Components of leaf elongation rate and their relationship to specific leaf area in contrasting grasses


  • J. Tulio Arredondo,

    Corresponding author
    1. Chair of Grassland Sciences, Technische Universität München, D-85350 Freising-Weihenstephan, Germany
      Author for correspondence: J. Tulio Arredondo Tel: +52 4448 335409 Fax: +52 4448 335412 Email:
    Search for more papers by this author
  • Hans Schnyder

    1. Chair of Grassland Sciences, Technische Universität München, D-85350 Freising-Weihenstephan, Germany
    Search for more papers by this author

Author for correspondence: J. Tulio Arredondo Tel: +52 4448 335409 Fax: +52 4448 335412 Email:


  • • In grasses the leaf growth zone is the main site of shoot growth where anatomical and chemical characteristics of leaves originate. Yet, there is insufficient information to generalize as to whether the leaf growth zone reflects habitat characteristics, whether leaf growth traits are regularly interrelated, and whether they coincide with characteristics of mature leaves.
  • • Here the contribution of both length of the leaf growth zone (region where cell division and expansion occurs) and relative elemental growth rate to the variability in leaf elongation rate (LER) were examined in eight grass species from habitats with different soil fertility. Further, we examined the relationship of the above traits with specific leaf area (SLA) and its components.
  • • Growth zone length differed significantly among species (P < 0.05) and it was the trait contributing the most to LER. Using LER and derived components it was possible to classify seven out of eight species into two groups related to soil fertility. Leaf elongation rate exhibited a positive correlation to SLA and a negative correlation to leaf dry matter content. A significant relationship existed between size of the growth zone and leaf dry matter content.
  • • The results suggest that the leaf growth zone is a critically important leaf trait that explains inherent differences in LER and other plant characteristics of grass species.


Shoot growth in grasses depends on several interrelated processes including the participation of apical, intercalary, and axillary meristems. These three meristem types contribute to overall shoot growth through leaf growth and tillering. Leaf growth in grasses is initiated by the division of cells at the base of leaves; it starts as a linear process by which cells are displaced in parallel longitudinal rows by the continuous production and expansion of cells (MacAdam et al., 1989). The location where cells stop expanding marks the end of the leaf growth zone and the initiation of the differentiation zone. As the leaf tissue is expanding, it is also differentiating, thus acquiring distinct anatomical, chemical, and physiological properties (Boffey et al., 1980). These processes are associated with the consumption of substantial amounts of organic substrate within the leaf growth zone (Schnyder et al., 2000).

Clearly, leaf expansion rate is controlled by factors affecting the leaf growth zone. Studies on leaf expansion in grasses have usually focused on the axial component of leaves, that is, on the leaf elongation rate (LER; mm h−1), the rate of increase in leaf length (Begg & Wright, 1962; Robson, 1973; Kemp, 1980; Schnyder et al., 1987; Fiorani et al., 2000). This is because LER is the dominant component of leaf expansion and can be measured relatively easily. Leaf elongation rate is the product of two leaf growth zone properties: length of the growth zone (mm) and the (average) relative elongation rate of tissue in the growth zone (mm mm−1 h−1). The latter has been termed relative elemental growth rate (REGR, Erickson, 1976) or strain rate (Silk, 1984). Thus, variation in LER may be a result of the variation in REGR, length of the leaf growth zone (from now on LGZ) or in both.

Thus far, the contributions of LGZ and REGR to inherent differences in LER have been studied in only a relatively narrow range of species and/or genotypes: one fast- and one slow-LER genotype of Festuca arundinacea (Volenec & Nelson, 1982), two Aegilops species (Van Volkenburg et al., 1998) and four Poa species (Fiorani et al., 2000). In these studies, differences in LER were mostly accounted for by differences in LGZ. However, to date no systematic comparative study of LGZ characteristics has been performed for species growing along environmental gradients, for example, along a gradient of soil fertility. Bultynck et al. (1999) suggested comparing species from contrasting environments to gain insight into those cellular mechanisms leading to high and low leaf growth potentials.

