Trade-offs in resource allocation that favour resprouting affect the competitive ability of woody seedlings in grassy communities


*Correspondence author. E-mail:


1.  Differences in the competitive ability of plant functional groups at early life-history stages can have important consequences for community structure. In particular, trade-offs in allocation to roots by woody plant seedlings may influence competitive ability with grasses in fire-prone vegetation.

2.  We followed post-fire survival of seedlings of facultative resprouter and obligate seeder (fire-killed) shrubs for 3 years in adjacent communities with a grassy/graminoid ground stratum (54 plots, 20 m2) or a non-graminoid ground stratum (54 plots, 20 m2).

3.  The competitive effect of a grass (Poa) on seedlings of three congeneric pairs of resprouters and obligate seeder shrubs was tested in a factorial experiment where nutrients and the grass competitor were manipulated. The effects of grass (+,−) and nutrients (+,−) on the growth response, biomass allocation and root carbohydrate storage were measured after harvest at 26 weeks and the relative neighbour effect calculated.

4.  Post-fire shrub seedling survival was high with about 50% (2163 seedlings) surviving over 3 years, but this varied between habitats and functional groups. In the grassy/graminoid ground layer communities 27% of shrub seedlings survived, whereas in the habitats with a more open ground stratum 55% of seedlings survived. In grassy habitats, obligate seeder survival was lower (23% survival) than that of resprouter seedlings (35% survival). Similarly, in open habitats, obligate seeder seedling survival was lower (51%) than that of resprouter seedlings (64% survival).

5.  Growth of both resprouters and obligate seeders in our manipulative experiment was strongly reduced in the presence of a grass competitor. Moreover, the addition of nutrients increased the relative difference in mass and height between those seedlings exposed to a grass competitor and those grown without a competitor. Resprouter species allocated more to roots under competition and were less affected by grass competition than obligate seeders.

6.Synthesis. The results of seedling survival and of the experiment on the effects of grass competition on woody plant seedlings suggest that early life-history trade-offs in allocation influence seedling survival. Allocation to resprouting appears to enhance the ability of shrub seedlings to survive grass competition. We propose that grass competition across productivity gradients plays an important role in influencing landscape-level distribution patterns of woody resprouters.


Interactions between grasses and woody vegetation have been studied extensively, but the effects of variation in life history (i.e. resprouting ability) in woody plants have rarely been examined in the context of grass competition. Trade-offs in resource allocation that favour persistence (resprouting) over seed-based regeneration (adults killed by fire) occur in woody plants (Bond & Midgley 2001), and this can have a profound influence on community dynamics (Clarke & Dorji 2008). Taxa that resprout after fire allocate resources to structures that enhance their ability to survive the next fire, whereas those that are killed by fire allocate resources to ensure they are reproductively mature before the next disturbance (Bell 2001; Bond & Midgley 2003; Lamont & Wiens 2003). Many studies have found that resprouting woody species allocate more resources to root mass and have higher levels of non-structural carbohydrate in the roots than those species that are killed by fire (e.g. Pate et al. 1990; Bell & Ojeda 1999; Verdaguer & Ojeda 2002; Knox & Clarke 2005; Schwilk & Ackerly 2005). In addition, experimental manipulations have shown that both biomass allocation and carbohydrate accumulation are also under strong environmental control in resprouting species, but less so in species killed by fire (Knox & Clarke 2005). As a consequence, seedlings of resprouters are predicted to be better at capturing below-ground resources, whilst seedlings of obligate seeders are predicted to be better at capturing above-ground resources. If allocation trade-offs appear early in ontogeny, they should also lead to different competitive responses to neighbours (sensuGoldberg 1990) between seedlings of resprouters and obligate seeders.

