RGRMAX and Plant Distribution

Bradshaw et al. (1964) were among the first to establish a relationship between the RGR_max of wild species, as measured under laboratory conditions, and the characteristics of the habitat they originated from. Others followed, but generally the number of species was rather low (less than 10) to infer strong conclusions. Grime and Hunt (1975) determined RGR_max of 130 species from England and classified them according to habitat. Fast-growing species were found relatively more often in fertile habitats, whereas species with a low potential growth rate tended to occupy infertile habitats. It is not always easy and straightforward to quantitatively classify a specific habitat along a fertility scale. A semiquantitative approach has been used by Ellenberg, who assigned so-called ‘‘N-numbers’’ to a wide range of species from Central Europe (Ellenberg 1988). The higher the value, the higher the fertility of the habitats in which such a species would generally occur. Plotting RGR_max data of herbaceous perennials against the N-number of Ellenberg generally yields positive relationships (Figure 3.2a). A positive relation between RGR_max and nutrient availability is also likely for woody species (Cornelissen et al. 1998). However, annuals seem to have a high RGR_max independent of soil fertility (Fichtner and Schulze 1992).

Is inherent variation in potential RGR of species also related to other environmental gradients? Evidence is less well-documented than in the case of nutrient availability. Alpine species have lower RGR_max under laboratory conditions than lowland species (Figure 3.2b). Dry (Rozijn and Van der Werf 1986, Figure 3.2c) or saline habitats (Figure 3.2d) harbor species that grow more slowly under optimal conditions, and the same relation between RGR_max and plant occurrence was found for sites with heavy metal pollution (Wilson 1988, Verkleij and Prast 1989). Disturbance regime may also be a source of variation in RGR_max: annuals that have to complete their life cycle in periodically disturbed habitats, display a higher RGR_max than congeneric perennials from more stable habitats (Garnier 1992), and species from early stages of secondary succession tend to have higher RGR_max than those from more advanced stages (Gleeson and Tilman 1994, Vile et al. 2006). The intensity of trampling has also been identified as selecting for species with different RGR_max: plants from trampled places have a lower RGR_max than those from nontrampled sites (Figure 3.2e). Finally, when determined at relatively high-light levels, species from strongly shaded habitats have a lower RGR than those from light-exposed environments. In most of the cases shown in Figure 3.2b through Figure 3.2f, however, it seems that differences in RGR_max between species from favorable and unfavorable habitats are not as clear as in the case of species adapted to habitats differing in fertility.

RGRMAX and Ecosystem Productivity

Most of this chapter focuses on the comparison between plant species. However, in recent years there has been much attention for ecosystem functioning and a possible link with species composition. The fact that inherently fast-growing species are more often found in nitrogen-rich habitats (sensu Ellenberg 1988), and nitrogen-rich habitats often show a higher productivity, makes it likely that there is in this case a positive correlation between ecosystem behavior and the RGR of the composing species as determined in the laboratory. Vile et al. (2006) tested this correlation in a Mediterranean habitat, following secondary succession in abandoned vineyards. The aboveground net primary productivity, expressed per gram of biomass present at the beginning of the growing season, varied fourfold between sites and was negatively correlated with field age. The decrease in productivity was correlated with a change in species composition, such that species more abundantly present at later succes- sional stages were those that were showing lowest RGR_max in growth room experiments (Figure 3.3a).

Figure 3.2   RGRmax of species originating from habitats differing in: (a) Nutrient availability, as indicated by the N-number of Ellenberg; (b) Altitude, in meters above sea level; (c) Rainfall; (d) Salt concentration; (e) Trampling intensity; and (f) Light intensity. Each study is represented by a regression line. Numbers refer to the following studies: A: data from (1) Rorison, I.H., New Phytol., 67, 913,1968; (2) Grime, J.P. andHunt, R., J. Ecol., 63, 393,1975; (3) Boorman, L.A., J. Ecol., 70, 607,1982; (4) Poorter, H., Causes and Consequences of Variation in Growth Rate and Productivity of Higher Plants, H. Lambers, M.L. Cambridge, H. Konings, and T.L. Pons, eds, SPB Academic Publishing, The Hague, 1989; (5) Poorter, H. and Remkes, C., Oecologia, 83, 553,1990; (6) Reiling, K. and Davison, A.W., New Phytol., 120, 29, 1992; (7) Stockey, A. and Hunt, R., J. Appl. Ecol., 31, 543, 1994; (8) Van der Werf, A., Geerts, R.H.E.M., Jacobs, F.H.H., Korevaar, H., Oomes, M.J.M., and de Visser, W., Inherent Variation in Plant Growth. Physiological Mechanisms and Ecological Consequences, H. Lambers, H. Poorter, and M. van Vuuren, eds, Backhuys Publishers, Leiden, 1998; and (9) Schippers P. and Olff, H., Plant Ecol., 149, 219, 2000; B: data from (1) Woodward, F.I., New Phytol., 82, 385, 1979; (2) Woodward, F.I., New Phytol., 96, 313, 1983; (3) Graves, J.D. and Taylor, K., New Phytol., 104, 681, 1986; (4) Atkin, O.K. and Day, D.A., Aust. J. Plant Physiol., 17, 517, 1990; and (5) Atkin, O.K., Botman, B., and Lambers, H., Funct. Ecol., 10, 698,1996a; C: data from (1) Mooney, H.A., Ferrar, P.J., and Slatyer, R.O., Oecologia, 36, 103, 1978; (2) Wright, F.I. and Westoby, M., J. Ecol., 98, 85, 1999; and (3) Warren and Adams, Oecologia, 144, 373, 2005; D: data from (1) Ball, M.C., Aust. J. Plant Physiol., 15, 447, 1988; (2) Van Diggelen, J., A Comparative Study on the Ecophysiology of Salt Marsh Halophytes, PhD Thesis, Free University, Amsterdam, 1988; (3) Schwarz and Gale, J. Exp. Bot., 35, 193, 1984 and (4) Ishikawa S.I. and Kachi, N., Ecol. Res., 15, 241, 2000; E: data from (1) Dijkstra, P. and Lambers, H., New Phytol., 113, 283, 1989; (2) Meerts, P. and Garnier, E., Oecologia, 108, 438, 1996; and (3) Kobayashi, T., Ikeda, H., and Hori, Y., Plant Biol., 1, 445, 1999; F: data from (1) Pons, T.L., Acta Botanica Neerl., 26, 29, 1977; (2) Corré, W.J., Acta Botanica Neerl., 32, 49, 1983a; (3) Corré, W.J., Acta Botanica Neerl., 32, 185, 1983b; (4) Kitajama, K., Oecologia, 98, 419,1994; (5) Osunkoya, O.O., Ash, J.E., Hopkins, M.S., and Graham, A.W., J. Ecol., 82,149, 1994; and (6) Poorter, L., Funct. Ecol., 13, 396, 1999.

