1. The effects of previous-year environment on current growth response were tested in seedlings of Mountain Birch (Betula pubescens Ehrh. ssp. czerepanovii[Orlova] Hämet-Ahti), a subarctic tree species with indeterminate shoot growth.
2. Mountain Birch seedlings were pot-grown outdoors in subarctic Sweden for 2 years and showed a clear delay in growth response when fertilization and temperature were reduced after the first year. The seedlings were grown under four experimental treatments (two temperatures and two nutrient availabilities) in 1994, and under low-temperature/low-nutrient conditions in 1995.
3. When nutrient supply and/or temperature was reduced in 1995 compared to 1994, the seedlings maintained the high relative growth rates (RGR) of the previous growing season, although the internal plant N accumulation rate was lower than in 1994. This resulted in decreasing plant N concentration (PNC), and a poor relationship between RGR and PNC during 1995. The high RGR in 1995 was achieved in response to phenotypic adjustments (e.g. number of foliar buds) to a more favourable environment in the past, and by dilution of the internal nutrient storage.
4. The effects of delayed responses found in this study indicate problems for the interpretation of results from growth studies performed under any climate with great year-to-year variability, such as the subarctic, because a delay in growth response could distort the relationships between plant growth, resource availability and climate. Predictions of current growth are therefore meaningless if the previous history of the plants is not taken into account.
Organisms cannot predict future environments and are not matched to the present or future circumstances. Rather, their properties are entirely consequences of the past (‘ab-aptations’; Begon, Harper & Townsend 1996). For example, in tree species with determinate shoot growth the number of buds determines growth potential during the current year, but reflects the environmental conditions of the previous year when the buds were formed (Kramer & Kozlowski 1979). Similarly, phenotypic adjustments in terms of root versus shoot allocation in plants may be the result of adaptations to past environments where the supply of either nutrients or carbon had been the major limiting factor (resource balancing; Bloom, Chapin & Mooney 1985). These examples indicate adjustments of the phenotype to past environments and could predispose plants to delay their growth responses to current environmental conditions. Delayed responses may, in many cases, affect the interpretation of results from growth studies. Thus adjustments of the phenotype, such as allocation patterns and the formation of foliar buds in plants, may turn out to be more or less disadvantageous if the environmental conditions change in an irregular, ‘unpredicted’ way. For example, delayed responses may result in disproportional increase of the biomass and N content of plants and, consequently, in changing plant N concentration over time. In contrast to plant growth under steady-state conditions (cf. Hirose 1988; Ingestad & Ågren 1995), these plants may not adjust their growth rate according to the internal N accumulation rate, and such a growth pattern will be referred to as unbalanced. Growth imbalances due to delayed responses are probably typical features of most plants growing naturally under temperate-arctic climates. They could be particularly common in plants within the subarctic, where the climate is characterized by particularly strong environmental unpredictability and great year-to-year variability (Blüthgen 1952; Ferguson & Messier 1996). Furthermore, growth rates are generally low under the cold climate of the subarctic and it may take plants longer than those under a warm climate to erase any adjustments to past environments.
Mountain Birch (Betula pubescens Ehrh. ssp. czerepanovii[Orlova] Hämet-Ahti) is an abundant tree species in large parts of the European subarctic. It is regarded as a species with typical indeterminate shoot growth, and the growth rate of Mountain Birch is strongly modified by various environmental factors (Junttila & Nilsen 1993; Karlsson & Weih 1996). Weih & Karlsson (1997) studied the growth of Mountain Birch seedlings exposed to different experimental treatments during 2 years differing in climate. We found indications of significant resource balancing in response to interannual climate variation, and of delayed growth responses. After-effects of previous treatments on growth rates of plants are occasionally reported (Richards & Caldwell 1985; Shaver & Chapin 1995), and delayed responses of growth rate to a change in nutrition are discussed, especially for species with determinate shoot growth such as Pinus sylvestris (Fagerström & Lohm 1977; Karlsson & Nordell 1987, 1989). However, little is known about the magnitude of the influence that such ‘memory’ effects could have on growth rates of tree species with indeterminate shoot growth, in particular when growing slowly under a cold climate.
