japonica at the upper distribution limit in the altitudinal ecotone (Takahashi et al. 2003). Examination of growth responses to climatic conditions for both species at the upper and lower distribution limits of an altitudinal ecotone is important to understand the effects of global warming on altitudinal plant distributions. However, a few studies have compared growth responses to climatic conditions between tree species at their upper and lower distribution limits in an altitudinal ecotone.

In the subalpine zone on Mount Norikura, central Japan, Abies veitchii Lindl. and Abies mariesii Mast. are dominant conifers at the lower (1,600–2,000 m a.s.l.) and upper parts (2,000–2,500 m a.s.l.), respectively (Miyajima et al. 2007). The altitude around 2,000 m a.s.l. is the transitional vegetation ecotone of the two Abies species along the altitude. In the case of conifers, climatic factors affect not only tree-ring width but also maximum latewood density (Hughes et al. 1984; Briffa et al. 1998; Barber et al. 2004). Wide tracheids with a thin secondary wall are formed in the early growing season (earlywood), and then narrow tracheids are formed with a thick secondary wall (latewood). Wood density is wood mass per unit volume, which is usually greater in tracheids formed in the later growing season (Yasue et al. 2000).

In this study, we investigated the two Abies species at their upper and lower distribution limits on Mount Norikura by attempting to answer the following questions:

  1. 1.

    Are tree-ring width and maximum latewood density of the two Abies species more sensitive to low temperature at the upper distribution limit than at the lower distribution limit?

  2. 2.

    Are tree-ring width and maximum latewood density of A. veitchii at the upper distribution limit reduced by low temperature more than those of A. mariesii at the lower distribution limit of an altitudinal ecotone?

Materials and methods

Study site

This study was done in the subalpine forest zone on the east slope of Mount Norikura (36°06′N, 137°33′E, 3,026 m above sea level) in central Japan. The mean annual temperature recorded at Nagawa Weather Station (1,068 m a.s.l., approximately 12 km in horizontal distance from the summit) was 8.2°C between 1979 and 2005. The mean monthly temperatures in January and in August were −3.6 and 20.4°C, respectively. The mean annual precipitation was 1,968 mm.

Three major vegetation zones can be recognized along the altitude on Mount Norikura: a montane deciduous broad-leaved forest zone between 800 m and 1,600 m a.s.l., a subalpine coniferous forest zone between 1,600 m and 2,500 m a.s.l. and an alpine dwarf pine (Pinus pumila Regel) scrub zone between 2,500 m and 3,000 m a.s.l. (Miyajima et al. 2007; Miyajima and Takahashi 2007). The timberline is at about 2,500 m a.s.l. on the east slope of Mount Norikura (Takahashi 2003).

Dominant tree species are A. veitchii and Tsuga diversifolia Mast. between 1,600 m and 2,000 m a.s.l., and A. mariesii, B. ermanii, Sorbus commixta Hedl. and Sorbus matsumurana Koehne between 2,200 m and 2,500 m a.s.l. Anthropogenic effects on vegetation were negligible from 1,600 m a.s.l. to the summit.

Sampling and measurements

This study was done at three altitudes corresponding to the lower distribution limit of A. veitchii (1,600 m a.s.l.), the upper distribution limit of A. mariesii (2,400 m a.s.l.) and the altitudinal ecotone from A. veitchii to A. mariesii distributions (1,900 m a.s.l.). To include as long a common time period as possible, wood cores (5 mm in diameter) were taken from trunks of canopy trees (about 20 m in height) at breast height (1.3 m) at 2,400 m a.s.l. in 2003 and at 1,600 m and 1,900 m a.s.l. in 2006. The number of cores was between 36 and 44 for each species at each altitude (Table 1). The trunk diameter at breast height (DBH) measured for each tree was 27–61 cm (Table 1).

Table 1 Basic statistics of residual chronologies for Abies veitchii (Av) and A. mariesii (Am) at three altitudes on Mount Norikura in central Japan
Full size table

Increment cores were cut transversely into 1.5 mm thick strips by using a twin-bladed saw. The strips were oven-dried and then they underwent soft X-ray analysis at 20 kV and 14 mA for 210 s from a distance of 2.2 m; a calibration wedge was also used in the soft X ray analysis. The resultant radiographs were scanned at resolution 2,400 dpi. The tree-ring widths and maximum latewood densities were measured by using a WinDENDRO (Regent Instruments Inc., Quebec City, Canada).

Chronology development

All cores were cross-dated visually by matching characteristic wide and narrow rings that were synchronous within sample trees. Visual cross-dating was statistically verified by using the COFECHA program (Holmes 1983, 1994) that tests each individual series against a master dating series (mean of all series) from correlation coefficients.

