Structural and physiological plasticity in response to light and nutrients in five temperate deciduous woody species of contrasting shade tolerance

Authors

  • A. PORTSMUTH,

    1. Department of Plant Physiology, University of Tartu, Riia 23 Tartu 51010, Estonia,
    Search for more papers by this author
  • Ü. NIINEMETS

    Corresponding author
    1. Department of Plant Physiology, University of Tartu, Riia 23 Tartu 51010, Estonia,
    2. Institute of Agricultural and Environmental Sciences, Estonian University of Life Sciences, Kreutzwaldi 64, Tartu 51014, Estonia
      †Author to whom correspondence should be addressed. E-mail: ylo.niinemets@emu.ee
    Search for more papers by this author

†Author to whom correspondence should be addressed. E-mail: ylo.niinemets@emu.ee

Summary

  • 1Plants encounter a variety of light and nutrient availabilities during succession. However, there is an ongoing debate to which extent light-dependent structural and physiological plasticity is associated with species shade tolerance.
  • 2Seedlings of five species, Betula pubescens Ehrh., B. pendula Roth, Populus tremula L., Quercus robur L. and Acer platanoides L. (from most intolerant to most shade-tolerant), were grown at four different light and nutrient availabilities to test the hypotheses that intolerant species have higher physiological and tolerant species higher structural plasticity to light and also that there is an overall increase in plasticity with increasing nutrient availability. Two replicate experiments in different years were conducted. Plasticity was characterized by four estimates: (1) the range of variation of the components of relative growth rate (RGR), leaf area ratio (LAR) and net assimilation rate (NAR) (RGR = LAR·NAR) at common RGR; (2) average standardized slopes of physiological (RGR, NAR, i.e. physiological plasticity, ΠP) and structural (LAR, leaf dry mass per unit area, biomass allocation traits, i.e. structural plasticity, ΠS) traits vs. irradiance relationships; (3) standardized difference of plant traits measured at low to medium irradiance; (4) coefficient of variation across different irradiance treatments.
  • 3Plant growth was more strongly associated with NAR than with structural traits, but shade-intolerant species had a greater range of variation in both NAR and LAR at a common RGR. RGR, NAR and structural characteristics also responded more strongly to increases in irradiance in shade-intolerant species, but at low irradiance RGR and NAR were similar among all species. Owing to higher biomass fraction in leaves, the intolerant species produced less woody biomass. In nonfertilized plants, both ΠP and ΠS were negatively associated with shade tolerance. The plasticity was enhanced by nutrient addition, but the nutrient-dependent enhancement in plasticity was greater in more tolerant species. Therefore, differences in plasticity among species of varying tolerance were lower at higher nutrient availability.
  • 4Our study does not support the hypothesis of a trade-off between structural and physiological plasticity. Shade-tolerant species are generally less plastic than intolerant species, but increases in nutrient availability during succession reduce the differences in plasticity. Despite similar RGR in low light, first-year seedlings of shade-tolerant species produce more woody biomass, favouring survival and growth in subsequent years.

Introduction

Modifications in temperate forest composition during succession crucially depend on species shade tolerance and species responsiveness to fluctuations and disturbance. Species dominating in the early stages of succession are characterized by high physiological capacities for light use in photosynthesis (Walters, Kruger & Reich 1993; Reich et al. 1998; Walters & Reich 1999; Ellis, Hubbell & Potvin 2000; Delagrange et al. 2004), contributing to their high growth rates and establishment in initial stages of succession. In turn, adjustments in crown architecture and overall leaf area may explain the increasing dominance of shade-tolerant species in later stages of stand development (Canham 1988; Wayne & Bazzaz 1993; Valladares et al. 2002; Sterck et al. 2003). Light availability in understory decreases during primary succession in a pre-determined manner, and accordingly, strong evolution of light-dependent plasticity (plant ability to respond to changing resource availability sensu Bradshaw 1965) is expected in species colonizing a variety of habitats (Schlichting 1986; Sultan 1992; Schlichting & Pigliucci 1998; Givnish 2002). As a collection of different traits seems to determine species performance in varying light environments, differential plasticity in physiological traits that modify growth rate, and in structural traits that affect light harvesting is expected between shade-intolerant and -tolerant species.

There are conclusive data indicating that shade-tolerant species have lower physiological plasticity than intolerant species (Strauss-Debenedetti & Bazzaz 1991, 1996; Rincón & Huante 1994; Berntson, Farnsworth & Bazzaz 1995; Groninger et al. 1996; Valladares et al. 2002; Bloor 2003; Delagrange et al. 2004; Sánchez-Gómez, Valladares & Zavala 2006a). Greater physiological plasticity allows shade-intolerant species to achieve rapid growth rates and thereby rapidly colonize early successional habitats (Walters & Reich 1999). While the superior performance of shade-intolerant species in high light is well understood, how shade-tolerant species compensate for limited physiological plasticity and do better in low light than intolerant species has been a long-standing enigma. Negative correlation between the plasticity in traits that improve light harvesting and that enhance carbon gain have been postulated for species differing in shade tolerance (Lortie & Aarssen 1996; Henry & Aarssen 1997), but so far, the information of the correlations between structural plasticity and shade tolerance is limited and the evidence is contrasting. Some data do suggest that morphological plasticity is larger in shade-tolerant species, partly offsetting lower physiological plasticity (Chen 1997; Valladares et al. 2000; Valladares et al. 2002; Paz 2003; Delagrange et al. 2004; Niinemets & Valladares 2004). In contrast, lower structural plasticity of shade-tolerant species has been observed in other studies (Groninger et al. 1996; Bloor & Grubb 2004; Sánchez-Gómez et al. 2006a). While it is often assumed that more plastic species have greater growth rates and outcompete less plastic species in heterogeneous environments (Rice & Bazzaz 1989; Givnish 2002), plasticity is not always strongly related to fitness, especially if species are compared over wide-ranging gradient of specific environmental factor (Sultan 2001; González & Gianoli 2004). Another explanation for the discrepancies among the studies can be that low light survivorship per se is selected in shade-tolerant species. Survival in low light can depend on traits that minimize respiratory losses (Walters & Reich 2000b; Craine & Reich 2005) and maximize long-term storage in stems and roots rather than harvesting of resources via plastic expenditure of carbon to track environmental changes (Kobe 1997; Canham et al. 1999; Reich et al. 2003). In the current study, we focus on plasticity, because understanding species-specific variations in plasticity makes it possible to assess the role of trait variations in species performance along light gradients.

In addition to light, plants can encounter large variations in nutrient availability during community development. In particular, light and soil nutrient availability are often negatively associated along gap–understory gradients (Tilman 1993; Bazzaz & Wayne 1994; Coomes & Grubb 2000). This is important as data demonstrate that shade-tolerant species perform poorly on infertile sites (Keddy & MacLellan 1990; Franklin et al. 1993; Kobe 2006), possibly because of large nutrient costs for constructing extensive leaf areas (Lusk & Contreras 1999). It has further been suggested that expression of light-dependent plasticity depends on nutrient availability (Latham 1992; Lortie & Aarssen 1996), providing an alternative explanation for contrasting patterns among the studies exploring the plasticity variation. If so, improvement of nutrient availability is expected to enhance the light-dependent plasticity more in shade-tolerant species. As light decreases faster in nutrient-rich sites due to faster leaf area development (Bergh et al. 1999; Sampson & Allen 1998), high potential for plastic adjustment to light modifications can confer an important adaptive strategy.

Available evidence demonstrates significant modification of light-dependent plasticity by nutrient availability, but whether the responsiveness of plasticity to nutrients depends on species shade tolerance is not clear (Latham 1992; Burton & Bazzaz 1995; Walters & Reich 1996, 2000a). Nevertheless, light and nutrients do affect growth (Latham 1992; Walters & Reich 2000a; Schreeg, Kobe & Walters 2005; Portsmuth & Niinemets 2006) and plasticity (Latham 1992; Burton & Bazzaz 1995) in an interactive manner, and these interactive effects can importantly alter species competitive potential. Overall, this discussion highlights the paramount importance of gaining more conclusive insight into species-dependent responses to interactive light and nutrient availability gradients.

