Interspecific variation in the size-dependent resprouting ability of temperate woody species and its adaptive significance
Summary
- Resprouting of woody species after above-ground damage may help plants to persist longer at a given site and quickly reoccupy disturbed sites, thereby strongly influencing forest dynamics. Resprouting has been discussed from two adaptation perspectives: recovery from damage by catastrophic disturbance and survival in frequently disturbed shaded understorey. However, few studies have comprehensively dealt with both adaptation types to understand resprouting strategies.
- To understand the adaptive significance of resprouting, we assessed the size dependence of resprouting ability after stem clipping for 24 deciduous broad-leaved species, including shrubs, sub-canopy and canopy trees, in a cool-temperate forest in Japan. The community assembly includes species adapted to past catastrophic disturbances (e.g. fire, logging) and to stable forest with intermittent treefall (currently the dominant disturbance). We correlated resprouting ability with life-history strategies based on demographic parameters and plant functional traits, such as leaf mass per area (LMA), leaf toughness and wood density.
- All the studied species could resprout in juveniles, and resprouting ability increased as stump size increased. Most sub-canopy and canopy trees lost their ability to resprout after attaining a particular stump size, whereas shrub species retained the ability to resprout throughout their lifetimes.
- The relative growth rate, LMA and foliar nitrogen did not greatly influence the resprouting ability of a species. In contrast, species with smaller maximum size, lower leaf toughness and lower wood density had better juvenile resprouting ability. This better resprouting ability may have evolved because these characteristics make them more vulnerable to shaded understorey. However, species with larger maximum size and lower leaf toughness retained their ability to resprout to a larger size.
- Synthesis. A better resprouting ability is related to the ability to survive frequent disturbances, in juveniles, which are characteristics of both forest understorey and frequent fire or drought. To retain resprouting ability until grown seems to be an adaptation to survive infrequent large disturbances. Light-demanding species, which generally have better resprouting ability than shade-tolerants both in juveniles and adults, are adapted to disturbances of various scale and frequency; however, shade-tolerants could survive well in the understorey due to a combination of stronger physical defences and resprouting ability.
Introduction
Resprouting is a trait that plants use to restore their above-ground biomass after above-ground parts are damaged by disturbances such as fire (Bell 2001) or a hurricane (Bellingham, Tanner & Healey 1994). Even in the absence of catastrophic disturbances, many woody plants resprout after physical damage caused by falling debris, browsing by animals, chronic coastal wind and dieback on the forest floor (Hara 1987; Paciorek et al. 2000; Ickes, Dewalt & Thomas 2003; Nzunda, Griffiths & Lawes 2007, 2008). Resprouts can grow much faster than similar-sized seedlings because they have the advantage of a large, pre-existing root system, and this allows damaged trees to quickly restore their above-ground biomass after a disturbance. This ability to resprout clearly differs among species, and the differences in their life-history strategy could therefore play a considerable role in a forest's dynamics and community assembly (Bellingham & Sparrow 2000; Loehle 2000; Bond & Midgley 2001; Poorter et al. 2010).
Resprouting has mostly been studied in relation to large-scale disturbance such as hurricanes and fires (Bellingham, Tanner & Healey 1994; Bellingham & Sparrow 2000; Bond & Midgley 2001). In case of such events, resprouting can extend the life span of individuals (Bond & Midgley 2001; Del Tredici 2001). However, resprouting ability comes with a cost (e.g. Bond & Midgley 2001). Species with better resprouting ability tend to allocate more biomass to their roots, resulting in slower growth of above-ground parts and a delayed start of reproduction (Bell 2001; Bond & Midgley 2001; Fig. 1c). On the other hand, several studies show that light-demanding species retain their resprouting ability until a larger size, and could regenerate after catastrophic events (Del Tredici 2001; Masaki 2002; Fig. 1a). The average time between catastrophes therefore strongly determines the selective advantage of good resprouting ability and the relationships between size or age and resprouting ability seems especially important.