Figure 3.3   (a) Aggregated RGRmax and (b) aggregated SLA of plant species grown in the field plotted as a function of the specific aboveground net primary productivity (aboveground RGR) of 12 abandoned vineyards at different stages of secondary succession. Laboratory-derived data of the species present in the succession were averaged according to the proportion of biomass they represented in the field. (From Vile, D., Shipley, B., and Garnier, E., Ecol. Lett., 9,1061,2006 and Garnier, E., Cortez, J., Billés, G.,Navas, M.-L., Roumet, C., Debussche, M., Laurent, G., Blanchard, A., Aubry, D., Bellman, A., Neill, C., and Toussaint, J.-P., Ecology, 85, 2630, 2004.)

RGRMAX and Plant Strategies

We have shown earlier that in a variety of cases, there is a link between the potential growth rate of a species and its occurrence in a given habitat. As such, RGR_max forms one of the cornerstones in the plant strategy theory formulated by Grime (1979). According to this theory, plant strategies are shaped by the possible combinations of two factors experienced by plants: stress and disturbance. Stress in this sense is defined as the extent to which a combination of environmental variables retards growth (e.g., low nutrient availability, low or high temperature, low water availability). Disturbance is defined as the degree of physical disruption of the plant’s biomass (e.g., grazing, trampling). Species from habitats with a high degree of stress and a low degree of disturbance are called ‘‘stress-tolerators.’’ They are generally perennials with a low RGR_max (Figure 3.4). Species from sites with a high disturbance but with low stress are called ‘‘ruderals’’. They are mostly annuals that have a high RGR_max, enabling them to complete their life cycle quickly. This would ensure that seeds are produced before a disturbance event takes place, which kills the plant. Habitats in which both stress and disturbance are low are favorable for plant growth. According to the plant strategy theory, these are sites where a strong competition between plants is expected; consequently species that thrive here are called ‘‘competitors’’. They are generally perennials, and also have a high RGR_max. Sites with a high level of both stress and disturbance (volcanoes, nutrient- poor and strongly drifting sand dunes) do not bear plants, because there is no feasible strategy to cope with such an environment.

Given the importance that is generally attached to biomass gain and the relative fitness of a plant (discussed in McGraw and Garbutt 1990), it may be hypothesized that there has been a selection pressure in fertile (and favorable, even if only temporarily) habitats toward plant species with a high RGR, and toward a low potential RGR in places which are unfavorable (Grime 1979, Chapin 1980). However, RGR is a parameter that is the result of a combination of many physiological, morphological, anatomical, and biochemical traits. Alternatively, it could well be that it is one or more of these traits underlying RGR that has been the target of selection, rather than RGR itself (cf. Grime 1979, Coley 1983, Lambers and Dijkstra 1987). RGR would then merely be a by-product of selection. Before evaluating these contrasting hypotheses, we first have to investigate the components underlying RGR.

Figure 3.4   Relation between stress and disturbance and the growth strategy, life form, and RGR of the three types of strategies. (Adapted from Grime, J.P., Plant Strategies and Vegetation Processes, John Wiley & Sons, Chichester, 1979.)

Growth Parameters

Growth is more than photosynthesis. It is the balance between the carbon gain per unit leaf area (ULR) and the carbon losses in the plant (which depend on the respiration rate but also on the relative proportion of the assimilatory and nonassimilatory organs, and in the longer run on biomass turnover), corrected for the C-concentration of the newly formed biomass. Evaluating the relative importance of each of these factors requires a top-down approach, in which RGR is broken down into components. A common way to do so is factorizing RGR into the increase in mass per unit leaf area and time (ULR) and the leaf area per unit plant mass (LAR, leaf area ratio). LAR can be factorized further into the components leaf area: leaf mass (SLA, specific leaf area) and leaf mass: plant mass (LMF, leaf mass fraction). A definition of these components of RGR, as well as an explanation of the concept of growth parameters underlying RGR is given in Box 3.2.

 

BOX 3.2

Components of RGR

A simple framework to factorize RGR was developed at the beginning of the twentieth century (Blackman 1919, West et al. 1920). The basic assumption underlying this framework is that plant growth is dependent on photosynthesis and that leaf area is the plant variable driving total C-gain. RGR is then factorized into two components: ULR and LAR (Hunt 1982). ULR is defined as the increase in biomass per unit time and leaf area

3.6 $$\mathrm{U}\mathrm{L}\mathrm{R}=\frac{1}{A}\frac{\mathrm{d}M}{\mathrm{d}t},$$

where A is the total leaf area of the plant, dM the increase in mass over period dt. LAR is defined as the total leaf area per unit total plant mass

3.7 $$\mathrm{L}\mathrm{A}\mathrm{R}=\frac{A}{M}$$

and consequently

3.8 $$\mathrm{U}\mathrm{L}\mathrm{R}\mathrm{.}\mathrm{L}\mathrm{A}\mathrm{R}=\frac{1}{A}\frac{\mathrm{d}M}{\mathrm{d}t}\frac{A}{M}=\mathrm{R}\mathrm{G}\mathrm{R}\mathrm{.}$$

By determining leaf mass and stem and root mass separately, one is also able to break down LAR into two components: specific leaf area (SLA) and leaf mass fraction (LMF). SLA is the amount of leaf area per unit leaf mass (ML):