On the other hand, it is well established that species differences in leaf tissue chemical composition and anatomy are reflected also in plant growth differences (Garnier & Laurent, 1994; Van Arendonk & Poorter, 1994; Poorter & De Jong, 1999). Slow-growing grass species generally contain more sclerenchyma tissue and consequently have a higher concentration of cell wall material and a lower concentration of cytoplasmic and vacuolar compounds than fast-growing species (Van Arendonk & Poorter, 1994). Fast-growing species, on the other hand, exhibit higher concentrations of organic N-compounds, organic acids and minerals (Poorter & Bergkotte, 1992). Important components or constituents of these characteristics of mature leaf tissue are determined at early stages of the leaves during tissue expansion in the leaf growth zone. Groeneveld et al. (1998) showed that 50% and 75% of the ultimate cell wall polysaccharide content was produced when cells resided in the growth zone in the fast-growing grass Holcus lanatus and in the slow-growing grass Deschampsia flexuosa, respectively. For a number of species these tissue anatomical and chemical characteristics are correlated with specific leaf area (SLA, ratio of leaf area to leaf dry mass) (Garnier & Laurent, 1994).

From all the above, it can be seen that leaf meristematic processes may determine the length of the growth zone and the relative elemental growth rates, which are both components defining LER. In this study, we compared eight grass species from habitats differing in soil fertility to have a range of plant growth rates and leaf growth rates that could allow us to examine the basic components associated with inherent species differences in LER. We tested the general hypothesis that both size of the elongation zone and REGRmax explain most of the variations in LER among grasses from habitats differing in soil fertility. In addition, leaf meristematic processes may contribute to define leaf tissue characteristics such as leaf dry matter concentration (leaf dry matter per volume of leaf) and leaf anatomy (e.g. leaf thickness) which are traits defining SLA. Hence, we ask the question whether inherent differences in length of the growth zone, REGR, and LER among grasses from contrasting habitats coincide with inherent differences in leaf dry matter content (equivalent to leaf dry matter concentration, Shipley & Vu, 2002), leaf thickness, and SLA. We are unaware of any studies having addressed this question; however, there are a handful of studies that offer partial evidence supporting our view (Horst et al., 1978; Lambers, 1998; Van Volkenburgh et al., 1998). Thus, we also tested the hypothesis that inherent differences in elongation zone, REGR and LER among species coincide with inherent differences in leaf dry matter content, leaf thickness and SLA.

Materials and Methods

Seeds of eight grass species Lolium perenne L., Festuca pratensis Huds., Dactylis glomerata L., Arrhenatherum elatius (L.) P. Beauv., Avena pubescens Huds., Brachypodium pinnatum (L.) P. Beauv., Bromus erectus Huds., and Festuca ovina L., were obtained from a commercial source (Saatzucht Steinach GMBH, Steinach, Germany). Although it seems unlikely that the seed source included the genetic diversity that exists within the respective species in natural conditions, we attested that the plants from all our species exhibited the morphological characteristics which are typical for them. The selected grass species are normally associated with nutrient-poor and nutrient-rich habitats according to the Ellenberg N indicator values (Table 1) which rank the species along a scale from one to seven corresponding to low to high soil fertility (mainly soil nitrogen availability) of the habitat in which they thrive best, respectively (Ellenberg et al., 1991). The first-four grasses of the list grow on fertile soils while the last-four occur mainly on soils of moderate to low fertility (Ellenberg et al., 1992). We selected species from a gradient of soil nitrogen availability since these usually exhibit inherent differences in growth rates and other plant growth components such as tissue chemistry and anatomy. We expected that plants adapted to different levels of soil fertility have evolved inherently different meristematic characteristics.

Table 1.  Ellenberg nutrient indicator value (N), and average values for leaf elongation rates (LER), length of leaf growth zone (LGZ), sheath length (SHT), specific leaf area (SLA), leaf dry matter content (LFDM), leaf thickness (LFTK), and maximum relative elemental growth rates (REGRmax) for eight grass species. Different letters indicate significant differences between values in a column at P < 0.05
mmmmmm h−1m2 kg−1g g−1cm−1mm mm−1 h−1
Arrhenatherum elatius725.4 ab54.8 a0.68 b19.8 ab0.30 abc0.017 b0.41 abc
Avena pubescens417.0 d30.7 c0.38 f12.4 c0.34 a0.027 f0.27 c
Bromus erectus320.9 cd45.3 b0.88 a17.6 b0.26 bcd0.022 de0.44 ab
Brachypodium pinnatum413.2 e27.6 c0.47 de19.7 ab0.35 a0.014 a0.54 a
Dactylis glomerata624.2 b41.7 b0.43 ef16.8 b0.32 ab0.019 bc0.35 bc
Festuca ovina113.1 e24.6 d0.40 ef16.8 b0.29 abc0.021 cd0.48 ab
F. pratensis627.6 a54.5 a0.59 c19.8 ab0.23 cd0.022 de0.39 bc
Lolium perenne720.4 bc45.4 b0.58 cd22.2 a0.20 d0.023 e0.48 ab

