Plant stored reserves do not drive resprouting of the lignotuberous shrub Erica australis

Authors

  • Alberto Cruz,

    1. Facultad de Ciencias del Medio Ambiente, Universidad de Castilla-La Mancha, 45071 Toledo, Spain
    Search for more papers by this author
  • Beatriz Pérez,

    1. Facultad de Ciencias del Medio Ambiente, Universidad de Castilla-La Mancha, 45071 Toledo, Spain
    Search for more papers by this author
  • José M. Moreno

    Corresponding author
    1. Facultad de Ciencias del Medio Ambiente, Universidad de Castilla-La Mancha, 45071 Toledo, Spain
      Author for correspondence: José M. Moreno Tel: +34 925268800 Fax: +34 925268840 Email: JoseM.Moreno@uclm.es
    Search for more papers by this author

Author for correspondence: José M. Moreno Tel: +34 925268800 Fax: +34 925268840 Email: JoseM.Moreno@uclm.es

Summary

  • Lignotuberous plants store carbohydrates and mineral nutrients within the lignotuber. Resprouting vigour may depend on stored reserves, as well as on the availability of soil mineral nutrients and water.
  • Here the role played by plant reserves and soil resources on the resprouting response of Erica australis was analysed after clipping plants in 13 different stands, varying in soil resource availability and in plant reserves.
  • There were significant among-site differences for resprout biomass and maximum length, but not for resprout number, 1 yr after clipping. Plant reserves at the time of clipping were not significantly correlated with resprout number, length or biomass. However, resprouting variables were significantly correlated with soil nitrogen or extractable cations, or plant water potentials. Resprout biomass and maximum length were negatively correlated with lignotuber size.
  • These findings indicate that the assumption that resprouting vigor in lignotuberous plants is primarily dependent on the amount of reserves stored in the lignotuber must be revised, as well as the overall role of lignotubers in resprouting.

Introduction

The ability of a plant to resprout after its above-ground parts are killed is a main feature of many plants from disturbance-prone, terrestrial ecosystems, like the Mediterranean ones in which fire plays a dominant role (Bond & Midgley, 2001). Many plants resprout from special organs, such as the lignotuber, partially buried into the soil, that act as a reservoir of dormant buds, as well as of carbohydrates and mineral nutrient reserves (Mullette & Bamber, 1978; James, 1984; Cruz & Moreno, 2001a). The carbohydrates stored in the lignotuber, or other below-ground parts, are thought to be mobilised during resprouting, thus acting as the main supply of carbon for regrowth at the early stages after a disturbance (Jones & Laude, 1960; DeSouza et al., 1986; Miyanishi & Kellman, 1986; Bowen & Pate, 1993; Van der Heyden & Stock, 1996; Canadell & López-Soria, 1998). The role of lignotuber mineral nutrients (N, P) over resprouting has been little explored, but evidence suggests that they may be mobilised to some extent and can potentially limit regrowth (Miyanishi & Kellman, 1986; Canadell & López-Soria, 1998). It has often been considered that the vigour of the resprouting response after fire would be primarily determined by the damage caused by heat to the dormant bud bank in interaction with the pool of reserves previously stored by the plant (Jones & Laude, 1960; Bowen & Pate, 1993). For instance, the level of carbohydrate reserves stored by the plant fluctuates largely during the year and between years (Cruz & Moreno, 2001a), which would explain the variations observed in the resprouting vigour depending on the time of disturbance (Rundel et al., 1987; Malanson & Trabaud, 1988). However, a lack of correlation between the amount of reserves and subsequent regrowth has often been reported (Richards & Caldwell, 1985; Hogg & Lieffers, 1991a,b; Erdmann et al., 1993; Sparks & Oechel, 1993). It has been suggested that not all reserves in the storage organs may be mobilizable (Chapin et al., 1990). In addition, underground demands from roots and mycorrhizas may limit mobilization towards above ground parts, remaining as strong sinks (Langley et al., 2002). It has also been argued that reserves may be stored in excess (Van der Heyden & Stock, 1995; Hoffmann et al., 2000; Cruz & Moreno, 2001a), hence other factors may be limiting. For instance, resprouting vigour of co-occurring populations of the shrub Erica australis was related to variations in the availability of soil resources (mineral nutrients and water) (Cruz et al., 2003), presumably due to differences in resource supply for the regrowth of resprouting tissues. It is not known whether variations in the level of reserves in the below-ground parts of the plant between co-occurring populations of the same plant species could in fact cause a differential resprouting response in the case of fire or other defoliating disturbance.

