Resprouting of the Mediterranean-type shrub Erica australis with modified lignotuber carbohydrate content

Authors

  • Alberto Cruz,

    1. Departamento de Ciencias Ambientales, Facultad de Ciencias del Medio Ambiente, Universidad de Castilla-La Mancha, Avda. Carlos III s/n, 45071, Toledo, Spain
    Search for more papers by this author
  • Beatriz Pérez,

    1. Departamento de Ciencias Ambientales, Facultad de Ciencias del Medio Ambiente, Universidad de Castilla-La Mancha, Avda. Carlos III s/n, 45071, Toledo, Spain
    Search for more papers by this author
  • José M. Moreno

    1. Departamento de Ciencias Ambientales, Facultad de Ciencias del Medio Ambiente, Universidad de Castilla-La Mancha, Avda. Carlos III s/n, 45071, Toledo, Spain
    Search for more papers by this author

José M. Moreno (tel. +34 925 26 88 00, fax +34 925 26 88 40, e-mail JoseM.Moreno@uclm.es).

Summary

  • 1The vigour of plant resprouting after fire may be driven by the amount of stored non-structural carbohydrates (NC). However, the extent to which NC reserves limit this response in woody plants has not been established.
  • 2This study analyses the effects of modifying NC concentrations in the lignotuberous Mediterranean-type shrub Erica australis, on resprouting after experimental burning. A factorial experiment with two treatments (shade and burn) was carried out, with three levels per treatment. Plants were shaded (exposure to 100%, 50% or 20% of incident radiation for 1 year), following which they were clipped (control, no fire), or clipped and burned at one of two levels of severity. After this, they were allowed to resprout and grow for 2 years.
  • 3Shading modified NC concentrations in the lignotuber, but not in the roots. Two years after burning, plants subjected to the greatest shading, which reduced their NC concentrations to 44% (sum of NC) or 19% (starch) of that of control plants, had suffered higher mortality after resprouting, had a significantly lower number of resprouts, of shorter length, and had produced lower biomass per plant than less shaded or unshaded plants. However, plants subjected to intermediate shade, which suffered a reduction in NC concentrations to 70% of that of control plants, did not differ in resprouting response from control plants.
  • 4Burning caused more direct mortality, and a severe reduction in both number or biomass of resprouts, than just clipping. There were no significant shade × burning interaction effects.
  • 5While reductions in NC may limit resprouting, such limitation may only occur when NC is reduced to much lower concentrations than caused by commonly experienced conditions. The role of NC reserves in limiting resprouting of lignotuberous, woody plants such as E. australis might therefore not be as important as is commonly assumed.

Introduction

Resprouting from below-ground parts is a widespread mechanism by which many plants regenerate after partial or total defoliation (Bond & Midgley 2001). Carbon supply for regrowth of above-ground parts during the early stages of resprouting depends on stored, non-structural carbohydrates (NC) in the below-ground organs (White 1973). Resprouter species have a significantly greater amount of stored NC in their below-ground parts than non-sprouting congeners (Pate et al. 1990). Below-ground NC concentrations usually decrease during regrowth of the above-ground parts (Jones & Laude 1960; Donart & Cook 1970; Kays & Canham 1991; Van der Heyden & Stock 1995; Canadell & López-Soria 1998), suggesting that NC reserves are mobilized during resprouting to support new growth. Thus, the vigour of resprouting after a defoliating disturbance would be determined by the amount of mobilizable NC (Miyanishi & Kellman 1986; Bowen & Pate 1993). Differences in resprouting vigour have frequently been interpreted as caused by differences in NC reserves. NC concentrations vary seasonally (Cruz & Moreno 2001a), and higher plant mortality and less vigorous resprouting usually coincide with the time of the year at which stored NC is at its lowest (Jones & Laude 1960; Rundel et al. 1987; Bradstock & Myerscough 1988; Trabaud 1991; Bowen & Pate 1993). Furthermore, resprouting plants are sensitive to recurrent fires (Hedrick 1951; Zedler et al. 1983), presumably due to insufficient recovery of NC reserves between successive disturbances.