Woody species are generally regarded as ineffective competitors for below-ground resources when establishing. In particular, grass neighbours are known to induce large reductions in woody seedling growth (Aerts, Boot, & van der Aart 1991; Wilson 1998; Nano & Clarke 2009), and increasing soil fertility reinforces the competitive superiority of grass vegetation (Aerts, Boot, & van der Aart 1991; Bloor, Barthes, & Leadley 2008). In fire-prone ecosystems, seedlings of both facultative resprouters and non-resprouting species initially escape the competitive effect of neighbours by rapid germination and establishment. This occurs because fires remove the herbaceous above-ground biomass, release nutrients and stimulate seed dispersal and/or germination. Predicting the subsequent outcome of woody–grass interactions at the seedling stage is complex because resource allocation in woody plants differs with resprouting ability (resprouters vs. obligate seeders) and with soil fertility (Knox & Clarke 2005). Landscape-scale studies have shown that resprouting shrubs are more common in grassy landscapes such as savanna woodlands, temperate grassy woodlands and cerrado (Frost 1985; Hoffmann 2000; Clarke et al. 2005; Bond 2008; Hoffmann et al. 2009). In contrast, obligate seeders are often more common in nutrient-poor communities such as chaparral, fynbos and sclerophyllous woodlands, where grasses rarely dominate the ground stratum (Keeley 2000; Pausas et al. 2004; Clarke et al. 2005; but conversely see Kruger, Midgley, & Cowling 1997; Bellingham & Sparrow 2000; Bell 2001). These landscape patterns suggest that the interactive effects of resprouting ability, grass competition and soil fertility may influence the composition and evolution of woody plants in fire-prone landscapes. More broadly, these trait patterns align with the theory of allocation trade-offs along productivity gradients (Tilman 1988).

In this study, we examined the post-fire survival of seedlings of resprouter (facultative resprouters) and obligate seeder (fire-killed) shrubs for 3 years in adjacent sclerophyllous communities with ground strata dominated by either graminoids (grasses and sedges) or non-graminoids. We then used a glasshouse experiment to assess how grass competition and nutrient availability interacted to affect growth, biomass allocation and investment in root carbohydrates in congeners with resprouter versus obligate seeder life histories.

Materials and methods

Study region

The study region is the New England Tableland Bioregion of eastern Australia that has a cool temperate climate (850–1200 mm mean annual rainfall) with predominantly summer rainfall but no winter drought. The study sites, all of which occur in conservation reserves, range in altitude from 950 to 1100 m a.s.l. and have similar local climates with mild summers and cool winter temperatures. A range of fire-prone evergreen vegetation formations occurs in the Bioregion: dry sclerophyllous forests and woodlands, grassy woodlands and forests, rocky outcrop heaths and sedgeland heaths. These broad groups form distinct floristic formations that are related to lithology and local physiography rather than climate. As a generalization, nutrient-poor siliceous soils provide habitats for scleromorphic shrub-dominated woodlands and forests, whilst the more fine-textured soils derived from metasediments and basalts support grassy woodlands and mesic forests. The mean total nitrogen (N) and phosphorus (P) concentrations in the surface soils for the ‘grassy’ habitats were 0.47 mg g−1 and 250 μg g−1 whereas for the ‘open’ habitat they were 0.23 mg g−1 and 74 μg g−1. All habitats are prone to fires, and in 2002 high intensity, landscape-scale fires burned into mesic wet sclerophyllous forests and rainforests margins.

Seedling survival in the field

Seedling survival of woody plants was measured for 3 years after concurrent wildfires in the Bioregion during October 2002. Three communities with graminoids dominating the ground stratum, ‘grassy’ habitat, (sedgeland heaths, grassy woodlands and grassy forests) and three communities with sparse ground stratum, ‘open’ habitat (sclerophyll forests, sclerophyll woodlands and rocky outcrops) (see Clarke et al. 2005) were sampled. Seedling emergence for all woody plants was recorded in plots (20 × 1 m). Commencing in February 2003, seedling demography was measured using coloured pin addition for emergences of different taxa and pin removal upon mortality. Each community was sampled at six sites and each site had three replicate plot samples. Overall, there were 54 plots for grassy communities and 54 plots where the ground stratum was open. Seedling censuses were conducted at 2-month intervals in the first year, then at 6-month intervals during the second and third year. One major cohort resulted from germination in 2003; minor cohorts of seedlings from later years were excluded from the analyses because of low numbers and survival. Seedlings of shrub species were classified into two classes: those where the adult population was recorded as resprouting after crown scorch and those where the adult population was killed by crown scorch. A two-factor anova was used to test the hypothesis that seedling survival of resprouter species was higher in the ‘grassy’ communities than in the non-grassy communities. Conversely, we tested whether the seedlings of species killed by fire had lower survival in the ‘grassy’ communities.