3.9 $$\mathrm{S}\mathrm{L}\mathrm{A}=\frac{A}{M_{\mathrm{L}}}$$

and LMF is the fraction of total plant mass that is invested in the leaves

3.10 $$\mathrm{L}\mathrm{M}\mathrm{F}=\frac{M_{\mathrm{L}}}{M}$$

and consequently

3.11 $$\mathrm{S}\mathrm{L}\mathrm{A}\mathrm{.}\mathrm{L}\mathrm{M}\mathrm{F}=\frac{A}{M_{\mathrm{L}}}\frac{M_{\mathrm{L}}}{M}=\mathrm{L}\mathrm{A}\mathrm{R}\mathrm{.}$$

The advantage of this approach is that it requires only data on progressions in leaf area and plant mass to obtain a good indication about the causes of variation in growth rate. This is because differences in ULR are often due to differences in the area-based rate of photosynthesis. However, ULR is, in fact, the net balance of total plant carbon gain and carbon losses, expressed per unit leaf area and corrected for the C-concentration of the newly formed biomass. If one really wants to obtain insight in the relation between the various C-fluxes and RGR, the following formula can be used

3.12 $$\mathrm{R}\mathrm{G}\mathrm{R}\mathrm{=}\frac{\mathrm{P}{\mathrm{S}}_{\mathrm{a}}\mathrm{F}\mathrm{C}\mathrm{I}}{\mathrm{P}\mathrm{C}\mathrm{C}}\mathrm{S}\mathrm{L}\mathrm{A}\mathrm{L}\mathrm{M}\mathrm{F}$$

Poorter (2002), where PSa is the total amount of C fixed per unit leaf area integrated over a 24 h period, FCI the fraction of that daily fixed carbon that is not respired in leaves, stems, or roots but invested in growth (so 1-RE/PS). The first four terms in the right-hand side of Equation 3.12 determine the net amount of C fixed per unit of biomass and per day. The denominator (PCC) converts this net amount of C into a biomass increase.

There are slightly different approaches, in which Equation 3.12 is written as the difference between carbon gain in photosynthesis and daily rate of respiration in leaves, stems, and roots. However, by using the form currently presented in Equation 3.12, with FCI (also termed carbon use efficiency), the equation is the product of five entities, which makes it amenable to the GRC analysis explained in the “Physiological Parameters’’ section on the next page. Note that in this equation C-losses due to exudation, or leaf and root turnover are considered to be negligible.

To what extent is inherent variation in RGR_max caused by variation in the components ULR and LAR? A wide variety of results has been published; some experiments found ULR to be the factor determining growth, others found LAR to be the cause of variation in growth, whereas others found intermediate results. Variation may be due to the choice of the species as well as growth conditions used and the experimental procedure followed (e.g., duration of the experiment and number of harvests). To enable a quantitative analysis of the cause of variation in RGR within a given experiment, we use the growth response coefficient (GRC). This coefficient can be calculated after determining the linear regression between the growth parameter X (which can be ULR, LAR, SLA, or LMF) as the dependent variable and RGR as the independent variable. GRC _X then is defined as the relative increase in growth parameter X divided by the relative increase in RGR

$$\mathrm{G}\mathrm{R}{\mathrm{C}}_X=\frac{dX/X}{\mathrm{d}\mathrm{R}\mathrm{G}\mathrm{R}\mathrm{/}\mathrm{R}\mathrm{G}\mathrm{R}}$$

The sum of GRC_ULR and GRC_LAR for any experiment should be 1, and this is also the case for the sum of GRC_ULR, GRC_SLA, and GRC_LMF. A value of 1 for GRC_ULR indicates that species variation in RGR within an experiment fully scales with variation in ULR, whereas a value of 0 indicates no effect of this parameter at all.

What is the overall picture that emerges from the literature? Poorter and van der Werf (1998) analyzed a total of 111 experiments on herbaceous C₃ species, and calculated the average GRC for the various growth parameters. They found that, on average, variation in SLA is by far the most dominant factor explaining variation in inherent RGR. ULR is second, and LMF is, on average, the quantitatively least important variable (Table 3.3). For a more elaborate analysis on GRC and the literature compilation see Poorter and van der Werf (1998).

It is not all that clear how the above-mentioned relationship between RGR of different species and the underlying growth parameters depend on environmental conditions or functional types. There have been suggestions that at high irradiance growth variation between species is due more to ULR than to SLA (Shipley 2006). Comparing variation in RGR between C3 and C4 species, or between sun and shade species, this seems indeed the case (Poorter 1989, Poorter 1999). However, for other species there is no indication that the relative importance of ULR and SLA shift with light environment (Poorter and van der Werf 1998). Growth temperature affects the relationships such that in cooler environments GRC_ULR gains in importance, in warmer environments GRC_SLA plays a dominant role (Loveys et al. 2002).

Physiological Parameters

The aforementioned technique of growth analysis has the advantage of being simple. Moreover, technical requirements to conduct such an experiment are low. A drawback is that ULR is a parameter that integrates various aspects of plant functioning (photosynthesis, respiration, chemical composition) and therefore cannot be related directly to a specific physiological process. One may assume that a considerable part of the variation in ULR is due to differences in the area-based rate of photosynthesis (Konings 1989). To obtain more insight into the physiological basis of variation in RGR, however, a more mechanistic approach has to be followed, in which growth is analyzed in terms of the plant’s carbon (C) economy. This requires knowledge of the C-gain of the plant in photosynthesis, and C-losses in leaf, stem, and root respiration, all integrated over the day. The net increase in C over the day can be converted into a dry mass increase, if the C-concentration of the newly formed material is known. A formula to relate the C-fluxes to RGR is presented in the second part of Box 3.2.