Nevertheless, several studies have found no correlation between the amount of stored NC at the time of disturbance and resprouting vigour (Richards & Caldwell 1985; Hogg & Lieffers 1991a,b; Erdmann et al. 1993; Sparks & Oechel 1993). In herbaceous plants, high photosynthetic rates, and rapid development of resprouts probably limit use of stored NC to sustain resprout growth to just a few days after a defoliating disturbance (Richards & Caldwell 1985; Danckwerts 1993). In woody plants, the role of stored NC in resprouting has not been well quantified, and is based on the observation that depletion of NC reserves is accompanied by failure to resprout (Miyanishi & Kellman 1986; Bowen & Pate 1993; Canadell & López-Soria 1998). However, after repeated defoliation, plants may not resprout, even if substantial amounts of NC remain in the plant (George & McKell 1978). Furthermore, plants may store more carbon than is needed to support a single resprouting episode (Van der Heyden & Stock 1995; Hoffmann et al. 2000; Cruz & Moreno 2001a). Some studies have shown that differences in resprouting vigour between different species (Van der Heyden & Stock 1995) or populations of the same species (Cruz et al. 2003) were not related to stored NC. In the case of plants affected by fire, resprouting vigour may also depend on the damage inflicted by heat on the bud bank, which, in turn, may be related to fire severity (Moreno & Oechel 1991, 1993). However, it is not known how this would interact with stored NC.

The objective of this study was to determine the effects of experimental modification of stored, non-structural carbohydrates and fire on plant survival, and on resprout production and growth in the Mediterranean shrub, Erica australis. This species has a well developed lignotuber (Cruz & Moreno 2001b), and resprouts vigorously after defoliation (Ojeda et al. 1996; Calvo et al. 1998; Cruz et al. 2002). Lignotubers are viewed as organs evolved to increase storage of NC and/or the number of dormant buds (James 1984; Zammit 1988).

Methods

study site

The study was carried out at the Quintos de Mora Range Station, in the Montes de Toledo mountains, Central Spain (39°23′ N, 4°07′ W). We selected a stand that had not been burned for at least 40 years, at 900 m above sea level, on a south-facing slope with 17% inclination. Soils at the study site are shallow, acidic and nutrient-poor and are derived from quartzite. The climate is Mediterranean-type, with a mean annual temperature of 14.5 °C and mean annual precipitation of 707 mm, of which only 55 mm (7%) falls between July and September (Los Cortijos station, 39°18′ N, 4°04′ W).

In a 150 × 50 m area, 132 plants of E. australis, covering a wide range of sizes, were randomly selected. Plants were sufficiently far apart that they would not interfere with each other because of the imposed treatments. As a measure of plant size we used an index of lignotuber area (π × D1 × D2/4, where D1 and D2 are the two largest diameters of the lignotuber, measured perpendicular one to the other). This index is highly correlated with above-ground total plant biomass (above-ground total biomass [ln g] = 1.30 × lignotuber area [ln cm2] − 1.45; R2 = 0.95; n = 21, Moreno et al. 1999). Plants included in the experiment had lignotuber areas ranging from 38 cm2 to 865 cm2.

shade treatment

A shade treatment was used to produce different NC concentrations in the below-ground organs of the plants by forcing them to use stored NC for maintenance or other functions (Danckwerts 1993; Wyka 1999). Plants were randomly assigned to one of three groups (44 plants per group): control (no shading, 100% of solar radiant flux, from now on L100 plants); 50% light (L50) and 20% light (L20). L50 and L20 shade levels were achieved by covering the plants with green shade-cloth to produce the specified levels of incident radiation. This was confirmed by measuring the amount of light passing through the cloth with a PAR sensor (LI-190SA, LI-COR, Lincoln, NE, USA). Shade treatments were established in spring (late May), and lasted for 1 year in total. After this time, the shade-cloths were removed, predawn shoot water potentials were measured, lignotuber and root samples were taken to determine NC concentrations, and the above-ground biomass of each plant was clipped at the lignotuber level and removed from the site prior to imposition of experimental burning.