Competition experiment

Three pairs of long-lived shrub species from three families were used as a model system to represent the phylogenetic diversity of plants with either resprouter or non-resprouter (obligate seeder) life histories (Table 1). The competitor grass, Poa sieberiana Spreng, is a widespread perennial tussock grass that is more abundant in grassy woodlands and forests on higher-fertility soils than in communities growing on infertile soils. Tussocks of Poa resprout after fire. Seeds of the shrub species were collected fresh from the field, placed in a germination cabinet and germinated on pads moistened with deionized water. Dormancy of Hovea spp. was broken by scarification of the seed coat with a razor blade. Following germination, seedlings were transferred to a glasshouse soon after radicle emergence and planted into experimental pots.

Table 1.   Focal shrub species used in the competition experiment and their families, seed mass, fire response and communities in which they commonly grow. Nomenclature according to New South Wales National Herbarium, Australia
Species and authorityFamilySeed mass (mg)Adult fire responseSeed bank location and recruitment processCommunity
Allocasuarina brachystachya (Miq), L.A.S JohnsonCasuarinaceae1.2Resprouter (basal shoots)Fire-stimulated release from serotinous fruits and post-fire seedling germinationRocky outcrop and sclerophyllous forests
Allocasuarina rigida (Miq), L.A.S JohnsonCasuarinaceae1.0Obligate seederFire-stimulated release from serotinous fruits and mass post-fire seedling germinationRocky outcrop
Hakea laevipes subsp. laevipes Gand.Proteaceae20.1Resprouter (basal shoots)Fire-stimulated release from serotinous fruits and post-fire seedling germinationDry sclerophyllous forests
Hakea salicifolia subsp. salicifolia (Vent.) B.L.Burtt.Proteaceae17.1Obligate seederFire-stimulated release from serotinous fruits and mass post-fire seedling germinationWet sclerophyllous forests
Hovea heterophylla (Smith) R.BrFabaceae14.8Resprouter (basal shoots)Fire-stimulated soil seed bank and germinationDry sclerophyllous and grassy forests
Hovea graniticola I. Thomps.Fabaceae42.5Obligate seederFire-stimulated soil seed bank and germinationRocky outcrop and sclerophyllous forests

A fully factorial, randomized design was used to assess the effects of nutrient addition and neighbour on three pairs of shrub species in a glasshouse experiment. Two nutrient treatments were applied: addition of nutrients in the form of 1 g L−1 of low phosphorus Osmocote™ mixed into coarse granitic river sand and no addition of nutrients. The coarse sand was derived from lithologies very similar to those in which the shrubs grow; they had very low levels of N and P and were not sterilized to remove soil biota. Two competition treatments (with or without grass) were applied. For the grass competition treatment, a large tussock of Poa sieberiana was dug up from the field and 4–8 tillers were transplanted into half of the experimental pots (25 cm diameter, 25 cm depth) filled with sand. These tillers were allowed to establish for 6 weeks, before the leaf blades were burned with a hand-held propane gas burner to simulate a fire. This was done because the recruitment of shrub seedlings is strongly fire-cued and they would not normally compete with unburned grass tussocks. Two weeks after the grass tillers had resprouted, two or three newly germinated shrub seedlings (<1 cm in height) were transplanted into the pot adjacent to the grass tillers or transplanted into pots that did not have the grass competitor. These seedlings were allowed to establish for 2 weeks before they were thinned to one seedling per pot. The remaining seedlings were left to grow for 26 weeks, and all pots were watered regularly because no seasonal droughts occur in the study region.

At approximately monthly intervals, the height, leaf number and basal girth of each plant were measured. Leaf numbers for Allocasuarina species were not measured because they have minute leaf scales and their cladodes cannot be easily counted. At harvest, the shrub seedlings were carefully disentangled from the Poa roots and divided into roots, stems and ‘leaves’ (cladodes and leaf scales) and were dried for 24 h at 80 °C for dry weight determinations. Additional response variables for each plant were derived from primary measures: specific leaf area (SLA = leaf area per leaf dry mass), leaf area ratio (LAR = leaf area per total plant dry mass), stem mass ratio (SMR = stem dry mass per total plant dry mass), leaf mass ratio (LMR = leaf dry mass per total plant dry mass), root mass ratio (RMR = root dry mass per total plant dry mass). For Allocasuarina, the term leaf refers to the cladodes because they are functionally equivalent to leaves. The total dry mass of Poa was also measured for each replicate and did not differ among competition treatments.