Given that 85%-95% of plant dry matter is composed of carbon-based compounds (for a review see Poorter and Villar 1997), it is beyond doubt that almost all newly formed biomass is fixed during the photosynthetic process. However, that does not necessarily imply that variation in RGR must be due to variation in the rate of photosynthesis, measured per unit leaf area and per unit of time (PSa). Only few attempts have been made to quantify both whole shoot photosynthesis at growth conditions (rather than determining photosynthetic capacity for the youngest expanded leaf) and RGR. In a number of cases no relationship at all has been observed (Dijkstra and Lambers 1989, Poorter et al. 1990, Van der Werf et al. 1993, Atkin et al. 1996a), but in others a positive relation was found (Garnier et al., unpublished). If indeed the growth parameter ULR is correlated well with the rate of photosynthesis per unit leaf area (Konings 1989, Poorter and van der Werf 1998), we might derive from the GRC_ulr value in Table 3.3 that, in the average experiment, there is a modestly positive relationship between PS_a and RGR.

It may well be that PS_a (and ULR) is not the best variable to consider if one seeks to understand the physiological basis of growth. Traditionally, in physiological research, the rate of photosynthesis is expressed per ULR. In this way the flux of C can be related to the flux of incoming photons and the efflux of water. In analyzing growth, however, it may be more relevant to consider how much C is fixed by 1 g of leaf biomass, which is the photosynthetic rate per unit leaf mass (PS_m). This would relate much better to RGR, which is the increase in biomass per unit of biomass and time. PS_m is the product of PS_a and SLA (Box 3.2), and this parameter is strongly associated with RGR (Poorter et al. 1990, Reich et al. 1992, Van der Werf et al. 1993, Walters et al. 1993, Atkin et al. 1996a, Garnier et al. unpublished results).

Both shoot and root respiration are generally positively correlated with RGR (Poorter et al. 1990, Van der Werf et al. 1993, Walters et al. 1993, see also Atkin et al. 1996a). This is partly a reflection of the higher rate of growth of the fast-growing species, which requires extra amounts of energy (ATP) and reducing power (NADH) per unit dry mass. As far as roots are concerned, the higher respiration rate is also a reflection of a much higher uptake rate of ions (Van der Werf et al. 1994).

An alternative approach to the carbon balance decomposition just described can be followed, focusing on the nitrogen economy of the plant. RGR can then be expressed as the product of the root mass fraction (RMF) and the nitrogen uptake rate divided by the mean plant nitrogen concentration (Garnier 1991, Lambers and Poorter 1992). Breaking down the RGR of fast- and slow-growing species in this way shows that faster growth is associated with a relatively lower allocation of biomass to roots, a higher plant nitrogen concentration, and a higher nitrogen uptake rate (NUR). In fact, NUR is the parameter that shows the widest variation and the strongest correlation with RGR in the fast-slow continuum (Garnier 1991).

Chemical and Anatomical Parameters

There is a wide range of other parameters for which fast- and slow-growing species differ under more or less optimal growth conditions. Most notably, fast-growing species have higher concentrations of reduced nitrogen in all organs, as well as higher amounts of minerals (Poorter and Bergkotte 1992, Reich et al. 1992, Van der Werf et al. 1993, Garnier and Vancaeyzeele 1994, Atkin et al. 1996b), observations that also hold for fast- and slow- growing tree species (Villar et al. 2006). They tend to have slightly lower concentrations of C (but see Garnier and Vancaeyzeele 1994 and Atkin et al. 1996a), but these differences are marginal for species within the same life form. Differences in C-concentration may become substantial when the growth differences between herbaceous and woody species are analyzed (Poorter 1989).

Given the importance of SLA in explaining variation in RGR_max (Table 3.3), it is appropriate to factorize this parameter further. SLA, or rather its inverse, 1 /SLA (expressed in g m^—2), is the product of leaf thickness and leaf density (see Box 3.3). Generally, the lower SLA of slow-growing species is due more to a higher leaf density (LD) than that it is caused by a higher leaf thickness (Van Arendonk and Poorter 1994, Garnier and Laurent 1994, but see e.g., Korner and Diemer 1987 and Shipley 1995). The density of a leaf or root is strongly related to its water content per unit dry mass (Garnier and Laurent 1994). This can easily be understood by envisaging a cell as a box. A higher density is often caused by an extra deposition of cell wall material (lignin, cellulose). A doubling of the amount of cell wall material hardly affects the cell size or the amount of water in the cell (the volume of the box). However, the amount of water relative to the dry mass decreases, whereas the density, the dry mass per unit volume increases. It can therefore be expected that fast-growing species, with a low density of tissues, will have a high water content per unit biomass. This turns out to be the case and not only applies to leaves but also to stems and roots (Garnier 1992, Ryser 1996). Another factor that strongly affects the density is the relative volume of intercellular spaces.

 

BOX 3.3

Components of Specific Leaf Area

SLA, leaf area per unit leaf mass, is a parameter that scales linearly and positively with RGR and is, in this respect, an easy parameter to use in growth analyses. However, if one would like to analyze what factors play a role in determining SLA, it is more appropriate to consider its inverse, 1/SLA, for which many other terms have been used (SLW, SLM, LMA). This is because components of the leaf, like leaf thickness, or anatomical and biochemical features, increase linearly with the inverse of SLA. Witkowski and Lamont (1991) factorized 1 /SLA into two components: leaf thickness (LTh) and leaf density (LD). Leaf density is defined as the mass of a leaf per unit leaf volume:

3.13 $$\mathrm{L}\mathrm{D}=\frac{M_{\mathrm{L}}}{\mathrm{L}\mathrm{T}\mathrm{h}A}$$

and therefore,

3.14 $$\frac{1}{\mathrm{S}\mathrm{L}\mathrm{A}}=\mathrm{L}\mathrm{D}\mathrm{L}\mathrm{T}\mathrm{H}=\frac{M_{\mathrm{L}}}{\mathrm{L}\mathrm{T}\mathrm{h}A}\mathrm{L}\mathrm{T}\mathrm{h}$$

A leaf can also be separated into its underlying anatomical tissues: epidermis, mesophyll, sclerenchyma, vascular tissue, and intercellular spaces. The 1/SLA is then the sum of the densities of the various tissues i, weighted by the volume per unit leaf area taken by each tissue (Garnier and Laurent 1994):

