Optimal use of leaf nitrogen explains seasonal changes in leaf nitrogen content of an understorey evergreen shrub

Abstract

Background and Aims

Understorey evergreen species commonly have a higher leaf nitrogen content in winter than in summer. Tested here is a hypothesis that such changes in leaf nitrogen content maximize nitrogen-use efficiency, defined as the daily carbon gain per unit nitrogen, under given temperature and irradiance levels.

Methods

The evergreen shrub Aucuba japonica growing naturally at three sites with different irradiance regimes in Japan was studied. Leaf photosynthetic characteristics, Rubisco and leaf nitrogen with measurements of temperature and irradiance monthly at each site were determined. Daily carbon gain was determined as a function of leaf nitrogen content to calculate the optimal leaf nitrogen content that maximized daily nitrogen-use efficiency.

Key Results

As is known, the optimal leaf nitrogen content increased with increasing irradiance. The optimal leaf nitrogen content also increased with decreasing temperature because the photosynthetic capacity per Rubisco decreased. Across sites and months, the optimal leaf nitrogen content was close to the actual leaf nitrogen content and explained the variation in actual leaf nitrogen by 64 %. Sensitivity analysis showed that the effect of temperature on optimal nitrogen content was similar in magnitude to that of irradiance.

Conclusions

Understorey evergreen species regulate leaf nitrogen content so as to maximize nitrogen-use efficiency in daily carbon gain under changing irradiance and temperature conditions.

INTRODUCTION

Nitrogen (N) is a limiting resource for plant growth in most natural ecosystems (Aerts and Chapin, 2000). Of plant organs, leaves are the largest sink of N and more than half of leaf N is invested in the photosynthetic apparatus (Evans, 1989; Hikosaka, 2004). Consequently the light-saturated rate of photosynthesis (P_max) of leaves is strongly correlated with leaf N content per area (N_area). The photosynthetic performance of plants thus depends on the allocation of N to the leaf as well as environmental conditions.

Plants alter their N_area in response to changes in environmental conditions. It is well known that N_area is higher in sunny than in shady conditions (e.g. Field, 1983; DeJong and Doyle, 1985; Hirose and Werger, 1987; Niinemets, 1997). N_area also increases when plants are exposed to low temperatures (e.g. Weih and Karlsson, 2001; Hikosaka, 2005; Yamori et al., 2009). In a temperate forest understorey, both irradiance and temperature change greatly. A deciduous tree canopy brings large variation in irradiance with unfolding and shedding of leaves; irradiance is higher during the winter to spring period than in summer. Temperature usually changes from around 25 °C in summer to around freezing point or below in winter. Leaves of evergreens in the understorey are exposed to such climate rhythms during their lifetime. It has been shown in several understorey evergreen species that the N_area increases in winter when irradiance is high and temperature is low (Muller et al., 2005, 2009; Katahata et al., 2007). Why does the leaf increase its N content under such conditions?

Daily carbon gain may increase with increasing N_area, but excessive N investment may not confer a larger benefit on the leaf. It has been suggested that there is an optimal N_area that maximizes daily carbon gain per unit leaf N (daily nitrogen-use efficiency; daily NUE) (Hirose, 1984, Hirose and Werger, 1987). Previous studies have shown that the optimal N_area accounts well for the regulation of leaf N with changes in growth irradiance: the optimal N_area is higher at higher light availabilities (Mooney and Gulmon, 1979; Hirose and Werger, 1987; Hikosaka and Terashima, 1995). Anten et al. (1996) showed that the optimal N_area was close to the actual N_area in an understorey shrub in a tropical rain forest. These results suggest that plants have evolved to efficiently use N for photosynthesis.

When leaf temperature decreases to suboptimal levels, the photosynthetic rate decreases (Berry and Björkman, 1980) and the slope of the P_max–N_area relationship will also decrease. The negative effect on the photosynthetic rate can be offset by an increase in N_area (Sage and Kubien, 2007; Medek et al., 2011). Weih and Karlsson (1999) have argued that the N_area should increase at lower temperatures to compensate for the decreased slope of the P_max–N_area relationship, which may explain the increase in N_area in winter. Schieving (1998) theoretically showed that optimal N_area increased with decreasing slope of the P_max–N_area relationship. However, there seems to be no study that quantitatively compared the optimal and the actual N_area under a regime where both air temperature and irradiance were varying seasonally.