Seeds of each of the species were germinated in pots containing a 50 : 50 mixture of sand and topsoil. One plant of each species was selected for vegetative propagation 3 months after seeding. Twenty-five third generation clones per species were transplanted to 10 × 10 × 15 cm pots containing sand. The transplanted plants were maintained under glasshouse conditions for an additional 2–4 wk depending on the experiment. Irrigation and fertilization were carried out with an automatic system which supplied approx. 50 ml of a modified half strength Hoagland-nutrient solution (Hewitt, 1966) per pot and per day. In weekly intervals an excess volume of distilled water was applied to leach out the accumulated salts.

Plants with at least 10 tillers were moved to growth chambers for a two-wk period of acclimation. Conditions in the growth chamber were set to 70% rh, a constant temperature of 21°C measured at the base of tillers, a photoperiod of 16 h, and 390–430 µmol m−2 s−1 photosynthetic photon flux density.

Experiment I: determination of length of leaf growth zone, REGR, and LER

Leaf elongation rate (LER, mm h−1) was recorded on eight randomly selected ramets (replicates) per species after 2 wk of acclimation. The length of the leaf lamina between the ligule of the youngest fully expanded leaf and the tip of an emerging leaf was measured daily. Criteria for tiller selection were the following: tillers had to have both a recently emerged leaf tip extending 1–3 cm above the encircling sheath of the last fully expanded leaf lamina and a similar sheath size. Based on reports suggesting that tiller size appears to influence the length of the leaf growth zone (LGZ) (Tonkinson et al., 1997; Bultynck et al., 1999), we established arbitrarily three sheath size categories based on the length of the last fully expanded leaf sheath. For the taller species (L. perenne, F. pratensis, D. glomerata, A. elatius, B. erectus) the three sheath size categories considered were 3.5–4.4 cm, 4.5–5.4 cm, and 5.5–6.5 cm. For the shorter species (A. pubescens, B. pinnatum, and F. ovina) the sheath size categories included 2–2.4 cm, 2.5–2.9 cm, and 3–3.5 cm. These ranges covered most of the sheath lengths observed in all plants. The largest average sheaths lengths were observed in A. elatius and F. pratensis (54.8 and 54.5 mm) whereas the shortest were observed in B. pinnatum and F. ovina (27.6 and 24.6 mm). Further, we selected only those tillers which had formed at least four leaves. This selection process allowed us to compare developmentally similar tillers. In all cases, we collected tillers located at the perimeter of the plant to reduce bias due to different light conditions within the ramet.

Leaf elongation was measured on four tillers per ramet (32 tillers per species) during a 4-d period before tillers were harvested for further evaluation. Leaf elongation rates were calculated as the average increment in leaf length per hour estimated for the 4-d period. Due to the labour intensive harvest, tiller evaluation was carried out on only one replicate per species on each sampling date, so that tiller evaluation per species was spread over 8-d (eight replicates). Although, the measurements were conducted over 12 d, in any case tillers were scheduled for evaluation of leaf growth components immediately after the 4th day of LER measurements. After the harvest of selected tillers the plants were returned to the growth chamber to maintain the same growth conditions for the remaining plants.

On day 4 of leaf length monitoring, the spatial distribution of leaf growth was determined on three tillers (one tiller per sheath size category) per species using a pinning technique (Schnyder et al., 1987). Briefly, holes were pinned through intact tillers using a fine needle (0.2 mm diameter); the first hole was placed near the base of the tiller and the consecutive holes were placed approximately 3 mm apart (2 mm in the short-statured grasses) in acropetal direction. We followed the recommendations of Peters & Bernstein (1997) to mark leaf segments of approx. 10% of the length of the growth zone. Leaves had generally reached < 50% of their mature size before being pinned. After pinning (24 tillers in approx. 1 h), leaves were allowed to grow for 6 hours before they were dissected from the plant. Once dissected, tillers were placed in hermetically closed plastic bags (Ziploc type) from which most of the air was removed immediately to allow for rapid humidity equilibration. Bagged tillers were stored in the cold (2–4°C) until evaluation that occurred within 3 h after the harvest. In a preliminary test, we did not detect any differences in LER and leaf growth distribution between refrigerated and nonrefrigerated tillers (data not presented) after 12 h of cold storage.

