A standardized protocol for the determination of specific leaf area and leaf dry matter content



  • 1 The impact of sample preparation, rehydration procedure and time of collection on the determination of specific leaf area (SLA, the ratio of leaf area to leaf dry mass) and leaf dry matter content (LDMC, the ratio of leaf dry mass to fresh mass) of mature leaves was studied in three wild species growing in the field, chosen for their contrasting SLA and LDMC.
  • 2 Complete rehydration was achieved 6 h after samples were placed into water, but neither of the procedures tested – preparation of samples before rehydration or temperature applied during rehydration – had a significant effect on the final values of SLA or LDMC.
  • 3 As expected, water-saturated leaves had a lower LDMC than non-rehydrated leaves; more surprisingly, their SLA was also higher. The impact of rehydration on SLA was especially important when the SLA of the species was high.
  • 4 There was no significant effect of time of sampling on either trait in any species over the time period covered (09·00–16·30 h).
  • 5 These results suggest that SLA and LDMC obtained on water-saturated leaves (SLASAT and LDMCSAT) can be used for species comparisons. We propose a standardized protocol for the measurement of these traits. This would allow for better consistency in data collection, a prerequisite for the constitution of large databases of functional traits.


The renewed interest in classifying species into groups relating to function rather than to taxonomy (e.g. Keddy 1992; Lavorel et al. 1997; Westoby 1998; Weiher et al. 1999) has triggered the search for traits that express meaningful differences in ecological behaviour among plant species. In this context, specific leaf area (SLA, the ratio of leaf area to leaf dry mass) and leaf dry matter content (LDMC, the ratio of leaf dry mass to fresh mass) reflect a fundamental trade-off in plant functioning between a rapid production of biomass (high SLA, low LDMC species) and an efficient conservation of nutrients (low SLA, high LDMC species; see Poorter & Garnier 1999 for a recent review). It has thus been proposed that these traits – or one of them – be measured routinely in screening programmes (Westoby 1998; Weiher et al. 1999; Wilson, Thompson & Hodgson 1999).

The measurement of these traits in many species from different habitats and microclimatic conditions, etc., requires standardization. Some authors have proposed that sampled leaves should be young and fully expanded, taken from full light positions and without serious herbivore and pathogen damage (Reich, Walters & Ellsworth 1992; Westoby 1998; Weiher et al. 1999). However, LDMC and/or SLA are sensitive to at least two other factors: leaf water status and the time of day at which leaves are sampled.

Spatio-temporal differences in leaf water status will obviously affect LDMC (= 1 – leaf water content), and several experiments suggest that this is also true for SLA (e.g. Cutler, Rains & Loomis 1977; Retuerto & Woodward 1993; Tardieu, Granier & Muller 1999). To correct for these differences, which are not the focus of interest in screening programmes, Eliáš (1985) has suggested that measurements be made on leaves at full hydration. A literature survey conducted on 246 papers (reference list available on request) shows that some kind of rehydration prior to the determination of leaf traits is generally applied when LDMC alone or both LDMC and SLA are studied (89 and 78% of the cases, respectively), while rehydration is conducted only 24% of the time when only SLA is considered. The rehydration procedure itself varies substantially. It may be conducted: (i) on stems bearing leaves or single leaves with their petioles placed in water, on whole leaves placed in a water-saturated environment (moist paper or humidified plastic bag) or on immersed whole leaves or leaf discs; (ii) for different periods of time, from 2 to 24 h, and (iii) at different storage temperatures, from 3 °C to ambient (25–30 °C). The impact of these procedures on leaf water content (and LDMC) are relatively well described and understood (e.g. Barrs & Weatherley 1962; Millar 1966; Evans, Black & Link 1990), but to our knowledge, their effects on SLA have been poorly investigated.

The time of day when leaves are sampled may also affect LDMC (e.g. Barrs 1968; Girma & Krieg 1992) and SLA (Chatterton, Lee & Hungerford 1972; Bertin et al. 1999). Diurnal variations in these traits are generally attributed to short-term imbalances between transpiration and water uptake for LDMC, and between source and sink activities for SLA. As discussed earlier, rehydrating leaves prior to determination of LDMC will correct for these short-term imbalances in water content. However, there is currently no simple treatment or measurement that can correct for the rapid fluctuations of carbohydrate levels in leaves, thought to be responsible for the diurnal variations in SLA. This has led some authors to standardize the time of day at which sampling is conducted (e.g. Niinemets 1997; Picon, Ferhi & Guehl 1997; Reich et al. 1999). However, when many plants are to be sampled – as in screening programmes – sampling within a narrow period of time is impracticable. The timing and length of period during which leaf traits remain reasonably stable should thus be established, and sampling synchronized accordingly.

