- Top of page
- Theory and model
It is widely believed that maximization of net photosynthesis (gross photosynthetic rate − dark respiration rate) drives the development of many plants. Consequently, many formulations of optimal canopies have been based on optimization of photosynthesis with respect to different plant properties, such as leaf nutrient concentration (e.g. Field 1983); leaf area per leaf mass (SLA, e.g. Schieving & Poorter 1999); and canopy leaf area per ground area (LAI, e.g. Anten et al. 1995a). This study explores the consequences for canopy development of adding two connected processes, senescence and nutrient resorption.
Because photosynthesis is strongly dependent on tissue nutrient concentration, particularly N, and irradiance, the distribution of N in the canopy with respect to light should be an important plant property. Field (1983) proposed an optimal canopy N distribution, where N is distributed such that no transfer of N between different positions in the canopy increases canopy net photosynthesis. This leads to leaf N concentrations proportional to the irradiance. This hypothesis has been tested many times and, although it does not fully explain the observations, it is a main determinant for canopy N distribution in many plants (e.g. Anten, Schieving & Werger 1995b; Field 1983; Hirose & Werger 1987b).
For a given amount of canopy N, it is possible to calculate the optimal LAI, i.e. the LAI that maximizes photosynthesis. Theoretical predictions of optimal canopy LAI and experimental measurements were compared by Anten et al. (1995a), who found that observed canopy LAIs were larger than predicted for a given amount of canopy N. A possible explanation for these results, suggested by Schieving & Poorter (1999), was that higher LAI and lower N concentration than optimal results from competition between plants differing in SLA. However, none of the above-mentioned studies includes leaf senescence or N resorption as optimizing processes. The question is, how will these processes affect plant properties such as N concentration, LAI and resorption efficiency?
Nutrient resorption during leaf senescence is often expressed as nutrient resorption efficiency (fraction of N resorbed at senescence relative to green leaf N concentration) or resorption proficiency, the lowest litter N concentration that can be reached by resorption (Killingbeck 1996). A complete understanding of the controlling factors is lacking, and their relation to plant nutrient status is unclear (Aerts 1996). In a review of many studies, Aerts (1996) found both positive and negative relations between resorption efficiency and plant N concentration within groups of plants of the same type (e.g. deciduous trees). The only significant relation between resorption efficiency and N concentration was a positive one for forbs. Unfortunately, resorption has rarely been studied in relation to canopy properties such as LAI and N distribution.
The aim of this study was to combine the concept of optimal canopy N distribution with simple mathematical descriptions of principles for the onset of leaf senescence and N resorption. Implications for relations between plant N concentration, LAI and resorption efficiency were investigated.
- Top of page
- Theory and model
Anten et al. (1995a) found that monospecies canopy LAIs were larger than optimal canopy LAIs predicted by maximization of canopy photosynthesis (cf. Rf = 1 in Figs 2 and 7). Schieving & Poorter (1999) proposed an explanation where competition between species of different SLA results in an evolutionary stable LAI higher than the optimal LAI. We present here an alternative explanation that does not rely on assumptions of multispecies competition to shape the canopy. Instead of being a direct result of optimization of photosynthesis at given levels of resources, LAI grows until it is constrained by litter formation. LAI may then be constant, although new leaves and litter are formed, and plant height and shape may change. This continuous growth of the canopy height is a means of competing for light. New leaves are formed in the top, while bottom leaves receive less light, become less productive, and eventually are shed. However, N cannot be redistributed from old to new leaves without losses, unless resorption is complete (cf. Rf = 1, eqn 8). Thus under N-limited conditions LAI is a result of a dynamic process, and not solely of optimization of photosynthesis at a fixed amount of canopy N.