The effect of phenotypic memory on growth rate of Mountain Birch was tested, and an experiment performed in which birch seedlings were grown under four different treatments in their first year and under identical treatments in their second year. Compared to the environmental conditions during the first year, the treatments of the second year implied a deterioration in environmental conditions in most cases, except one control treatment. The hypothesis was tested whether the previous year's environment could affect the current growth response of Mountain Birch seedlings in terms of changed relationships between the growth rate and internal N accumulation rate and, in general, plant growth, climate and resource availability.
Materials and methods
The experiment was conducted outdoors in the vicinity of the Abisko Scientific Research Station, northern Sweden (68°21′ °N, 18°49′ °E, 380 m above sea level; Andersson, Callaghan & Karlsson 1996). The mean air temperature at Abisko is −0·8 °C at the annual average and 9·7 °C during the growing season (June to August); the annual precipitation sum is 304 mm (SMHI 1961–90). Due to the proximity of the Atlantic Ocean, the climate of the Abisko region is warm in comparison to other areas at the same latitude, and is characterized by large year-to-year variations in temperature.
Seeds of Mountain Birch were randomly collected within the local population at Abisko. The seeds were sown on moist filter paper on 17 May 1994 and grown for 2 weeks in the greenhouse until the beginning of the experiment. The experiment lasted from 5 June 1994 until 15 August 1995. All seedlings were pot-grown in peat during the experiment.
During the 1994 growing season, the seedlings were grown under four experimental treatments including two temperature regimes, ambient and elevated temperature, and two fertilization regimes (Table 1). The seedlings were fertilized weekly and exposed to natural fluctuations of temperature and light. The two fertilization treatments supplied each seedling with ≈ 12 (low) and 120 (high) mg N per season, and simulated nutrient conditions in a Mountain Birch woodland from poor to very rich in nutrients (Weih & Karlsson 1999). The composition of the fertilizer solution and the design of the elevated-temperature treatment (open greenhouse) were adopted from Weih & Karlsson (1997). The seedlings were kept outdoors from September 1994 until August 1995, and all seedlings were replanted in fresh peat after the 1994/95 winter. All seedlings were exposed to ambient temperature and were fertilized with the low nutrient dose during the 1995 growing season. Mean ambient air temperatures from June to August of 1994 and 1995 were 9·4 and 9·2 °C, respectively.
Table 1. Summary of experimental conditions for four treatments during the 1994 and 1995 growing seasons. Experimental conditions differed among treatments during 1994 only. TL temperature low; TH temperature high; FL fertilization low; FH fertilization high
Soil temperature (°C)
Fertilization (g N m−2 year−1)
Soil temperature (°C)
Fertilization (g N m−2 year−1)
Fourteen replicate plants per treatment were assigned to each of four harvests throughout the 1994 and 1995 growing seasons. Eight seedlings were lost during the 1994/95 winter due to animal browsing. The harvests were carried out at the start of the growing season (early June) and after the main growing season in mid-August (peak biomass, Karlsson & Nordell 1989) of 1994 and 1995. All harvested plants were fractionated into leaves, stems and roots. Oven-dried leaves, stems and roots were weighed and their total N content determined (Weih & Karlsson 1997). The content of total non-structural carbohydrates in leaves of the August harvest 1995 was determined using the p-hydroxybenzoic acid hydrazide method (Cottrell & Dale 1986; Lever 1972). Buds of the seedlings assigned to the 1995 harvests were counted in September 1994.
In this experiment the periods between the initial and final harvests of each year represented periods of continuous growth, and functional growth analysis was used to compare seedling growth and N economy between the two harvests (Hunt 1982). The growing season relative growth rates (RGR) and N accumulation rates (RNA) of the seedlings were calculated from the differences between the treatment means of the loge-transformed biomass and N content at initial and final harvest, divided by the time difference between the harvests. Four-way analysis of variance (anova GLM procedure, SPSS version 7·0 for Windows, SPSS Inc., Chicago, IL) was used to assess the effects on plant biomass, N pool size and N concentration of the factors year, time (of year), temperature in 1994 and fertilization in 1994, and the interactions between them.