The aim of the present study was to assess the impact of rehydration and time of sampling on the determination of SLA and LDMC, and to propose a standardized sampling and measurement procedure for these traits. This was performed on three wild species growing in the field in southern France: an annual forb, a perennial legume and a sclerophyllous tree – chosen for their contrasting SLA and LDMC. Three successive experiments were carried out, in which we assessed the effects of (i) the way samples are prepared for rehydration, (ii) the length of rehydration period and storage temperature, and (iii) the time of sampling, on SLA and LDMC. The dependency of SLA on leaf water status was further investigated by examining the relationship between the two traits, using (i) the pooled data from the three experiments, and (ii) data taken from three other studies, where the two traits were measured on a large number of species.

Materials and methods


The three experiments (Table 1) were conducted between 20 July and 3 August 1999 on the CNRS campus of Montpellier (France) and its immediate vicinity (43°38′ N, 3°52′ E; 15 km from the Mediterranean coast; altitude 56–70 m a.s.l.). The site is subject to a subhumid Mediterranean climate (Daget 1977) with cold winters; the soil is derived from a calcareous substrate.

Table 1.  Summary of procedures and experimental conditions in the three experiments. I: initial treatment (value measured immediately upon return to the lab, no further hydration); PR: storage of leaves in moist paper in a cool chest until return to the lab, where the stem was re-cut under water; PN: same as PR, but stem not re-cut under water; TR: stems placed immediately in a tube filled with water and put into a cool chest until return to the lab, where they were re-cut under water; TN: same as TR, but stem not re-cut under water. See Materials and methods for further details
ExperimentProcedure and conditions
Preparation of samples for rehydrationTemperature during rehydration (°C)Length of rehydration period (h)Time of day at harvest (h)
1I, PR, PN, TR, TN42410·30–13·00
2I, PR, TR4, 260, 6, 24, 4810·00–12·30
3I, TR42409·00, 11·30, 14·00, 16·30

The species selected are the annual Kickxia spuria (L.) Dumort. (Scrophulariaceae: low LDMC, high SLA), the herbaceous perennial Psoralea bituminosa L. (Fabaceae: intermediate LDMC and SLA) and the evergreen tree Phillyrea latifolia L. (Oleaceae: high LDMC, low SLA).

Each data point was obtained from the measurement of 10 replicate samples per species; a total of 780 samples were collected from robust, well-grown plants. For the two herbaceous species, the 10 replicate samples were taken from 10 different fully illuminated individual plants (i.e. not under tree cover). For the tree species, two replicates were taken from each of five individuals, from the part of the canopy hit by direct sunlight at the time of sampling. Stems or twigs bearing 5–10 leaves were severed from a plant. One set of samples was quickly wrapped in moist paper and brought back to the laboratory for immediate (i.e. within 30 min of collection) determination of leaf traits (initial treatment). The other sets were treated in various ways, as described below.

Upon the completion of these treatments (or immediately upon return to the lab in the case of the initial treatment), the youngest fully expanded leaves (generally two per sample) free from herbivore or pathogen damage were severed from the stem or twig. All subsequent manipulations were carried out on leaf blades only, i.e. after the removal of petioles (and rachis in the case of P. bituminosa). Leaves were dried with tissue paper to remove any surface water, and weighed immediately to determine their saturated fresh mass. Their projected area (one side of the leaves) was determined with an area meter (model MK2, Delta-T Devices, Cambridge, UK). Samples were then oven-dried at 60 °C for at least 2 days and their dry mass measured.


When a leaf or a twig is severed from a transpiring plant, the water column in the xylem vessels retracts from the cut edge, and some of these vessels fill with air. If the sample is stored in a dry and warm atmosphere, transpiration continues, the water content of the leaf decreases and the length of the air-filled portion of the vessels increases. Since water movement in air-filled vessels is very difficult (e.g. Tyree & Sperry 1989), this may impair the subsequent rehydration of the sample. In this experiment, we assessed whether (i) the way samples are stored between collection in the field and processing in the lab and (ii) re-cutting the stem or twig under water at the onset of rehydration to remove the air-filled portion of the vessels affected the measured SLA and LDMC.