We studied light- and nutrient-dependent modifications in growth, biomass allocation and foliage architecture in five temperate deciduous species of contrasting shade tolerance to test the hypotheses that shade-tolerant species have greater morphological plasticity and that the plasticity to light increases with increasing nutrient supply, in particular, in shade-tolerant species. The experiments were replicated in different years to enhance the generality of our results.

Materials and methods

plant material

Five broad-leaved wide-spread species of contrasting shade tolerance were studied. Acer platanoides is a late-successional forest component, Quercus robur is intermediate, while Betula pendula, Betula pubescence and Populus tremula are early successional forest species. The seed size is largest in Q. robur, followed by A. platanoides and the three shade-intolerant species. The shade tolerance of temperate tree species was recently reviewed by Niinemets & Valladares (2006) and a common shade tolerance scale was developed for East Asian, European and North American species. There are many possible definitions of shade tolerance (survival, growth, completion of life cycle, optimal physiological performance, etc.) (e.g.: Grime 1979; Woodward 1990; Kobe et al. 1995; Grubb 1998; Reich et al. 2003; Valladares et al. 2005). Given that the ability to grow under low light is necessary for long-term survival, shade tolerance was defined as the capacity for growth in the shade in deriving these shade tolerance rankings (Niinemets & Valladares 2006). Given that plants are continuously loosing biomass due to mechanical damage and herbivory and need to replace the old leaves and fine roots, maintenance of sufficiently high gross growth rate is essential for long-term survival. According to this scale (1 = very intolerant, 5 = very tolerant), the species rank as Acer platanoides (average ± SE shade tolerance index according to reviewed studies = 4·20 ± 0·37) > Quercus robur (2·45 ± 0·28) > Populus tremula (2·22 ± 0·07) > Betula pendula (2·03 ± 0·09) > Betula pubescens (1·85 ± 0·07).

The study was accomplished in Tallinn, Estonia (59°22′ N, 24°42′ E). Two independent growth experiments in consecutive years were conducted with A. platanoides, B. pendula and B. pubescens in growing seasons 1999 and 2000 and with Q. robur in 2000 and 2001 to estimate the variation in plant responses to light environment and nutrients and thus, increase the robustness of our conclusions. Owing to limited seedling availability, P. tremula was studied only in 2000. The current-year seedlings (emerged in the year of experiment) of A. platanoides, Betula spp. and Populus tremula were excavated a week prior to the experiments in the beginning of growing season in mid-June (Table 1 for seedling sources) and kept in large containers filled with a mixture (5 : 1 v/v) of garden peat and sand (fraction 0·63–2·0 mm) and under moderate shade (30% of incident light) to ameliorate the initial differences in growth conditions. Quercus robur seedlings were grown from acorns collected and sown in the autumn of the year preceding the experiments (Table 1 for acorn sources). In all cases, the seedlings were collected from open locations (more than 60% of full sunlight) to minimize the initial differences among the seedlings. Table 1 provides information of seedling sources, and average plant sizes for all species. Further details for Betula species are provided in Portsmuth & Niinemets (2006).

Table 1.  Seedling sources, average ± SE plant dry mass and survivorship at low light at the end of experiment in five temperate deciduous species of contrasting shade tolerance*
SpeciesYear of experimentSite of collectionPlant dry mass (g)Survivorship at low light
  • *

    Species shade-tolerance ranking is A. platanoides > Quercus robur > Populus tremula > Betula pendula > Betula pubescens (Niinemets & Valladares 2006 for a revised shade-tolerance estimates).

  • 5% of incident irradiance.

  • means with the same letter are not statistically different (P > 0·05, anova followed by Bonferroni test). At every year, the number of seedlings was 101 for A. platanoides, B. pendula and B. pubescence and 53 for P. tremula and Q. robur.

  • §

    Site of acorn collection in the autumn of the year prior to the measurements.

Acer platanoides199959°22′ N, 24°38′ E 0·98 ± 0·08a1·0
Acer platanoides200059°22′ N, 24°38′ E0·387 ± 0·020bc0·958
Betula pendula199959°22′ N, 24°38′ E 1·03 ± 0·09a0·875
Betula pendula200058°58′ N, 26°02′ E 0·66 ± 0·06c0·875
Betula pubescens199959°28′ N, 25°40′ E
58°27′ N, 24°05′ E
0·365 ± 0·047c0·75
Betula pubescens200059°28′ N, 25°40′ E
58°27′ N, 24°05′ E
0·423 ± 0·045c0·75
Populus tremula200057°49′ N, 27°12′ E 3·75 ± 0·48d0·917
Quercus robur200058°23′ N, 26°43′ E§ 2·13 ± 0·11e1·0
Quercus robur200158°39′ N, 25°58′ E§ 2·84 ± 0·19e1·0

light availability treatments

For both Betula species and A. platanoides 101 seedlings and for Q. robur 53 seedlings were available in both years, while 53 seedlings of P. tremula were available (altogether 765 seedlings during all experiments). After seedling stabilization, five seedlings per species were harvested for determination of initial dry mass in each year. The initial seedling dry masses averaged across the years (± SE) were 0·0367 ± 0·007 g in B. pubescens, 0·126 ± 0·030 g in B. pendula, 0·281 ± 0·009 in P. tremula, 0·45 ± 0·10 in Q. robur and 0·350 ± 0·051 in A. platanoides.

The remaining 96 seedlings (A. platanoides and Betula ssp.) or 48 seedlings (P. tremula and Q. robur) per species were replanted in 0·85 dm3 clay pots filled with the mixture of peat and sand (5 : 1 v/v), and transferred to shade houses situated in an open location in Tallinn. The shade houses with dimensions 1 m × 1·1 m × 0·8 m (length × width × height) were made of glass and were painted with white oil paint to obtain spectrally uniform reduction in irradiance. Variation of irradiance in shade houses was achieved by varying the number of paint layers. The relative quantum flux densities estimated at mid-day by LI-190SA quantum sensor (Li-Cor, Inc., Lincoln, NE, USA) (n = 23) in the shade houses were 95% (nonpainted), 34%, 10% and 5%. The pots were randomly re-arranged in the shade houses every fourth day to reduce the effect of microheterogeneity in the environmental conditions in the chamber. During the growing season, the average daily integrated quantum flux density (Qint) corresponding to 100% light was 35·7 mol m−2 day−1 in 1999, 28·4 mol m−2 day−1 in 2000 and 30·0 mol m−2 day−1 in 2001 (Tõravere Actinometric Station, 58°16′ N, 26°28′ E). Several studies exploring seedling establishment in temperate forests have used lower minimum irradiances, on the order of 1–3% (Walters & Reich 1996, 2000a; Kobe 1997). However, these studies were conducted at latitudes around 45° N, where the growing season is longer and daily peak irradiances higher than in northern temperature forests. At our latitudes, typical daily average growing season minimum irradiances in the understories of deciduous mid- to late-successional stands are on the order of 1·5–4 mol m−2 day−1 (Niinemets 1998; Niinemets et al. 1999). Thus, we conclude that the minimum irradiance of 5% used in our study is representative of northern temperate forests.