The process of carbohydrate storage depends upon the balance between photosynthesis and growth and respiration, which, in turn, may depend on soil moisture and nutrient supply (Chapin et al., 1990; Steinlein et al., 1993). However, little is known about the factors that control reserve content of the plant. In arid environments, the process of carbohydrate storage and use in the roots of Prosopis glandulosa was partially controlled by water content (Wan & Sosebee, 1990). Variations in soil-resource availability might affect the amount of reserves stored by the plant in its below-ground organs prior to the disturbance. Furthermore, soil-resource availability might also determine the relative size of the storage organ (Cruz & Moreno, 2001b), which in turn, may affect the capacity of the plant to resprout. Therefore, the relationship between stored reserves of carbohydrates or mineral nutrients and resprouting vigour might be indirectly related to one another, to the extent that the first might be driven by soil-resource availability and, eventually, mediated by the relative size of the resprouting organ. Unravelling these interactions is important in order to have a clear understanding of the mechanisms that control resprouting from lignotuberous species in many areas of the world; this is notably so of the Mediterranean-type species.

This paper analyses whether the content of carbohydrates and nitrogen of the lignotuber and roots as well as the relative size of the lignotuber in the shrub E. australis drove resprouting by this species. Resprouting by E. australis and other Erica species is a critical component of the regeneration process after fire in many shrublands of the Mediterranean Region (Mesléard & Lepart, 1989; Calvo et al., 1998).

Materials and Methods

Species and study sites

E. australis L. has a well-developed lignotuber ( Moreno et al., 1999 ) from which it resprouts vigorously after fire. It is spread mainly over the western half of the Iberian Peninsula ( Rivas-Martínez, 1979 ), being one of the dominant elemants where it grows. We selected 13 sites in the province of Cáceres (central-western Spain) in which E. australis was present. The shrublands of these sites were dominated by E. australis , together with Cistus ladanifer L., Rosmarinus officinalis L. or Erica umbellata L., among other. The selected sites covered a wide range of substrate types in which E. australis was present: fluvial sands (A1), granites (G1, G2, G3), quarzits (C1, C2, C3), slates (P1, P2, P3) and ‘rañas’ (Pliocene deposits) (R1, R2, R3). All the sites were selected to be as homogeneous as possible with respect to elevation, slope and aspect. Maximum distance between sites was c. 70 km from N to S, and 25 km from W to E. The climate of the study area is Mediterranean-type, with a mean annual temperature of 15–16°C and a mean annual rainfall of c. 800–1100 mm, mostly concentrated in autumn and spring. The year following plant clipping was drier than average, with c. 69% of average rainfall. The spring following clipping was particularly dry, with less than 60 mm rainfall compared to an average of 225 mm in the period 1941–1990.

Plant manipulations

At each site we randomly selected six individuals of E. australis (78 plants in total). These plants covered a wide range of sizes, as assessed by an index of lignotuber area (La) (La = π * D1 * D2/4), obtained from the two largest diameters of the lignotuber perpendicular one to the other (D1 and D2). The lignotuber area index is highly and significantly correlated with the size of the whole plant (Moreno et al., 1999). The plants included in this study had a mean (± SE) lignotuber area of 255.8 ± 25.8 cm2 (from 44 to 1432 cm2). In mid-summer (19 and 30 August), the aerial parts of all plants were clipped and removed. This procedure tried to simulate the effect of removal of above-ground biomass caused by a fire with null severity and no ash fertilization. The plants were then fenced with chicken wire to avoid herbivory by rabbits or ungulates. To diminish differential competitive interactions with neighbours, the biomass of all plants in a 0.5-m radius circle around each target E. australis plant was cut and removed.