After severing the stem or twig from the plant, four treatments were imposed (Table 1): (i) storage in moist paper in a cool chest until return to the lab, where the stem was re-cut under water (treatment PR); (ii) same as (i), but stem not re-cut under water (treatment PN); (iii) stems placed immediately in a tube filled with water and put into a cool chest until return to the lab, where they were re-cut under water (treatment TR); (iv) same as (iii), but stem not re-cut under water (treatment TN). In all cases, samples were subsequently placed into tubes with the cut end in deionized water to allow rehydration (i.e. immediately upon arrival at the lab for treatments PN and TN, and after re-cutting for treatments PR and TR). They were then put into a dark, cold room maintained at 4 °C, and measurements were taken after 24 h.


In recent ecological work, plant samples were usually rehydrated in a cold environment overnight (Shipley 1995; Poorter & de Jong 1999; Wilson et al. 1999). Samples are stored at low temperatures to decrease respiration rate and limit dry mass loss during rehydration, but since the temperature applied during rehydration may significantly affect saturated fresh mass (Barrs & Weatherley 1962; Millar 1966), this potential source of error must be assessed. Furthermore, although a 12 h rehydration period seems sufficient to reach full turgor (Barrs & Weatherley 1962; Millar 1966), it is important to know the time-frame during which measurements can be made reliably, especially within large screening programmes where tight experimental control may not always be possible. We tested the effects of rehydration period length and of temperature during rehydration on SLA and LDMC.

After severing the stems or twigs from the plants, one set of samples was subjected to the PR treatment described above, while another was subjected to the TR treatment. For each of these, we varied the length of time during which the samples were left in water (0 [initial treatment], 6, 24 or 48 h) and the temperature at which they were stored (4 °C or ambient temperature [≈ 26 °C]) (see Table 1). Rehydration took place in the dark.


The aim of this experiment was to assess the importance of diurnal trends in SLA and LDMC, and to determine whether rehydration of leaves prior to measurement could help minimize the errors when leaves are not sampled at a fixed time of day.

Leaves were harvested at 09·00, 11·30, 14·00 and 16·30 h (true solar time +2 h) on 2 August 1999. Initial treatment values of SLA and LDMC were determined for each species at each harvest time. These were compared with values taken on samples prepared with the TR method and rehydrated for 24 h at 4 °C in the dark (Table 1).

Data on short-wave radiation, air temperature and vapour pressure deficit for this particular day were taken from the CNRS meteorological station, located on the campus.


SLA was calculated as the ratio of leaf area to leaf dry mass (m2 kg−1), and leaf dry matter content as the ratio between leaf dry mass and fresh mass (mg g−1). These were considered as the dependent variables in the statistical treatments. To better approximate normality, SLA was transformed to its natural logarithm, while LDMC was transformed to its log-odds (i.e. ln[LDMC/(1 – LDMC)]) since it is a proportion. The effects of the various treatments on these dependent variables was evaluated using anova with fixed factors. Post hoc comparisons of the different hydration treatments with the initial treatment were carried out using Dunnett’s method. Post hoc comparisons of all possible combinations of treatments were performed using Tukey’s method.



SLA differed widely between the three species, with the value for K. spuria being almost twice that for P. bitumosa and four times larger than that for P. latifolia (Fig. 1a). anova detected significant differences in SLA between the rehydration procedures in K. spuria (F4,45= 3·69, P = 0·01) and in P. bituminosa (F4,45= 5·18, P = 0·002), but not in P. latifolia (F4,45 = 0·97, P = 0·44). In general, the initial treatment produced the smallest SLA and the TR treatment produced the largest. When the initial treatment was removed from the analysis, no further significant differences were detected in K. spuria (F3,36 = 1·48, P = 0·24) or in P. latifolia (F3,36 = 0·11, P = 0·96), but there was still a significant contrast in P. bituminosa between the PN and TR treatments, for which the difference was 1·8 m2 kg−1. The differences between the initial and the TR treatments varied greatly and increased with SLA. For instance, the initial SLA (and percentage increase of SLA in the TR treatment) was 24·4 (increase of 52%) for K. spuria, 10·4 (increase of 13%) for P. bituminosa and 6·3 m2 kg−1 (increase of 7%) for P. latifolia.

Figure 1.