For the evaluation of the spatial distribution of leaf growth, the length of the sheath of the youngest fully expanded leaf was measured; then the leaf growth zone of the emerging leaf was carefully freed by removing the surrounding sheaths. Distances between holes were measured for both a mature nongrowing sheath and the growing leaf using a 8× magnifying glass with a micrometer (0.1 mm accuracy). Relative elemental growth rates (REGR mm mm−1 h−1) were calculated as described by Erickson and Sax (1956):

REGR = (df − di)/(di × Δt);

df, length of a leaf segment (as delimited by two neighbouring pinholes) in the growing leaf at the time of observation; di, length of the leaf segment at pinning, and Δt is the time period between pinning and observation. The parameter di was taken as the length of the same segment in the nongrowing sheath surrounding the growing leaf of interest. As pinning reduces LER but does not affect the relative spatial distribution of growth along the leaf growth zone (Schnyder et al., 1987; Hu & Schmidhalter, 2000), we corrected the data by multiplying REGR times the ratio of the LER of unpinned leaves to LER of pinned leaves (pinning caused on average 30% decrease in LER). The position (relative to the leaf base) at which REGR declined to < 10% of the maximum REGR was considered as the distal limit of the leaf growth zone. The contribution of LGZ and REGRmax (the leaf segment were maximum REGR was observed) to the interspecific variation in LER was calculated using the growth response coefficients (GRC) proposed by Poorter & Van Der Werf (1998). In the present case, GRC components were calculated as the slope of a linear regression with ln(LGZ) or ln(REGRmax) as the independent variable, and ln(LER) as the dependent variable. GRC values of 1 for a particular component (LGZ or REGRmax) indicate that this trait is accounting for all of the observed variability of LER.

Experiment II: determination of SLA and its components

Sixteen plants per species were established in growth chambers after vegetative propagation (as described above). Clones consisted of 3–4 trimmed tillers and the fresh mass of each clone was recorded at transplanting time. The growth conditions inside the chamber were the same as for Experiment I. Four ramets (replicates) per species were harvested every 5 d for a total of four harvests. At harvest, plants were separated into roots, leaf lamina, sheaths, and the crown (consisting mainly of the interconnected bases of tillers) and both f. wt and d. wt was determined for each fraction. Additionally, the leaf areas of mature and emerging leaf lamina of each individual plant were measured with a leaf area meter (LI-3100, Li-Cor Inc., Nebraska, USA). Leaf dry matter content was calculated as the ratio of leaf dry mass to leaf fresh mass. Leaf dry matter content was used as an estimate of leaf dry matter concentration (Shipley & Vu, 2002), one of the two components of SLA. Mean leaf thickness, the second SLA component, was estimated as the ratio of leaf volume (m3) to leaf area (m2) under the assumption that leaf volume (neglecting air spaces) is approximately equal to leaf fresh mass (106 g fresh mass ≈ 1 m3). This assumption is based on the fact that water (tissue) is largely incompressible and has a near constant density of 1. Similarly, the density of cellulose and other biomass components is near 1.3, thus yielding a density of total fresh mass slightly larger than 1. A close correlation between the fresh mass to leaf area ratio and the measured leaf thickness was confirmed for a very large number of species for leaf thicknesses ranging between 0.1 and 1.1 mm (E. Garnier, personal communication). The present estimate of leaf thickness corresponds to the average of the thickness of veinal and valley parts of lamina.

Finally, specific leaf area (SLA, the ratio of leaf area to leaf dry mass, m2 kg−1) was calculated.

Statistical analysis

Observations for each variable were examined for normal distribution using the Shapiro-Wilk statistic (Sokal & Rohlf, 1981). Deviations from normality were corrected by logarithmic transformation. Relationships among variables were established using both type I and type II regression analysis. Examination of correlation between plant traits usually included the means (eight replicates) for each of the three sheath length categories (n = 24). In the type II regression analysis we used the reduced major axis-method (RMA) to examine trait associations. Regression coefficients determined by RMA regression (rRMA) equal regression coefficients calculated by least square regression divided by the coefficient of correlation (rRMA = rLS/r) (Sokal & Rohlf, 1981). In this case, we compared RMA regression coefficients by examining the overlap between confidence intervals (C.I.) for the regression coefficient. Since we examined variables (LGZ, REGR, SLA, etc.) which are subjected to natural variation and measurement error they can not be considered truly independent variables (Niklas, 1994). Still, we define LER as the dependent variable predicted by LGZ and REGRmax. When we correlated traits belonging to different scales of organization (e.g. SLA) we used the trait from the higher scale of organization as dependent variable (e.g. LER vs SLA) assuming that this would derive from the lower scale process (e.g. LER).