Growth of trees is affected not only by climatic factors but also by age, disturbance and competition between neighboring trees. To reduce the variations caused by such non-climatic factors, all raw ring-width series were standardized by fitting smoothing splines (Cook and Peters 1981) with a 50% frequency–response cutoff of 40 years by using the ARSTAN program (Cook 1985; Holmes 1994). After standardizing each individual series, standard chronologies of the tree-ring width and maximum latewood density were obtained by averaging the standardized individual series in each year. We used at least five cores for each chronology in each year.

A standardized series usually shows autocorrelation that negates the assumption of independence necessary for most statistical analyses (Fritts 1976; Monserud 1986). To remove the effects of autocorrelation, we transformed each standardized series to a residual series through pooled autoregressive modeling by using the ARSTAN program.

Five statistics were calculated for residual chronologies: (1) mean correlation between trees represents the chronology signal strength; (2) mean sensitivity and (3) standard deviation are measures of interannual variation; (4) signal-to-noise ratio expresses the strength of the observed common signal among trees; (5) expressed population signal quantifies the degree to which a particular sample chronology portrays the hypothetically perfect chronology, which may in turn be regarded as the potential climatic signal. A threshold of 0.85 was empirically suggested as an acceptable statistical quality (Wigley et al. 1984).

Statistical analysis

Relationships of residual chronologies of tree-ring width and maximum latewood density with monthly climatic data were analyzed by using a bootstrapped correlation function for the common interval (1944–2001, Table 1) between the two Abies species at the upper and lower distribution limits by using the DENDROCLIM2002 program (Biondi and Waikul 2004). DENDROCLIM2002 uses 1,000 bootstrapped samples to compute correlation coefficients, and to test their significance at the 0.05 level.

Monthly mean temperature and the sum of precipitation data were used for the analysis. The nearest weather station to the study site was Nagawa (1,068 m a.s.l., approximately 11 km from the study site at 2,400 m a.s.l.). However, the available meteorological data at Nagawa started in 1979, whereas long-term records from 1898 were available at Matsumoto (610 m a.s.l., approximately 40 km from the study site at 2,400 m a.s.l.). Correlations of climatic data between Matsumoto and Nagawa stations were highly significant, i.e., correlation coefficients ranged between 0.89 (August) and 0.98 (April) for monthly mean temperature and between 0.59 (May) and 0.94 (November) for monthly precipitation. Thus, the general temporal trends of temperature and precipitation were consistent between Matsumoto and Nagawa, indicating that the climatic trend of Matsumoto represents that of the study region. Thus, we used long-term climatic data recorded at Matsumoto in this study for the bootstrapped correlation function.

The bootstrapped correlation function was done by using climatic data from the start of the previous growing season to the end of the current growing season because the growth of many tree species is affected not only by climatic conditions of the current year, but also by those of the previous year (cf. Eshete and Ståhl 1999; Takahashi et al. 2001; Speer et al. 2004). The approximate growing season at 1,600 m and 1,900 m a.s.l. was estimated as May to October, because the mean monthly temperatures exceeded 5°C, the effective heat for plant growth (Kira 1948), during this period, and so the correlation analysis was done using climatic data from May of the previous year to October of the current year (18 months total). The growing season at 2,400 m a.s.l. was estimated as June–September, so the correlation analysis was done using climatic data from June of the previous year to September of the current year (16 months total).

Results

Basic statistics

Wood cores were taken from the similar DBH range (ca. 30–60 cm) of the two Abies species at the upper and lower distribution limits (Table 1). However, the chronology range was considerably shorter for A. veitchii at 1,600 m a.s.l. than the other chronology ranges because they were young and the growth rate was high (Table 1, Fig. 1). Mean correlation coefficients between trees were 0.273–0.353 for the two Abies species at the upper and lower distribution limits. The mean sensitivity and standard deviation were lower in the four chronologies of the maximum latewood densities than those of the tree-ring widths (ex., 0.037–0.049 vs. 0.132–0.206 for mean sensitivity, 0.034–0.046 vs. 0.126–0.186 for standard deviation). The signal-to-noise ratio and expressed population signal were lowest for A. mariesii at 2,400 m a.s.l. among the four chronologies for both tree-ring width and maximum latewood density (Table 1). The expressed population signal was greater than the empirical threshold of 0.85, indicating a good estimation of the population value for all chronologies, except for the maximum latewood density of A. mariesii at 2,400 m a.s.l.