Resprouting vigour

Resprouting vigour was monitored three times: in January (4 months after clipping), April (7 months) and September (12 months) of the following year. At each sampling date we counted the number of resprouts in all plants (RN, number) and measured the length of the longest resprout (resprout maximum length, RML, cm). From these measurements we estimated resprout biomass (RB, cm), by applying a regression equation obtained in a different set of plants (ln RB =−3.25 + 0.12 * sqrt RN + 1.81 * ln RML; R2 = 0.72). A single RML measurement had probed to be more tightly related to resprout biomass than resprout mean length obtained from multiple measurements. Twelve months after clipping (September of the next year), plants from two of the sites (G2 and C1) had been browsed, and were discarded for further analysis. Consequently, the number of sites from which data was available was 13 (4 and 7 months after clipping) and 11 (12 months after clipping).

Plant carbohydrate and mineral nutrient content

Samples from lignotubers and roots were obtained from three plants of E. australis at each site, for determinations of nonstructural carbohydrates and nitrogen concentrations. Previous trials indicate that phosphorus concentrations in the lignotuber were very small and this element was not included in the study. Sampling was done concurrently with clipping the target E. australis plants. Because of the semidestructive nature of sampling (see below, this section), plants sampled for TNC and N determinations were not later used in monitoring of resprouting. Lignotuber samples were obtained with a Pressler-type borer. Root samples were obtained from roots excavated close to the lignotuber. All samples were immediately immersed in dry ice, taken to the laboratory and pulverized after being oven-dried. Determinations were made for glucose (plus fructose), sucrose and starch concentrations, following the enzimatic method proposed by Azcón-Bieto & Osmond (1983) and described in Cruz & Moreno (2001a). The sum of glucose, sucrose and starch was termed total nonstructural carbohydrates (TNC). Lignotuber and root samples were also analysed for nitrogen (N) content in a CHN auto-analyser (Perkin-Elmer 2400 CHN, Shelton, CT, USA). From the three plants sampled at each site we calculated the mean population value of TNC, starch and N concentrations, respectively, which were expressed as mg g−1 d. wt.

No statistically significant correlation was detected between lignotuber area and TNC or N concentration (for TNC, r = 0.25, P > 0.05; for N, r = 0.15, P > 0.05; n = 39). Thus, concentrations were not related to plant (lignotuber) size.

Soil resource availability

Soil fertility was evaluated at each site after collecting five soil samples (0–10 cm depth) and mixing them into a single sample (Cruz & Moreno, 2001b). The following analyses were made: pH (1 : 2.5 soil : water), extractable cation concentration (sum of Na, K, Ca and Mg, after extraction by shaking 5 g of soil in 100 ml of ammonium acetate at pH of 7.0, and determination of Ca and Mg by atomic absorption and of Na and K by flame photometry), total nitrogen, in a CHN autoanalyser (Perkin-Elmer 2400 CHN) after total combustion, and available phosphorus, after extraction with sulfuric and hydrochloric acid (Olsen & Summers, 1982).

Soil water availability was evaluated in a comparative way. Shoot predawn water potential (Ψpd) were measured in terminal branches of each four additional, nondisturbed, individuals of E. australis per site, with a Scholander-type pressure chamber. Measurements were made during summer (August), at the same time when plants were clipped. The period without rains lasted for at least 30 d before water potential measurements. We used the mean values of predawn water potentials at each site as comparative indices of water availability.

Lignotuber relative size

The biomass, or other measure of the size of the lignotuber, can be expressed as an absolute value. However, the amount of biomass allocated to the lignotuber may also be expressed with respect to that allocated to other plant structures (leaves, roots, etc.). For the same lignotuber dimension, the capacity to resprout by a plant may depend on the amount of leaves held prior to the disturbance. This ‘relative’ size of the lignotuber with respect to the biomass of leaves may vary between populations (Cruz & Moreno, 2001b). It was calculated by obtaining the mean value of the individuals of each population from the xy (leaf biomass, lignotuber biomass) linear regressions. For this, 10 individuals were sampled at each site, from which we obtained the mean value of the lignotuber biomass (Lb), adjusted for the mean value of foliar biomass (Fb) (Cruz & Moreno, 2001b), which will be denoted as Fb-Lb. The greater the value of Fb-Lb, the greater the biomass of the lignotuber with respect to the same foliar biomass.