A good resprouting ability can present an advantage to species growing in the shaded understorey, not only under catastrophic events but also in mature forest understorey (Paciorek et al. 2000; Poorter et al. 2010). The shaded understorey is often unsuitable for the growth and survival of woody species because of suppression by canopy trees and physical damages by falling debris or by pathogens or herbivores (Hara 1987; Gartner 1989). After such small-scale disturbances, resprouts have an advantage over seedlings because they can use their well-developed root system for a high growth rate of the above-ground parts (Cooper-Ellis et al. 1999; Dietze & Clark 2008). In Bolivian tropical forests, Poorter et al. (2010) showed that the juveniles of slow-growing shade-tolerant species had a better resprouting ability than those of fast-growing light-demanding species, even though nearly all species resprouted after stem clipping (Fig. 1b). Because small disturbances may occur at relatively high frequency in these mature forests, the resprouting ability at juvenile stages seems especially important in these systems.
To comprehensively understand a species' resprouting ability in the context of its overall life-history strategies, it is necessary to investigate the size dependence of resprouting ability and to relate the resprouting ability with other life-history strategies during a particular life stage of co-existing species. However, there have been no such studies. We therefore determined the size dependence of resprouting ability for deciduous broad-leaved species in cool-temperate forests in and around the Ogawa Forest Reserve, in central Japan. In this reserve, catastrophic disturbances such as wildfire and logging have been eliminated for 100–200 years, and the stands are now shifting towards a stable mature forest, where intermittent treefall to create canopy gaps is the dominant form of disturbance (Masaki 2002; Nakashizuka & Matsumoto 2002). Thus, the tree community assembly includes species adapted to catastrophic disturbances and to stable forests, and therefore provides an ideal system for understanding the resprouting strategy of woody species.
In our analysis of the life-history strategies of these species, we related the resprouting ability at different life stages with several key plant traits: the maximum basal area (as an index of maximum plant size), minimum basal area at first reproduction (as an index of the minimum tree size for first reproduction), relative growth rate during the juvenile stage, leaf mass per unit area (LMA), leaf toughness, foliar nitrogen concentration, wood density and the individual-seed mass. These traits should differently affect the trade-offs between growth and survival. On the basis of the previous studies, we hypothesized that (i) species with a relatively small maximum size (e.g. shrubs) would have better resprouting ability than species with larger maximum size (e.g. trees) when they are compared at similar size to improve their regeneration success and life span in a shaded understorey with frequent disturbance (Fig. 1b,c). In contrast, species that retain their resprouting ability until they are old or reach a large size would be adapted to catastrophic disturbances with a long return interval (Kamitani 1986) (Fig. 1a). (ii) As to the resprouting ability while plants are small, slow-growing shade-tolerant species would have better resprouting ability because they may be more frequently damaged than fast-growing light-demanding species by falling debris, as well as by herbivores and pathogens in shaded understorey (Gartner 1989; Clark & Clark 1991; Poorter et al. 2010; Fig. 1b). Shade-tolerant species generally have higher LMA, leaf toughness and wood density because these traits are also associated with stronger physical defences against pathogens and herbivores (Poorter et al. 2010). They would also have larger seeds than light-demanding species. Larger seeds may be unable to escape the unfavourable conditions in shaded understorey by means of long-distance dispersal, but tend to increase the seedling survival by providing a larger nutrient reserve (Poorter et al. 2008). (iii) In contrast, fast-growing light-demanding species would have lower LMA and a higher foliar nitrogen concentration, which results in a higher photosynthetic rate (Wright et al. 2004), and lower leaf toughness, wood density and seed mass than slow-growing shade-tolerant species. They grow quickly and would retain better resprouting ability until a larger size to survive catastrophic disturbances such as wildfire or tree logging (Del Tredici 2001; Masaki 2002; Fig. 1a).
To test these hypotheses, we addressed the following questions: (i) How can we characterize the resprouting ability of a species, and how does the relationship between stump size and resprouting ability differ among species? (ii) Are there any relationships between a species' resprouting ability and other life-history traits such as maximum size, growth or defence? On the basis of our results, we discuss how the resprouting strategies of various species change through their life history and how they relate to different disturbance regimes and shade-tolerance gradients, eventually leading to the coexistence of species.