Means and standard errors (n = 10) of specific leaf area (a) and leaf dry matter content (b) in the three species as a function of the method of collecting and storing the leaves for rehydration. Initial: value taken immediately upon return to the lab (no further hydration); PR: storage in moist paper in a cool chest until return to the lab, where the stem was re-cut under water; PN: same as PR, but stem not re-cut under water; TR: stems placed immediately in a tube filled with water and put into a cool chest until return to the lab, where they were re-cut under water; TN: same as TR, but stem not re-cut under water. See Materials and methods for further details.

Large differences between species were also found for LDMC (F2,135 = 1270·03, P < 10−9), with values for P. latifolia 2·5 times larger than those for K. spuria and intermediate values for P. bitumosa (Fig. 1b). Rehydration always decreased LDMC, and significant differences in this trait between treatments were also detected (F4,135= 52·65, P < 10−9). However, this was almost exclusively because of the initial treatment. When this treatment was removed from the analysis and each species was analysed separately, only P. bituminosa showed a significant difference in LDMC between the remaining treatments (F3,36 = 3·00, P = 0·04). The decrease in LDMC during rehydration differed according to the species: for the TR treatment, it was 32% in K. spuria (initial value: 250 mg g−1), 19% in P. bitumosa (initial value: 370 mg g−1) and 23% in P. latifolia (initial value: 640 mg g−1).


Since anova did not detect any significant effect of the two rehydration procedures (PR and TR) used in this experiment, the data for both treatments were pooled.

Although there were obvious differences in SLA between species (F2,324 = 50·71, P < 10−7), the only other significant term in the anova was a marginally significant interaction, suggesting that the effects of length of rehydration period across the three species differed according to the storage temperature (F4,324= 3·38, P = 0·02) (Fig. 2a). However, when separate analyses were conducted for each species, no significant effects were detected (P > 0·05). SLA increased during the first 6 h of rehydration, after which it remained relatively stable (the slight effects of time or temperature observed in K. spuria were not significant [P = 0·06]).

Figure 2.

The effects of storage duration (0, 6, 24 or 48 h) and temperature (○ = ambient [≈ 26 °C]; ● = 4 °C) on specific leaf area (a) and leaf dry matter content (b). For each temperature, graphs were drawn using the pooled data from the PR and TR treatments (see Materials and methods for further details). Data are means ± SE (n = 20, except initial treatment, where n = 10).

LDMC was unaffected by the temperature (P > 0·05) during the rehydration period (Fig. 2b; F1,114 = 3·56, 0·29 and 0·96 for K. spuria, P. bituminosa and P. latifolia, respectively). During the first 6 h after rehydration, it decreased by 19, 27 and 25% for K. spuria, P. bituminosa and P. latifolia, respectively. It changed significantly (but only slightly) between 6 and 48 h only for P. latifolia (F2,114 = 5·09, P = 0·004).


Neither SLA nor LDMC tracked temperature, irradiance or vapour pressure deficit of the air (Fig. 3). No significant diurnal differences in SLA were detected for any species, either for the initial or for the rehydrated treatment (Fig. 3b; F1,72 = 0·08, 0·62 and 1·34 for K. spuria, P. bituminosa and P. latifolia, respectively). The initial treatment always produced the smallest SLA (Fig. 3b). Differences between the initial and the rehydrated treatment (TR) increased with SLA, significantly so in K. spuria and P. bituminosa (F1,72 = 8·65, 28·73 and 1·08 for K. spuria, P. bituminosa and P. latifolia, respectively).

Figure 3.

(a) Diurnal courses of short-wave radiation (▴), air temperature (▵) and vapour pressure deficit (□) on 2 August 1999 at the CNRS campus, Montpellier. The two lower panels show the effects of the time at which leaves were collected and of rehydration on specific leaf area (b) and leaf dry matter content (c); ● = initial treatment; ○ = leaves rehydrated with the TR procedure (see Materials and methods for further details). Means and standard errors are shown (n = 10).

There were clear differences in LDMC between the initial and the hydrated treatment in all three species (F1,72 = 23·39, 757 and 234 for K. spuria, P. latifolia and P. bituminosa, respectively), but none displayed significant diurnal differences in LDMC (Fig. 3c).