Statistical analysis

The significance of differences among-sites for RN, RML and RB were tested separately at each sampling date by ANCOVA tests. In all cases, La was used as the covariable. In order to correct for normality, before analysis RN was square-root transformed, La was ln transformed, and RML and RB were ln(x + 1) transformed. The significance of differences among-sites for TNC, starch, and N concentrations in lignotubers or roots were tested by one-way ANOVA. The relationships between mean site values of plant reserve concentrations (N, starch and TNC) and soil resource indices (soil pH, total N, available P and extractable cation concentrations, and plant Ψpd) were determined by Pearson correlation.

The relationships between resprouting and the potentially explanatory variables (plant reserve content and soil resources) were ascertained using site as the reference unit. Plants from each site may differ in size, and resprouting response is known to depend on plant size. To avoid a confounding interpretation of site effect caused by variations in plant size, we used a two-step procedure in order to isolate the effect of a differential plant size. First, we regressed RN, RML and RB on La, separately for each sampling date. This determined the amount of variance of the resprouting variables accounted by lignotuber size, La being a surrogate of this. Second, we calculated for each site the mean value of the residuals from the previous regressions of the six individuals per site. The magnitude of the residual indicates whether a plant had a more or less vigorous resprouting once it has been corrected by its size. We then calculated the correlation coefficients between the mean values of the residuals of the resprouting variables for each site and the variables characterizing soil resource availability (soil extractable cations, total nitrogen, available phosphorus and plant Ψpd) and plant stored carbohydrates and mineral nutrients (TNC, starch and N). Mean values of plant stored reserves must also be corrected for plant size, in order to be correlated with mean residuals of resprouting variables. However, no statistically significant correlations were detected between carbohydrate or N reserves and plant size (see above, Plant carbohydrate and mineral nutrient content). Consequently, plant concentrations were used as the size-independent magnitude of plant reserves.

In order to determine the relationship between the resprouting response and the predisturbance relative lignotuber biomass, we calculated Pearson correlation coefficients between the resprouting variables (mean site values of the residuals of the La-RN, La-RML and La-RB regressions) measured at the different sampling dates and the measure of the relative size of the lignotuber at each site (Fb-Lb adjusted means).

Results

Plant stored nutrient concentrations and soil fertility

Mean TNC concentrations of E. australis in August 1994 were almost double in the roots than in the lignotuber (mean of 122 vs. 67 mg g−1 d. wt., Fig. 1a–b). Plant TNC concentration varied approximately two-fold among the sites, and these differences were statistically significant both in lignotubers (ANOVA, F12,26 = 3.44, P < 0.01) and roots (ANOVA, F12,26 = 2.41, P < 0.05). Starch concentrations comprised, on average, 44% of TNC in the lignotuber, and 52% in the roots. Starch concentration was more variable among the sites in the lignotubers (almost six-fold) than in the roots (Fig. 1a–b). Indeed, starch concentrations were statistically significantly different among the sites only in the lignotubers (ANOVA, F12,26 = 7.43, P < 0.001). Plant N concentrations in the roots and lignotubers were low, rarely exceeding 3 mg g−1 d. wt. (Fig. 2). Mean N concentrations were statistically significantly different among sites only in the roots (ANOVA, F12,26 = 6.37, P < 0.001). The sites also varied with respect to some characteristics of soil resource availability, such as total nitrogen, available phosphorus, extractable cation concentration (Cruz et al., 2002) or plant water potentials during summer (it ranged from −0.56 to −2.59 MPa). Carbohydrate concentrations of the lignotuber were not statistically significantly correlated with any of the soil variables, whereas starch concentrations in the roots were statistically significantly and negatively correlated with soil pH and available phosphorus (Table 1). Nitrogen concentrations into the lignotuber showed a positive and statistically significant correlation with soil total nitrogen (Table 1).

Figure 1.