Fig. 1
figure 1

a–d Residual tree-ring width and e–h maximum latewood density chronologies and sample depths of a, e Abies mariesii at 2,400 m a.s.l., b, f A. mariesii at 1,900 m a.s.l., c, g A. veitchii at 1,900 m a.s.l., d, h A. veitchii at 1,600 m a.s.l. on Mount Norikura, central Japan

Full size image

Correlations between climatic factors and tree-ring widths

The tree-ring width chronology of A. veitchii at 1,600 m a.s.l. negatively correlated with July and August temperatures of the previous year and positively correlated with March and August precipitation of the previous year (Fig. 2). Responses of the tree-ring width chronology of A. veitchii at 1,900 m a.s.l. considerably differed from those at 1,600 m a.s.l. The tree-ring width of A. veitchii at 1,900 m a.s.l. positively correlated with May precipitation of the previous year and October precipitation of the current year and negatively correlated with July precipitation of the current year. Tree-ring width of A. veitchii at 1,900 m a.s.l. positively correlated with temperatures of 8 months between November of the previous year and October of the current year. Thus, tree-ring width of A. veitchii positively responded to temperatures more at 1,900 m a.s.l. than at 1,600 m a.s.l.

Fig. 2
figure 2

Correlation coefficients between tree-ring width chronologies and monthly climate data (temperature [left] and precipitation [right]) for Abies mariesii at 2,400 m a.s.l., A. mariesii at 1,900 m a.s.l., A. veitchii at 1,900 m a.s.l., and A. veitchii at 1,600 m a.s.l. on Mount Norikura, central Japan. Shaded bars indicate significant correlation (P < 0.05). PGS, DS and CGS mean previous-year growing season, dormant season and current-year growing season, respectively

Full size image

The responses of the tree-ring width chronology of A. mariesii at 1,900 m and 2,400 m a.s.l. were similar to those of A. veitchii at 1,900 m a.s.l. (Fig. 2), although the number of months with significant correlations was less in A. mariesii at 1,900 m and 2,400 m a.s.l. than A. veitchii at 1,900 m a.s.l. Especially, precipitation did not correlate with the tree-ring width chronology of A. mariesii at 2,400 m a.s.l. The tree-ring widths of A. mariesii at 1,900 m and 2,400 m a.s.l. positively correlated with temperatures of the beginning of the dormant season and ad summer temperatures.

Correlations between climatic factors and maximum latewood densities

The maximum latewood densities of the two Abies species at the upper and lower distribution limits negatively correlated with July temperature of the previous year, and positively correlated with temperatures between late spring and early autumn in the current year (Fig. 3). Correlation patterns of precipitation were the reverse of those of temperature for the two Abies species at the upper and lower distribution limits. The maximum latewood densities were positively correlated with summer precipitation in the previous year and negatively correlated with precipitation between late spring and summer in the current year. Thus, high temperatures and little precipitation between late spring and early autumn in the current year increased the maximum latewood densities of the two Abies species irrespective of altitude.

Fig. 3
figure 3

Correlation coefficients between maximum latewood density chronologies and monthly climate data (temperature [left] and precipitation [right]) for Abies mariesii at 2,400 m a.s.l., A. mariesii at 1,900 m a.s.l., A. veitchii at 1,900 m a.s.l., and A. veitchii at 1,600 m a.s.l. on Mount Norikura, central Japan. Shaded bars indicate significant correlation (P < 0.05). PGS, DS and CGS mean previous-year growing season, dormant season and current-year growing season, respectively

Full size image

Discussion

Responses of tree-ring widths

The responses of tree-ring width chronology of A. veitchii at 1,600 m a.s.l. differed from those of the three other tree-ring width chronologies at 1,900 m and 2,400 m a.s.l. Although A. veitchii sampled at 1,600 m a.s.l. was younger than A. veitchii at 1,900 m a.s.l. and A. mariesii at 1,900 m and 2,400 m a.s.l., this study developed tree-ring width chronologies assuming an age-invariant growth–climate relationship. However, some studies reported that such an assumption is invalid because growth–climate relationships change with tree age (Szeicz and MacDonald 1994; Rozas 2005; Rossi et al. 2008; Vieira et al. 2009). For example, the tree-ring width of young Juniperus thurifera is prone to respond to precipitation in a continental Mediterranean region, probably because of undeveloped roots (Rozas et al. 2009). In this study, positive correlation with summer precipitation was recognized for only A. veitchii at 1,600 m a.s.l. However, whether the root of A. veitchii at 1,600 m was less developed compared with the other trees examined at 1,900 m and 2,400 m a.s.l. is difficult to consider, because trunk height and the range of DBH of trees examined in this study were similar among the two Abies species at the upper and lower distribution limits. Carrer and Urbinati (2004) found that sensitivity of tree-ring width of Larix deciduas to climatic conditions was greater in old trees than young trees, and suggested that the high sensitivity of old trees was ascribed to the strong hydraulic resistance due to the tall trunk height. Thus, the effect of tree age on the growth–climate relationship among the four chronologies would be small in this study because of the similar trunk height among the two Abies species at the upper and lower distribution limits.