Sub-samples of fine and coarse root material of shrubs were ground for starch and soluble sugars determination. Following a technique adapted from Grimmer, Bachfischer, & Komor (1999), soluble sugars were extracted from 20 mg of dried, finely ground root samples by addition of 0.5 mL 70% ethanol and 5.5 mL 0.0001 HEPES buffer pH 7.4 and incubated at 70 °C for 45 min. The extract was centrifuged at 4 °C at 1780 r.p.m. for 15 min and the supernatant was collected and stored at −20 °C before being analysed for glucose enzymatically using hexokinase and glucose phosphate dehydrogenase (Bergmeyer 1963). The pellet was washed twice with water and once with 70% ethanol, the starch content was analysed by resuspending the pellet in 1 mL of water and autoclaving the suspension for 2 h followed by incubation with 1 mL 1.0 M Na-acetate pH 4.8 with 0.5 mL amyloglucosidase (147.6 μL−1), 0.5 mL alpha-amylase (264.5 μL−1) and 0.5 mL water at 37 °C for 45 min. The liberated glucose was then analysed as described above.

Statistical analyses and competition indices

The first hypothesis to be tested was that survival of seedlings differed between habitats (grassy vs. open) and between functional groups (resprouter vs. obligate seeder). This was tested for plot-based data in a two-factor anova where the response variable was the proportion of seedlings surviving after 3 years. All proportion data were arcsine-transformed prior to analyses and were checked for homogeneity of variance by plotting the residuals versus predicted values as recommended by Quinn & Keough (2002).

The second hypothesis to be tested was that glasshouse plant growth (stem length, leaf number, leaf area, SLA), patterns of allocation (LMR, RMR, SMR), and root carbohydrates varied with resprouting ability, nutrient addition and competition. This was tested using a four-factor mixed-model anova with sprouting ability, genus, nutrients and competition as factors. Genus was a random effect, whereas all other factors were fixed effects in the model. The dependent variables were ln-transformed except for ratio data where the arcsine squareroot of the variable was used. All data were checked for homogeneity of variance as above.

The third hypothesis to be tested was that the outcome of glasshouse competition varied with resprouting ability and nutrient addition. The index of competition used as the response variable was the relative neighbour effect (RNE). This index compares the total biomass of the shrub species in the presence of the grass (bmix) with that grown in isolation (biso), where inline image (Markham & Chanway 1996; Oksanen, Sammul, & Magi 2006). The third hypothesis was tested using a three-factor mixed anova with sprouting ability and nutrients as fixed factors and genus as a random factor. This model was reduced to two factors when the main effect of genus and its interactions was non-significant at > 0.5 because the power of this test is very low with four replicates for each treatment combination.


Seedling survival in the field

Of the initial cohort of 4301 shrub seedlings, about 50% survived over 3 years, but survival varied between habitats and function group (Table 2). More seedlings of obligate seeders (3120) than resprouters (1181) emerged after fire and more seedlings emerged in open habitats (3605) than grassy habitats (696). In the communities dominated by grassy/graminoid ground layer, 23% of obligate seeder seedlings survived, whereas in the habitats with a more open ground stratum 51% of obligate seeder seedlings survived. Over all sites, seedlings of obligate seeders had lower survival (47%) than resprouter seedlings (58%) (Table 2), but the total number of seedlings surviving was similar for obligate seeders (101) and resprouters (89) in open habitats. In contrast, more obligate seeders (1379) than resprouters (594) remained present in the open habitats. Throughout the field observations, no chlorosis or obvious pathogenic symptoms were observed on any of the seedlings.

Table 2.   Total number of seedlings recorded in both habitats and functional groups and proportion surviving after 3 years. The fates of all seedlings were followed individually
Seedling numbersGrassy habitatOpen habitatTotal
Obligate seedersResproutersObligate seedersResprouters
% survival in 200623.135.551.464.350.3

At the plot scale, mean seedling survival was higher in the open habitats than in adjacent habitats with a dense graminoid ground stratum (F1,102 = 26.78, P < 0.001, Fig. 1), whereas in both habitats, seedling survival was higher for resprouting species than in those species killed by fire (obligate seeders) (F1,102 = 5.03, P < 0.05, Fig. 1). Seedling mortality of obligate seeders appeared to be greater in the grassy habitats than in the open habitats, although no significant statistical interaction was detected (F1,102 = 0.48, P > 0.05).