3.15 $$\frac{1}{\mathrm{S}\mathrm{L}\mathrm{A}}=\underset{i=1}{\overset{5}{\mathrm{\Sigma }}}\frac{V_i}{A}\frac{M_i}{V_i}$$

Leaf biomass can also be separated into its various biochemical fractions. A simple grouping of the wide range of compounds is: Lipids, lignin, soluble phenolics, protein, structural carbohydrates (cellulose, hemicellulose, pectin), nonstructural carbohydrates (glucose, fructose, sucrose, starch), organic acids, and minerals (for a review see Poorter and Villar 1997). 1 /SLA is then the sum of the masses of each of the eight classes of compounds expressed per unit area:

3.16 $$\frac{1}{\mathrm{S}\mathrm{L}\mathrm{A}}=\underset{i=1}{\overset{8}{\mathrm{\Sigma }}}\frac{M_i}{A}$$

The three ways of breaking down 1 /SLA are interrelated: a high leaf density, for example, can be the result of a high proportion of sclerenchyma, which shows up in the biochemical analysis as a high concentration of cell walls (lignin and structural carbohydrates).

Selection for RGR

As discussed in the previous section, RGR is a parameter that is associated with a suite of traits. Based on the negative correlations between RGR_max and the harshness of the environment (Figure 3.2), it has been suggested that RGR was the target of selection (Grime 1977, Chapin 1988). Alternatively, it may have been one or more of the components of RGR that has been selected for (Grime 1977, Coley 1983, Dijkstra and Lambers 1987).

What are the arguments in favor of selection for a high RGR? For ruderals and other annual species, a fast completion of the life cycle seems of paramount importance (Grime 1979). A high RGR may be of help to reach the size required for high seed production (discussed by Benjamin and Hardwick 1986, McGraw and Garbutt 1990, and Garnier and Freijsen 1994). Competitive species, be it woody or herbaceous, seem to have an advantage by occupying quickly the available space within the vegetation. The acquisition of resources, both above and belowground, depends on how much of the volume of soil and air is occupied by roots and leaves. A high RGR may enable this.

Figure 3.6   (a) SLA; (b) leaf nitrogen concentration, and (c) leaf density measured on the same species in field and laboratory conditions. Each study is represented by a regression line and contains several species. (Data from: a: (1) Cornelissen, J.H.C. et al. (unpublished, laboratory, and field); (2) Garnier, E. and Laurent, G., New Phytol., 128, 725, 1994 (laboratory) and Garnier, E., Cordonnier, P., Guillerm, J.-L., and Sonie, L., Oecologia, 111, 490, 1997 (field); (3) Mooney, H.A., Ferrar, P.J., and Slatyer, R.O., Oecologia, 36, 103, 1978 (laboratory and field; (4) Poorter, H. and Remkes, C., Oecologia, 83, 553, 1990 (laboratory and Poorter, H. and de Jong, R., New Phytol., 143, 163, 1999 (field); (5) Walters, M.B., Kruger, E.L., and Reich, P.B., Oecologia, 96, 219, 1993, and P.B. Reich (unpublished) (laboratory) and Reich, P.B., Walters, M.B., and Ellsworth, D.S., Ecol. Monogr., 62, 365, 1992 (field); (6) Bloor J.M.G., J. Trop. Ecol., 19, 163, 2003; b: (1) Cornelissen, J.H.C., Werger, M.J.A., Castro-Diez, P., van Rheenen, J.W.A., and Rowland, A.P., Oecologia, 111, 460, 1997; (2) Garnier, E. and Vancaeyzeele, S., Plant, Cell Environ., 17, 399, 1994 (laboratory) and Garnier, E., Cordonnier, P., Guillerm, J.-L., and Sonie, L., Oecologia, 111, 490, 1997, (field); (3) Hull, J.C. and Mooney, H.A., Acta Oecologica, 11, 453, 1990 (laboratory and field); (4) Mooney, H.A., Ferrar, P.J., and Slatyer, R.O., Oecologia, 36,103,1978 (laboratory and field); (5) P.B. Reich (unpublished, laboratory) and Reich, P.B., Walters, M.B., and Ellsworth, D.S., Ecol. Monogr., 62, 365, 1992 (field); C: Garnier, E. and Laurent, G., New Phytol., 128, 725, 1994 (laboratory) and E. Garnier and J.-L. Cordonnier (unpublished); (2) Poorter, H. and Bergkotte, M., Plant, Cell Environ., 15, 221, 1992 (laboratory) and Poorter, H. and de Jong, R., New Phytol., 143, 163, 1999 (field).)

What would be the adaptive value of a low RGR? Some explanations have been offered, focusing on plant species from nutrient-poor habitats (Chapin 1980, 1988). However, these suggestions are questionable (Poorter 1989, Lambers and Poorter 1992):

1. If plants grow slowly, they are less likely to deplete the available nutrient resources early in the season. This does not seem to be an evolutionarily stable strategy. As soon as one species or genotype starts to take up as much nutrients as possible early in the season, the others miss out.
2. Plants in nutrient-poor areas will never grow fast. If they have a low potential RGR, they will be closer to their optimum in the field. For all plants it applies that their physiological optimum differs from the conditions they experience in the field (see the second section, pp. 69-73). However, a clear difference between physiological and ecological optimum does not necessarily imply a disadvantage. Fast- as well as slow-growing species have a large flexibility, morphological as well as physiological, to cope with varying conditions (Reynolds and d’Antonio 1996). At low nutrient supply, for example, it is found that species with a high RGR_max still grow faster or at similar rates than those with a low RGR_max (Shipley and Keddy 1988, Poorter et al. 1995).
3. If plants grow slowly, they can accumulate sugars and nutrients during favorable times, so as to enable growth in later times when nutrients are less easily available. As far as sugars are concerned this explanation does not hold. Plants experiencing nutrient stress are limited more in growth than in photosynthesis. They fix more C than can be used in growth, resulting in accumulation of nonstructural sugars (mainly starch; Poorter and Villar 1997). Therefore, it does not seem necessary to store sugars for times with a low-nutrient availability. For nutrients however, this reasoning could be valid. A prerequisite is that nutrients become available in flushes. Such flushes have been shown, for example during freeze-thaw events that lyse microbial cells or during drought-wet cycles (for discussion see Bilbrough and Caldwell 1997). However, although these processes have been shown to occur, there is at present not enough evidence to conclude that species found in sites of different nutrient availability (or those with contrasting RGR_max) differ in this respect. For plants grown hydropon- ically in the laboratory at a nonlimiting nutrient supply, we found fast-growing species to accumulate 4–5 times more NO₃ ^— than slow-growing species.
4. A last hypothesis does not explain why plants in low-resource environments do have a low RGR, but rather why they do not have a high potential RGR: A high RGR cannot be realized in low-resource environments and therefore RGR is a selectively neutral trait. Although a very high RGR would not be reached, a genotype with a slightly higher RGR could still occupy some extra space and consequently would be able to acquire some extra nutrients. Thus, in those cases RGR would not necessarily be a selectively neutral trait.