Before clipping, predawn shoot water potentials were measured in terminal branches (one per plant) from 39, 37 and 33 individuals of L100, L50 and L20 plants, respectively, with a Scholander-type pressure chamber. Predawn shoot water potentials ranged from −0.5 to −0.7 MPa, and there were no statistically significant differences (anova, P > 0.05) between plants in different shade treatments at the time of burning. Lignotuber and root samples for carbohydrate analysis were taken from five randomly selected plants from each level of shade. Samples were immersed in dry ice, taken to the laboratory and analysed for monosaccharides (glucose plus fructose), sucrose and starch concentrations (mg g−1 dry wt.) using an enzymatic method described in Cruz & Moreno (2001a). All plants sampled for carbohydrate analysis were then discarded. The statistical significance of differences in starch or NC (the sum of all fractions) concentrations between plants from the different shade treatments was tested by one-way anova. After the plants had been burned and allowed to resprout and grow for 2 years, three plants from each of the nine possible combinations of shade and burning treatments (see below) were selected and sampled for new analyses of NC concentration. Differences in starch or total NC concentrations treatments were tested by two-way anova.

burning treatment

The 39 clipping plants from each shade treatment remaining after carbohydrate sampling were divided into three equal groups and randomly assigned to one of the following treatments: no burning (S0), burning (S1) and severe burning (S2). The S0 treatment simulated the effects of biomass removal caused by a fire, but without the effects of heat. S1 and S2 treatments were applied using a propane torch, after enclosing each plant within a metal cylinder (50 cm diameter and 150 cm height) in order to ensure homogeneous heating over the soil and lignotuber surfaces. Temperatures were continuously monitored with two k-type copper-constantan thermocouples, connected to a data-logger (Campbell CR-21X). The S1 treatment exposed the plants to the flame until the temperature reached 400 °C, after which the flame was maintained for an additional 30 s. The S1 treatment reached a maximum temperature of 646 ± 19 °C (mean ± SE, n = 37), and temperatures were above 400 °C, on average, for 100 s (Fig. 1). The S2 treatment maintained exposure to the flame for 60 s after reaching 400 °C, and reached a maximum temperature of 725 ± 20 °C (n = 37) with temperatures in excess of 400 °C for 150 s (Fig. 1). S1 and S2 simulated time-temperature exposures that may be reached during severe wildfires in E. australis vegetation (E. Zuazua, J. R. Quintana, personal communications).

Figure 1.

Mean temperature (± SE) plotted against time during experimental burning of plants in treatments S1 (filled circles, n = 37) and S2 (open circles, n = 37). Each point represents the mean temperature measured with a type-k thermocouple at the centre of the lignotuber.

monitoring of resprouts

Following burning, either 12 (S1/L20, S1/L50, S2/L100, S2/L20) or 13 (S0/L100, S0/L50, S0/L20, S1/L100, S2/L50) plants per treatment could be followed for resprouting. Both the entire experimental plot and each individual plant were fenced to avoid herbivory by mammals. The number of resprouts (defined as a leafy shoot, originating from a different point on the lignotuber) produced by each plant was counted twice, at the end of the first summer (in September, 3 months after burning) and again in spring (May) 2 years after burning. Resprout maximum length was measured at both times, and 9 months after burning (March). At the end of the experiment (May, 2 years after burning), all resprouts were cut, counted, their length measured, and the total biomass and leaf biomass of each plant determined after drying in a forced-air oven at 70 °C for 48 hours. Mean resprout length (cm) per plant was calculated.