The influence of sheath length categories on LER and LGZ was analysed with an anova using a nested design with species and sheath size category as classification variables (Proc GLM, SAS, 1988). A one-way anova was used to examine species differences in REGRmax.

To determine whether linear combinations of characters could be used to discriminate traits depending on habitat characteristics we implemented a discriminant analysis on the different grass species. Canonical scores were constructed using species identity as classification variables and sheath length, LGZ, and LER as quantitative variables using PROC DISCRIM of SAS (SAS, 1988).


The anova showed significant differences among species for the average of LGZ, sheath length (SHT), LER, and REGRmax (Table 2, P < 0.05). Avena pubescens, B. pinnatum, and F. ovina, exhibited the smallest LGZ and SHT as well as the lowest LER among all the grasses (Table 1). Remarkable differences (Table 1, P < 0.05) were observed in LGZ with B. pinnatum and F. ovina having the shortest (about 13 mm) and F. pratensis and A. elatius (approx. 27 mm) having the longest leaf growth zone. Leaf elongation rates differed significantly within species and this depended on sheath length category (sheath categories nested within species, Table 2, P < 0.05).

Table 2.  Associated mean squares and F-values for length of leaf growth zone (LGZ), sheath length (SHT), and leaf elongation rates (LER) analyzed as a nested design using species and SHT nested in species as main factors. Also, presented is the associated mean square and F-value of one-way anova for maximum relative elemental growth rates (REGRmax) with species as the main factor. * and ** indicate significant differences between means of species and SHT nested in species at P < 0.05 and P < 0.01, respectively
Sources of Var.LGZSHTLERREGRmax
dfMSF-value dfMSF-valuedfMSF-valuedfMSF-value
Species  70.291614.89**  70.2992374.8**  70.2916 2.16** 70.0002 2.93*
SHT (Species) 150.0293 1.61 160.0388 48.6** 160.035615 
Error1150.0181 1430.0007 1460.0164 150.00007 

There was an overall positive and significant association between the SHT and LGZ (Fig. 1a; P < 0.01, r 2 = 0.75, n= 24) where longer sheaths yielded longer leaf growth zones. anova of LGZ showed no differences within species attributable to sheath size category (Table 2), although, the observed value of probability (P = 0.08) suggests marginal differences in LGZ. Nevertheless, except for A. elatius and B. pinnatum, each species showed a trend towards increasing LGZ with increases in SHT (Fig. 2), and least square regressions within species showed significant relationships between LGZ and SHT in A. pubescens, F. ovina, and F. pratensis (data not presented).

Figure 1.

Plots of length of leaf growth zone vs sheath length (a), leaf elongation rate vs sheath length (b), and leaf elongation rate vs length of leaf growth zone (c) for eight grass species. Symbols are means for each sheath size category. Symbols stand for Arrhenatherum elatius (open circles), Dactylis glomerata (open squares), Festuca pratensis (open triangles), Lolium perenne (open diamonds), Brachypodium pinnatum (solid squares), Avena pubescens (solid circles), Festuca ovina (solid triangles), and Bromus erectus (crosshair).

Figure 2.

Averages and standard errors of the leaf growth zone observed in three categories of sheath length for eight grass species. Solid bars correspond to the shortest sheath category (3.5–4.5 cm and 2–2.5 cm, for tall and short species, respectively), grey-tone bars correspond to the middle sheath category (4.5–5.5 cm and 2.5–3 cm), and the open bars correspond to the largest sheath category (5.5–6.5 cm and 3–3.5 cm).

Leaf elongation rates varied significantly with SHT and LGZ (Fig. 1b,c; P < 0.01, r2 = 0.46 and r2 = 0.19, respectively; n= 24) including all three sheath categories. In both cases, LER increased with SHT and LGZ. The scaling regression coefficient (rRMA) between LER and SHT (LER = 0.005 SHT4.5, r2 = 0.47, rRMA = 9.2 ± 0.15, C.I. = 4.5–18.6) and LER and LGZ (LER = 0.015 LGZ2.7, r2 = 0.19, rRMA = 9.6 ± 0.19, C.I. = 3.9–23.4) were similar; reflecting indirectly a relationship between SHT and LGZ. The estimation of growth response coefficients (GRC) showed that LGZ explained about three-quarters of the variation in LER (0.757 ± 0.486 and 0.279 ± 0.282 for LGZ and REGRmax, respectively). There was a significant relationship between SLA and LER (Fig. 3a) although the coefficient of determination was relatively low (r2 = 0.18). No relationship existed between SLA and LGZ. Leaf dry matter content exhibited a negative relationship with LER (Fig. 3b; P < 0.001). Leaf dry matter content also maintained a negative relationship with LGZ (Fig. 3c). Both SLA and leaf dry matter content varied proportionally with SHT (Table 3). Overall, REGRmax exhibited a marginally positive association with SLA, and a significant relationship with leaf thickness, while no association was detected with leaf dry matter content (Table 3).