Figure 7. Litter formation-constrained LAI (W) vs. canopy N, Nc, for different species (Table 2). Squares are measured values at different N treatments from Anten et al. (1995a). Rf* has been chosen for best fit to data and corresponds to litter N concentrations NL = 17·4, 23·1, 20·3, 9·25 for Amaranthus cruentus, Oryza sativa, Glycine max and Sorghum bicolor, respectively.
Download figure to PowerPoint
Comparing the model predictions of the relationship between LAI (W) and canopy N (Nc) with the experimental results by Anten et al. (1995a) for four different plant species (A. cruentus, Glycine max, Oryza sativa and Sorghum bicolor) and high- and low-N treatments shows that, for all species and treatments except one, the LAI values lie almost exactly on the predicted curve for resorption factor Rf = 0·7 (Fig. 7). The deviant, S. bicolor at high-N treatment, should not be seen as a deviation from the model prediction; rather, it may be caused by such a high N availability relative to growth rate that the observed LAI occurs at an N concentration where litter formation is not yet beneficial (cf. Figure 1), below the Rf = 0·7 curve in Fig. 7. The lines for Rf* (constant litter N concentration) fit the observed change in canopy LAI between high- and low-N treatment slightly less well than the curves for resorption fraction (Rf). Furthermore, significantly different values of Rf* are required for fitting the different species in Fig. 7.
This study has focused on changes in N and LAI at constant irradiance (I0) and light extinction coefficient, k. Changes in k can be used to represent different plant densities – a lower density leads to more light dispersing between plants, reducing k. Halving k (Fig. 8) shows that for small plants with Rf ≥ 0·7, the effect on W is marginal, although the effect is larger for higher Nc and LAI, when light limitation increases in importance. In accordance with this prediction for small plants, Bazzaz & Harper (1977) found that leaf senescence in Linum usitatissimum was initiated at a constant plant biomass of 13 g for equally sized pots, independent of planting density. Furthermore, they found that the biomass at initiation of senescence was increased only by addition of nutrients, which is also predicted by our model (Fig. 1).
Area-based resorption efficiency is predicted to increase with LAI and Nav, which seems to have been observed only for forbs (Aerts 1996). This could indicate that additional processes and metabolic costs associated with senescence and N retranslocation (Field 1983) play a more important role for other plant types, such as trees.
If these costs are comparable to costs for N uptake, it could be hypothesized that the different costs and benefits of N acquisition should be balanced. Furthermore, assuming that N resorption costs increase with increasing resorption fraction (Rf), another possible relation between N resorption efficiency and N availability emerges. A reduction in soil N availability would increase N acquisition costs, leading to an increase in Rf. This implies a negative relation between plant N concentration and resorption efficiency. However, because an increase in plant N concentration Nav and an accompanying decrease in Rf are seen to change resorption efficiency in different directions (Fig. 4), it is difficult to predict the combined effect. This could be one explanation for the wide variation in observed resorption efficiency and its relation to nutrient status (Aerts 1996; Killingbeck 1996). Furthermore, estimation of mass-based resorption efficiency is complicated by changes in SLA accompanying changes in N per area (N). If the predicted variation in N was entirely due to variation in SLA, our mass-based resorption efficiency would be equal to Rf = 0·7 independently of mean canopy N.
Other factors that influence resorption efficiency could be seasonal climate changes, leaf age and nonleaf nutrient pools. In a resorption model of Kull & Kruijt (1999), plant N and C are separated between leaves and a nonleaf common pool, and where resorption depends on the transfer between these pools. They conclude that herbaceous species have a small common pool compared to trees, implying that our model is more appropriate for herbs.
In conclusion, senescence and resorption should be viewed as properties of the canopy, rather than at the level of single leaves. Although the present model represents a highly simplified view of a plant, the suggested principles provide a simple mechanistic basis for interpreting observed N–resorption–LAI relations under N-limited conditions. The model should be applicable for single-species herbaceous plant canopies. Additional factors, such as costs of resorption, probably should be invoked to gain further insight into, for example, N availability–resorption relations in trees.