Carbohydrate concentrations (TNC: open columns, starch: filled columns; mean ± SE, mg g −1 d. wt.) of (a) lignotubers, and (b) roots of E. australis at the different sites at the time of clipping (mid-summer) ( n = 3).

Figure 2.

Nitrogen concentrations (mean ± SE, mg g −1 d. wt.) of (a) lignotubers, and (b) roots of E. australis at the different sites at the time of clipping (mid-summer) ( n = 3).

Table 1.  Correlation coefficients between the variables indicative of soil resource availability and nitrogen and carbohydrate concentrations in lignotubers and roots of E. australis
 Soil fertilityWater availability
 Total NAvailable PCationspHΨpd Aug
  1. (n.s) P > 0.05; (*) 0.05 = P > 0.01; (**) 0.01 = P > 0.001; (***) P= 0.001.

Lignotuber
Starch−0.39−0.16−0.15-0.31−0.26
TNC−0.30−0.25−0.16-0.35−0.34
N 0.72 (**)−0.190.29−0.21−0.11
Roots
Starch0.07−0.69 (**)−0.11−0.72 (**)−0.26
TNC0.34−0.54−0.25−0.34−0.29
N0.150.07−0.010.02−0.06

Number and growth of resprouts

RN reached a maximum for most sites at 7 months after clipping (mean of 153 resprouts per plant), and then declined to a mean of 115 resprouts per plant 1 yr after clipping (Fig. 3). Site effect for RN was statistically significant at 4 and 12 months after clipping (ANCOVA, P < 0.05, Table 2). RML showed a continuous increase during the study period, reaching a mean maximum of 54 cm 1 yr after clipping (Fig. 4). RB, estimated from measures of RN and RML, reached a mean of c. 213 g per plant 1 yr after clipping. (Fig. 5). Growth rates of the resprouts showed a consistent pattern of among-site variation, as indicated by the significant differences found at all sampling dates for RML and RB (ANCOVA, P < 0.001, Table 2).

Figure 3.

Resprout number (RN, mean ± SE, no.) of E. australis at the different sites at (a) 4 months (b) 7 months, and (c) 12 months after clipping (mid-summer) ( n = 6).

Table 2.  Results of the ANCOVA tests for significance of the among-site differences in the variables indicative of resprouting vigor of Erica australis ( RN , RML and RB ), measured at different time ( n = 13). La was used as the covariable
ANCOVACovariableSite effect
d.f.FPd.f.FP
  1. (n.s) P > 0.05; (*) 0.05 ≥ P > 0.01; (**) 0.01 ≥ P > 0.001; (***) P≤ 0.001.

RN
4 months134.48***122.46*
7 months157.80***120.89n.s.
12 months137.14***102.39*
RML
4 months116.16***128.14***
7 months117.01***125.29***
12 months11.31n.s.105.66***
RB
4 months131.30***127.94***
7 months150.94***124.46***
12 months118.38***103.98***
Figure 4.

Resprout maximum length (RML, mean ± SE, cm) of E. australis at the different sites at (a) 4 months (b) 7 months, and (c) 12 months after clipping (mid-summer) ( n = 6).

Figure 5.

Resprout biomass (RB, mean ± SE, g) of E. australis at the different sites at (a) 4 months (b) 7 months, and (c) 12 months after clipping (mid-summer) ( n = 6). RB was estimated from measurements of RN and RML.

Relationships between number and growth of resprouts and plant resources

Correlation coefficients corresponding to the La-RN, La-RML and La-RB regressions are shown in Table 3. The amount of variance of the different resprouting variables accounted by lignotuber area varied with time, but they were generally moderate for RN and RB and much lower for RML. The residuals from these regressions showed correlation coefficients with starch, TNC or N concentrations of lignotubers and roots which were, in all cases, low and statistically nonsignificant (Table 4). On the contrary, residuals of the resprouting-La regressions showed some significant correlations with some characteristics of soil resource availability, such as extractable cations, total nitrogen or plant water potentials during summer (Table 4). The coefficients of such relationships were, in most cases, positive, suggesting a stimulatory effect of a larger availability of soil resources on resprouting. Correlation coefficients between La-RN residuals and Fb-Lb adjusted means were always low and statistically nonsignificant. However, La-RML and La-RB residuals became statistically significant and negatively correlated with the lignotuber relative size when measured 1 yr after clipping (Table 4). This suggests that resprout production was not affected by the previous relative size of the lignotuber, but that the growth of the resprouts was higher in those sites in which the plants carried a larger number of leaves per unit of lignotuber. When resprouting variables were regressed by steps against all possible explanatory variables, only some variables related to soil fertility and water availability were selected, along with the relative size of the lignotuber (Table 5). These results suggest that the among-site variation in resprouting vigour was mostly accounted for by differences in soil resource availability and plant structure, rather than by differences in reserve concentrations stored by the plants among the sites.