Figure 3.

Relationships between specific leaf area and leaf elongation rate (a), leaf dry matter content and leaf elongation rate (b), and leaf dry matter content and length of the leaf growth zone (c). Symbols represent mean values of final harvest. Symbols stand for Arrhenatherum elatius (open circles), Dactylis glomerata (open squares), Festuca pratensis (open triangles), Lolium perenne (open diamonds), Brachypodium pinnatum (solid squares), Avena pubescens (solid circles), Festuca ovina (solid triangles), and Bromus erectus (crosshair).

Table 3.  Intercepts, slopes, coefficients of determination and probability values of least square regressions between leaf biomass allocation traits and leaf elongation traits. Initials stand for specific leaf area (SLA), leaf dry matter content (LFDM), length of growth zone (LGZ), leaf elongation rate (LER), sheath length (SHT), and maximum relative elemental growth rate (REGRmax)
Leaf traitsInterceptSloper2nPvalue
Var. 1 Var. 2
SHTvsLGZ 5.25        0.37 ± 0.030.75230.01
SHTvsLER 0.0170.0009 ± 0.00020.44240.01
LGZvsLER 0.0270.0013 ± 0.00050.17310.021
REGRmaxvsLER 0.025  0.26 ± 0.100.26200.018
REGRmaxvsSLA 0.0190.0013 ± 0.00060.16230.053
REGRmaxvsLFDM 0.042 −0.002 ± 0.040.001230.96
REGRmaxvsLFTK 0.018 0.005 ± 0.0020.18230.041
LERvsSLA 0.0190.0047 ± 0.0020.18230.041
SHTvsSLA12.7  85.6 ± 30.40.22310.01
LGZvsSLA12.9 0.113 ± 0.050.13310.048
LERvsLFDM 0.35  −1.18 ± 0.50.16320.02
SHTvsLFDM 0.37 −0.003 ± 0.00070.29300.01
LGZvsLFDM 0.35−0.003 ± 0.0010.12310.062

Discriminant analysis using mean values of SHT, LGZ, and LER as predictor variables revealed affinities among species that could be assigned to one of three groups. The first canonical discriminant function discriminated B. erectus from all the other grass species (Fig. 4). Both SHT and LER contributed the most to the classification of this function. The second function discriminated a group of species consisting of F. pratensis, D. glomerata, A. elatius, and L. perenne, all species usually occurring in fertile habitats from a group of species consisting of F. ovina, B. pinnatum, and A. pubescens, species typically found on unfertile soils. In the second function, B. erectus was classified with the group of species that prosper in fertile habitats. Examining the canonical structure, LER exhibited the strongest correlation with the first discriminant function whereas, LGZ and sheath length showed the strongest correlation with the second function (0.76 and 0.57, −0.003 and 0.92, and −0.20 and 0.98, for the first and second function of LER, SHT, and LGZ, respectively). Therefore, the first function appears to be related to leaf growth and the second function to size (tiller and meristem).

Figure 4.

Plot of canonical scores using LGZ, sheath length, and LER to discriminate among eight grass species. Standardized coefficients for the first canonical discriminant function (−3.04 (SHT), 3.95 (LER), and 0.55 (LGZ)) discriminated Bromus erectus from the other grasses. The second function (−0.04 (SHT), 0.47 (LER), and 1.77 (LGZ)) separated species from fertile and unfertile habitats. The coefficients indicate the partial contribution of variables to the function. Symbols stand for Arrhenatherum elatius (open circles), Dactylis glomerata (open squares), Festuca pratensis (open triangles), Lolium perenne (open diamonds), Brachypodium pinnatum (solid squares), Avena pubescens (solid circles), Festuca ovina (solid triangles), and Bromus erectus (crosshair).