The glass cover of the shade house was raised by 4 cm for free air movement in the chamber. The difference between the growing chambers in temperature was less than 2 °C, and the average difference in humidity was within 5%; the temperature and humidity inside the chambers differed from ambient values only on one occasion during the study years (Portsmuth & Niinemets 2006 for details).

experimental modification of nutrient availability

Simultaneously with light treatments, six plants of A. platanoides, B. pendula and B. pubescens and three plants of Q. robur and P. tremula were subject to each of the four different nutrient supplies at every light level as detailed in Table 1. Plant nutrient additions were carried out according to Lajtha and Klein (Lajtha 1994; Lajtha & Klein 1988). Phosphorus was given in the form of NaH2PO4, while nitrogen in the form of NH4NO3 and Ca(NO3)2·4H2O (molar ratio of salts in the solution 1 : 1·3 : 2·2). The maximum salt concentrations (high nutrient treatment) were 2·0 mmol L−1 NaH2PO4, 2·65 mmol L−1 NH4NO3 and 4·49 mmol L−1 Ca(NO3)2. The nutrients were supplied on every fourth day, altogether 16 times during the growing season. Each time, 50 mL nutrient solution or distilled water (controls) was added to each pot, and the total amount of added nutrients during the growing season was 199·7 mg N and 62 mg P per pot for the high, 99·9 mg N and 31 mg P per pot for the medium, and 66·7 mg N and 20·7 mg P per pot for the low nutrient treatment (Portsmuth & Niinemets 2006 for further details). As fertilization is most effective without additional stresses (Graciano, Guiamét & Goya 2005), the plants were additionally watered every second day to substrate field capacity.

analysis of biomass, leaf areas and sapling growth

After 63 days of growth in the shade houses, the seedlings were carefully washed out from substrate to avoid loss of fine roots. Every plant was partitioned between leaves, stem, coarse (diameter ≥ 1 mm) and fine roots (diameter < 1 mm). All fresh leaves per plant were scanned (Plustek Opticpro Scanner), and the areas of the leaves were determined with home-made computer software.

Dry mass of all biomass fractions was estimated after oven-drying at 75 °C for at least 48 h. These measurements were further used to calculate total plant biomass (MD, g), leaf mass per unit area (MA, g m−2), leaf area per total plant dry mass (leaf area ratio, LAR, m2 kg−1), shoot to root mass ratio (S : R, g g−1), leaf mass ratio (FL, g g−1), stem mass ratio (FS, g g−1), shoot mass ratio (fraction of above-ground biomass, FSh, g g−1), root mass ratio (FtR, g g−1), and fine root mass ratio (FfR, g g−1).

The relative growth rate (RGR, g g−1 day−1) was estimated as:

image( eqn 1)

where t2 is the time at harvesting and t1 at the transfer of plants to shade houses, MD2 is the total plant dry mass at harvesting and MD1 is the dry mass in the beginning of the experimental treatments. Net assimilation rate (NAR, g m−2 day−1) was calculated as:

image( eqn 2)

where LA1 is the initial plant leaf area, and LA2 is the final area. Thus, in our study, RGR, LAR and NAR are independent measures. From a conceptual perspective, average NAR, NARA, can also be expressed as:

image( eqn 3)

providing a functional linkage between plant relative growth rate and the overall biomass allocation for LAR, and the physiological activity of this leaf area (NAR).

The ratio of total leaf area at the end of the experiment relative to the beginning of experimental treatments was on average (± SE) 8·0 ± 0·3 for all data pooled, and was on average 4·17 ± 0·45 for the treatment with lowest growth rate (5% light, no nutrient addition), being in no cases less than 1·50. Given further that plants with young developing leaves were transferred to shade houses, we conclude that the length of experimental treatments was sufficient for full expression of light- and nutrient-dependent plasticity.

chemical analyses

Foliage N and C contents were estimated by an elemental analyser (CHN-O-Rapid, Foss Heraeus GmbH, Hanau, Germany), while P content was determined by molybdenum blue method after standard Kjeldahl digestion (Grimshaw, Allen & Parkinson 1989).

determination of the plasticity to light

Phenotypic plasticity was characterized by four different estimates. The first plasticity estimate was based on the degree of variation in the components of RGR, NAR and LAR relative to the variation in RGR (eqn 3). The variation in NAR characterizes physiological plasticity and the variation in LAR the structural plasticity. Increases in irradiance generally result in enhanced NAR, and lower LAR, suggesting a trade-off between these characteristics. The rationale of this analysis was that the same value of RGR may be achieved by different combination of NAR and LAR, and more plastic species have larger variation in either LAR or NAR or in both at the same degree of variation in RGR. We developed the regressions of RGR vs. NAR and RGR vs. LAR for every species at each combination of light and nutrients, and used the regression slopes as the estimate of physiological (RGR vs. NAR) and structural (RGR vs. LAR) plasticity. In these analyses, data from both years were pooled for specific treatments as the year effect on these relationships was not significant according to ancova analyses (data not shown). Because the relationships were curvilinear (Fig. 1 for a sample relationship of RGR vs. NAR), NAR and LAR were log-transformed, and the slopes were calculated for RGR = aLn(NAR) + b and RGR = cLn(LAR) + d relations. The correlations between RGR and NAR were generally stronger (r2 > 0·9) than the correlations between RGR and LAR (r2 = 0·1–0·9). The standardized slope (slope divided by the standard deviation of the sample) scales directly with r2 (Sokal & Rohlf 1995), and provided the overall variation within the treatments (standard deviation) is similar, a lower value of r2 also implies a lower nonstandardized slope. In our study, we found that the correlation between the slopes of RGR vs. Ln(LAR) and r2 values determined for each specific relationship was highly significant (r2 = 0·86, P < 0·001), suggesting that even statistically nonsignificant relations provided information of the plasticity (variation/constancy) of specific traits. The concept of RGR vs. Ln(LAR) and RGR vs. Ln(NAR) is analogous to growth response coefficient analysis, which determines the slopes log-transforming also RGR (Poorter & Nagel 2000; Poorter 2001). However, we preferred the slopes based on nontransformed RGR, as this resulted in larger range in the slopes, allowing us to more efficiently discriminate between the species.

Figure 1.

Sample dependence of relative growth rate (RGR) on net assimilation rate (NAR) for Acer platanoides seedlings grown at 34% of full light at low nutrient addition rate. The slope, a, of linearized regression equations in the form of RGR = aLn(NAR) + b was used as an estimate of physiological plasticity.

The second estimate of plasticity was based on responsiveness of a large number of plant traits to light (slope-based estimate of plasticity). We first calculated the regressions between specific plant traits and seasonal integrated quantum flux density (Qint). Qint for each shade house was determined as the year-specific value of incident Qint as given above (Light availability treatments) times the fraction of light transmitted by each shade house. As plant structural and physiological characteristics often respond curvilinearly to light (Poorter 1999; Montgomery 2004) as was also observed in our study (e.g. Appendix SI for growth variables), Qint was log-transformed in these relations (LogQint). The traits used were LAR, MA, FL, FtR, FS, NAR, RGR, and we first determined the regression slopes for all these characteristics. As the regression slope of specific variable characterizes the overall light-dependent change in the trait for a certain change in light, it provides a measure of the trait plasticity. The slope of the reaction norm of the trait is commonly used as a measure of plasticity (Schlichting & Pigliucci 1998; Pigliucci & Kolodynska 2002).

Characteristics varying more strongly with light have larger slopes and larger degrees of explained variance, while less variable characteristics have smaller slopes and lower degrees of explained variance. The slopes are negative for characteristics decreasing with increasing Qint (generally, LAR, FL and FS), and positive for traits increasing with light (MA, NAR, RGR and FtR). Because the absolute slope values also depend on the unit of the specific characteristic, a standardized value of plasticity of trait i, Πi, was calculated as:

image( eqn 4)

where Si is the regression slope and is the average of trait value over all light levels. Πi values standardized this way can be directly compared for various traits, and generally varied from 0 to 1, being in rare cases slightly above 1. Sánchez-Gómez et al. (2006a) suggested an analogous standardization to directly compare the plasticities for different traits.