Figure 1.

 Mean (+SE) proportion of seedlings surviving at plot scales for 3 years after post-fire recruitment between two broad habitat types and between species that have high resprouting ability (dark shade) and obligate seeders with no resprouting ability (light shade).

Competition experiment

Plant growth variables were strongly affected by the addition of Poa to pots, with total mass, stem length and leaf number all being reduced in the presence of Poa (Table 3, Fig. 2). There was also an interactive effect of nutrient addition and competition from Poa: plant dry mass and height were more strongly suppressed in the presence of Poa when nutrients were added (Table 3, Fig. 2). No interactive effects of life history (sprouting ability) and competition were detected for growth variables (Table 3) and overall obligate seeder species gained more plant mass and grew taller than resprouters (Fig. 2). Neither competition nor nutrient addition affected specific leaf area (Table 3).

Table 3.   Results of four-factor analyses of variance (anova) for the effect of resprouting ability (resprouters vs. obligate seeders), genera (Allocasuarina, Hakea, Hovea), nutrients (addition vs. no addition) and competition (grass vs. no grass) for eight plant response variables
Factord.f.Total massShoot lengthd.f.Leaf number† SLA†d.f.LMRRMRSMRTotal root starch
  1. *< 0.05, **< 0.01, ***< 0.001; NS, not significant.

  2. † Note that Allocasuarina was not included as a genus in these analyses.

Sprouting × genera20.49NS3.62NS10.60NS2.54NS26.86**2.05NS19.5NS7.55**
Sprouting × nutrients10.10NS1.79NS10.01NS1.26NS10.22NS0.64NS0.33NS0.64NS
Sprouting × competition10.37NS1.79NS10.92NS2.64NS16.77*6.04*8.87**0.39NS
Genera × nutrients20.66NS0.85NS16.31*3.73NS21.14NS2.30NS1.42NS0.12NS
Genera × competition24.34**0.90NS10.75NS0.79NS22.58NS1.51NS0.53NS2.10NS
Nutrients × competition12.52**5.42*19.23**3.06NS10.01NS0.87NS0.96NS2.78NS
Sprouting × genera × nutrients20.96NS2.99**11.22NS1.16NS20.07NS0.06NS0.21NS0.18NS
Sprouting × genera × competition20.84NS2.23NS10.03NS1.23NS20.20NS0.15NS0.38NS1.58NS
Sprouting × nutrients × competition10.18NS0.43NS10.09NS0.75NS12.62NS1.68NS2.62NS0.49NS
Genera × nutrients × competition20.80NS1.674NS12.59NS1.12NS21.68NS3.53NS4.50NS1.58NS
Sprouting × genera × nutrients × competition21.02NS1.55NS10.86NS0.79NS20.54NS0.53NS0.78NS0.88NS
Residual d.f. 6270 4850 626262113
Transform Ln + 1Ln + 1 Ln + 1Ln + 1 Arcsin sqrtArcsin sqrtArcsin sqrtNone
Figure 2.

 Mean values (+SE) for (a, b) plant dry mass, and (c, d) stem height with and without Poa as a neighbour (a, c) and among resprouters and obligate seeders where nutrient and competition effect are pooled (b, d). Allocas. = Allocasuarina.

Resource allocation was strongly affected by the nutrient treatment with increased allocation to leaves when nutrients were added (Table 3). The effect of competition interacted with sprouting ability (Table 3, Fig. 3) in that a pronounced increase in allocation to roots was observed for resprouter species in the presence of Poa. Conversely, there was a decrease in allocation to leaf material in resprouter species, whilst obligate seeders showed little response to allocation in the presence of Poa (Fig. 3). Starch concentrations in the roots of both life-history types were not affected by either competition or nutrient treatments (Table 3), but resprouting species had significantly higher concentrations of starch in their roots than obligate seeders (Fig. 4).

Figure 3.

 Mean values (+SE) for (a) leaf mass ratio and (b) root mass ratio with and without Poa as a neighbour, nutrient and genera levels pooled. OS, obligate seeders; R, resprouters.