Probably, the different growth–climate relationship of A. veitchii at 1,600 m a.s.l. was mainly caused by altitudinal change in climatic conditions. Generally, precipitation is less and temperature is higher at lower altitudes. Especially, the precipitation of August (104 mm at Matsumoto) is less than the other summer months (July 130 mm, September 153 mm), which causes drought stress in August at 1,600 m a.s.l. (Takahashi et al. 2003). By contrast, the tree-ring width of A. veitchii at 1,900 m a.s.l. was positively correlated with temperatures between late spring and early autumn. Such altitudinal changes in growth–climate relationships were also observed in other studies (Adams and Kolb 2005; Wang et al. 2005). Thus, the different growth–climate relationship of A. veitchii is suggested to be ascribed to altitudinal change in climatic conditions.

The tree-ring width of A. mariesii positively correlated with July and August temperatures at 2,400 m a.s.l. and June and July temperatures at 1,900 m a.s.l. in the current year, like A. veitchii at 1,900 m a.s.l. Fujiwara et al. (1999) also observed that the tree-ring width of A. mariesii positively correlated with summer temperatures of the current year at about 2,000 m and 2,200 m a.s.l. on Mount Norikura. Cambial activity (cell division) is generally slow under cold conditions (Gricar et al. 2007). Therefore, tree-ring width is often positively correlated with summer temperatures at high altitudes (Buckley et al. 1997; Oberhuber 2004; Savva et al. 2006).

Tree-ring widths of A. veitchii (1,900 m a.s.l.) and A. mariesii at (1,900 m and 2,400 m a.s.l.) showed positive correlations with temperatures of the beginning of the dormant season (October to December) of the previous year. Although radial growth already finishes by October, high temperature is beneficial for photosynthetic production in this period. Thus, much photosynthetic production due to warm temperatures between October and December is suggested to increase tree growth in the next year.

Responses of maximum latewood densities

The maximum latewood densities of the two Abies species at the upper and lower distribution limits showed significant positive correlations with temperatures during the growing season of the current year. Maximum latewood densities also negatively correlated with precipitation during the growing season in the current year. However, this would not mean that much precipitation directly reduced maximum latewood densities because of negative correlation between temperature and precipitation during the growing season (Takahashi et al. 2005).

Cell division of the cambium of the two Abies species starts from early June to early July at 1,600 m to 2,400 m a.s.l. on Mount Norikura, and thickening of the secondary wall continues until mid-September (Koike 2009). Yasue et al. (2000) reported that interannual variation in maximum latewood density of Picea glehnii Mast. is due to interannual variation in cell wall thickness of the last-formed cells, which varies depending on the climatic conditions in summer. In this study, positive correlations of maximum latewood density with current-year April and May temperatures before the start of cell division were also found. It is well known that warm spring temperature cause earlier onset of shoot growth (Gostev et al. 1996), which might lead to higher photosynthetic production due to earlier development of current-year needles. Hence, more carbon might be available for cell wall growth and lignification of latewood cells. Therefore, maximum latewood density would respond to climatic conditions in late spring and during the growing season.

In this study, July temperature of the previous year was negatively correlated with maximum latewood densities of the two Abies species at the upper and lower distribution limits. Unfortunately, we do not have enough ecophysiological information to explain the relationships. Possibly, July temperature is not a causal factor if the temperature correlates with other climatic factors that correlate with maximum latewood densities. However, July temperature of the previous year did not correlate with other climatic factors, except for the positive correlations with temperatures of the other months during the growing season of the same year. However, these temperatures did not correlate with the maximum latewood densities of the two Abies species at the upper and lower distribution limits. Thus, the negative correlation of July temperature of the previous year was, at least, not indirectly caused by correlation with other climatic factors.

Conclusion

This study concludes that (1) tree-ring widths of the two Abies species above 1,900 m a.s.l. are increased by high temperatures of the previous-year late autumn and current-year summer, and maximum latewood densities of the two Abies species at the upper and lower distribution limits are increased by high temperatures between spring and summer and redued by high summer precipitation in the current year, (2) responses of tree-ring width and maximum latewood density to climate were similar between A. veitchii at the upper distribution limit and A. mariesii at the lower distribution limit of the altitudinal ecotone (1,900 m a.s.l.). Therefore, global warming is suggested to increase maximum latewood densities and tree-ring widths of the two Abies species, except for the tree-ring width of A. veitchii at the lower distribution limit. The issue whether global warming will alter chronologies of the two Abies species in old-growth stands is useful to predict effects of global warming on stand dynamics along the altitude.