The present paper aimed to elucidate the importance of N_area that is regulated in response to seasonal changes in temperature and irradiance. An evergreen shrub Aucuba japonica, which grows naturally under different irradiance regimes, was used. Daily photosynthetic carbon gain in relation to N_area was determined and the optimal N_area that maximized daily NUE calculated. Sensitivity analysis was applied to distinguish the effect on N_area of irradiance from that of temperature.

MATERIALS AND METHODS

Species and growth conditions

Aucuba japonica Thunb., Aucubaceae, is a common evergreen understorey shrub in Japan. Plants growing naturally at three sites different in irradiance in the same forest area (Botanical Garden, Tohoku University, Sendai, 38 °14'N, 140 °50'E) were used: (1) under a deciduous Quercus serrata-dominated canopy (deciduous site); (2) under an evergreen Cryptomeria japonica-dominated canopy (evergreen site); and (3) in a gap in a deciduous forest (gap site). The mean leaf life span of an A. japonica shrub is 2·1 and 2·6 years in a deciduous and evergreen forest, respectively, in central Japan (Yamamura, 1986). New leaves unfold in April when 2- or 3-year-old leaves are shed in about 1 month. Fully exposed leaves at the top (100–150 cm) of the shrub that unfolded in spring 2001 throughout this study were used. In May and June 2002 these leaves were sometimes partly shaded by new leaves. Growth conditions at the study sites were characterized by monitoring photosynthetic photon flux density (PPFD) and air temperature at the average height of the studied individuals every 15 min through the study period. For detailed description of the study sites and growth conditions see Muller et al. (2005).

Photosynthetic, leaf nitrogen and Rubisco measurements

Photosynthetic rates were measured monthly from August 2001 to July 2002 using a portable LI-6400 photosynthesis system (Li-Cor, Lincoln, NE, USA) on attached leaves at an ambient CO₂ concentration of 370 µmol mol⁻¹. Leaf temperature was set at the ambient average temperature that was determined from measurements for 2 weeks before each photosynthetic measurement. The photosynthetic rate was measured first at the saturating irradiance (1000 µmol m⁻² s⁻¹; P_max) and then at irradiance levels of 40, 30, 20 and 10 µmol m⁻² s⁻¹ to determine the initial slope of the irradiance response curve (quantum yield) by linear regression. The respiration rate was measured after leaving the leaf for 5 min in the dark. After photosynthetic measurements, the leaves were cut at the petioles, put in plastic bags and taken within 1 h after cutting to the laboratory, where five to ten leaf discs with a diameter of 1 cm were punched out for N and Rubisco analysis. Three leaf discs per leaf were dried at 80 °C in an oven for >3 d and N content was determined with an NC-analyser (NC-80; Shimadzu, Kyoto, Japan). The remaining leaf discs were stored at –80 °C for determination of Rubisco by SDS–PAGE as described in Muller et al. (2005).

Calculation of daily carbon gain and optimal N content

Photosynthesis–irradiance response curves were expressed with a non-rectangular hyperbola (Johnson and Thornley, 1984):

(1)

where P is the instantaneous net photosynthetic rate (μmol CO₂ m⁻² s⁻¹), I the photon flux density (μmol m⁻² s⁻¹), φ the apparent quantum yield (μmol CO₂ μmol⁻¹), θ the curvature factor (dimensionless) and R the respiration rate (μmol CO₂ m⁻² s⁻¹⁾. The daily carbon gain (P_day)

(2)

was obtained as an average carbon gain calculated using the actual PPFD measured at 15-min intervals for 2 weeks and the actual P_max, φ and R determined for three to six leaves per site in each month. φ was obtained from a linear regression of the photosynthetic rate on PPFD between 0 and 40 µmol m⁻² s⁻¹. The curvature factor was assumed constant at 0·9, the value previously determined for A. japonica (O. Muller, unpubl. res.). Leaves were assumed to be displayed horizontal during the year and daily changes in temperature were not taken into account.