We differentiated between ‘direct mortality’ (plants that never resprouted), and ‘indirect mortality’ (plants that resprouted but died later in the experiment). The effect of the shade and burning treatments on direct and indirect mortality was tested by a contingency test (χ2). The effects of the shade and burning treatments on resprout number (square-root transformed), maximum length (ln + 1 transformed), mean length (ln), and total or leaf biomass (ln) per plant were tested by two-way ancova, with lignotuber area (ln) as the covariate. Multiple comparisons of the means for each level within treatments were made using Scheffét-tests. To differentiate between the effects of treatment on plant mortality and on resprouting vigour, only the group of plants that were alive at the end of the experiment (n = 93) was included in the ancovas at any sampling date.

lignotuber carbohydrate reserves after resprouting

To evaluate the possible role played by NC reserves, we calculated the reduction in lignotuber carbohydrate pools that would have been caused by the growth of resprout tissue during the first 9 months after burning (i.e. to the beginning of March of the year following burning), assuming that this would have used only the NC reserves in the lignotuber. This date was chosen because it would mark the end of the period following burning when photosynthetic activity might have been minimal, after burning the plants. The biomass of resprouts produced by each plant by this sampling date was estimated by linear regression based on resprout maximum length. These estimates were based on the measurements made at the end of the experiment (2 years after burning), assuming that the relationships found at this time also held 9 months after burning. We assumed that E. australis plants could produce 0.52 g of biomass per 1 g of mobilized stored carbon, following calculations by Merino (1987) for the leaves of Erica scoparia. The depleted level of reserves expected to result from such a regrowth of E. australis was estimated by subtracting the carbon equivalent contained in the regrowth from the starch or sum of NC reserves stored in the lignotuber of each plant at the time of burning. The size of the initial NC pool contained within the lignotuber of each plant was estimated by multiplying lignotuber biomass (estimated by regression based on lignotuber area), by the percentage of the lignotuber devoted to storage (66%) (Cruz & Moreno 2001a) and by the concentration of NC (of plants from the appropriate shade treatment).

Results

carbohydrate concentrations

At the time of burning, shading had significantly affected the sum of NC and starch concentrations in the lignotubers (anova, P < 0.05). L20 plants had lower concentrations of both carbohydrate fractions in their lignotubers than L100 plants, with L50 plants in an intermediate position (Table 1). However, there was no significant effect of shading on the starch or sum of NC concentrations in the roots (anova, P > 0.05).

Table 1.  Starch and sum of non-structural carbohydrate (NC) concentrations (mean ± SE, mg g−1 dry wt.) in lignotubers and roots of E. australis plants (a) after 1 year of shading (i.e. just before burning, only shade treatment applied) and (b) 2 years after burning and resprouting (shade and burn treatments). Measurements were taken in May of the respective years. Significant differences (P < 0.05) of one-way anovas (before burning) or two-way anovas (2 years after burning) between levels within treatments (shade or burn) for either carbohydrate fraction and organ, respectively, are indicated by different letters
 LignotuberRoots
StarchSum of NCStarchSum of NC
(a) Before burning
Shade
 L10039.3 ± 7.6a55.8 ± 7.9a18.7 ± 2.6a48.5 ± 9.1a
 L5023.7 ± 6.0ab39.4 ± 6.5ab20.3 ± 2.7a50.4 ± 5.7a
 L20 7.4 ± 1.3b24.7 ± 3.6b17.9 ± 3.8a50.0 ± 10.8a
(b) Two years after burning
Shade
 L10013.0 ± 2.6a26.3 ± 3.6a16.7 ± 4.8a33.2 ± 5.8a
 L5020.1 ± 5.5a33.0 ± 6.2a19.0 ± 4.2a35.3 ± 4.7a
 L2017.6 ± 3.7a32.7 ± 5.1a21.9 ± 5.4a36.6 ± 6.9a
Burning
 S019.0 ± 4.7a32.3 ± 5.2a20.8 ± 4.3a35.7 ± 5.4a
 S118.2 ± 4.5a32.2 ± 6.1a15.7 ± 4.4a30.8 ± 5.8a
 S213.4 ± 3.1a27.4 ± 3.9a21.5 ± 5.5a38.9 ± 5.9a