Table 3.  Correlation coefficients between the variables indicative of resprouting vigor of E. australis ( RN, RML and RB ), measured at different time, and the lignotuber area ( La, ln cm 2 ). Regressions were calculated at the individual level
 Correlation (r) with LanSignificance
  1. (n.s) P > 0.05; (*) 0.05 ≥ P > 0.01; (**) 0.01 ≥ P > 0.001; (***) P ≤ 0.001.

RN (No) [sqrt]
5 months0.5978(***)
8 months0.7078(***)
13 months0.5866(***)
RML (cm) [ln + 1]
5 months0.3678(**)
8 months0.5078(***)
13 months0.3066(*)
RB (g) [ln + 1]
5 months0.5278(***)
8 months0.6778(***)
13 months0.5666(***)
Table 4.  Correlation coefficients between the mean site residuals of the La -RN, La -RML and La -RB relationships, measured at different time, and mean site values of soil resources, plant carbohydrates and nutrient concentrations and lignotuber relative size ( Fb-Lb ) at each site in several populations of E. australis (4 and 7 months after clipping: n = 13; 12 months after clipping: n = 11)
 Soil resource availabilityLignotuber stored reservesRoot stored reservesLignotuber relative size
 NPCatpHΨpd AugNStarchTNCNStarchTNCFb-Lb
  1. (*) 0.05 ≥P > 0.01; (**) 0.01 ≥ P > 0.001.

Residuals of La-RN
4 months 0.59 (*)−0.11 0.64 (*)−0.150.200.33−0.25−0.160.100.280.17−0.05
7 months−0.11−0.10 0.56 (*)0.10−0.30−0.180.110.160.050.270.020.47
12 months−0.05−0.440.32−0.01−0.71 (*)−0.150.150.19−0.300.230.060.49
Residuals of La-RML
4 months0.430.020.42−0.220.510.17−0.28−0.22−0.020.350.19−0.42
7 months 0.59 (*)0.050.22−0.100.380.19−0.26−0.150.140.260.30−0.26
12 months0.280.21−0.30−0.25 0.69 (*)0.140.050.070.310.190.27−0.74 (**)
Residuals of La-RB
4 months0.49−0.060.47−0.320.440.29−0.21−0.120.110.360.19−0.41
7 months0.540.040.33−0.090.370.15−0.27−0.160.120.270.28−0.21
12 months0.300.07−0.21−0.240.500.100.120.150.230.300.25−0.64 (*)
Table 5.  Stepwise regression models of the mean site residuals of the La -resprout variables relationships of E. australis measured at different times on some variables related to soil resource availability, plant nutrient concentrations and lignotuber relative size of sites
 Independent variablesRegression modelR2P
Y = Residuals of La-RN
4 monthsX1 : Soil cations Y = −2.79 + 0.03 ( X1 ) 0.400.020
7 monthsX1 : Soil cations Y = −1.16 + 0.01 ( X1 ) 0.310.047
12 monthsX1 : Ψ pd summer Y = −2.17 – 1.44 ( X1 ) 0.500.015
Y = Residuals of La-RML
4 months
7 monthsX1 : Soil nitrogen Y = −0.70 + 3.07 ( X1 ) 0.350.034
12 monthsX1 : Fb-LbY = 1.50 – 0.20 ( X1 ) + 0.18 ( X2 ) 0.730.006
 X2 : Ψ pd summer    
Y = Residuals of La-RB
4 months
7 months
12 monthsX1 : Fb-LbY = 2.26 – 0.37 ( X1 ) 0.410.035