Materials and methods
Research Site
Our study was conducted in the Ogawa Forest Reserve and the surrounding area, in the southern part of the Abukuma Mountains of central Japan (36°56′N, 140°35′E, elevation 610–660 m a.s.l.). The mean annual temperature is 12.4 °C; the mean monthly temperature is highest in July (23.5 °C) and lowest in January (1.9 °C). Annual precipitation is 1750 mm, and snow depth occasionally reaches 50 cm. The reserve is an old-growth, cool-temperate deciduous forest. Quercus spp. and Fagus spp. dominate the canopy layer, and Acer spp. and Carpinus spp. are abundant in the sub-canopy layer (Masaki 2002). This reserve had been affected by natural fire with a long return interval (i.e. typically longer than 100 years) for presumably more than 1000 years, and by human activities such as cattle grazing and selective cutting in the past. These disturbances were eliminated in the reserve between 100 and 200 years ago; as a result of this, the youngest stands are about 100 years old and the oldest stands are more than 200 years old. The surrounding secondary-growth stands had been clear-cut with a rotation of 20–30 years until the 1980s (Suzuki 2002), but now they are mostly left untouched. The dominant form of disturbance in this area is currently intermittent treefall that creates canopy gaps, because the local forest office has strictly protected the forest against fires in recent decades. Therefore, the tree community in this reserve has been affected both by catastrophic disturbance such as fire or logging in the past and by the creation of small canopy gaps by treefall, which is currently the dominant disturbance (Nakashizuka & Matsumoto 2002).
We measured the diameter at breast height (1.3 m above the ground; d.b.h.) of all adult trees (> 5 cm d.b.h.) in a 6-ha plot at the Ogawa Forest Reserve every 2 years from 1987 to 1993 and every 4 years thereafter. We measured the heights of the saplings for all studied species in more than 100 quadrats (5 × 5 m) in gaps in and around the 6-ha plot in 1990, 1992, 1994 and 1996 (Nakashizuka & Matsumoto 2002).
Some of the secondary forests around the reserve are periodically logged to provide wood for the production of mushrooms or wood chips. We collected resprouting data in clear-cut areas in second-growth stands that resulted from this logging. The logged stands were mixed deciduous temperate forests ranging from 50 to 80 years old and were dominated by Quercus spp. and Carpinus spp. mixed with many other species. Trees were cut at 5–20 cm above the ground using chainsaws. The severity of the resulting damage to each tree was considered constant.
Species Selection and Trait Measurements
We investigated 24 tree and shrub species from 17 genera in 13 families, most of which were common species in the secondary forests (Table 1). Among these species, 17 can reach d.b.h. >10 cm at their maximum size, and the other seven are shrubs. These species vary in their ecological characteristics, including their maximum size, light requirements and other life-history traits.
| Species | Family | Authority | Life-form | Estimated model | R 2 | Rjuv (cm2)a | Rmax (cm2)a | R rate |
|---|---|---|---|---|---|---|---|---|
| Carpinus tschonoskii | Betulaceae | Maxim. | Canopy | Decline | 0.60 | −1.12 (0.18) | 1.32 (0.17) | 0.82 (0.22) |
| Carpinus laxiflora | Betulaceae | (Sieb. et Zucc.) Bl. | Canopy | Decline | 0.59 | −1.49 (0.17) | 0.97 (0.21) | 1.21 (0.47) |
| Corylus sieboldiana | Betulaceae | Bl. | Shrub | Simple increase | 0.47 | 0.03 (0.12) | b | 0.34 (0.12) |
| Quercus serrata | Fagaceae | Thunb. ex Murray | Canopy | Decline | 0.27 | −0.94 (0.07) | 2.24 (0.23) | 0.34 (0.07) |
| Quercus crispula | Fagaceae | Blume | Canopy | Decline | 0.60 | −0.79 (0.08) | 3.06 (0.06) | 0.41 (0.07) |
| Castanea crenata | Fagaceae | Sieb. et Zucc. | Canopy | Decline | 0.88 | −0.64 (0.10) | 2.53 (0.29) | 0.61 (0.07) |
| Magnolia obovata | Magnoliaceae | Thunb. | Canopy | Decline | 0.78 | −0.79 (0.18) | 2.01 (0.15) | 0.88 (0.15) |
