1Resprouting in woody plants is a trait that facilitates persistence in disturbance-prone environments. Patterns of allocation of resources for resprouting may depend on the severity of disturbance to which the plants are adapted. Fire-adapted plants allocate more resources to below-ground structures for resprouting after destruction of above-ground structures. However, plants that resprout in response to disturbances where above-ground structures survive may remobilize above-ground resources for resprouting.
2In coastal sand dunes, stem leaning and partial uprooting of trees result in high frequency of resprouting (38·9% of individuals; 90·6% of species). We tested whether ‘good’ and ‘poor’ resprouters differed in allocation to root biomass and root carbohydrate reserves. Species were assigned to categories of resprouting ability based on the frequency of multi-stemmed individuals in the local population. To control for phylogenetic effects, we contrasted poor and good resprouter species pairs from three families.
3We tested whether plants stored more reserves in roots or stems and whether above-ground resources were remobilized for resprouting. The latter was measured from the number and dry mass of sprouts produced by trees cut to stump heights of 10 and 150 cm above-ground level.
4Good resprouters had larger seedling root:shoot ratios and higher stem and root carbohydrate concentrations than poor resprouters. Both good and poor resprouters maintained higher carbohydrate concentrations in stems than in roots.
5For both good and poor resprouters, 150-cm stumps produced more sprouts than 10-cm stumps. At each stump height, good resprouters produced more sprouts than poor resprouters.
6Resource allocation in coastal dune trees appears to be a bet-hedging strategy. After low-severity disturbances, resprouting occurs by remobilization of above-ground resources. Below-ground resources may be more costly to remobilize but may allow recovery from occasional more severe disturbances.
The pattern of allocation of resources to resprouting in fire-adapted plants is expected to differ from the pattern in habitats where fire is not an important disturbance (Bond & Midgley 1995; Bellingham & Sparrow 2000). For instance, Euptelea polyandra, a tree species that resprouts readily on steep hillslopes in Japanese old-growth temperate forests, lacks the allocation and reserve formation strategies exhibited by fire-prone resprouters (Sakai, Sakai & Akiyama 1997). Sakai et al. (1997) hypothesized that E. polyandra resprouted using resources remobilized from above-ground parts after leaning or partial uprooting. Hence E. polyandra does not need larger root and stem reserves of carbohydrates than the poor resprouters with which it co-exists (Sakai et al. 1997; Sakai & Sakai 1998).
Most studies on storage of reserves for resprouting have analysed root reserves only, rather than both root and shoot reserves (Knox & Clarke 2005; Schwilk & Ackerly 2005). These studies have largely been conducted in fire-prone systems where carbohydrate reserves are typically stored in the roots. The few studies that have analysed reserves in both roots and shoots suggest that the reserves in different plant parts could play different roles in determining the resprouting ability of a species, depending on the disturbance severity to which resprouting is adapted (Bell, Pate & Dixon 1996; Sakai et al. 1997). Where disturbance is for the most part at a low level, occasional severe selection may maintain both above- and below-ground reserves to allow maximum flexibility of sprouting response to disturbance.
We explored biomass and carbohydrate reserve allocation patterns for poor and good resprouters in a coastal dune forest, where trees mainly resprout in response to chronic low-severity disturbances that cause leaning and partial uprooting (Nzunda et al. 2007a,b). We examined whether good and poor resprouters differed in their seedling root:shoot ratios, and stem and root carbohydrate concentrations. We also conducted an experiment by cutting trees creating stumps in two height classes to determine the effect of the remobilization of above-ground resources on number and biomass of sprouts produced.
We predicted that
1If resprouting can be achieved through resource remobilization from above-ground structures, then good resprouters will not differ from poor resprouters in their root biomass allocation or carbohydrate reserves.
2If trees subjected to low-severity disturbances resprout primarily by remobilizing above-ground resources, taller stumps will produce more sprouts than shorter ones (Sakai & Sakai 1998).