Figure 4.

 Mean values (+SE) for concentration of root starch in roots of three congener obligate seeders and resprouters. Allocas = Allocasuarina.

The RNE was strongly influenced by the nutrient treatment and resprouting ability, but no interactive effect of nutrients and sprouting ability was detected (Table 4, Fig. 5). The addition of nutrients enhanced the competitive effect of Poa, whilst obligate seeders were more strongly affected by competition with the grass than resprouters (Table 4, Figs. 5). These effects are close to being marginally significant at < 0.05 due to the low power of the experiment (power = 0.5) with four replicate plants for each treatment.

Table 4.   Results of three-factor analyses of variance (anova) for the effect of resprouting ability, genera and nutrients for relative neighbour effect (RNE) response variable (see Materials and methods section for index calculations)
  1. *< 0.05. NS, not significant.

  2. †Main effects results are for a reduced two-factor model where genus and the interaction terms (sprouting × genera, sprouting × genera × nutrients) have been pooled because they were not significant. For genera the P-values exceeded 0.5 when factors were pooled.

Sprouting ability10.40†4.15†*0.5†
Sprouting × genera20.212.02NS 
Sprouting × nutrients10.03†0.34NS0.1†
Genera × nutrients20.050.53NS 
Sprouting × genera × nutrients20.030.25NS 
Residual36, 42†0.09†   
Figure 5.

 Mean values (+SE) for relative neighbour effect on shrub seedlings by a grass neighbour (Poa). Means are pooled across three congeners Allocasuarina, Hakea and Hovea. Treatment effects are for nutrient addition. A more negative value indicates a stronger competitive effect of the grass neighbour, Poa.


Seedling survival

Seedling survival in grassy communities was lower at landscape and patch scales than in adjacent habitats that lacked a dense ground cover. We believe this difference is due to competition for below-ground resources after germination in the post-fire environment. Above-ground competition may also be important as pre-existing vegetation regrows and casts shade on seedlings, but most mortality occurred in the first 6 months prior to light limitation through overgrowth of seedlings (P. J. Clarke & K. J. E. Knox, personal observation). Our study supports the general view that post-fire grass competition plays an important role in influencing the woody-grass ratios in savanna and woodland ecosystems (Kocky & Wilson 2000; Davis et al. 2005; Bond 2008; Nano & Clarke 2009). Although, as Bond (2008) highlights, few studies show complete seedling failure due to grass competition.

Much of the recent literature on the competitive effect of grasses on woody plants has focused on different grass functional groups (exotic vs. native, C4 vs. C3) (e.g. Davis et al. 2005; Zavaleta 2006). Conversely, experimental studies on the response of woody functional groups to grass competition have been lacking. In particular, the effects of grass competitors on seedling survival in savannas has been examined intensively (see Bond 2008), but it appears that the effect of grass competition on weak vs. strong resprouters has been overlooked. Our field results of shrub seedling survival suggest that the resprouting ability of different woody plants strongly influences the outcome of competition with the herbaceous ground layer: those species with the ability to resprout have higher seedling survival across all landscapes.

In Mediterranean-type climates, as distinct from temperate climate, post-fire seedlings of shrubs have high mortality in their first year of life. In these systems, summer drought and competition from surrounding vegetation is thought to be the main cause of seedling mortality (Keeley & Zedler 1978; Fraser & Davis 1988; Odion & Davis 2000). Strong negative effects of grasses on post-fire recruitment of chaparral shrub seedlings are also well-known where exotic grasses are sown into the post-fire environment to ameliorate erosion (Beyers 2004). Selection for water stress tolerance is thought to be greatest among those species that are dependent on seedling recruitment for population survival (i.e. obligate seeders) because of higher population turnover and possible shorter generation times. Hence, the expectation from ecophysiological studies in Mediterranean climates is that obligate seeders should be more tolerant of competition for below-ground resources than resprouters. This is because obligate seeders in seasonally stressed biomes have higher water stress resistance at the leaf, stem and root level (Ackerly 2004; Paula & Pausas 2006; Pratt et al. 2007, 2008). This is the converse to what we found because, obligate seeders and resprouter congeners have no known trait differences in relation to water stress in temperate, fire-prone, sclerophyll biomes (Knox & Clarke 2005).