Discussion

This study documents a significant among-site variability in the resprouting vigour by E. australis, confirming the results obtained in previous studies (Cruz & Moreno, 2001c; Cruz et al., 2002). Indeed, during the first year of resprouting, the different populations of E. australis studied here showed a 2-fold variation in either resprout number or maximum length. As plants were subjected to identical disturbance severity in all cases, that is, complete removal of above-ground parts without any other damage to the lignotuber, such variability in resprouting response might be due to a great extent to differences in resource supply among the sites. These resources may be supplied from the reserves stored by the plant within its below-ground organs, or may be obtained from the soil when regrowth occurs. In a previous study (Cruz et al., 2002), we found that a significant amount of variance in resprouting of E. australis among the sites was accounted for by some variables related to water and nutrient availability in the soil, suggesting that the amount of resources supplied from the soil may be a determining factor of resprouting. As the amount of resources stored within the plant may also be a function of soil resource availability, it was necessary to determine the amount of variance of resprouting specifically explained by each group of resources.

How much did intersite variations in carbohydrate storage contribute to cause differences in resprouting?

Fluctuation in the carbohydrate content of the resprouting organs through the year may be responsible for temporal variations in the subsequent vigour of the resprouting response (Rundel et al., 1987; Malanson & Trabaud, 1988; Castell et al., 1994). However, it has not been established whether differences in carbohydrate content between neighbouring plants may cause differences in their resprouting response when disturbed at the same time. Sparks & Oechel (1993) observed that the capacity of the shrub Adenostoma fasciculatum to produce new resprouts after an experimental fire had no correlation with the carbohydrate content in the lignotuber. However, these authors measured the carbohydrate concentrations in samples collected 1 yr before burning, so that the status of stored reserves in the plant at the time of burning was unknown. In the present study we found a 2-fold variation among the sites in the mean values of lignotuber and root carbohydrate concentrations of E. australis (almost 6-fold in the case of lignotuber starch), after measuring the concentrations just at the time of clipping the plants. Simple calculations based on the range of carbohydrate concentrations measured here lead us to the conclusion that an average medium-sized plant, with a lignotuber biomass of 1500 g, of which c. 66% can be devoted to storage, could have stored in this organ from 44 to 100 g of TNC, depending on the site. Considering that stored carbohydrates are thought to constitute the main carbon source for regrowth during the early stages of resprouting, such a variation should have caused a substantially different resprouting vigour among the sites studied here. However, once corrected for the differences in lignotuber size, there were no statistically significant correlations between any of the variables indicating the carbohydrate concentrations of the lignotubers or roots at the different sites and those related to resprouting vigour. In other words, we found no evidence suggesting that the significant variation found among the sites in the mean values of starch and total carbohydrate concentrations in the below-ground organs would be the main responsible factor of the variation observed later in the variables describing the vigour of the resprouting response. Thus, the amount of resources contained within the plant may be less relevant than suspected as a driving factor of resprouting. The populations studied here included a relatively wide range of carbohydrate concentrations. This supports the hypothesis that this species may store reserves, particularly of carbohydrates, in greater abundance than needed for supporting a single resprouting event (Cruz & Moreno, 2001a).

What is the role played by the nitrogen reserves contained in the lignotuber on resprouting?

The soils of the study sites are acidic and moderately poor in mineral nutrient (N, P) concentrations. We would expect that accumulation of nitrogen, or other mineral nutrients, in the lignotubers or roots of E. australis would have been relevant for ensuring mineral nutrient supply for regrowth in such a nutrient-deficient environment. Despite some authors finding significant depletion in the pool of mineral nutrients stored by lignotubers and roots during resprouting episodes (Miyanishi & Kellman, 1986; Canadell & López-Soria, 1998), it has not been clearly demonstrated that the mineral nutrients stored by the plants would contribute significantly to sustain resprout growth. Resprouting plants that inhabit nitrogen-deficient soils, such as in the Mediterranean-type ecosystems of Australia, do not accumulate nitrogen in their underground organs in greater amounts than the nonresprouting congeners (Pate et al., 1990). This suggests that mineral nutrient storage is not very relevant for ensuring resprouting. In the case of the nitrogen concentrations that we measured in this study for E. australis, they can be considered as relatively low in comparison with those described by Shaver (1983) for other woody species from Mediterranean-type areas of California and Chile, and those reported by Canadell & López-Soria (1998) for the resprouter Erica arborea. In addition, the concentrations of this nutrient measured at the time of clipping in the lignotubers and roots were not statistically significantly correlated with the resprouting vigour in the following year. We therefore found no evidence that lignotubers of E. australis may constitute a significant source of nitrogen for supporting resprout growth, and a differential nitrogen storage does not seem to be a contributing factor in explaining the variation in the resprouting response by this species.