Over all treatments, the mean RGR and RNA of the seedlings was slightly lower during 1995 than 1994 (Table 2[time × year]; Fig. 1). In parallel, the shoot length growth generally was greater during 1995 than 1994, and more so when the seedlings were exposed to increased temperature during the first year (Fig. 2a). The biomass increment during the 1995 growing season was closely related to bud number during autumn 1994, but not to the mean plant N concentration (PNC) during the 1995 growing season (Fig. 3).
Table 2. anova for effects of time, year, temperature in 1994 and fertilization in 1994 on total plant biomass (loge; mg), plant N pool (loge; mg), plant N concentration (PNC, %), root as well as shoot and leaf biomass (loge; mg) of Mountain Birch seedlings grown in northern Sweden during the 1994 and 1995 growing seasons. Time × factor interaction effects on biomass and N pool indicate factor effects on relative growth rates and relative N accumulation rates, respectively (Poorter & Lewis 1986). SS, sums of squares; df, degrees of freedom; P, significance level
Plant N pool
Shoot + leaf biomass
Source of variation
Within + residual
Time × Year
Time × Temperature 94
Time × Fertilization 94
Year × Temperature 94
Year × Fertilization 94
Temperature 94 × Fertilization 94
Time × Year × Temperature 94
Time × Year × Fertilization 94
Time × Temperature 94 × Fertilization 94
Year × Temperature 94 × Fertilization 94
Year × Time × Temperature 94 × Fertilization 94
Under similar experimental conditions during both years (TLFL in Table 1), the RGR, RNA and PNC of the seedlings were similar (Fig. 1a). The temperature and fertilization treatment during the first year significantly affected the RGR and RNA of the seedlings during both years (Table 2[time × temperature 94] and [time × fertilization 94]). For RGR, no significant differences in the magnitude of treatment effects were found between the two years (Table 2[time × year × temperature 94] and [time × year × fertilization 94]). In contrast, the effect of the 1994 fertilization treatment on RNA varied between 1994 and 1995 (Table 2[time × year × fertilization 94]). The effects of both 1994 treatments on PNC varied between the two years (Table 2[year × temperature 94] and [year × fertilization 94]). Thus PNC decreased during the second year when the seedlings had been well fertilized during the first year (Fig. 1b, d).
The RGR of roots was generally lower during 1995 than 1994 (Table 2[time × year], Fig. 2b). However, the RGR of roots was similar between the two years when nutrient availability had been reduced 1995 compared to 1994 (Table 2[time × year × fertilization 94], Fig. 2b). In the same treatments (TLFH and THFH), RGR of shoots and leaves was significantly lower in 1995 than 1994 (Table 2[time × year × fertilization 94], Fig. 2b).
In August 1995, mean ± SE contents of total non-structural carbohydrates in leaves were 3·6 ± 0·4 (TLFL), 3·3 ± 0·6 (TLFH), 4·7 ± 0·8 (THFL) and 3·4 ± 0·5 g m−2 (THFH); no significant differences were found between treatments (two-way anova, P > 0·050, n = 6).
Seedlings of Silver Birch (Betula pendula), grown at 20 °C in a growth chamber for 2 months, showed no signs of delayed response in growth rate after a strong stepped decrease in nutrient availability (McDonald, Lohammar & Ericsson 1986). In fact, within less than a single growing season seedlings of tree species with indeterminate shoot growth rapidly adjust their growth rate to changed environmental conditions, as numerous studies have indicated (Ericsson 1995; Junttila & Nilsen 1993; Weih 1998). However, the seedlings in the present study were allowed to go through an entire growth cycle, including bud formation and resorption of resources from senescing leaves, before nutrient availability and/or temperature conditions were changed compared to the previous year. Internal cycling of nutrients and carbohydrates to storage during autumn and winter has been documented for many trees (Chapin & Kedrowski 1983; Millard & Neilsen 1989). A further increase in the nutrient content of plants may occur due to winter N uptake (Weih & Karlsson 1997). After the 1994/95 winter, the growth potential of the seedlings during the growing season was apparently determined by phenotypic adjustments (e.g. number of leaf buds; Fig. 3) according to the previous history of the plants. Although N availability was strongly reduced during 1995 compared to 1994 in the FH treatments, the plants maintained the high growth rates of the previous year and might have relied on the internal reserves of carbohydrates and mineral nutrients.