Physiological plasticity (ΠP) was defined as average of Πi values for RGR and NAR. Several of the structural characteristics were not entirely independent and the Πi estimates for these were strongly correlated with each other. Therefore, for structural plasticity, we first calculated separately the average Πi values for allocation traits –FL, FS and FtR– and for morphological traits – LAR and MA. These averages were further averaged to get an estimate of structural plasticity (ΠS). The overall plasticity to light (ΠT) was defined as the average of ΠS and ΠP. For every species, plasticity estimates were calculated for both years using year-specific average values of Qint, allowing us to estimate the variation (SE) in plasticity and analyse the consistency in relationships between shade tolerance and species-specific plasticity estimates. The values of different plasticity estimates were calculated separately for control (no nutrient addition) and high nutrient treatments to determine the nutrient-dependent variation in plasticity. The relative difference in plasticity between control and high-nutrient grown plants, R, was calculated as:

image( eqn 5)

where ΠH is either ΠP, ΠS or ΠT of plants grown at high N, and ΠC the corresponding estimate of the control plants. The values of R calculated this way provide a standardized estimate of plasticity change. The results were statistically identical for absolute differences in plasticity (ΠH − ΠC) (data not shown).

The slope-based method of plasticity calculation characterizes plant response to the full light gradient, and may be biased by the highest light levels. To characterize plant response to low to intermediate irradiances, we calculated the third estimate of plasticity using the values of traits at 5% and 10% and 5% and 34% irradiance (point-based plasticity estimates). For the 5% and 10% irradiance, the point-based estimate of plasticity for trait i, Di, was calculated as:

image( eqn 6)

where X10% is the trait value at 10% light and X5% the value at 5% light, and ΔL is the difference in irradiance (10% − 5%). Analogously, the values of Di were calculated for the trait values at 5% and 34% light. Physiological (DP), structural (DS) and total (DT) plasticities were defined using the same set of traits as for slope-based plasticity estimates. DP, DS and DT were estimated for each species for every year, and species averages were calculated. In addition to our data for five species, we calculated the values of DP, DS and DT for three North American temperate deciduous species of varying shade tolerance (Betula alleghaniensis, Betula papyrifera and Populus tremuloides) using the data of Reich et al. (1998) to enhance the generality of our conclusions. In the study of Reich et al. (1998) the same set of traits has been measured for seedlings of comparable age at 5% and 25% of light, and we calculated the values of Di analogously as for our data.

As the plasticity estimates based on 5% vs. 10% and 5% vs. 34% light were strongly correlated (e.g. r2 = 0·64, P = 0·009 for physiological plasticity), we show only the data with the plasticity estimates calculated using the values at 5% and 34% irradiance.

In many studies exploring the plastic response of plants to light, plasticity has been characterized by coefficient of variation (standard deviation divided by the sample mean) across the light treatments (Valladares et al. 2000; Balaguer et al. 2001; Bloor & Grubb 2004; Sánchez-Gómez, Valladares & Zavala 2006b). Implicit in this analysis is that the variation of specific traits within the data set is mainly driven by plastic responses to light, while the slope- or point-based plasticity analyses basically only consider the light-driven component in the overall variation. To make our results more comparable with other studies, we also calculated the plasticity estimates based on the coefficient of variation for every trait as in Valladares et al. (2000). After calculating trait-specific values of variation coefficients, physiological (CP), structural (CS) and total (CT) plasticities for control and high nutrient treatments were calculated using the same traits as for the slope- and point-based plasticity estimates. Although the slope- and variation-based estimates of plasticity relay on different assumptions, we observed a strong correlation between the values of CP and ΠP (r2 = 0·66), CS and ΠS (r2 = 0·69) and CT and ΠT (r2 = 0·64) (P < 0·001 for all relationships). Given these correlations and that both of these estimates characterize the phenotypic plasticity in the same sample (differently from Di, that is the plasticity in low irradiance only), we present graphically only the results with slope-based plasticity, while stating the correlations with variation-based plasticity estimates in the text. All statistical relationships with species shade tolerance rank were qualitatively identical for both slope- and variation-based estimates, as shown in the Results.

data analyses

We consider all statistical tests significant at P < 0·05. The species and treatment effects were investigated by analyses of variance (anova) using systat 8·0 (SPSS Inc.). Only very few plants died at the two highest light regimes, while at the lowest irradiance, the survivorship was 0·88 in B. pendula and 0·75 in B. pubescence in both years, 1·0 in 1999 and 0·96 in 2000 in A. platanoides, 0·83 in Populus in 2000, and 1·0 in Q. robur in both years. Because of different seedling survival in different treatments, the statistical design was slightly unbalanced. As suggested by Potvin (1993), we used Type III sums of squares in our anova analyses to reduce the influence of design unbalance on our statistical conclusions.

For four species, independent replicate experiments were conducted in different years to test the generality of our findings. Year of experiment was included in the initial statistical models as a factor. Owing to differences in seedling sources and plant size (Table 1) as well as the differences in climate between the years, the year effect is complex and characterizes the combined influence of several different sources of variation. While the year of experiment was generally significant in the statistical models exploring the statistical significance of the experimental treatments on plant traits, the interaction terms, year × light and year × nutrition as well as higher-order interactions such as year × light × nutrition were generally not statistically significant, indicating that the experimental treatments affected plant traits similarly in both years. Therefore, year effect was suppressed in the general statistical model (Table 2). However, in comparison of the means at specific nutrient and light availability treatments (Table 3), year was included in the statistical models to more efficiently separate the means at various treatments.

Table 2.  Summary of the species and treatment (light, nutrient availability) effects on the traits* in the first-year seedlings of Acer platanoides, Betula pendula, B. pubescens, Populus tremula and Quercus robur: the P-values of three-way anovas
FactorNutrient concentrationsRelative growth indicatorsBiomass fractions (g g−1)
N (%)P (%)MA (g m−2)LAR (m2 kg−1)RGR (g g−1 day−1)NAR (g m−2 day−1)FLFSFShFtRFfRS : R
  • *

    Relative plant growth descriptors: leaf dry mass per unit area (MA), leaf area ratio (LAR), relative growth rate (RGR), mean net assimilation rate (NAR); Biomass ratios: leaf mass ratio (FL), stem mass ratio (FS), shoot mass ratio (FSh), total (FtR) and fine root (FfR) mass ratio, and shoot to root mass ratio (S : R). Bold type outlines effects that are significant (P < 0.05).

Species0·0000·0000·0000·0000·0000·0000·0000·0000·0000·0000·0000·000
Light0·0000·0000·0000·0000·0000·0000·0000·0000·0000·0000·0380·000
Nutrition0·0000·0110·4220·0000·0000·0010·0000·3630·0000·0000·0000·000
Species × light0·0000·0000·0000·0000·0000·0000·0050·2420·0170·0310·0000·002
Species × nutrition0·0000·0000·0000·1000·2940·6100·3590·6420·0520·0250·4320·008
Light × nutrition0·0030·0080·4740·2980·0000·0000·8620·3210·3460·3390·1640·813
Species × light × nutrition0·0000·0000·9090·7620·4000·2620·9720·5340·4930·5490·0600·983
Model r20·6150·8020·7290·6750·6640·5900·5250·4240·6420·6260·3930·493
Table 3.  Average (± SE) values of relative growth rate (RGR, g g−1 day−1), net assimilation rate (NAR, g m−2 day−1), leaf area ratio (LAR, m2 kg−1), leaf dry mass per unit area (MA, g m−2), the fraction of plant dry mass in leaves (FL, g g−1) and total root fraction (FtR, g g−1) in five species of contrasting shade tolerance at high (95% of incident irradiance) and low light (5%) and at two different nutrient supplies (Contr. high)*
CharacteristicSpeciesHigh light, Contr.Low light, Contr.High light, high NLow light, high N
  • *

    Means with the same small letter are not different between the treatments within species, while the means with the same capital letter are not significantly different between the species within specific treatments according to anovas followed by Bonferroni tests.