| Prunus verecunda | Rosaceae | (Koidz.) Koehne | Canopy | Decline | 0.40 | −0.78 (0.22) | 2.55 (0.13) | 0.34 (0.15) |
| Prunus grayana | Rosaceae | Maxim. | Sub-canopy | No decline | 0.45 | −0.49 (0.16) | b | 0.74 (0.32) |
| Hydrangea paniculata | Saxifragaceae | Sieb. et Zucc. | Shrub | Simple increase | 0.83 | −0.35 (0.10) | b | 0.72 (0.10) |
| Acer amoenum | Aceraceae | Carr. | Canopy | Decline | 0.55 | −1.26 (0.16) | 1.89 (0.38) | 0.52 (0.15) |
| Acer rufinerve | Aceraceae | Sieb. et Zucc. | Sub-canopy | Decline | 0.20 | −0.39 (0.58) | 1.81 (0.53) | 0.44 (0.46) |
| Acer mono | Aceraceae | Maxim. | Canopy | Decline | 0.45 | −1.09 (0.16) | 2.44 (0.27) | 0.36 (0.12) |
| Swida controversa | Cornaceae | (Hemsl.) Soják | Canopy | Decline | 0.14 | −0.86 (0.20) | 1.86 (0.49) | 0.35 (0.19) |
| Acanthopanax sciadophylloides | Araliaceae | Franch. et Savat. | Sub-canopy | Decline | 0.57 | −0.59 (0.19) | 2.00 (0.18) | 0.64 (0.18) |
| Aralia elata | Araliaceae | (Miq.) Seemann | Shrub | Decline | 0.95 | −0.07 (0.04) | 1.12 (0.03) | 0.74 (0.07) |
| Clethra barbinervis | Clethraceae | Sieb. et Zucc. | Sub-canopy | Decline | 0.31 | −0.29 (0.17) | 0.94 (0.24) | 0.38 (0.26) |
| Styrax japonica | Styracaceae | Sieb. et Zucc. | Sub-canopy | No decline | 0.75 | −0.66 (0.11) | b | 0.83 (0.16) |
| Styrax obassia | Styracaceae | Sieb. et Zucc. | Sub-canopy | Decline | 0.56 | −0.64 (0.16) | 2.01 (0.05) | 0.65 (0.15) |
| Fraxinus lanuginosa | Oleaceae | Koidz. | Sub-canopy | Simple increase | 0.76 | −0.83 (0.09) | b | 0.72 (0.11) |
| Callicarpa japonica | Verbenaceae | Thunb. | Shrub | Simple increase | 0.73 | −0.59 (0.08) | b | 0.60 (0.11) |
| Lonicera gracilipes | Caprifoliaceae | Miq. | Shrub | Simple increase | 0.79 | −0.43 (0.06) | b | 0.67 (0.11) |
| Viburnum dilatatum | Caprifoliaceae | Thunb. ex Murray | Shrub | Simple increase | 0.79 | −0.62 (0.05) | b | 0.58 (0.08) |
| Viburnum wrightii | Caprifoliaceae | Miq. | Shrub | Simple increase | 0.74 | −0.61 (0.05) | b | 0.59 (0.10) |
- a Log10-transformed.
- b The species did not lose its ability to resprout within the range of sizes that we examined.
We investigated resprouting ability in the first summer after clear-cutting in the previous winter, from June to early August. This period is immediately after the first shoot elongation by the resprouts, which depends mostly on root storage, and before the start of the second growth phase, which mainly uses photosynthate to sustain the resprouts (Van Nieuwstadt 2002). For each species, we selected 10–38 stumps of different sizes. We measured the diameter of each stump in two directions at right angles to each other and calculated the basal area by assuming an elliptical cross section. Diameters of all resprouts at their base were measured to the nearest 0.1 mm using callipers. If an individual had two or more stumps, we pooled all the resprouts and calculated the sum of the basal area of the stumps (BAs) and that of resprouts (BAr) for the individual. We accumulated data for 565 individuals between 1995 and 2011.
For each species, we evaluated five functional traits: LMA (g m−2), foliar nitrogen concentration (% w/w), the specific force required to punch through the leaf as an index of leaf toughness (MN m−2), wood density (g cm−3) and seed mass (g). LMA is one of the components of the leaf economics spectrum that relates to photosynthetic rates and leaf longevity (Wright et al. 2004). Foliar nitrogen concentration is positively correlated with leaf photosynthetic rate (Reich et al. 1998). Leaf toughness is positively correlated with leaf life span and with defence against herbivores (Kitajima & Poorter 2010). High wood density is correlated with a low mortality rate (Poorter et al. 2008) because it confers high physical strength (Muller-Landau 2004) and defence against attacks by fungi and other pathogens (Augspurger 1984). Seed mass is negatively correlated with the seed-dispersal distance (Augspurger & Franson 1987) and negatively correlated with the seedling mortality rate (Poorter et al. 2008).