To determine the optimal N_area, photosynthetic rate was modelled as a function of N_area. P_max was expressed as a function of Rubisco and Rubisco as a function or N_area (Figs 3–5)

(3)

(4)

where [Rubisco] is the Rubisco content per unit leaf area, k indicates the specific activity of Rubisco, i.e. the slope of the P_max–Rubisco relationship (Fig. 4A–C) and is also defined as the Rubisco-use efficiency (Hikosaka and Shigeno, 2009). a_b and b_b are the slope and the intercept of linear regression of Rubisco content on N_area, respectively (Fig. 3B). R was expressed as a function of N_area:

(5)

where a_r and b_r are the slope and the intercept of linear regression, respectively (Fig. 3A). Values of φ were adopted from measured values and constant θ of 0·9 was used. Using determined PPFD, P_day was expressed as the function of N_area. The optimal N_area (N_opt) that maximizes daily N-use efficiency, defined as the ratio of P_day to N_area, was calculated numerically.

Fig. 3.

(A) Linear regression of respiration rate against N_area for August–November (y = 0·0008x + 0·1175, r² = 0·20, P = 0·007), December–February (y = 0·0007x + 0·063, r² = 0·14, P = 0·014) and March–June (y = 0·0011x + 0·1468, r² = 0·11, P = 0·003), as indicated in the key. (B) Linear regression of Rubisco against N_area for A. japonica growing in a gap, under a deciduous canopy or evergreen canopy, as indicated (y = 0·02x – 0·69, r² = 0·81, P = 0·000).

Fig. 5.

(A) Linear relationship of P_max and N_area for August and January, as indicated. (B) The daily carbon gain and (C) daily NUE as function of N_area calculated for the deciduous forest conditions in August (low PPFD, high T) and January (high PPFD, low T) and for January conditions except for the k of August (high PPFD, high T). Circles indicate when the daily NUE is maximal (at the tangent of the origin for the daily carbon gain versus N_area).

Fig. 4.

(A) Linear regression for P_max and Rubisco content, forced through the origin, for August, October and January, as indicated. (B) The slope of the linear regression for P_max and Rubisco content (k) as a function of monthly temperature, with the fitted line obtained from the calculated temperature dependence of Rubisco-limited photosynthesis (Bernacchi et al., 2001) at a constant yearly average intercellular CO₂ concentration of 244 µmol mol⁻¹ adjusted to the k of November. (C) k as a function of time.

Sensitivity analysis was performed to evaluate the effect of irradiance level and temperature on optimal N_area. It was assumed that irradiance directly influenced the P_day, and that temperature influenced the P_day through changes in k. Optimal N_area (N_pAug) was calculated with PPFD kept at August values of each site while all other parameters were taken from the values observed from each of months and sites. If the variation in PPFD has no effect on the variation in N_opt, N_pAug would be similar to N_opt. Similarly, optimal N_area (N_tAug) was calculated keeping k constant at August, values while all other parameters were taken from the observed values in each month and site.

Statistical analysis was done with R (R ver. 2·11·1; R development core team, 2005).

RESULTS

Seasonal changes in the environment and leaf traits have been reported previously (Muller et al., 2005). As they are essential for the present study, they are described briefly here. The daily average PPFD at the level of the top of the crown of Aucuba japonica was highest in the spring when all deciduous trees dropped their leaves (Fig. 1A). The gap site had the highest PPFD and the evergreen site the lowest. The daily average temperature was lowest in February and highest in August without a significant difference between the three sites (Fig. 1B). Leaf N_area and Rubisco content increased from summer to the highest values in winter and/or early spring (Fig. 1C, D). They were highest at the gap and lowest at the evergreen site, corresponding to their irradiance. P_max was highest at the gap, but it decreased gradually from August to June with its lowest value in February. P_max was lower at the deciduous site and still lower at the evergreen site, both having peaks in summer and spring. The quantum yield maintained high values until autumn/winter and then decreased to June with large variations between sites in winter (Fig. 1F).