Materials and methods
The study was conducted in subtropical Indian Ocean coastal dune forest at Cape Vidal, South Africa (28°05′32″S, 32°33′40″E). Cape Vidal is part of the Greater St. Lucia Wetland Park. The mean annual rainfall is c. 900 mm, spread evenly throughout the year (range 80–90 mm per month) (Schulze 1997). Mean annual minimum and maximum temperatures are 17·8 °C in July and 25·6 °C in January, respectively, with a mean annual temperature of 21·5 °C. The topography consists of steep (slopes up to 55°) longitudinal sand dunes that are orientated parallel to the coastline. The loose sand substrate, steep slopes, and seasonal winds create a low-level but chronic disturbance regime that makes trees susceptible to leaning and partial uprooting (Nzunda et al. 2007a,b).
The dune forest at Cape Vidal forms part of a narrow strip of forest (0·1–4 km wide) extending in a contiguous belt for 240 km along the KwaZulu–Natal coast (Tinley 1985). Important tree species in decreasing order include Diospyros natalensis (Ebenaceae), Mimusops caffra (Sapotaceae), Drypetes natalensis (Putranjivaceae), Celtis africana (Celtidaceae) and Ochna natalitia (Ochnaceae) (Nzunda et al. 2007a). The incidence of resprouting, which results in multi-stemmed individuals, is high in this forest, with 38·9% of trees being multi-stemmed (Nzunda et al. 2007a). Of 53 species recorded, 48 exhibited multi-stemming (90·6%). Because of the high incidence of resprouting in this forest and variation in resprouting ability among species, this is a good site for examining resource allocation in resprouters.
Because most species are capable of both vegetative (resprouting) and sexual reproduction, species were designated as either poor or good resprouters based on their relative frequency of multi-stemmed individuals in the forest (Kruger, Midgley & Cowling 1997; Holness 1998; Nzunda et al. 2007a). Poor resprouters in our study differ from non-resprouters in fire-prone ecosystems in that they are not obligate reseeders. The categories of sprouting were matched in three species pairs from three families to control for the effect of phylogeny on resource allocation characteristics (Table 1).
Ten seedlings per species were excavated to their full rooting depth. Seedlings were 10–20 cm in height. The seedlings were divided into roots, stems and leaves and oven dried at 70 °C for 96 h. Dry mass of the root and shoot portions was determined to 0·001 g. Subsamples of root and stem were ground for determination of total non-structural carbohydrate (TNC) concentrations. TNC was not determined for leaves because the levels in leaves fluctuate daily in response to photosynthetic activity and export of starch out of the leaves (Graham et al. 2003; Myers & Kitajima 2007).
TNC was measured as the sum of starch and three sugars: fructose, glucose and sucrose (Sakai et al. 1997). Starch is the main food and reserve compound in woody plants, while fructose, glucose and sucrose are the most essential assimilation products (Zimmermann & Brown 1971; Sakai et al. 1997). Concentrations of fructose, glucose and sucrose were determined using gas chromatography (Varian 3800 Chromatograph, Walnut Creek, CA) based on standard procedures (Sweeley et al. 1963; Grob 2001). The concentration of starch was determined as glucose equivalents in the residues of the ethanol extracts (Rose et al. 1991; Sakai et al. 1997). Starch was extracted and hydrolyzed to glucose by perchloric acid (Rose et al. 1991). After neutralization by ammonia solution, the extract was subjected to gas chromatography to determine the glucose concentration.
We then tested the effect of stump height on the number and dry mass of sprouts produced. For each of the six species, 10 trees were cut to a stump height of 10 cm above the ground (hereafter referred to as short stumps) and 10 trees to a stump height of 150 cm (referred to as tall stumps). The 120 trees used in this experiment ranged from 2·5 to 11·1 cm diameter at breast height (d.b.h.). Only single-stemmed individuals were used. The number of stumps that resprouted and the number and dry mass of sprouts per stump were determined after one year of growth. Stumps were caged to prevent herbivory by large browsers.