Contribution of LGZ and REGRmax to LER

Inherent variations in LER were analysed in terms of the contribution of its two components LGZ and REGR (Skinner & Nelson, 1995) to the overall variation between species. Based on the growth response coefficients (GRC) we confirmed our hypothesis that both LGZ and REGRmax explained most of the variations in LER, however, the contribution of LGZ was more important than the contribution of REGRmax (GRCLER = 0.757 ± 0.486, GRCREGR max = 0.279 ± 0.282). These results are in agreement with the outcome of the studies by Volenec & Nelson (1982), Paolillo & Sorrells, (1992) and Fiorani et al. (2000) which showed that the inherent variation in LER was also more closely associated with the variation in LGZ than with REGR. In our study, REGRmax contributed only 25% of the variation in LER, however, the relationship was significant between these two traits (Table 3).

Notably, grasses from infertile habitats tended to have higher REGRmax than grasses from fertile habitats (Table 1, e.g. in these grasses LGZ was shortest). Similarly, Fiorani et al. (2000) reported that Poa alpina and P. compressa, two grasses common to harsh alpine and subalpine habitats, exhibited the highest cell strain rates (equivalent to our REGRmax) but shortest LGZ, whereas P. trivialis and P. annua, species from mild temperate conditions exhibited long LGZ but low cell strain rates. Also a similar response, characterized by a shortening of the elongation zone and increases of REGRmax was observed in roots of Maize (Sharp et al., 1990). Our results and the Fiorani study suggest that habitat environmental conditions may have played a role in shaping the characteristics of REGR. This way, plants evolved under harsh conditions may compensate the reduction in meristem size with a greater capacity to increase REGRmax.

In the work of Fiorani et al. (2000) the differences in LER and LGZ among four Poa species were primarily due to differences in the number of cells produced per cell file and per unit time, which led to differences in the number of cells in the growth zone and consequently in the length of the leaf growth zone. These differences in cell production rate arose from differences in the number of dividing cells rather than from differences in rates of cell division. The parallels between Fiorani's results and our data regarding the contribution of REGR and LGZ to LER may be taken to suggest that differences in LGZ and LER in our grasses may have arisen from both differences in meristem size and differences in cell production rate.

Additionally, it has been suggested previously that the length of sheath or tiller size may influence LER and LGZ (Volenec & Nelson, 1983; Skinner & Nelson, 1995), but up to now the generality of this relationships over a range of contrasting species has not been tested. The present results show that sheath length was indeed near-proportional to LGZ and LER among a significant range of species (Fig. 1a,b). These relationships also appeared to hold for hierarchical classes of tillers within a plant for most species except B. pinnatum and A. elatius (Fig. 2). The causal mechanisms for this relationship between sheath length and both LER and LGZ are not clear. One possibility could be that spatial constraints for the meristem size decreased with tiller size, so that larger tillers favour larger meristems. However, while such a mechanism is consistent with the relationships observed in most species it would not explain the relationships observed in B. pinnatum and A. elatius.

Leaf width is another trait that may have had an effect on LER and its components. It is possible that differences in leaf width occur concomitantly with increases in cell row number and consequently with assimilate demands in leaf growth zones. Thus, if assimilate demand is high then LER and its components could be affected negatively. Accordingly, Dactilys glomerata, the species with the widest leaves in our study, showed LER values that were comparatively similar to those observed in some species from unfertile soils (Figs 1b,c and 3a,b). Apparently the relatively low LER of D. glomerata occurred as a consequence of decreasing REGRmax rather than as a consequence of a shortening in the LGZ (Table 1).

Do variations in LER, LGZ, and REGRmax coincide with variations in SLA and its components?

The present results show significant relationships between LGZ and LER with leaf dry matter content and SLA (Fig. 3) in grasses adapted to different soil fertilities. This may indicate that both leaf growth zones and mature leaf traits originate from similar meristematic processes, and thus, that species-specific differences in SLA and dry matter content are causally related to the processes determining inherent variation in LER and LGZ. Unfortunately, this sort of interpretation is not necessarily straightforward, since there are operational problems involved in the interpretation of trait relationships, such as the scale of processes of traits being correlated (Bultynck et al., 1999; Gunn et al., 1999). For instance, inherent plant characteristics such as SLA are the result of multiple integrated processes, such as secondary cell wall deposition which only occurs after the cessation of cell expansion (MacAdam & Nelson, 1987; Maurice et al., 1997). Therefore, relationships of SLA with either LGZ or LER, which represent a single organ trait or process, may be coincidental and not a result of similar origin as pointed out by Bultynck et al. (1999) and Gunn et al. (1999). Still it is possible, however, that mechanisms that underlie variations in LGZ or LER are the same that provoke variations in SLA (Lambers, 1998).