Fig. 1.

Monthly changes from August 2001 to June 2002 in environment and leaf traits: (A) average daily integrated PPFD, (B) average daily temperature, (C) leaf nitrogen content per area, (D) Rubisco content per area, (E) light-saturated photosynthetic rate, P_max, and (F) quantum yield for A. japonica growing in a gap, under a deciduous canopy or evergreen canopy, as indicated. Standard error bars are shown. (A–E) are redrawn with permission from Muller et al. (2005).

The daily carbon gain changed seasonally and differed between sites (Fig. 2). It was highest at the gap, lowest at the evergreen and intermediate at the deciduous site except in February when it was lower in the gap than at the deciduous site. At the deciduous and the evergreen site the daily carbon gain increased from September to its highest values in March. At the gap the highest gain was found in August and decreased until February and increased again to another maximum in March (Fig. 2). As respiration showed considerable variation, its relationship to N_area was analysed on pooled data divided into three seasons: August to November, December to February and March to June. In each season respiration increased with N_area (Fig. 3A). The respiration per unit N was highest in March to June and lowest in December to February.

Fig. 2.

The daily carbon gain from August 2001 to June 2002 for A. japonica growing in a gap, under a deciduous canopy or evergreen canopy, as indicated. Standard deviation bars (n = 3–6) are shown.

Across months Rubisco content was linearly related to N_area and the relationship did not differ among the three sites (ANCOVA with site as factor and Rubisco as variable and N_area as covariate P > 0·05 among slopes and among intercepts; Fig. 3B). A linear regression line through the origin was fitted to the relationship between P_max and Rubisco in each month (Fig. 4A). The Rubisco activity, k, i.e. the slope of P_max to Rubisco content, decreased from summer to winter, slightly increased from February to April, and then decreased to June (Fig. 4B). k tended to be higher at higher temperatures (Fig. 4C; note that P_max was measured at the average temperature in each month).

The daily carbon gain as a function of N_area was calculated according to eqns (1)–(4). Figure 5A shows examples of P_max as a linear function of N_area relationship in the warmest and coldest months (August and January). The slope was different between months due to the change in k (Fig. 4C). Figure 5B shows how variation in k affects the P_day–N_area relationship: continuous and broken lines show the P_day–N_area curve calculated using the same high PPFD (deciduous site in January) with the k value from August and January, respectively. The tangent from the origin shows the optimal N_area (N_opt) that maximizes P_day per unit of leaf nitrogen (daily NUE; Fig. 5C) (Hirose, 1984). Decreasing k (or the slope of the P_max–N_area relationship) leads to an increase in N_opt because the P_day–N_area curve is more flattened. The dotted line was calculated using k at August and the low PPFD at the deciduous site in August: the difference between continuous and dotted lines reflects the difference in PPFD. As has been shown in previous studies (e.g. Hirose and Werger, 1987), N_opt was higher at higher PPFD.

The N_opt determined monthly for the three sites is shown along with seasonal dynamics of actual N_area (Fig. 6). The N_opt was highest and lowest in the gap (Fig. 6A) and the evergreen site (Fig. 6C), respectively, like the actual N_area. The changes over the year were larger than the actual N_area, especially in the deciduous site (Fig. 6B), but followed a similar pattern that was tested by linear regression of optimal N versus actual N (Fig. 7 and Table 1). In gap and deciduous sites, for example, both actual N_area and N_opt increased from summer to autumn and were relatively constant during winter. On the other hand, they continuously increased from summer to spring in the evergreen site. Actual N_area was strongly correlated with N_opt; although the actual N_area approached N_opt with a slope of 0·6 and small intercept, most of the points distributed around the 1 : 1 relationship. N_opt explained 64 % of the variance of actual N for all the months and three sites pooled. Per site the amount of variance explained was highest for the deciduous site (87 %) and lowest in the gap site (30 %). Per site the slope was lower (≤0·4) and the intercept larger (Table 1).

Fig. 6.