We measured LMA, leaf toughness and the foliar nitrogen concentration for sound mature leaves growing under full sunlight from three individuals per species in July 2010. LMA was determined for 5–10 fresh leaves from each individual. The force to punch through the leaf was estimated as the mean of two punch tests performed with a digital penetrometer (2.0 mm in diameter, model RX-1; Aikoh, Osaka, Japan) on two fresh leaves per individual, avoiding both primary and secondary veins. The leaf toughness was calculated by dividing the force by the leaf lamina thickness, which was measured with a thickness gauge (model 547-401; Mitutoyo, Kawasaki, Japan), avoiding both primary and secondary veins. Foliar nitrogen concentration was determined using an NC analyser (Vario EL III; Elementar, Hanau, Germany). Seed mass and wood density were determined from published sources: for seed mass, we used Nakayama, Inokuchi & Minamitani (2000), Shibata et al. (2010), and Masaki et al. (2011); for wood density, we used Aiba & Nakashizuka (2009). When no published data for wood density were available, we measured wood density in November 2011, following the method of Aiba & Nakashizuka (2009).
To analyse the relationship between resprouting ability and the tree size at first reproduction, we determined the minimum basal area at first reproduction for each species. To do so, we searched for flowering individuals of all studied species in open roadside areas around the study sites and measured the d.b.h. (or the diameter at 10 cm above the ground if they did not reach breast height) when we found individuals with flowers or fruits. We searched a total of 50 km of roadside and forest edge in an area of 1000 ha around the Ogawa Forest Reserve. We summed the basal area of all stems for multi-stemmed individuals and used that value as the total basal area for the individual. We looked for the smallest flowering or fruiting individual that we could find, and stopped looking when we could no longer find any individuals more than 10% smaller than the smallest previously recorded individual.
(eqn 1)
(eqn 2)
Statistical analysis
To compare the resprouting ability among species, we first estimated the relationship between the log10-transformed BAs (the sum of the basal area of the stumps) and the log10-transformed BAr (the sum of the basal area of the resprouts) for each species. Before this transformation, we added 0.0079 to each BAr so that the BAr of non-resprouted individuals could be log-transformed; that is, we converted values of 0 (no resprout) to a diameter of 1 mm. The added value (a diameter of 1 mm) was equivalent to the smallest size of the resprout in all of our samples and was considered to represent a failure of resprouting because such a small amount of resprouting generally leads to the eventual death of individuals (R. Shibata, pers. obs.). According to Del Tredici (2001), resprouting ability starts to decline for some species at a single breakpoint of tree size, so we regressed the relationship between this ability and size using a piecewise regression model to detect the breakpoint for each species (sensu Toms & Lesperance 2003). We then calculated the juvenile resprouting ability (Rjuv), which represents the intercept of the best-fit models by either simple linear regression or piecewise regression; this represents the BAr when BAs is 1 cm2. Because resprouting ability could change with changing stump size, we selected 1 cm2 as the reference stump size for Rjuv to compare the resprouting ability among the species at a constant stump size. We selected this reference size (a basal area of 1 cm2) because it was almost identical to the largest stem size of first-year seedlings of the studied species. We also calculated the maximum stump size at which resprouting ability was retained (Rmax), which equalled the BAs of the breakpoint in the piecewise regression models, and the rate of increase in the resprouting ability (Rrate), which equals the slope of the regression (Fig. 2). To determine the breakpoint in the functions for resprouting ability, we selected the best-fit model based on Akaike's information criterion (AIC); the model with the smallest AIC was considered the best-fit model for each species (Johnson & Omland 2004). We used the package ‘segmented’ in the r software (Muggeo 2011) to perform the piecewise regression.