Two years after burning the plants, and subsequent resprouting, L100 plants had lower starch and sum of NC concentrations in the lignotuber than at the time of burning, whereas carbohydrate concentrations of L20 plants were higher than 2 years before (Table 1). There were no significant main effects or interaction effects of the shading or burning treatments for the final concentrations of the two carbohydrate fractions in either the lignotubers or roots (two-way anova, P > 0.05) (Table 1).

effects of treatments on plant mortality

Three weeks after burning, all S0 (clipped only) plants had visible signs of resprouting. Burned (S1 + S2) plants showed a more delayed response, but 80% of them were resprouting by 6 weeks after burning. Ten plants (10.7% of total) never resprouted during the 2 years following burning. Direct mortality was significantly affected by the burning treatment (χ2 = 6.45; P < 0.05) and all dead plants were in the S1 or S2 treatments (Table 2). Direct mortality was not significantly different between the levels of the shade treatment (χ2 = 0.91; P > 0.05). Ten additional plants (12% of the resprouted plants) died after having resprouted. Death of resprouted plants occurred between August and March in the first year after burning, which suggests that it was not related to a particular climatic event, such as drought in the summer or frost in the winter. Indirect mortality was not significantly affected by the burning treatment (χ2 = 4.72; P > 0.05) but it was by the shade treatment (χ2 = 7.47; P < 0.05). L20 plants were particularly prone to indirect mortality (Table 2). As a consequence of direct or indirect mortality, only 93 of the original 113 plants were alive at the end of the experiment.

Table 2.  Number of dead plants for each level of burning and shade treatments
Level of treatmentDirect mortalityIndirect mortality
Number of dead plantsnNumber of dead plantsn
Burning treatment
 S0039239
 S1637631
 S2437233
Shade treatment
 L100238236
 L50438134
 L20437733

effects of treatments on resprouting

The effect of burning on the number of resprouts per plant was highly significant both 3 months and 2 years after burning (ancova, P < 0.001, Fig. 2b). At both dates, S1 and S2 plants had fewer resprouts than S0 plants (P < 0.05) but there were no significant differences between S1 and S2 plants. The shade treatment also significantly affected resprout number (Fig. 2a, ancova, P < 0.05); 3 months after burning only L50 plants had more resprouts than L20 plants, whereas after 2 years both L100 and L50 plants had produced more resprouts than L20 plants. The mean number of resprouts was never significantly different between L100 and L50 plants and there was no interaction between the effects of shade and burning.

Figure 2.

Number of resprouts (mean ± SE) measured 3 months after burning (September) (open bars) and 2 years after burning (May) (filled bars) for each level of (a) shade and (b) burning. Different letters indicate significant differences (P < 0.05) between levels within treatments.

Burning significantly affected resprout maximum length at all sampling dates (ancova, P < 0.001, Fig. 3b). During the first year of resprouting, S0 (unburned) plants showed a significantly greater (P < 0.05) resprout maximum length than either burned treatment, which did not differ significantly among them. Two years after burning, however, we found no significant differences for the resprout maximum length between S0 and S2, but both were significantly longer than S1 plants (Fig. 3). The effect of shade on maximum resprout length was not significant 3 months after burning (ancova, P > 0.05). At both 9 months and 2 years after burning, L100 and L50 had a maximum resprout length significantly higher (Fig. 3a. P < 0.05) than L20 plants. However, there were no significant differences between L100 and L50 plants (Fig. 3). Significant interactions between burning and shade treatments on maximum resprout length were not observed at any sampling date.

Figure 3.

Maximum resprout length (mean ± SE, cm) measured 3 months after burning (September) (open bars), 9 months after burning (March) (hatched bars) and 2 years after burning (May) (filled bars), for each level of (a) shade and (b) burning. Different letters indicate significant differences (P < 0.05) between levels within treatments.