Selection for Components of RGR

Is there evidence that components of RGR have been selected for, rather than RGR itself? This is not a question that can be easily answered. If all of the components scale positively with RGR, it becomes impossible to separate between the two alternatives. Moreover, it is difficult to single out one component if the different traits have not been selected independently, or if they are functionally related. For example, a high rate of photosynthesis is achieved with a high concentration of protein. A high concentration of protein may imply a high rate of protein turnover, and therefore a high maintenance respiration. In addition, a high rate of photosynthesis may result in a high rate of export to the phloem, which also causes an increase in shoot respiration (cf. Figure 3.5). Given these mechanistic interrelations, it is not easy to break the correlation between photosynthesis and respiration.

Nevertheless, we believe that there are good reasons to think that selection in a low- resource environment has been for components related to a low SLA, rather than for RGR per se. Generally, SLA is the most important growth parameter associated with inherent variation in RGRmax (pp. 76–83). Moreover, differences in SLA observed in the laboratory are preserved in the field (p. 84), and in the case of the old-field succession described in the ‘‘RGR_max and Ecosystem Productivity,’’ pp. 73–75, the correlation between the RGR of the vegetation and the aggregated SLA of the composing plant species, which was determined in the field, was almost as good as the correlation with the aggregated RGR derived from lab measurements (Figure 3.3b). What are the arguments in favor of this alternative hypothesis?

Selection for SLA-Related Traits in Adverse Environments

Plants from extremely nutrient-poor environments are often evergreens, with a high leaf longevity. As has been previously discussed, availability of photosynthates is not a growth-limiting factor in these environments, but the availability of nutrients is. A first problem of plants in a low-nutrient environment is to acquire nutrients; a second problem is to use them efficiently. Berendse and Aerts (1987) showed that an efficient use of, for example, N [nitrogen use efficiency (NUE); gram of growth per unit N taken up or lost by the plant] depends on two components: the biomass increase per unit N and per unit of time (NP, the mean annual nitrogen productivity) and the average time that a unit of N stays in the plant (MRT, mean residence time of nitrogen; see Box 3.4). They argue that there is a trade-off between these two components: a plant cannot achieve both a high NP and a high MRT (Figure 3.7). Such a trade-off has indeed been shown in several cases (e.g., Eckstein and Karlsson 1997; Figure 3.7), but is less evident in other experiments (reviewed by Garnier and Aronson 1998). The putative trade-off between NP and MRT is most likely due to the suite of characters discussed on pp. 82–83: a high nitrogen productivity is strongly correlated with both a high SLA and a low density of organs (characteristics found at the right-hand side in Figure 3.5). Plants with those characteristics are mainly directed toward attaining a high rate of resource capture (carbon, nutrients). This strategy is further discussed on p. 90. The other extreme is formed by the species with a high MRT. Species with such a strategy can be found in nutrient-poor environments. Why is this so? In nutrient-poor environments, it may be a disadvantage for an individual plant to lose acquired nutrients, as it is very questionable whether that individual is able to take them up again once they have entered the nutrient cycling process. Therefore, there is a premium to increase the residence time of nutrients. Theoretically, this can be achieved in two ways (Box 3.4): either a plant resorbs nutrients from senescing organs very efficiently, or it restricts the turnover of organs and thus the loss of biomass per unit of biomass present (Aerts 1990, Garnier and Aronson 1998). Although there is variation in resorption efficiency among species, it does not differ strongly among life-forms or species originating from habitats differing in fertility (reviewed by Aerts and Chapin 2000). Thus, the way species from nutrient-poor habitats achieve a high MRT is by a long life span of both leaves and roots (Aerts 1995, Ryser 1996, Eckstein and Karlsson 1997, Garnier and Aronson 1998, Hikosaka 2003).

 

BOX 3.4

Components of Nitrogen Use Efficiency

In Box 3.2, RGR was factorized into components based on the assumption that leaf area is the important plant variable driving photosynthesis, and thus growth. An alternative approach is to consider plant (organic) N as the driving variable, as proteins play a vital role in C-fixation, nutrient uptake, and in most other physiological processes of the plant. RGR is then factorized into the components NP (Nitrogen Productivity) and PNC (Plant Nitrogen Concentration; preferably restricted to organic nitrogen, as NH₄ ⁺ and NO₃ ^– play less of a physiological role). NP is defined as the increase in biomass per unit time and plant nitrogen (Ingestad 1979):

3.17 $$\mathrm{N}\mathrm{P}=\frac{1}{N}\frac{\mathrm{d}M}{\mathrm{d}t},$$

where N is the total amount of nitrogen in the plant. PNC is the concentration of nitrogen in the plant

3.18 $$\mathrm{P}\mathrm{N}\mathrm{C}=\frac{N}{M}$$

and consequently

3.19 $$\mathrm{N}\mathrm{P}\mathrm{P}\mathrm{N}\mathrm{C}=\frac{1}{N}\frac{d\mathrm{M}}{d\mathrm{t}}\frac{N}{M}=\mathrm{R}\mathrm{G}\mathrm{R}$$