Competitive response of sprouters vs. obligate seeders

Growth of both resprouter and obligate seeder seedlings in our experiment was strongly reduced in the presence of a grass competitor that had been burned prior to the establishment of shrub seedlings. Furthermore, the addition of nutrients increased the relative difference in mass and height between those seedlings exposed to grass competitor and those grown without a competitor. These results support the notion that competitive importance of the herbaceous layer increases along soil fertility gradients (Bloor, Barthes, & Leadley 2008). This may reflect the reported shift from root-to-shoot competition when light becomes a limiting factor under increased grass growth, due to the asymmetry of an established grass clump vs. that of a seedling. We detected no plant mass, leaf trait or height differences in the way seedlings of obligate seeders and resprouters responded to the grass competitor, but allocation, absolute and relative competitive indices were strongly affected. Whilst growth rates of congener seeder and sprouter seedlings are reported to be similar (Knox & Clarke 2005; Schwilk & Ackerly 2005), resprouter leaf trait analyses suggest that carbon assimilation may be enhanced to meet the demands of resprouting (Paula & Pausas 2006). Allocation to roots was increased in the resprouter species when exposed to grass competition and, hence, increased their competitive ability for below-ground resources. These results correspond to the findings of Aerts, Boot, & van der Aart (1991) for northern hemisphere resprouter shrub species and align with previous studies of allocation and resprouting ability where nutrient stress increased root allocation and the concentration of non-structural carbohydrates in resprouters (Knox & Clarke 2005). More broadly, our results illustrate how resource constraints, allocation and competition interact to influence life histories of plants as predicted by theory (Tilman 1988).

Linking functional traits and demography

Allocation to roots increased in resprouter species in the presence of a competitor, but not in the obligate seeder congeners. When combined with greater total non-structural carbohydrate concentrations in resprouter roots, the initial ‘storage success’ provides resources for maintenance when shoot competition with grasses becomes more intense as grasses grow. Conversely, the taller seedlings of obligate seeders must escape the light trap by allocating to leaves at the expense of roots. However, the fibrous root system of established grasses that surrounds seedlings will inevitably induce a strong competitive effect. The outcome of these interactions is that in the early growth stage seedlings of obligate seeders are more likely to be strongly suppressed than seedlings of resprouters, as suggested by Knox & Clarke (2005). These trait differences at the early seedling stage are consistent with our field data, showing proportionally more survival in resprouting seedlings in the 3 years following a fire. The difference in the early competitive response in resprouters and obligate seeder seedlings also aligns with observations in Mediterranean landscapes where seedlings of obligate seeders survive in more open habitats, but are susceptible to drought stress (Keeley 2000). Recruitment failure of obligate seeders, caused by drought and/or short fire intervals, probably allows resprouter species to coexist with obligate seeders in habitats without strong competition.

Trade-offs between persistence and reproduction are well-known for shrubs in fire-prone environments (Bond & Midgley 2001; Knox & Clarke 2005) and are thought to extend to trade-offs concerning improved site occupancy as plants mature and disperse seed (Falster & Westoby 2005; Higgins, Flores, & Schurr 2008). We have demonstrated the potential for very early allocation differences in seedlings to influence the outcome of competition with neighbours. Furthermore, when obligate seeders are competing with grasses they may fail to reach reproductive maturity, or if they do, their potential reproductive output may be curtailed. In these circumstances, the innate ability for improved site occupancy in obligate seeders can be suppressed in grassy environments without having to invoke fire or herbivory as a limiting factor. Clearly, however, a fire event prior to reproductive maturity poses an ‘immaturity risk’ to obligate seeders and potentially to the survival of resprouter seedlings if they have not acquired enough resources to resprout. In our system, and in other biomes, seedlings of resprouter species appear to be able to resprout at a very early age (P. J. Clarke, personal observation; Hoffmann 2000). Hence, the capacity for obligate seeders to dominate grassy landscapes may be constrained by soil fertility and grass competition interacting with high fire frequency.


We thank Ian Simpson for assistance in the glasshouse and Merideth Powell for chemical and plant analyses. We also thank Ross Bradstock, David Bowman, Jeremy Midgley, William Bond William Hofmann, editors and referees for robust comments and discussions. This research was supported by Australian Research Council infrastructure funding and the University of New England.