The effect of the shade treatment on the mean resprout length measured 2 years after burning was significant (ancova, P < 0.001, Fig. 4a). Mean resprout lengths of L100 and L50 plants were not significantly different, but were higher than those of L20 plants. The effect of burning on mean resprout length at this sampling date was not significant (Fig. 4b, ancova, P > 0.05).

Figure 4.

Mean resprout length (mean ± SE, cm) measured in May, 2 years after burning. Different letters indicate significant differences (P < 0.05) between levels within treatments.

The effects of the burning treatment on leaf and total resprout biomass were significant 2 years after burning (ancova, P < 0.001, Fig. 5b). S0 plants had produced a significantly greater total and leaf resprout biomass than burned plants, but S1 and S2 plants were not significantly different (P > 0.05). The shade treatment also affected resprout biomass significantly after 2 years (Fig. 5a, ancova, P < 0.001). Biomass of resprouts was significantly lowest in L20 plants (P < 0.05) although L50 and L100 were not significantly different. No significant interactions between burning and shade treatments were observed for resprout biomass.

Figure 5.

Total resprout biomass (mean ± SE, g, open bars) and leaf biomass (mean ± SE, g, filled bars) measured in May, 2 years after burning, for each level of (a) shade and (b) burning. Different letters indicate significant differences (P < 0.05) between levels within treatments.

depletion of lignotuber nc reserves caused by resprouting

The biomass of resprouts 9 months after burning was equivalent, on average, to 69% and 42% of the initial starch or sum of NC pools, respectively, of the lignotuber (Table 3). The depletion that would have been caused by the regrowth using only lignotuber-stored reserves depended on the combination of shade and burning. Stored reserves were generally lower in unburned (S0) than in burned (S1, S2) plants, due to the greater amount of resprout biomass produced by S0 plants. On average, complete depletion of lignotuber reserves (as indicated by negative ‘unused’ pools) would be caused for starch in L20 and L50 plants subjected only to clipping (S0), and for sum of NC in L20/S0 plants. However, similar estimations made for burned L20 plants suggested that considerable amounts of both starch and sum of NC would be contained in the lignotuber without being used to support regrowth. In other words, the biomass produced by L20 burned plants (either S1 or S2) could have been produced by mobilizing only 40% of the starch or 20% of the sum of NC of the lignotuber pools.

Table 3.  Mean (± SE) estimated values of carbohydrate (starch and sum of NC) pools in the lignotuber (g) at the time of burning, resprout biomass (g) 9 months after burning, and ‘unused’ carbohydrate pools (g)(initial pool minus carbohydrate cost of regrowth) in plants from the different treatments. Resprout biomass was estimated based on maximum resprout length measurements: for S0 plants [ln(biomass) = 2.84 × ln(max. length) − 5.84. (n = 37, R2 = 0.68, P < 0.001)]; for S1 plants [ln(biomass) = 2.83 × ln(max. length) − 6.48. (n = 25, R2 = 0.65, P < 0.001)]; for S2 plants [ln(biomass) = 2.96 × ln(max. length) − 7.10. (n = 31, R2 = 0.66, P < 0.001)]
TreatmentCarbohydrate poolsResprout biomass (g)Unused carbohydrate pools (%)
StarchSum of NCStarchSum of NC
Concentration (mg g−1 dry wt.)*Pool (g)Concentration (mg g−1 dry wt.)*Pool (g)Pool (g)Pool (g)
  • *

    See Table 1.

  • †A negative value indicates that the pool of carbohydrates would have been exhausted for resprout growth.