What is the role played by soil resources on resprouting?

Resprouting variables showed some high and statistically significant correlations with several characteristics of soil resource availability, such as total nitrogen content, extractable cation content or plant water potential during summer, used here as a relative index of water availability at the sites. Total nitrogen and extractable cation content showed positive correlations with resprout length or number, respectively, suggesting a positive effect of soil fertility on resprouting. The positive correlation of plant water potential measured during summer on resprout length can be interpreted as a positive effect of soil moisture on resprouting. Such effect was reported in a previous study to be relevant for a short period of time (Cruz et al., 2003), suggesting that early resprouting during drought periods would be more vigorous in sites with less severe water stress. Paradoxically, in the present study the effect of water potential contributed to explain a significant portion of the variation in resprout length not immediately, but 1 yr after clipping. This result might be explained by the severe drought of the spring following the treatment implementation, which may have caused the relative differences in soil moisture among the sites during the period of summer drought to be maintained during the following year. Plant water potential had a significant but negative effect over resprout number at the same sampling date. Resprout number tended to decrease during spring, apparently due to ‘self-thinning’. It seems that higher water availability, by promoting a more vigorous resprout growth, might have caused a more intense reduction in the subsequent number of resprouts.

Resprouting and the relative size of the lignotuber

Resprouting vigour of E. australis seems to be also largely controlled by the relative biomass allocation made by the plant to the lignotuber, as growth of the resprouts during the first year after clipping was more vigorous at the sites in which plants carried smaller lignotubers for the same leaf biomass. The relative size of the lignotuber is partially controlled by soil resources, resulting in the plant developing a proportionally larger lignotuber at sites apparently unfavourable for plant growth (Cruz & Moreno, 2001b). This finding suggests that soil resource availability is a powerful driving force of resprouting, either directly, presumably by determining the water and mineral nutrient supply to regrowth, or indirectly, affecting the relative size of the resprouting organ. However, we did not find evidence that the concentration of carbohydrate reserves stored by the plant were related to soil resource availability.

Conclusion

This study supports evidence suggesting that early resprouting of co-occurring populations of the Mediterranean-type shrub Erica australis is largely driven by the relative size of the resprouting organ (negatively) and the availability of some soil resources (mostly positively), particularly of water. Given the relationships between soil resources and relative size of the lignotuber (Cruz & Moreno, 2001b), it can be concluded that moist and moderately fertile soils will favour a more vigorous resprouting, either directly, by causing a higher water and mineral nutrient supply to regrowth, or indirectly, by causing a proportionally larger biomass allocation to leaves per unit of lignotuber. Soil resources showed a less evident relationship with the storage process, particularly that of carbohydrates. In any case, the concentration of reserves was unrelated to resprouting vigour. In conclusion this study found that resprouting is more vigorous the smaller the storage organ. Further, how much this is filled, that is, the concentration of reserves stored within the lignotuber, is not a relevant driver of resprouting. The role usually attributed to the lignotuber as an organ developed to enhance resprouting is not supported by this study.

Acknowledgements

Funding was provided by the EC (Project ENV-CT91-320). We thank Dr F. Fernández del Campo and Dr C. Fenoll, and Angeles Muñoz (Departamento de Fisiología Vegetal, UAM) for their advice and for carbohydrate determinations, respectively. We also thank Angel Velasco and Nuria Acevedo for help in the field.

Ancillary

Advertisement