RGRAcer platanoides0·0207 ± 0·0020aA0·0166 ± 0·0021aA0·0333 ± 0·0019bA0·0167 ± 0·0023aA
Betula pendula0·0327 ± 0·0030aBC0·0170 ± 0·0023bA0·0541 ± 0·0026cB0·0227 ± 0·0032bA
Betula pubescens0·0336 ± 0·0034aBC0·0256 ± 0·0044aA0·0653 ± 0·0020bC0·0209 ± 0·0045aA
Populus tremula 0·038 ± 0·009aB0·018 ± 0·007bA0·0554 ± 0·0042aB0·0172 ± 0·0020bA
Quercus robur0·0233 ± 0·0024aAC0·0162 ± 0·0007bA0·0314 ± 0·0011cA0·0164 ± 0·0013bA
NARAcer platanoides  3·03 ± 0·39aA  2·20 ± 0·27acA  4·61 ± 0·37bA  1·59 ± 0·22cAB
Betula pendula  3·80 ± 0·44aA  1·06 ± 0·15bB  5·75 ± 0·36cAB  1·31 ± 0·22bAB
Betula pubescens  3·04 ± 0·28aA  1·47 ± 0·33bAB   6·4 ± 0·6cB  1·23 ± 0·28bB
Populus tremula   7·2 ± 2·1aB  1·56 ± 0·72bAB   7·6 ± 1·2aAB   1·8 ± 1·0bAB
Quercus robur   5·0 ± 1·2aA  2·35 ± 0·08bA   7·4 ± 0·7cB  2·47 ± 0·24bA
LARAcer platanoides   4·8 ± 0·5aAB   8·3 ± 0·9bcAC  6·01 ± 0·35abA  10·1 ± 0·7cA
Betula pendula  5·41 ± 0·42aAB  14·9 ± 1·6bB  7·43 ± 0·8aAB  16·7 ± 1·1bB
Betula pubescens   7·1 ± 1·1aB  16·4 ± 1·4bB   8·5 ± 0·9aB  14·1 ± 1·0bB
Populus tremula  3·51 ± 0·31aAB  11·9 ± 1·8bA   6·2 ± 0·7aAB  15·1 ± 1·4bAB
Quercus robur  3·14 ± 0·42aA  5·43 ± 0·10bC  2·72 ± 0·15aC  4·91 ± 0·14bC
MAAcer platanoides  42·1 ± 2·2aA  35·2 ± 1·1bA  50·2 ± 3·0cA  38·3 ± 2·4abA
Betula pendula  60·0 ± 2·3aAB  28·5 ± 1·5bB  54·5 ± 2·2aA  29·2 ± 1·1bB
Betula pubescens  52·0 ± 2·2aB  25·9 ± 2·6bB  53·1 ± 2·0aA  30·0 ± 3·2bB
Populus tremula  54·3 ± 1·4aAB  30·0 ± 0·6bAB  55·2 ± 4·4aA  26·9 ± 1·1bB
Quercus robur  56·8 ± 2·2aB  49·2 ± 1·4bC  74·1 ± 1·0cC  54·9 ± 0·8abC
FLAcer platanoides 0·212 ± 0·035aA 0·296 ± 0·037aA 0·303 ± 0·027abAC 0·382 ± 0·032bAB
Betula pendula 0·321 ± 0·022aAB 0·410 ± 0·038bB 0·391 ± 0·028abBC 0·476 ± 0·021bA
Betula pubescens 0·352 ± 0·043aB 0·398 ± 0·021abB 0·436 ± 0·029bB  0·41 ± 0·05abAB
Populus tremula 0·191 ± 0·020aAB 0·357 ± 0·051bAB 0·336 ± 0·015bAB  0·44 ± 0·33bAB
Quercus robur 0·164 ± 0·020aA 0·267 ± 0·009bAB 0·194 ± 0·010aA 0·269 ± 0·005bB
FtRAcer platanoides 0·547 ± 0·028aA 0·476 ± 0·032aAC 0·500 ± 0·021aA 0·363 ± 0·024bA
Betula pendula 0·540 ± 0·010aA 0·380 ± 0·035bcB 0·412 ± 0·026bAB 0·331 ± 0·025cA
Betula pubescens 0·508 ± 0·030aA 0·362 ± 0·014bB 0·372 ± 0·027bB 0·350 ± 0·044bA
Populus tremula 0·578 ± 0·048aA 0·427 ± 0·043bAB 0·426 ± 0·049bAB  0·37 ± 0·21bA
Quercus robur 0·741 ± 0·026aB  0·56 ± 0·06bC 0·689 ± 0·023aC 0·591 ± 0·009bB

Because of the strong controls of plant size on relative growth rate, allocation and foliage morphology as demonstrated in several studies (Poorter 2001; Niinemets, Portsmuth & Truus 2002; Delagrange et al. 2004; Claveau, Messier & Comeau 2005; MacFarlane & Kobe 2006), we also analysed the influence of whole plant dry mass on statistical models using covariance analyses. As with the year of experiment, plant structural and physiological traits were significantly affected by plant size in most cases, but the significance of the treatment and species effects as well as the interaction terms was not biased by the inclusion of the plant dry mass, demonstrating that significant treatment effects were not mediated by concomitant changes in plant size.

Regression analyses were employed to investigate the relationships between plasticity and shade tolerance as well as the correlations between different shade-tolerance estimates. As the seedlings of shade-tolerant species tend to be initially larger due to larger seed size (Walters & Reich 2000a; Sack & Grubb 2003), and there were year-to-year differences in plant size (Table 1), and plant size effects were generally significant for most plant traits, plant dry mass was included as an additional explanatory variable in multiple regression analyses to rule out the possibility that the correlations between plasticity and shade tolerance resulted from overall differences in seedling size and year-to-year differences in size. To compare the average plasticity estimates among species, the plasticity estimates derived for single traits (eqn 4) were compared by paired t-tests.

Results

survivorship in relation to shade tolerance, plant size and year of experiment

The survivorship values of species at the lowest light regime (Table 1) were in general agreement with the observational estimates of species shade tolerance, overall demonstrating that we had a representative range of species with contrasting performance in low light environments.

However, the correlation between shade tolerance and survivorship was curvilinear (for both years pooled, r2 = 0·48, P < 0·05 for a regression in the form y =aLog(x) + b), and larger plants of similar shade tolerance apparently had a greater survivorship than smaller plants (Populus vs. Betula and Quercus vs. Acer). According to multiple linear regression analysis, both the average plant dry mass (P = 0·024) and shade tolerance (P = 0·013) were positively correlated with survivorship in low light (r2 = 0·76).

More tolerant species Acer platanoides and Quercus robur had greater seeds than less tolerant species (data not shown), contributing to larger survival of first year seedlings in shaded environments. In general, shade-tolerant species tend to possess larger seeds and greater initial dry mass than the tolerant species (Grubb & Metcalfe 1996; Leishman et al. 2000; Körner 2005), and seed size is often strongly correlated with seedling survival in low light (Leishman & Westoby 1994; Osunkoya et al. 1994; Walters & Reich 2000a), but not always (Augspurger 1984). In our study, differences in initial size also partly explained year-to-year differences in survival (Table 1), further underscoring the importance of enhanced carbon reserves for survival (Kobe 1997; Canham et al. 1999).

changes in foliage nitrogen and phosphorus contents in relation to light availability and nutrient supply

Increases in irradiance generally resulted in decreased nitrogen (NM) and phosphorus (PM) contents per unit dry mass in all species (Fig. A1 for all treatment combinations, Table 2 for general statistical relationships). However, there were significant nutrient supply–light interactions for both NM and PM, implying that the decrease in nutrient content at high light was lower at higher nutrient supply (Table 2, Fig. A1), or in specific cases, that the nutrient content was independent or slightly increased with light at higher nutrient supply (NM in Betula pubescens and Quercus robur). These species-dependent patterns also implied significant species interactions (Table 2).