Root:shoot ratio (a measure of relative biomass allocation to roots) and TNC data were analysed using a split-plot design in analysis of variance after ln-transformation. To avoid the risk of incurring Type I errors, multivariate analysis of covariance (mancova) was used to analyse the effect of resprouting, stump height and their interaction on number and dry mass of sprouts per stump simultaneously with stump d.b.h. as a covariate. These data were blocked by family. Stump d.b.h. was included as a covariate because resprouting ability may depend on tree size (Bellingham et al. 1994; Iwasa & Kubo 1997; Yamada & Suzuki 2004). Both number and dry mass of sprouts were ln-transformed. All data transformations were based on examination of the distribution of model residuals (Kéry & Hatfield 2003). Data were analysed using GenStat 9·1 (Lawes Agricultural Trust 2006).
Root:shoot ratios varied according to resprouting category (F1,56 = 36·97, P < 0·001). Controlling for the effect of family, seedlings of good resprouters had higher root:shoot ratios than poor resprouters (Fig. 1).
TNC concentration in seedlings was significantly influenced by resprouting ability and plant part (Table 2). Good resprouters had higher TNC concentrations than poor resprouters and stems had higher TNC concentrations than roots (Fig. 2). Good resprouters had higher TNC concentrations in their stems than poor resprouters. For good resprouters, stems had higher TNC concentration than roots, whereas there was no difference in TNC concentration between stems and roots for poor resprouters (Fig. 2).
Table 2. Split-plot analysis of variance (anova) for the effect of resprouting (poor or good resprouter) and plant part (root or stem), and their interaction on the concentration of total nonstructural carbohydrates
Source of variation
Family (Ebenaceae, Rutaceae or Salicaceae) was used as a blocking factor in the anova. TNC data were ln-transformed.
Sprouting × Part
The proportion of stumps that resprouted was greater for good than poor resprouters and for taller than shorter stumps (Fig. 3a). Stumps produced sprouts just below the cut end regardless of stump height. Controlling for the effect of stump d.b.h., there was a significant difference in the number and the dry mass of sprouts produced between resprouting categories and stump heights, but no interaction between resprouting category and stump height (Table 3). Good resprouters mostly produced more sprouts than poor resprouters, while tall stumps generally produced more sprouts than did short stumps (Fig. 3b). For good resprouters, tall stumps produced a significantly greater dry mass of sprouts than any other combinations of sprouting category and short stump height (Fig. 3c). For the mass of sprouts, there was no significant interaction between resprouting category and stump height (Table 3).
Table 3. Multivariate analysis of covariance (mancova) for the effect of resprouting (poor or good resprouter), stump height (short = 10 cm or tall = 150 cm) and their interaction on the number and dry mass of sprouts per stump
Source of variation
Data were blocked by family (Ebenaceae, Rutaceae or Salicaceae), and diameter at breast height (d.b.h.) of stumps was included as a covariate. Number and dry mass of sprouts per stump were ln-transformed.
Number of sprouts
Stump height (H)
R × H
Dry mass of sprouts
Stump height (H)
R × H
The pattern of larger seedling root:shoot ratios and root TNC concentration for good resprouters compared with poor resprouters is similar to the pattern of differences in root:shoot ratios and root carbohydrate concentration in fire-prone shrublands between resprouters and reseeders, respectively (Bowen & Pate 1993; Bell & Ojeda 1999; Verdaguer & Ojeda 2002; Knox & Clarke 2005). However, good resprouters in this study differ from resprouters in fire-prone habitats because they maintain higher concentrations of TNC in their stems than in their roots (Bell et al. 1996). Although in fire-prone systems there is a strong selective advantage to storing reserves below-ground than above-ground (Bond & Midgley 2001), higher TNC concentration in stems than in roots is favoured in good resprouters in this study, because they can usually remobilize above-ground reserves for resprouting after disturbance. Disturbances at our study site tend to cause leaning (Nzunda et al. 2007a,b). Leaning may break roots on the side opposing the lean, resulting in loss of TNC stored in the disconnected roots. Having higher TNC concentration in the stem could bet-hedge against this possibility.