The relatively high contents of non-structural carbohydrates in leaves in August 1995 suggest that there was no serious carbon limitation in any of the treatments at the time (Stitt & Schulze 1994; Weih & Karlsson 1997 for comparison). In the treatments comprising a reduction in nutrient supply during 1995, below-ground growth was given priority at the expense of above-ground growth in 1995 (Fig. 2b). This may indicate that during 1995 growth of the seedlings exposed to reduced fertilization was limited more strongly by nutrient availability than of seedlings in the other treatments (Bloom et al. 1985; Ericsson 1995). The strong nutrient limitation of growth in the seedlings of the FH treatments was also reflected by the increasing dilution of the internal N and unbalanced growth, which was indicated by the lack of relationship between RGR and PNC. Nevertheless, these seedlings had apparently accumulated sufficient resources for the RGR of the previous year to be almost maintained during 1995.
A small decrease in RGR and an increase in shoot growth were found in 1994 compared to 1995 for all treatments. These may be partly attributable to increasing proportions of the biomass allocated to support structures as seedlings grew larger (Pollard & Wareing 1968; Walters, Kruger & Reich 1993).
The mechanisms responsible for the delayed response in seedling growth may be connected to control of growth potential by bud demography and also hormonal aspects, but were not further investigated in this study. Nevertheless, the growth response of plants to improved environmental conditions, which is rapid and immediate in species with indeterminate growth, might be very different from the growth response to a deterioration in environmental conditions. Thus the response to improved conditions is an opportunistic feature, in order to rapidly exploit periods favourable for growth. In contrast, the response to deteriorated conditions could be primarily a response of plants to frequent periods of environmental stress, i.e. low-resource conditions. Delayed responses could be a way for stress-tolerant plants to cope with environmental unpredictability, because they could buffer current growth from extremes of weather. Hence the delayed response in growth rate to environmental deterioration might be advantageous to the growth of plants as long as periods of low-resource conditions are temporary. In this sense the delayed response in growth rate might be regarded as part of a suite of plant traits in response to temporary environmental stress (Chapin, Autumn & Pugnaire 1993). But if periods unfavourable for growth persist during a longer period, as simulated in this study, seedling growth probably would have suffered severely from nutrient limitation, and the delayed response of growth rate would inevitably result in strong growth reductions in the following years. It is an interesting question whether the delayed response of growth rate seen especially in the THFH treatment of this study could impair seedling growth performance to such an extent that recovery to ‘normal’ growth rates would take several years. If so, then delayed responses could be clearly disadvantageous for plant growth performance in the longer term.
In general, the dilution of internal resource pools should be accelerated in rapidly growing seedlings and retarded in slowly growing seedlings. Seedling RGR in this study ranged between 0·03 and 0·06 day−1. This range is low compared to the RGR of deciduous tree seedlings (including the congeneric B. pendula) grown under a warm-temperate climate (Cornelissen, Castro Diez & Hunt 1996) and represents the upper margin of RGR for Mountain Birch seedlings grown naturally in the subarctic (Weih & Karlsson 1999). It is therefore suggested that delayed responses similar to those found in this study are unlikely to influence growth rates of tree seedlings under a warm-temperate climate to the same extent as found in this study.
Strong after-effects of previous-year conditions on the current growth rate of seedlings, as described here, can be critical for the interpretation of plant growth data from experiments extending over several growing seasons, especially in a cold climate where growth rates are low. Although the delayed response in growth rate cannot override environmental constraints and resource limitations, it could distort or hide the relationships seen between plant growth, climate and resource availability. The results of growth experiments covering several years might therefore be difficult to interpret if there were large year-to-year variations in the environmental conditions and growth rates were low (Shaver & Chapin 1995). Special care must be taken in transplantation experiments, and replanted seedlings should be acclimatized to common environmental conditions for several years before they are used for any growth experiments. These and other precautions due to possible delays in growth responses are already known for species with determinate shoot growth (Fagerström & Lohm 1977). The results of this study suggest that delayed responses could also influence the growth rate of tree species with indeterminate shoot growth.
J. H. C. Cornelissen, M. Diekmann, P. S. Karlsson and three anonymous reviewers provided valuable comments on the manuscript. The Abisko Scientific Research Station and the Swedish Royal Academy of Sciences generously supported the research reported here. All persons are gratefully acknowledged.