Considered over a short and fixed period of time, plants use their internal nitrogen efficiently if they have a high NP. Considered over a longer time scale, an alternative way to make efficient use of a unit of nitrogen taken up is to increase the time this unit remains in the plant. This time span is called the mean residence time (MRT, expressed in years). Under steady-state conditions (i.e., when the amount of nutrients taken up by the plant equals that lost by leaf and root shedding, herbivory, etc.), MRT is equal to the ratio between the average amount of nitrogen in the plant (N) and that taken up or lost over a given period of time (dN/dt; see, e.g., Frissel 1981). Berendse and Aerts (1987) have proposed that it is the product of NP and MRT that determines the nutrient use efficiency (NUE)

3.20 $$\mathrm{N}\mathrm{P}\mathrm{.}\mathrm{M}\mathrm{R}\mathrm{T}=\frac{1}{N}\frac{\mathrm{d}M}{\mathrm{d}t}N\frac{\mathrm{d}t}{\mathrm{d}N}=\frac{\mathrm{d}M}{\mathrm{d}N}=\mathrm{N}\mathrm{U}\mathrm{E}$$

NUE is therefore the total amount of biomass produced per unit N taken up or lost. Note that in Equation 3.19, NP is generally determined over a short time scale (days to weeks), whereas in Equation 3.20 it is determined on an annual basis.

Under steady-state functioning MRT depends on two parameters in the following way (Garnier and Aronson 1998):

3.21 $$\mathrm{M}\mathrm{R}\mathrm{T}=\frac{1}{1-R_{\mathrm{e}\mathrm{f}\mathrm{f}}}\frac{1}{e},$$

where Reff is the nitrogen resorption efficiency during senescence (defined as the reduction in the amount of nutrient between mature and senesced organs, relative to the amount in mature organs: Aerts 1996, Killingbeck 1996), and e is the rate of biomass lost by the plant, which depends on the life span of organs. These two parameters are central to the definition of the nutrient conservation strategy of the various species.

Figure 3.7   The relation between Nitrogen Productivity (NP; on an annual basis) and the MRT of nitrogen of 14 species grown in the field. Squares, herbaceous species; circles, deciduous shrubs; triangles, evergreen shrubs. (Data from Eckstein, R.L. and Karlsson, P.S., Oikos, 79, 311, 1997.)

What determines the life span of a leaf or root? For perennial species, there is an effect of the environment, with increases in leaf life span under relatively predictable low-resource conditions (low temperature, low irradiance). However, most of the variation in leaf life span is due to inherent differences between species (Reich 1998). Leaves of some species live for less than 2 months, whereas leaves of others have been reported to function for more than 20 years (Ewers and Schmid 1981). The physiological mechanism determining the life span of a leaf is unknown. However, it is evident that a leaf can only become long-lived if it can withstand adverse periods. Therefore, compared with a leaf with a short life span extra investments have to be made to survive periods of drought or coldness. It should also be less attractive to herbivores, and not be so frail that it is damaged in storms. Finally, in a nutrient-poor environment one could expect extra investments of plants to prevent nutrients leaching out of the leaf. Drought tolerance may be achieved by leaf hairs, a thick cuticle, an increase in lipid concentration and possibly small cell sizes (Jones 1983). Cold resistance may be acquired by increases in osmotic solutes. Herbivory may be counteracted by the accumulation of phenolics or other secondary compounds. In all these cases extra lignification may occur. Extra lignification and thicker cell walls can also be expected for plants that have adapted to a high degree of mechanical disturbance, such as trampling or strong winds. Nutrient leaching could be prevented by extra investment in wax layers.

Compared with a basic leaf with a short life span, all of these additional investments increase the biomass per unit leaf area and thus decrease SLA. Therefore, we might expect a negative relation between SLA (and the suite of traits positively associated with SLA) and leaf life span. This has indeed been shown in an analysis of plant species across a wide range of habitats (Reich et al. 1992, Wright et al. 2004, Figure 3.8). Data on the life span of roots are far less abundant (Eissenstat and Yani 1997). As far as data are available, they seem to indicate that species with a higher density of root tissue may have longer life spans (Ryser 1996). It is this connection between the nutrient economy (in the form of a high MRT through a long leaf or root life span) and the carbon economy (in the form of a low SLA) that may explain the success of species with a low RGR_max in nutrient-poor environments.

Up to now, we have focused discussion on RGR components of plants from habitats low in nutrients. Basically, similar considerations could work for plants in other environments adverse to growth. In all cases where the production of biomass is difficult, one may expect a premium on maintaining existing biomass rather than replacing lost leaves or roots.

Figure 3.8   Leaf longevity as a function of SLA for a wide range of trees, shrubs, and herbaceous C3 species found across a range of contrasting climates. (Wright, I.J., Reich, P.B., Westoby, M., Ackerly, D.D., Baruch, Z., Bongers, F., Cavender-Bares, J., +26 others, Nature, 428, 821, 2004.)

Selection for SLA-Related Traits in Favorable Environments

What about a possible selection for RGR components in an environment favorable for plant growth? Vegetation is dense there, with a fast leaf area development during the growing season. Consequently, there is a strong competition for light, necessitating a fast increase in stem height and an efficient light interception per unit leaf biomass. In simulations of competition under agricultural conditions, it has been shown how important SLA is during the period before canopy closure: high-SLA plants have a higher light-interception per unit leaf biomass and therefore faster growth than competing plants with a low SLA (Spitters and Aerts 1983, Gutschick 1988). In a closed canopy, maximum photosynthetic carbon gain of the vegetation as a whole is highest at lower values of SLA (Gutschick 1988). However, also under those conditions there may be a selection pressure towards an increase in SLA. Using game theory, Schieving and Poorter (1999) analyzed mathematically the total carbon gain of two putative genotypes, which were similar in all traits, except for SLA. Simulating competition in a dense stand on the basis of functions for the distribution of light, nitrogen and leaf area, they found that the genotype with the higher SLA always replaced the lower SLA genotypes.