Another factor that might favour higher concentration of TNC in stems than in roots in sand dune forests is that leaning trees may use an alternative to resprouting that involves regaining of the vertical orientation of the growing tip of the leaning stem, a process we refer to as ‘turning up’ (Nzunda et al. 2007b). When the growing tip of a leaning stem regains its vertical orientation, the tree's survival is ensured and the TNC in the stem is not wasted. However, because turning up occurs by formation of reaction wood that requires more resources than formation of normal wood, extra TNC has to be used (Zimmermann & Brown 1971; Mattheck 1995; Jiang et al. 2006). During turning up it might be easier to redistribute TNC within the stem than to translocate it from roots, providing another explanation for higher TNC concentration in stems than in roots in these subtropical forests.
In spite of similarity in the severity of disturbance that causes resprouting, good resprouters in our study generally differed in root:shoot ratios, carbohydrate storage, and stump resprouting response, from E. polyandra studied in Japan by Sakai and colleagues (Sakai et al. 1995, 1997; Sakai & Sakai 1998). In our study, good resprouters had larger root:shoot ratios, root and stem TNC concentration and number and dry mass of sprouts produced by stumps than poor resprouters. In contrast, E. polyandra did not differ from a poor resprouter in root:shoot ratio, root and stem TNC concentration and number of sprouts, and had lower dry mass of sprouts produced by stumps. In fact, E. polyandra produced surprisingly few sprouts of low dry mass even from tall (150 cm) stumps, unlike good resprouters in our study. Sakai & Sakai (1998) suggested that much larger trunk volume or additional above-ground structures such as foliage and branches were necessary for resprouting in E. polyandra. Moreover, resprouting in E. polyandra may be achieved using resources remobilized directly from the disturbed photosynthesizing parts rather than using stored resources because the species maintains very low levels of reserves in both roots and shoots (Sakai et al. 1997, 1998).
Because E. polyandra resprouted abundantly in response to natural stem leaning and partial uprooting and resprouted only slightly after cutting, Sakai & Sakai (1998) concluded that resprouting in E. polyandra was uniquely adapted to disturbances that cause stem leaning and partial uprooting. In that case, resprouting is achieved using resources drawn from the disturbed, but, live shoot, unlike in cutting where most of the shoot is removed. Similarly, good resprouters in our study might be expected to have a poor resprouting response to cutting. However, a vigorous resprouting response to cutting by good resprouters in our study suggested that although resprouting and multi-stemming are strongly associated with stem leaning and partial uprooting at Cape Vidal (Nzunda et al. 2007a,b), resprouting ability is not uniquely adapted to disturbances that cause stem leaning and partial uprooting. Trees in our study area persist by resprouting in response to other disturbances as well and, depending on the severity of the disturbance, are clearly capable of using either or both below- and above-ground resources to resprout. This versatility of resource deployment for resprouting may be advantageous in the coastal dune forests where low-intensity background levels of chronic disturbance (e.g. wind; Nzunda et al. 2007a) are occasionally punctuated by high-intensity disturbances (e.g. dune slump).
Resource allocation in coastal dune trees appears to be a bet-hedging strategy. After low-severity disturbances, resprouting occurs by remobilization of above-ground resources. Below-ground resources may be more costly to remobilize but may allow recovery from occasional more severe disturbances. This study emphasizes the adaptive advantage of sprouting through remobilization of above-ground resources and the importance of the persistence niche (Bond & Midgley 2001) in this habitat.
We are grateful to Ezemvelo KwaZulu-Natal Wildlife and the Greater St Lucia Wetland Park authority for permission to conduct fieldwork at Cape Vidal. Financial support from the National Research Foundation of South Africa (Focus area: Conservation and Management of Ecosystems and Biodiversity; GUN: 2069339), the Andrew W. Mellon Foundation and the Mazda Wildlife Fund are gratefully acknowledged. We thank Zivanai Tsvuura for assistance with fieldwork, Colin Southway and Dashnie Govender for assistance with chemical analyses, and Robyn Wethered and Harriet Eeley for administrative support. Comments from two anonymous reviewers on an earlier draft are greatly appreciated.