L100/S039.351.6 ± 10.455.8 73.2 ± 14.825.8 ± 5.3  1.8 ± 14.523.5 ± 17.8
L50/S023.735.0 ± 6.839.4 58.2 ± 11.324.9 ± 3.6−13.1 ± 9.310.1 ± 12.7
L20/S0 7.4 7.3 ± 2.824.7 24.3 ± 9.213.8 ± 4.1−19.3 ± 8.6−2.3 ± 12.7
L100/S139.373.2 ± 14.155.8103.9 ± 20.0 8.1 ± 2.9 57.6 ± 14.688.3 ± 20.2
L50/S123.728.6 ± 4.039.4 47.5 ± 6.6 4.5 ± 1.3 19.9 ± 2.438.9 ± 4.7
L20/S1 7.4 6.3 ± 1.724.7 20.9 ± 5.8 1.9 ± 0.5  2.6 ± 2.117.3 ± 6.0
L100/S239.351.5 ± 13.355.8 73.2 ± 18.8 6.8 ± 1.8 38.5 ± 13.260.1 ± 16.6
L50/S223.727.3 ± 7.339.4 45.5 ± 12.1 7.3 ± 1.9 13.2 ± 8.031.3 ± 12.4
L20/S2 7.4 8.8 ± 4.024.7 29.4 ± 13.3 2.8 ± 1.1  3.3 ± 2.724.0 ± 11.7
Overall mean 33.9 ± 3.6  54.7 ± 5.112.0 ± 1.4 10.6 ± 4.231.5 ± 5.5

Discussion

effects of a reduction of reserves on resprouting vigour

The most severe shade treatment (L20) reduced NC concentrations in the lignotubers of E. australis at the time of burning to 44% (sum of NC) or 19% (starch) of that of control plants. Two years after burning, L20 plants showed a lower survival rate after resprouting, and had much less vigorous resprouting than control plants. This reduced vigour was presumably due to a reduction in mobilizable carbohydrates from the below-ground parts to support resprout growth and maintenance. These negative effects on resprouting of L20 plants were not apparent in the short term after burning (their ability to initiate resprouting was not affected, and direct mortality was not affected by shading), but L20 plants suffered considerably higher indirect (post-resprouting) mortality (21.2%) than L100 or L50 plants (5.5 and 2.9%, respectively).

The period during which photosynthesis by the new leaves may be comparatively low, and reserves are the main source of carbon to support resprout growth and respiration of the whole plant is thought to be critical. This period would end when the plant has produced sufficient photosynthetic tissue to maintain a positive carbon balance. In herbaceous species subjected to defoliation, this period does not exceed a few days (Richards & Caldwell 1985; Danckwerts 1993). For Mediterranean-type woody resprouters, low photosynthetic and growth rates, and the presumably large respiratory demands of below-ground tissues could lengthen this critical period to several months. In the case of E. australis, an average-sized plant had approximately 165 g of leaves before burning. The control (L100) plants subjected to burning (S1, S2) produced, on average, 105 g of leaves during the first 24 months after burning. Therefore, they presumably could reach a value of leaf biomass and leaf area close to the pre-fire level during the third year after fire. However, during the first 2 years after fire, L20 plants produced only 20 g of leaves. Hence, the critical period for such plants might be considerably longer.

In summary, a reduction of NC reserves in the lignotuber increased mortality and decreased the vigour of resprouting. However, the depletion of NC reserves necessary to cause such effects was rather severe. Plants subjected to 50% of incident light for 1 year, in which lignotuber sum of NC and starch concentrations had been reduced to 70% and 60%, respectively, of the control plant values, did not differ statistically from the control plants in resprout number, length or biomass. It seems that the critical level of NC reserves necessary to cause a significant effect on resprouting had not yet been reached in these plants. However, the carbohydrate concentrations measured in L50 plants prior to burning were much lower than the minimum values measured under normal conditions in several populations of E. australis (Cruz & Moreno 2001a). In fact, such low values might be infrequent under normal disturbances. This indicates that the resprouting capacity of E. australis is high, and that it might rarely be limited by low NC stored in the lignotuber. This has important implications for predictions of the impact of anthropogenic changes in atmospheric CO2 on plant regeneration after fire. Under elevated CO2, resprouting is expected to increase due to greater allocation of carbon to storage (Bond & Midgley 2000; Hoffmann et al. 2000). This study suggests that resprouting in lignotuberous species like E. australis may be less affected than expected by an increase in reserves due to elevated CO2.