Seasonal dynamics of the optimal leaf N content in the gap (A), deciduous (B) and evergreen site (C), shown together with the actual leaf N and results of sensitivity analysis with N_pAug (constant PPFD) and N_tAug (constant temperature, k), when PPFD or k would have no effect N_pAug or N_tAug would be the similar to optimal leaf N; see text for more details.

Fig. 7.

Linear regression of optimal versus actual leaf N content for A. japonica growing in a gap, under a deciduous canopy or evergreen canopy, as indicated. The broken line indicates the linear regression of all months and sites pooled, and the 1 : 1 line is also shown.

Table 1.

Linear regression of optimal nitrogen versus actual nitrogen

Model	Intercept	P	Slope	P	R2
All	52·3	<0·001	0·6*	<0·001	0·64
Gap	102·9	0·005	0·4*	0·046	0·30
Deciduous	75	<0·001	0·3*	<0·001	0·87
Evergreen	64·8	<0·001	0·4*	0·002	0·62

The optimal nitrogen is calculated for each month in each site at corresponding PPFD and photosynthetic parameters. Regresssion values, difference from 1 for the slope (*P < 0·05) and R² are shown for all sites and months pooled (all) and per site.

The predictive value of N_opt is quantified by the deviation from the 1 : 1 line. The deviation is calculated as the sum of squares of the difference between N_opt and N_area divided by the number of data points. The deviation for N_opt from N_area was 7·4 for all months and three sites pooled. Per site the deviation was highest (11·2) in the deciduous site and lowest (5·2 and 5·9) in the evergreen and gap sites, respectively. In the sensitivity analysis the importance of PPFD or k is shown also by the deviation of N_opt from N_area; the larger the deviation at a constant PPFD or k the higher its contribution. The deviation increased by approx. 70 % in both cases (12·6 and 12·3 at constant PPFD and k, respectively), suggesting that both seasonal changes in PPFD and k considerably influence N_area and that the effect of temperature was similar in magnitude to that of irradiance.

DISCUSSION

Variation in the actual N_area was well explained by the optimal N_area (N_opt; Figs 6 and 7 and Table 1), suggesting that N_area was regulated to maximize daily N-use efficiency in the seasonal environment where both irradiance and temperature were changing. The present result is consistent with previous studies that showed a strong correlation between N_area and N_opt across different irradiance levels (Hirose and Werger, 1987; Anten et al., 1996). Moreover it was shown that N_opt was higher when air temperature is low. The present sensitivity analysis demonstrated that the effect of air temperature on N_opt was as strong as that of irradiance (Fig. 6 and Table 1). This is the first paper to report that not only irradiance but also temperature is an important factor affecting the optimal N_area and that actual plants alter their N_area to maximize daily N-use efficiency of carbon gain in response to seasonal temperature changes. This is consistent with the result of a multiple regression analysis that both temperature and irradiance had significant effects on the N_area (Muller et al., 2005).

Effects on N_opt are different between light and temperature. When light availability decreases, photosynthetic rates become less sensitive to N_area because photosynthesis is limited by light rather than by N (Hirose and Werger, 1987). Moreover, respiration increases with increasing N_area (Fig. 3A), leading to a decrease in N_opt with decreasing light (Hirose and Werger, 1987; Fig. 5B). When temperature decreases, on the other hand, whilst the slope of P_max–N_area decreases (Fig. 5A), the response of P_day to N_area does not vary if N_area is scaled as a relative value. N_opt increases with decreasing temperature to compensate for a decrease in the slope of P_max–N_area relationship.

The theory of optimal N_area implicitly assumes a trade-off between N_area and leaf area development: if plants are given a certain amount of N, they may use it to increase leaf area instead of increasing N_area or to increase N_area instead of increasing leaf area (Hirose, 1984). If leaf area is increased, more light will be intercepted. If N_area per area is increased, P_max will increase. There is an optimal N use to maximize carbon gain. This situation may not be strictly applicable to an evergreen species like Aucuba japonica because its leaf area per plant is almost constant after the development of new leaves in early summer when it also sheds its leaves (Yamamura, 1986). However, leaf phenology of A. japonica is well co-ordinated with the optimal N_area theory. Aucuba japonica develops new leaves in early summer when N_opt is low. In summer and autumn, plants may absorb N from soil and N_area increases from summer to winter (e.g. Millard et al., 2001), when N_opt increases. When spring comes, N_opt decreases and part of leaf N is retranslocated to be used for development of new leaves. In other words, A. japonica produces new leaves utilizing N that becomes excessive in existing leaves. This might imply that leaf phenology, including the timing of leaf emergence are determined to attain N_opt.