Nutrient Availability

The most obvious difference in SLA is between woody deciduous and woody evergreen species. Evergreens have much lower SLAs (Villar and Merino 2001) and are generally found in nutrient-poor habitats (Monk 1966, Small 1972). Comparing productivity of various woody deciduous and evergreen species at the stand scale, a positive correlation has been found between SLA and annual productivity per unit leaf mass (Reich et al. 1997). Additionally, for herbaceous vegetations, there is a general trend that productive sites bear species with a higher SLA (Poorter and de Jong 1999). However, there is considerable variability around this trend, as only half of the total variation in SLA between species was explained by differences between sites, the other half due to differences in species within sites. This is in line with data of Van Andel and Biere (1989), who found a large variation in RGR_max and SLA for species co-occurring in the same habitat.

Species replacement has been observed in Venezuela, where two introduced C₄ grasses have outcompeted the original C₄ grass in most areas, but not in drier sites with low fertility (Baruch et al. 1985, Baruch 1996). Conforming to our expectations, the native species has a lower SLA, a higher proportion of sclerenchyma, and a lower concentration of N (Table 3.4).

A controlled experimental test to analyze the relation between growth parameters and field performance was carried out by Biere et al. (1996). They selected a range of families of Lychnis flos-cuculi and performed a growth analysis for each of these families in the glasshouse. At the same time, seeds of these families were sown in the field, along a fertility gradient. The aboveground biomass of all field plants at the end of the growing season, a good predictor of next year’s fecundity, was estimated and correlated with the growth parameters determined in the glasshouse. At the poorest site, families with a high RGR_max, LAR, and SLA fared worst (Table 3.5). However, the higher the site fertility, the more these parameters gained importance, and in the most productive site families with a high SLA (and LAR) were those that attained the highest biomass. These results were confirmed in an experiment where early plant growth characteristics were correlated with fitness in H. spontaneum accessions (Verhoeven et al. 2004). At high, but not at low nutrient supply, high SLA accessions had the highest seed mass output. RGR itself was not correlated with fitness.

Water Availability

Mooney et al. (1978) investigated Eucalyptus species along a rainfall gradient, and showed that, both for plants growing in the field and in the laboratory, the SLA of the low-rainfall species was lower than those of species growing at sites with higher rainfall. The high SLA species had the highest concentration of N per mass. In this case, the correlation between seedling RGR and rainfall was not very tight, whereas the correlation with SLA was tight. Similarly, the SLA of sun-lit leaves from Eucalyptus canopies, sampled along a climatic gradient in Australia was found to be inversely related to the potential evaporation at the various sites studied (Specht and Specht 1989). These results are not supported by Warren and Adams (2005) though, who showed a relationship between rainfall and biomass allocation, rather than with SLA for nine Eucalypt species. Encelia farinosa plants from dry places are reported to have lower SLA than those from wetter places, and lower SLA than Encelia frutescens plants growing close-by but tapping in on deeper water (Ehleringer and Cook 1984, Ehleringer 1988). The lower SLA of E. farinosa from the driest sites is completely due to a thick layer of leaf hairs, which reflect the sunlight and may take up 50% of the biomass of the leaf. Finally, Tsialtas et al. (2004) found low-SLA species to be more abundant in dry Mediterranean sites than high-SLA species.

Trampling and Wind Damage

A very clear case in which a low SLA is of survival value is trampling. Dijkstra and Lambers (1989) studied two subspecies of Plantago major. One is an annual, growing on occasionally flooded river banks, whereas the other is a perennial, occurring in frequently mown and trampled lawns. The leaves of the first subspecies have a high SLA and are erect. Those of the second are prostrate and have a low SLA, as well as a higher proportion of biomass in cell walls. An experiment showed that the subspecies with the low SLA survived trampling better (Table 3.6). Similarly, Meerts and Garnier (1996) found that Polygonum aviculare genotypes from trampled habitats show a lower SLA under laboratory conditions than those from nontrampled places, and so did Kobayashi et al. (1999) for trampling-resistant forbs and grasses.

Leaf destruction can also take place by strong winds. Pammenter et al. (1986) analyzed differences in leaf anatomy between two Agrostis species that occur at sub-Antarctic islands. Agrostis magellanica is native to these islands, Agrostis stolonifera has been introduced recently. The latter species has displaced A. magellanica in wind-sheltered areas, but not in more open terrain. This is most likely explained by the fact that the leaves of the introduced species are relatively thin and fragile. A. magellanica, with an unusual high fraction of the leaf volume occupied by sclerenchyma, a high leaf thickness, and a much lower SLA (Table 3.4), seems better able to withstand wind damage.

Alpine species have been found to have lower SLA than lowland species, both in the field (Korner and Diemer 1987) and in the laboratory (Atkin et al. 1996a). Under both sets of conditions, this was at least partly due to thicker leaves. It has been suggested that the increased wind speed measured at higher elevations could play a role here as well. In an experiment with an upland, low-SLA species and a lowland high-SLA species grown in a wind tunnel at high wind speed, Woodward (1983) found much greater leaf damage in the high-SLA species.

Herbivory

Herbivory by insects and mammals may have a large impact on the vegetation and strongly damage individual plants. Species that have a low attractivity to herbivores are those with a low water content, a low organic nitrogen concentration, high concentrations of lignin and other cell wall components, a high concentration of secondary compounds like tannins and tough leaves in general (Scriber 1977, Grubb 1986, Coley and Barone 1996). These are all traits associated with a low SLA.

An interesting case where species with an inherently high SLA perform less is in the understorey of tropical forests. Plants generally acclimate to a low light environment by an increase in SLA, enabling to capture more light per unit leaf biomass. Therefore, one might expect at first to find high-SLA species in the understorey. Indeed, high-SLA pioneer species have a higher RGR at low light intensities than shade-tolerant species (Veneklaas and Poorter 1998). Notwithstanding their initially higher growth rate, these plants suffer from a high mortality rate under these conditions compared with the shade-tolerant seedlings with low SLA (Kitajima 1994). Susceptibility to herbivory may play a role here as well.