The overall change in nitrogen content from the lowest (NL) to the highest irradiance (NH) > (NL − NH)/NL (all different nutrient supply treatments averaged), was 0·29 for Populus tremula, 0·25 for Acer platanoides, 0·18 for B. pendula, −0·06 for B. pubescens and Q. robur. The corresponding changes in P contents were 0·74 for A. platanoides, 0·31 for Q. robur, 0·27 for B. pubescens, and 0·23 for B. pendula and P. tremula (Fig. A1). Although the relative change was larger in P than in N, P : N ratio was generally large in our study varying from 0·047 g g−1 (slightly P-limited) to 0·54 g g−1 (N-limited) and averaged 0·163 ± 0·004 g g−1, implying that N was the primary limiting element.

light and nutrient effects on physiological characteristics: relative growth rate and net assimilation rate

Both RGR and NAR significantly increased with increasing irradiance and nutrient availability in all species (Fig. A2 for RGR and NAR values for all treatment combinations, and Table 2 for overall statistical tests). There were significant species–light interactions (Table 2), implying that the species responsiveness to light differed among the species (Table 3). Species effects on RGR vs. light responses were also evident in large species differences at high light, but smaller differences at low light (Fig. A2, Table 3). For instance, RGR did not differ at low light for either control plants (no nutrient addition) or high nutrient plants (Table 3), while shade-intolerant species Betula pendula, B. pubescens and Populus tremula possessed higher RGR at high irradiance than more tolerant species Acer platanoides and Quercus robur (Table 3). In contrast, the values of NAR tended to be larger in more tolerant species in low light, especially in control plants, while at high light, NAR values did not differ statistically in control plants or were occasionally larger in shade-intolerant species at high N (Table 3). These differences in the light responsiveness at different nutrient availabilities were further confirmed by statistical analyses that showed a significant nutrient supply–light interaction (Table 2).

structural traits in relation to light and nutrients

LAR decreased with increasing irradiance in most cases, and this decrease was associated with light-dependent increase in leaf dry mass per unit area (MA) and reduction in the fraction of plant biomass in leaves (FL, leaf mass ratio, LAR = FL/MA) (Tables 2, 3). The degree of light-dependent change in all these characteristics varied among the species (significant species–light interaction, Tables 2, 3). In particular, all these structural characteristics tended to vary more in shade-intolerant species B. pubescens, B. pendula and P. tremula than in more tolerant species A. platanoides and Q. robur. LAR tended to increase with increasing nutrient availability, mainly due to increased fractional biomass investment in leaves (Tables 2, 3). While the nutrient supply itself was not statistically significant for MA, the interaction term, species × light, was significant (Table 2), illustrating greater variation in MA in high nutrient availability in A. platanoides (Table 3).

The fraction of plant biomass in fine and coarse roots and total root fraction increased with increasing irradiance, while the fraction of plant biomass in stem and shoot : root ratio decreased with increasing light (Tables 2, 3). Increases in nutrient availability resulted in reduced fraction of roots and larger shoot : root ratio (Tables 2, 3). Again, significant species × light and light–nutrient interactions implied that the sensitivity of biomass allocation to light and nutrient supply differed among species. For instance, in plants without nutrient addition (control) total root mass fraction did not differ among A. platanoides, B. pendula, B. pubescens and P. tremula at high light, while significant differences were observed among species at low light (Table 3).

physiological and structural determinants of growth: the plasticity in nar and lar

When different treatments were pooled, RGR increased curvilinearly with NAR (Fig. 2A), while RGR was independent of LAR or decreased with increasing LAR (Fig. 2B), demonstrating that the variation in NAR more strongly affected growth rates. The species followed different curves of RGR vs. NAR, depending on specific LAR values and variation patterns (Fig. 2). Overall, the ranges of variation in RGR, NAR and LAR tended to be larger in shade intolerant than in more tolerant species (B. pendula, B. pubescens and P. tremula vs. Q. robur and A. platanoides in Fig. 2). As the same value of RGR may be achieved by different combinations of NAR and LAR (eqn 3), larger variation in LAR and NAR at a common RGR implies greater plasticity. As illustrated in Fig. 3, the slopes of NAR vs. RGR and LAR vs. RGR were positively correlated for the three shade-intolerant species, but not in more tolerant species, in which the range of variation was much smaller. Nevertheless, for all species pooled, these slopes were positively associated for both control (no nutrient addition, r2 = 0·22, P = 0·036) and for the high nutrient treatment (r2 = 0·30, P = 0·016), and these relationships were not statistically different between control and high nutrient treatments (P > 0·5 for the interaction term in the separate slope ancova model, and P > 0·7 for the main effect in the common slope ancova model).

Figure 2.

Relative growth rate in relation to its components, NAR (A) and leaf area ratio (LAR) (B), for five species of contrasting shade tolerance. RGR is the product of NAR and LAR (eqn 3). Depicted are the averages for every light and nutrient treatment for both years of the study. The data were fitted by nonlinear regressions as in Fig. 1. Nonsignificant regressions in B are shown by dashed lines. The shade tolerance of the species follows the sequence: Betula pubescens < Betula pendula < Populus tremula < Quercus robur < Acer platanoides (Niinemets & Valladares 2006 for a revised tolerance ranking).

Figure 3.

Correlations between the slope of RGR vs. Ln(NAR) (Fig. 1 for sample dependence) and the slope of RGR vs. Ln(LAR) for five species of contrasting shade tolerance. Every data point corresponds to the pair of slopes for specific nutrient and light availability treatment. The slope of RGR vs. Ln(NAR) was used as an estimate of physiological plasticity of growth, while the slope of RGR vs. Ln(LAR) as an estimate of structural plasticity of growth. Data were fitted by linear regressions and nonsignificant regressions (P > 0·05) are depicted by dashed lines.

light-dependent plasticity in species of differing shade tolerance: evidence from slope- and variation-based plasticity analyses

In control plants, the estimates of physiological (ΠP), structural (ΠS) and total (ΠT, average of ΠS and ΠT) plasticity calculated from the slopes of regressions describing the response of plant traits to irradiance (eqn 4) decreased with increasing species shade tolerance (Fig. 4A). With increasing nutrient availability, ΠP and ΠT increased in all species, while ΠS decreased in all species, except for A. platanoides (Fig. 4B). Overall, the response of plasticity to nutrient addition was larger in more shade-tolerant species, as the result of which the negative relationships of plasticity vs. shade tolerance were only marginally significant (Fig. 4B), and the relative difference of plasticity between control and high nutrition plants (eqn 5) increased with increasing shade tolerance (Fig. 4C). Although plant size varied among different species and in different years (Table 1), plant dry mass was not significant in multiple regressions (plasticity vs. shade tolerance and plant dry mass, data for all years included). For control plants, the probability of significant size effects P = 0·83 for ΠP, P = 0·96 for ΠS and P = 0·88 for ΠT. For high nutrition plants, P = 0·95 for ΠP, P = 0·39 for ΠS and P = 0·74 for ΠT.

Figure 4.

Variation in the physiological (ΠP), structural (ΠS) and total (ΠT, average of ΠP and ΠS) plasticity to light in seedlings grown without nutrient addition (control) (A) and at high nutrient regime (B), and the relative difference in plasticity between control and high nutrient treatments (eqn 5) (C) among five species of varying shade tolerance. The plasticity to light was calculated according to eqn 4, and is based on the slopes of linear regressions characterizing the dependence of specific traits on log-transformed seasonal average integrated quantum flux density (Log(Qint)). In A and B, the data were fitted by nonlinear regression equations in the form of y = axb, while the data in C by linear regressions. Nonsignificant regressions are depicted by dashed lines. The data are averages of two independent growth experiments conducted in different years, and error bars (± SE) are calculated using these independent experiments. The species are ranked according to shade tolerance as Betula pubescens < Betula pendula < Populus tremula < Quercus robur < Acer platanoides.