N_area increased at low temperatures as has been shown in previous studies (Weih and Karlsson, 2001; Hikosaka, 2005; Muller et al., 2005; Yamori et al., 2009). This response has been regarded homeostatic, where leaf N is increased to compensate for the decrease in the P_max–N_area slope at low temperatures to maintain P_max independent of growth temperatures (Yamori et al., 2009). In the present study, however, P_max decreased considerably in winter (Fig. 2E). It was shown that even though P_max decreased, NUE was maximized with an increase in N_area in winter. This result suggests that N_area may be regulated to maximize NUE rather than to maintain P_max constant.

One may consider that an increase in N_area is a passive result of nitrogen economy (Körner and Larcher, 1988; Körner, 1989): A. japonica does not develop any leaves after summer so N taken up during summer and autumn might be accumulated in leaves. If so, an increase in N_area would be similar among sites irrespective of PPFD. In the present study, however, the pattern of N_area change was different among sites; in gap and deciduous sites, for example, N_area increased from summer to autumn and was relatively constant during winter, while it continuously increased from summer to spring in the evergreen site. (Figs 1 and 6). Such patterns were also found in N_opt (Fig. 6). Yamamura and Kimura (1992) showed that in A. japonica, a steep increase in N_area in autumn is accompanied by a decrease in root N concentration. Weih and Karlsson (2001) showed for Betula pubescens growing at different combinations of soil and air temperature that an increase in leaf N at low temperature was not only a passive response but also acclimation to a lower air temperature. These facts suggest that N_area change is active acclimation to environmental change rather than a passive result.

The regression line between the actual and N_opt had the intercept greater than 0 and the slope smaller than 1 (Fig. 6). This pattern was more noticeable when regressions were taken separately for the three sites. This suggests that the plasticity in N_area was smaller than the optimization theory predicted. The limitation in the plasticity may come from the cost of acclimation. An increase in N_area involves protein synthesis and an increase in chloroplast volume, and the inverse is the case when leaf N decreases (Oguchi et al., 2003, 2005, 2006; Muller et al., 2009). If such changes are very costly, optimization may not be advantageous. Significant deviation of the actual N_area from the optimal suggests that the costs were not small.

It was assumed that the Rubisco activity, k, was affected by temperature only. To test this assumption, the temperature dependence of the photosynthetic rate was calculated using Farquhar's photosynthesis model (Farquhar et al., 1980) with published temperature dependence of Rubisco kinetics (Bernacchi et al., 2001). A line in Fig. 4B demonstrates the calculated temperature dependence of Rubisco-limited photosynthesis per unit Rubisco content at a constant CO₂ concentration, which is normalized at the k of November. It fits to most of the observed variation, validating the assumption that k varies with temperature. However, data obtained in June deviated from the line, suggesting that other factors are also involved in the variation of k. As the June leaf was the oldest of the leaves studied, the deviation may be attributable to an age-dependent decline in k, which has been observed in evergreen leaves (Kitajima et al., 2002; Escudero and Mediavilla, 2003; Han et al., 2003; Motomura et al., 2008).

Modelling optimal plant functioning was suggested to be a powerful tool to predict forest responses to global change (Dewar et al., 2009). Where A. japonica occurred naturally, there were complex interactions between temperature and irradiance in their effect on leaf photosynthesis and no clear correlation was found across sites and months (Muller et al., 2005). However, it was shown in the present study that optimal leaf N for maximizing NUE explained well the actual leaf N in the environment, where temperature and irradiance were changing seasonally. This demonstrates the functional advantage of seasonal changes in N_area as explained by an optimization theory. This approach may be extended to plant response to inter-annual and longer-term changes in climate.