We further investigated the sensitivity of SLA to leaf water status by plotting SLA against LDMC for our pooled data (Fig. 4a). The curve fitted to these data was superimposed on those taken from three other published studies (Fig. 4b). The trend, based on a combination of interspecific differences and short-term response to rehydration (this study), matches those observed in interspecific comparisons (Roderick, Noble & Berry 1999; Garnier et al. 2001) and in comparisons where differences between species were mixed with plastic responses to the environment (Meziane & Shipley 1999). This points to a general relationship between LDMC and SLA, and shows that the SLA of low-SLA leaves (below 6–7 m2 kg−1) is relatively independent of changes in LDMC, but that this dependency becomes substantial in leaves with SLA higher than 10–15 m2 kg−1.

Figure 4.

The relationship between specific leaf area and leaf dry matter content. (a): Data from this study (□ = K. spuria;○ = P. bituminosa;▵; = P. latifolia); the curve shows the resulting allometric regression. (b): The same regression curve superimposed on data from three other studies: ▿ = Roderick et al. (1999), field data on Australian species (n = 104); ○ = Meziane & Shipley (1999), herbaceous species grown at different levels of nutrient availability and light intensity (n = 86); ◊ = Garnier et al. (2001), field data on Mediterranean herbaceous and woody species (n = 57).



Species whose leaves have a high SLA and a low LDMC will be sensitive to the degree of rehydration that is permitted, while species with a low SLA and a high LDMC will be relatively insensitive to the way in which the leaf is treated prior to the measurement of SLA (Figs 1–4). Given that the relationship between SLA and LDMC is robust (Fig. 4) and applicable to very different floras, the error in not rehydrating leaves will be substantial for many species. Averaged over the three experiments of the present study, the SLA of non-rehydrated leaves (initial treatment) was 22, 15 and 5% smaller than that of fully rehydrated leaves in K. spuria (rehydrated SLA: 28·8 m2 kg−1), P. bituminosa (rehydrated SLA: 12·0 m2 kg−1) and P. latifolia (rehydrated SLA: 6·8 m2 kg−1), respectively. If the maximum acceptable error on the determination of SLA is 10%, these figures suggest that all leaves with a SLA > 10 m2 kg−1 should be rehydrated before measurement. The SLA of most herbaceous species exceeds this threshold, with means around 25 m2 kg−1 (e.g. Shipley 1995; Garnier et al. 1997; Poorter & de Jong 1999), but it also applies to many broadleaf shrubs and trees whose SLA can exceed 10 m2 kg−1 (e.g. Abrams & Kubiske 1990; Niinemets 1999; Reich et al. 1999). The groups of species that will be less affected by not rehydrating before SLA determination are needle-leaved and sclerophyllous species, which generally have low SLA values, as found here for P. latifolia (e.g. Cunningham, Summerhayes & Westoby 1999; Niinemets 1999; Reich et al. 1999).

The increase in SLA upon hydration may occur either because leaf cells swell with water and therefore increase the leaf surface area, or because of an actual loss of biomass (Delgado et al. 1992; Shipley 2000) because of the utilization of non-structural carbohydrates by respiration during storage in the dark (Stewart 1971). Our study was not designed to separate these effects, and more carefully controlled studies are needed to address this question.

In agreement with Eliáš (1985), we suggest that SLA and LDMC obtained on water-saturated leaves (abbreviated as SLASAT and LDMCSAT) be used for species comparisons. Although this is carried out in many (but not all) studies for LDMC, this is not the case for SLA (see Introduction). SLASAT and LDMCSAT do not necessarily reflect field values of SLA and LDMC, but rather ‘potential’ values for these traits. This standardization will allow more consistent comparisons of different species growing in different environments.


The preparation of samples before hydration has only a small impact on the final value of SLA and LDMC (Fig. 1). Storing the sample in a cool and wet environment during transportation to the lab appears to be sufficient. Although re-cutting the stem under water upon arrival to the lab led to slightly lower final LDMC and higher SLA, the effect was not significant.

Water saturation of the sample is completed within 6 h of the start of rehydration (Fig. 2b), in general agreement with previous studies (e.g. Barrs & Weatherley 1962; Turner 1981; Evans et al. 1990). However, Millar (1966) reported that longer periods were necessary for needle leaves, and Turner (1981) presented data showing that up to 20 h may be necessary to saturate very dehydrated leaves. After the period of rapid water uptake by the leaf, measurements can be conducted for at least 2 days without altering significantly either SLA or LDMC (Fig. 2). However, after 2 days of storage at ambient temperature (≈ 26 °C), fungi started to develop on the leaves of K. spuria. We suggest that if samples are to be stored for such a long period, rehydration should be conducted at a low temperature, especially if other measurements (e.g. chemical composition) are planned.