The relationship between LGZ and leaf dry matter content, a component of SLA, appears to be more directly related because a large part of dry matter deposition occurs in the leaf growth zone (Groeneveld et al., 1998). Therefore, the amount of dry matter needed and deposited in the leaf growth zone should maintain a proportional relationship to LGZ assuming that leaf width is more or less similar among species. For instance, comparing the leaf dry matter content of leaf growth zones and mature leaf laminas for L. perenne, A. elatius, A. pubescens, and B. pinnatum (Arredondo JT & Schnyder H, unpublished), we observed a close correlation between them (P = 0.06, r2 = 0.89, n= 4). Dry matter content in the leaf growth zones and mature leaves were nearly proportional. This result was (at least partially) because of the fact that a large fraction of cell wall polysaccharide deposition occurred during cell expansion (Groeneveld et al., 1998). The studies by Fiorani et al. (2000) and Lambers (1998) also suggest that differences in meristematic growth processes (cell division and elongation) may directly or indirectly determine plant growth (e.g. RGR) and the characteristics of growth parameters (e.g. SLA).

Are there inherent differences in LER?

Our results using LER, LGZ, and sheath size as predictor variables for eight grass species allowed us to group them according to habitat similarities (Fig. 4). Our classification (except for Bromus erectus) coincided with other studies in how they ranked grass species using relative growth rates (RGR) and biomass allocation patterns (SLA) (Grime & Hunt, 1975; Garnier, 1992; Poorter & Bergkotte, 1992; Garnier & Laurent, 1994; Lambers et al., 1998). Grass species typical of fertile and infertile habitats showed clear differences in LER, LGZ, and SHT (Table 1). For additional evidence that trait differences correspond to the habitat gradient (nitrogen availability), we examined whether LER, LGZ, and SHT were related to the Ellenberg N indicator values (Ellenberg et al., 1992). A significant positive relationship existed only between LGZ and N-number (Fig. 5; P < 0.05, r2 = 0.49, n= 8) with species from infertile habitats (low N indicator value) tending to have shorter LGZ; this in turn was related to the shorter sheath length and, hence, shorter stature. Whereas this relationship does not prove that resource availability selected for LGZ characteristics, it shows that LGZ has a distribution that corresponds with fertility gradients. Bromus erectus was a clear exception to this pattern of trait distribution. It is possible that part of this behaviour depends on its particular plant architecture characterized by tall robust tillers assembled in loosely packed tussocks (low tiller number per unit area). With a low tiller density it is possible that resources may become more easily available for individual tillers, hence fulfilling the demands that could develop from large leaf growth zones. These large leaf growth zones on the other hand were evidently able to maintain LER equivalent to the ones observed in species from fertile habitats. Conversely, other traits including SLA and leaf dry matter content corresponded better to values reported for this species from previous studies.

Figure 5.

Relationship between length of the leaf growth zone and Ellenberg's N indicator value for eight grass species. Symbols stand for Arrhenatherum elatius (open circles), Dactylis glomerata (open squares), Festuca pratensis (open triangles), Lolium perenne (open diamonds), Brachypodium pinnatum (solid squares), Avena pubescens (solid circles), Festuca ovina (solid triangles), and Bromus erectus (crosshair).

A difficulty in interpreting plant trait relationships from an adaptative point of view arises when phylogenetic relationships among species are ignored (Silvertown et al., 1997). In this case, trait relationships may arise from species sharing a common ancestry rather than as a result from present-day selection pressures (Silvertown & Dodd, 1997). Given the implicit connection between meristems, organs and whole plant processes, one aim of our study was to examine explicitly the generality of trait associations at different levels of organisation such as leaf meristems and whole leaves. In this paper we focus on showing trait ranks among species as well as how traits distribute on habitats differing in soil fertility avoiding interpretations from an adaptive or evolutionary point of view. Our study, similar to that of Fiorani et al. (2000), suggests that in order to determine the mechanism that defines inherent variability in growth and growth components we should focus on the meristem level.


The authors thank F. Fiorani, H. Poorter and E. Garnier for useful comments on an earlier version of this manuscript and two anonymous reviewers. We thank also Angelika Ernst-Schwärzli for her assistance during the study and Jeff Creque and Elisabeth Huber-Sannwald for editorial improvements on the manuscript. This study was supported by the Deutsche Forschungsgemeinschaft (SFB #607).