The correlations with shade tolerance were essentially the same for the coefficient of variation-based estimates of structural (CS), physiological (CP) and total plasticity (CT). The r2 values for the correlations between the variation-based estimates of plasticity with shade tolerance in control plants (analogous to correlations with slope-based estimates in Fig. 4A) were 0·88 (P = 0·018) for CS, 0·87 (P = 0·022) for CP, and 0·89 (P = 0·016) for CT. For high nutrient treatment (analogous to Fig. 4B), the r2 values for these correlations were 0·35 (P = 0·29) for CS, 0·78 (P = 0·047) for CP and 0·74 (P = 0·054) for CT. The r2 values for the correlation of shade tolerance with the difference between plasticity estimates between control and high nutrition site (analogous to Fig. 4C) were 0·88 (P = 0·019) for CS, 0·70 (P = 0·076) for CP and 0·83 (P = 0·033).

Although the regressions in Fig. 4 were strongly biased by A. platanoides, statistical comparisons of overall plasticity by paired t-tests (using the standardized slopes for LAR, FL, FtR, MA, NAR and RGR calculated by eqn 4 as replicates) demonstrated that the average plasticity was significantly lower in control plants of A. platanoides than in all other species × treatment combinations (P < 0·045, on average P = 0·022), and the average plasticity was lower in Q. robur than in P. tremula control plants (P < 0·05). Thus, these data further strengthen the conclusion of the lower plasticity in more shade-tolerant species.

Physiological and structural plasticity were positively correlated for control plants, while the correlation was not significant for plants at high nutrient supply (Fig. 5). According to separate-slope ancova, nutrition × ΠP term was not statistically significant for these relations (P = 0·12), but common-slope ancova demonstrated a larger ΠS at a common ΠP of control plants (P = 0·015).

Figure 5.

Correlations between structural and physiological plasticity in control plants and in plants grown at high nutrient supply (the same data as in Fig. 4). Data were fitted by linear regressions.

Values of ΠP and ΠT were correlated with the slope of RGR vs. Ln(NAR) (r2 = 0·51, P < 0·02 for the correlation with all treatments pooled), but not with ΠS (r2 = 0·24, P = 0·15) demonstrating that the slope of RGR vs. Ln(NAR) is a measure of plant physiological plasticity to light. However, the slope of RGR vs. Ln(LAR) was not correlated with ΠP, ΠS and ΠT (r2 < 0·12, P > 0·3).

further evidence of lower plasticity of shade-tolerant species from single point analyses

Examination of the plasticity estimates derived from measurements at low to medium irradiances (eqn 6) further demonstrated that the physiological (DP) and total (DT) plasticity were negatively associated with shade tolerance, but the correlation with structural plasticity (DS) was weaker (Fig. 6, pooled data for five species from our study and three North American species). When the various components of DS were examined separately, plasticity estimates based on leaf morphology and allocation (MA, FL and LAR) were not significantly associated with shade tolerance (P > 0·1), but the plasticity of root allocation (FtR) was negatively related to shade tolerance (r2 = 0·53, P = 0·041). The correlation between DP and DS was marginally significant (r2 = 0·47, P = 0·06).

Figure 6.

Physiological (DP), structural (DS) and total (DP) plasticity to light determined from the trait values at two different light availabilities (eqn 6) for the five European temperate deciduous species and for three North American temperate deciduous species investigated by Reich et al. (1998) (Betula alleghaniensis, Betula papyrifera and Populus tremuloides). The plasticity was estimated from measurements at 5% light and 34% light for the European species and from measurements at 5% and 25% light for the North American species. As eqn 6 standardizes the plasticity estimates with respect to the light range, the moderate difference in the higher irradiance level between the European and North American species had only minor effect on the plasticity estimates. Species shade tolerance is given according to the revised shade tolerance table of Niinemets & Valladares (2006). Data were fitted by linear regressions.

Discussion

species growth and allocation responses to light and nutrients

The results of this study agree with previous reports showing greater RGR, NAR, and larger fraction of biomass in leaves (FL) and LAR in first year-seedlings of shade-intolerant species in high light (Reich et al. 1998; Poorter 1999; Walters & Reich 2000a). Greater values of RGR have also been observed in shade-intolerant species in low irradiance (Reich et al. 1998; Poorter 1999), but not always (e.g. Walters & Reich 2000a). We found no difference in RGR between the species in the lowest irradiance (Table 3). At the same time, our data (Table 3) are consistent with previous studies showing lower FL and LAR and greater fractional investments of biomass in roots and stem in shade-tolerant species in low light (Groninger et al. 1996; Walters & Reich 1996, 2000a; Reich et al. 1998; Poorter 1999; Sánchez-Gómez et al. 2006a).

Enhancement of FL and LAR by low irradiance is a common response in many species (Grubb et al. 1996; Minotta & Pinzauti 1996; Reich et al. 1998; King 2003) and is interpreted as an adaptive response resulting in overall increase in the light capturing area. This increase was observed in all species in our study, yet, shade-tolerant species had a smaller LAR than light-demanding species (Table 3). How did shade-tolerant species achieve a greater woody biomass growth rate given that they had less leaves? First, there is conclusive evidence that the degree of leaf aggregation and overlap increases with total leaf area (Farque, Sinoquet & Colin 2001; Niinemets et al. 2004). In fact, having more leaves means that there are stronger light gradients within the canopy due to self-shading (Wayne & Bazzaz 1993), and accordingly a unit leaf area works less efficiently, in particular, in low light. Furthermore, there is evidence of architectural adjustments in shade-tolerant species that enhance light capture and reduce self-shading between neighbouring leaves. For instance, shade-tolerant species often have wider crowns with less leaf overlap than light-demanding species at common irradiance (Wayne & Bazzaz 1993; Niinemets 1998; Sterck et al. 2003; Delagrange et al. 2004).

Our study and most of the previous work has addressed the performance of first-year seedlings, but strong changes occur in biomass allocation and foliage architecture during ontogeny. In particular leaf dry mass per unit area increases and LAR decreases faster with increasing plant size in shade intolerant than in tolerant species (Sack & Grubb 2001; Delagrange et al. 2004; Niinemets 2006), leading to greater overall plant leaf area in shade-tolerant species (Lusk & Contreras 1999; Delagrange et al. 2004), and explaining the apparent advantage of tolerant species in low irradiance.

different methods to assess species plasticity

We used a series of estimates to characterize species phenotypic plasticity in response to light and nutrient gradients. The plasticity estimates based on the slopes of RGR vs. Ln(LAR) and RGR vs. Ln(NAR) characterize the overall variation in plant structural and physiological traits producing a certain variation in the growth rate (Fig. 3), while the slope-based (eqn 4) and point-based (eqn 6) estimates of plasticity directly describe the variation in specific traits due to light (Schlichting & Pigliucci 1998; Pigliucci & Kolodynska 2002) and variation coefficient-based estimates assumes that the variation in the sample is mainly due to phenotypic adjustment to light (Valladares et al. 2000; Balaguer et al. 2001; Bloor & Grubb 2004; Sánchez-Gómez et al. 2006b). The comparisons between different plasticity estimates have been rare. In a previous study, slope-based (eqn 4) and variation-based estimates of plasticity resulted in essentially similar conclusions with respect to the overall plastic response to light and species differences in plasticity (Niinemets, Valladares & Ceulemans 2003). In this study, we further found that the point- and slope-based estimates as well as the slope- and variation-based estimates were tightly correlated, collectively indicating that higher light-dependent plasticity is associated with greater coefficient of variation of the sample. Although different estimates of plasticity characterized different aspects of plant phenotypic response to light, they all showed plasticity differences among shade-intolerant and -tolerant species in a similar manner, demonstrating that our results are robust and not biased by analytical methods used to derive plasticity. Given that slope-based plasticity estimation does not require assumptions of the variation sources within the sample, we considered it a preferable method in our study.

structural and physiological plasticity in relation to species shade tolerance

The results of our study corroborate the previous evidence of greater physiological plasticity of shade-intolerant species (see Introduction). However, our results did not confirm a larger structural plasticity in shade-tolerant species that has been hypothesized in some studies (Valladares et al. 2002; Niinemets & Valladares 2004). In fact, the plasticity estimates based on the slopes of RGR vs. Ln(LAR) (structural plasticity) and RGR vs. Ln(NAR) (physiological variability) that merged both the light- and nutrient-dependent variation in LAR and NAR were positively associated (Fig. 3). The slope-based (eqn 4) and variation-based plasticity estimates (data not shown) further suggested that both structural and physiological plasticity are lower in shade-tolerant species (Fig. 4A), in particular in control plants, while point-based plasticity estimates (eqn 6) indicated that structural plasticity does not correlate with shade tolerance (Fig. 6). Collectively, these data suggest no overall trade-off between physiological and structural plasticity, but rather that structural and physiological plasticity scale positively or are independent (Figs 3–6).