effects of burning on resprouting vigour

Burning produced plants with significantly fewer resprouts, and less resprout biomass after 2 years. These results support findings in other species (Kayll & Gimingham 1965; Rundel et al. 1987; Zammit 1988; Canadell et al. 1991; Moreno & Oechel 1991; Lloret & López-Soria 1993). Heating may cause damage to meristematic tissue in the lignotuber, and hence induce a lower resprout number. Burning did not significantly affect resprout length, hence the reduced biomass production 2 years after burning was due solely to the effect on resproud density.

Differences in resprouting vigour between plants burned with different severity were much less apparent, 2 years after fire, resprout number, length and biomass were not significantly different between the two fire treatments. This suggests that E. australis is not very responsive to fire severity. However, the range of fire severity used in this study was relatively narrow, and the response to low-severity conditions cannot be inferred from this experiment. Therefore, extrapolation to wildfires, in which a greater range of fire severities is expected, must be made cautiously.

effect of resprouting on carbohydrate reserves

E australis may contain excess carbohydrates in its lignotuber for supporting resprouting. Regrowth during the first 9 months after burning was far from depleting its carbohydrate reserves. In L20 burned plants, only 20% of the sum of the NC in the lignotuber would have been required to construct the new growth, assuming no contribution from photosynthesis or use of root reserves. The lesser vigour of L20 plants was probably due more to a limited capacity to mobilize carbohydrates than to lower reserves. The ability of E. australis to mobilize carbohydrates for supporting regrowth after fire may be controlled by demands from other functions. During early summer, E. australis allocates considerable amounts of carbon, derived from other organs including the lignotuber, to roots (Cruz & Moreno 2001a). Roots constitute an important sink for stored carbon in resprouting plants (Langley et al. 2002). Furthermore, refilling of storage organs seems to be favoured over growth since, 2 years after burning, there were no significant differences in lignotuber NC concentrations between plants from the various treatments. This might be advantageous under recurrent disturbances, which are often experienced by plants in Mediterranean areas (Van der Heyden & Stock 1996).

Resprouting vigour of E. australis after fire might also be constrained by developmental limitations, such as an insufficient availability of buds (Richards & Caldwell 1985; Van der Heyden & Stock 1995) or a limited ability to activate them (Bilbrough & Richards 1993). The reduction in the number of resprouts in burned plants was not compensated by a greater growth rate, although each individual resprout might have had access to a proportionally larger share of reserves (Zammit 1988). It seems that the reserves that a single resprout can access for supporting its own growth are limited. If this is the case, burning the plants would cause a reduction in the number of resprouts produced, and an increase in the amount of carbohydrate reserves that remain immobilized.

concluding remarks

We found evidence that mortality and resprout vigour of E. australis is significantly affected by a severe reduction of the lignotuber NC store. However, it seems that this species stores much more carbon in this organ and in the roots than is needed to support resprouting after defoliation. Despite the combination of low carbon reserves and severe burning, most plants resprouted and grew, indicating that the potential of this species for enduring extreme conditions is considerable. Having large carbon reserves in the lignotuber may favour the persistence of this species in Mediterranean environments, where there are multiple stresses and disturbances. It appears, however, that, under normal conditions, limitations by carbohydrate reserves in the lignotuber might not be very critical.

Acknowledgements

Funding was provided by the EC (ENV-CT91-320). We thank the staff of Quintos de Mora Station. Thanks to F. Fernández del Campo and C. Fenoll, for their advice, and to A. Velasco, J. R. Quintana, B. Virumbrales, E. Zuazua, N. Acevedo and J. Torres for their help.

Ancillary