Apart from this, storage temperature during rehydration had little effect on the final values of either trait (Fig. 2). The lack of an effect of low temperature on LDMC conflicts with other studies showing that final fresh mass is usually less when rehydration occurs at a low temperature (Barrs & Weatherley 1962; Millar 1966). The interpretation given is that rehydration involves water entry in separate regions of the leaf, some of which depend on metabolic activity (Barrs & Weatherley 1962). We do not know why this was not observed in the present experiment. The slightly smaller final SLA value attained at low temperature in K. spuria (−14%: Fig. 2a) may be the consequence of lower respiration at this temperature compared with ambient.


Although the experiment was conducted on a clear, warm summer day (Fig. 3a), we found only little evidence of diurnal variations in either SLA (Fig. 3b) or LDMC (Fig. 3c) in any of the three species (for either non-rehydrated or rehydrated leaves). This contrasts with previous experiments showing that (non-rehydrated) SLA tends to decrease during the day, at least in crop species (Chatterton et al. 1972; Delgado et al. 1992; Bertin et al. 1999; Tardieu et al. 1999) and (to a lesser extent) in trees (Jurik 1986). These variations are attributed to transient increases of non-structural carbohydrates in leaves (Chatterton et al. 1972; Bertin & Gary 1998), reflecting a temporary imbalance between source and sink activities. This certainly occurs in the species studied, but two hypotheses can be put forward to explain why their impacts on SLA were not significant: (i) the highest rates of change in carbohydrate concentration may have occurred before and after the sampling period covered here (09·00–16·30 h), and (ii) imbalances between carbohydrate production and export may be less pronounced in wild species than in crops. Expanding the period of measurement from sunrise to sunset and the determination of non-structural carbohydrate concentrations would be required to test whether these hypotheses hold.

The small variations in LDMC of non-rehydrated leaves during the day actually correspond to significant decreases in relative water content – a quantity more often used to describe leaf water status (Barrs 1968; Turner 1981) – both in K. spuria (from 86 to 73%) and P. bituminosa (from 70 to 62%), but not in P. latifolia (constant value of 75–76%). In fact, since variations in leaf water status are usually pronounced following sunrise or preceding sunset (e.g. Jordan & Ritchie 1971; Girma & Krieg 1992), the variations recorded here between 09·00 and 16·30 h are probably smaller than the daily range actually spanned. This is also suggested by the systematically lower LDMC (higher water content) obtained after rehydration at all times of sampling (Fig. 3c).

The relative stability of SLA and LDMC of non-rehydrated leaves was also found for SLASAT and LDMCSAT (Fig. 3b,c), suggesting that sampling can be conducted safely from 2–3 h after sunrise to 3–4 h before sunset.



Both SLA and LDMC were sensitive to leaf hydration. The higher the SLA of the species, the more it was affected by leaf water content. There appears to be little practical difference between the rehydration procedures tested. Leaf saturation was completed within 6 h, and no differences were detected in leaves rehydrated at different temperatures or collected at different times (09·00–16·30 h).

We propose that SLA and LDMC obtained on water-saturated leaves (SLASAT and LDMCSAT, respectively) be used for species comparisons. We suggest the following protocol for their determination:

  • Select young, fully expanded and illuminated leaves, without serious herbivore or pathogen damage, as recommended previously (e.g. Reich et al. 1992; Westoby 1998; Weiher et al. 1999).
  • Sample plant material at least 2–3 h after sunrise and 3–4 h before sunset.
  • Store the cut stems in a cool and wet environment until their return to the lab.
  • Place the cut end of the stems into water upon arrival in the lab.
  • Store the samples in the dark for at least 6 h (if samples are to be stored for longer than 24 h before measurements, store at low temperature to avoid fungal development).

Such standardization is a prerequisite for the compilation of large databases of functional traits (Keddy 1992; Westoby 1998; Weiher et al. 1999).


Many thanks to the staff of the CEFE experimental garden for their help in the setting of the experiment, and to Christian Collin and Jean-Louis Salager for kindly providing us with the meteorological data. Comments by Marie-Laure Navas and Muhaymina Sari improved the quality of the manuscript.

Received 9 February 2001; revised 4 May 2001; accepted 7 May 2001