The previous suggestions of greater morphological plasticity in shade-tolerant species have been based on limited data (Valladares et al. 2002; Niinemets & Valladares 2004). In fact, Valladares et al. (2002) comparing structural and physiological attributes of Q. robur and Fagus sylvatica found that the more tolerant species F. sylvatica was morphologically more plastic. However, the data provided by Welander & Ottosson (1998) for the same two species suggest that the plasticity is similar if not larger in Q. robur. This suggests that more interspecific comparisons, preferably with replications should be conducted. Our experiments were replicated in different years, and consistent results were observed for species ranking according to plasticity.

Our data indicate that having more leaves and greater flexibility to change the amount of leaves implies greater variation in both LAR and NAR (Fig. 3), and overall covariation in physiological and structural plasticity (Figs 3–6). In frequently fluctuating and disturbed environments as is common in early stages of primary succession or in secondary succession, highly variable plant structure apparently is advantageous as it allows the plants to more efficiently use environmental resources and achieve higher growth rates (Givnish 1988). Fully mature leaves have a limited potential to re-acclimate to new conditions (Naidu & DeLucia 1997; Yamashita, Koike & Ishida 2002), and continuous production of leaves brings about a greater potential to adjust to rapid changes in environment. However, enhanced self-shading and lower amount of biomass available for the growth in the next year(s) are apparent weaknesses of the highly plastic strategy in low light environments. While self-shading can be alleviated by architectural modifications (Niinemets 1998; Bloor & Grubb 2004), these modifications require expensive investments in branch framework. In addition, continuous production of leaves and small twigs apparently depletes carbohydrate reserves, and this can compromise plant survival in long-term (Kobe 1997; Canham et al. 1999). Furthermore, as leaf pay-back time is longer in lower light (Williams, Field & Mooney 1989), higher investment of resources in defence compounds in shade-tolerant species has been postulated (Kitajima 1994). Such a higher leaf cost can further constrain the plasticity in shade-tolerant species.

While plasticity generally evolves in response to predictable changes in environment such as overall changes in light availability during succession (Schlichting 1986; Schlichting & Pigliucci 1998), light environment can also rapidly change in established communities due to lightflecks during the day as well as due to gap formation. Gap formation in the canopy is a random event, but it is important that greater plasticity to light achieved through rapid leaf area production also confers a greater potential to adjust to rapid changes in light conditions. This reasoning is consistent with the evidence demonstrating higher gap-responsiveness of light-demanding species (Kiama & Kiyiapi 2001; Krause et al. 2001; Coates 2002).

Lower plasticity of physiological and structural characteristics in tolerant species also has relevant implications for interpreting the values of structural and physiological traits at any given irradiance. Sack & Grubb (2001, 2003) analysing RGR vs. Log(light) relationships, suggested that whenever the slopes of these relationships differ for any given two species, there should be a cross-over of species RGR rankings. Given the lower slopes in shade-tolerant species, the values of RGR are expected to equalize as was observed in our study at 5% light (Table 3), or become higher in shade-tolerant species with further reductions in irradiance. Such cross-overs in species ranking with decreasing irradiance have been observed in a series of species (Latham 1992; Walters & Reich 2000a), further confirming that having greatest plasticity does not necessarily imply greatest fitness across the entire light gradient.

plasticity responses to enhanced nutrient availability

In natural environments, nutrient availability and irradiance gradients often interact (see Introduction). This is associated with the circumstance that nutrient requirement is larger and growth is more responsive to nutrients at higher light (Ingestad & McDonald 1989; Poorter & Nagel 2000). As noted by several researchers, shade-tolerant temperate species are rarely found in infertile sites (Keddy & MacLellan 1990; Franklin et al. 1993; Schreeg et al. 2005; Kobe 2006). The limited competitive potential of shade-tolerant species at low nutrient availabilities has been associated with large nutrient costs for constructing extensive leaf area (Lusk & Contreras 1999). Our study indicates that nutrient availability can crucially alter species responsiveness to light. In particular, the plasticity in shade-tolerant species A. platanoides was most responsive to changes in nutrient availability (Fig. 4C). In addition, increases in light availability resulted in most dramatic decreases in nutrient concentrations in control treatment in this species (Fig. A1). Similarly, Burton & Bazzaz (1995) observed greatest reductions in growth in shade-tolerant species at low fertility.

Overall, there is a large variation in species capacity to extract nutrients from soil at common nutrient availability (Portsmuth & Niinemets 2006), and this apparently can modify species plastic responses. Previous work suggests that growth is more sensitive to nutrients in fertile site adapted species (Lusk, Contreras & Figueroa 1997). In secondary succession, soil nutrient availability is often high, and large growth rates and NAR observed in fertilized shade-tolerant species at high light in our study can explain a large competitive potential of shade-tolerant species even in open patches of rich sites (Kloeppel & Abrams 1995; Anderson 1999). In secondary succession, light availability decreases faster at higher nutrient availability due to enhanced leaf area growth (Sampson & Allen 1998; Bergh et al. 1999). Thus, greater plasticity of shade-tolerant species at richer sites can be an important adaptive feature allowing these species to enhance light harvesting, establish under developing canopy and finally reach the canopy. Furthermore, late-successional forests are also characterized by certain dynamics in light environment (Nakashizuka & Iida 1996). Given that a gap in the canopy is closed faster in richer than in poorer sites, larger plasticity may be beneficial for plant establishment in dense late-successional forests as well.

The larger nutrient requirement for plasticity expression in shade-tolerant species is consistent with the hypothesis that species cannot be simultaneously tolerant to multiple environmental factors (Niinemets & Valladares 2006). This hypothesis can be especially pertinent for canopy tree species that have large inertia because of extensive biomass costs for formation of roots, stems and foliage. Although juveniles of canopy trees can be potentially more plastic, modification of allocation patterns to resist additional stresses in young plants can compromise the competitive ability of mature plants in the canopy. For instance, there is evidence that the asymptotic height of canopy trees and the physiological potentials of seedlings are tightly related (Thomas & Bazzaz 1999). Negative correlations of shade tolerance with drought and water logging tolerance have been reported (Battaglia & Sharitz 2006; Niinemets & Valladares 2006) (but see Sack 2004; Sánchez-Gómez et al. 2006b). Our study further suggests that limited tolerance of low nutrient availabilities exerts additional limitations to niche differentiation, especially in northern temperate forests where the growing season is short. Overall, the results of this study suggest that more experiments with interacting environmental factors are needed to determine the limits of plasticity and understand species differentiation along the gap–understory continuum.

Acknowledgements

We thank Prof. Richard K. Kobe for thoughtful comments on the manuscript and the Estonian Science Foundation (Grant 5702), the Estonian Ministry of Education and Science (Grant 0182468As03), the Estonian Academy of Sciences, and the Estonian National Culture Foundation for financial support.

Ancillary

Advertisement