The value added to BAr before log-transformation could affect our estimate of the resprouting parameters, such as Rjuv, Rmax and Rrate, when there were many non-resprouted individuals. Therefore, we compared the results when we added 0.0079 (basal area of a resprout with a diameter of 1 mm) or 10 times (0.0790) and 0.1 times (0.00079) this number to each BAr before the transformation (Tables S1, S2 and S3 in Supporting Information). We found that the difference in added value did not affect the Rmax estimation for any of the studied species. In addition, the effect of the added value was negligible for the estimates of Rjuv and Rrate for 23 of the 24 species because those species generally had a small number of non-resprouted individuals (0–16%; Fig. S1). We also found that the order of the species in terms of their resprouting ability changed little when we added the different values to BAr. We only found a large number of individuals that did not resprout for Carpinus laxiflora (38.9% of the observed individuals), and this may lead its Rjuv become smaller and its Rrate become larger as we added the smallest value to BAr. However, this did not greatly affect our subsequent analyses or our conclusions (Tables S1, S2 and S3). We therefore used the results for the log-transformed BAr after adding 0.0079 in the rest of the analyses.
We used a generalized linear model with a Gaussian error distribution and an identity link function (Faraway 2006) to investigate the effect of species traits on resprouting ability. We used the three resprouting ability parameters (Rjuv, Rmax and Rrate) as the response valuables, which were multiplied by the corresponding 1/variance value of the regression model as weights for an iteratively reweighted least squares fitting (Table 1). The full models included maximum basal area, minimum basal area at first reproduction, RGR95, LMA, leaf toughness, foliar nitrogen concentration, seed mass, wood density and SDI as the explanatory valuables. However, we did not include maximum basal area, minimum basal area at first reproduction and RGR95 in the same model because they are strongly and significantly correlated (Table 2), and this would lead to spuriously high correlations in the models. We selected the most appropriate model based on the smallest AIC (Johnson & Omland 2004). To identify the variables most strongly related to resprouting ability, we also calculated the Akaike weight for each explanatory variable following the method of Mason et al. (2010).
| BAmaxa | BAreproa | RGR95 | LMA | LT | N | WD | SMa | SDI | |
|---|---|---|---|---|---|---|---|---|---|
| BAmaxa | – | 0.79 (24) | 0.68 (21) | 0.17 (24) | 0.20 (24) | −0.10 (24) | −0.16 (24) | 0.42 (24) | 0.03 (17) |
| BAreproa | 0.68 (23) | – | 0.61 (21) | −0.17 (24) | 0.06 (24) | −0.04 (24) | −0.31 (24) | 0.19 (24) | −0.42 (17) |
| RGR95 | 0.52 (20) | 0.64 (20) | – | −0.19 (21) | −0.24 (21) | 0.36 (21) | −0.67 (21) | 0.09 (21) | 0.40 (17) |
| LMA | 0.08 (23) | −0.16 (23) | −0.34 (20) | – | 0.34 (24) | −0.23 (24) | 0.21 (24) | 0.32 (24) | 0.25 (17) |
| LT | 0.35 (23) | 0.22 (23) | −0.21 (20) | 0.21 (23) | – | −0.48 (24) | 0.40 (24) | 0.18 (24) | −0.20 (17) |
| N | −0.36 (23) | −0.19 (23) | 0.21 (20) | −0.32 (23) | −0.39 (23) | – | −0.37 (24) | 0.04 (24) | 0.41 (17) |
| WD | −0.08 (23) | −0.21 (23) | −0.74 (20) | 0.20 (23) | 0.42 (23) | −0.14 (23) | – | −0.02 (24) | −0.05 (17) |
| SMa | 0.37 (23) | 0.38 (23) | −0.11 (20) | 0.07 (23) | 0.28 (23) | 0.03 (23) | −0.08 (23) | – | 0.55 (17) |
| SDI | −0.33 (16) | −0.36 (16) | 0.32 (16) | 0.18 (16) | −0.18 (16) | 0.25 (16) | −0.07 (16) | 0.33 (16) | – |
- BAmax, maximum basal area; BArepro, minimum basal area at first reproduction; RGR95, 95% maximum relative growth rate in height during the sapling stage; LMA, leaf mass per area; LT, leaf toughness; N, foliar nitrogen concentration; WD, wood density; SM, seed mass; SDI, size distribution index.
- Numbers are Pearson's correlation coefficient (r), with bold values indicating significant correlations (P < 0.05). The number of species or PICs is given in brackets.
- a Log10-transformed.
We calculated Pearson's correlation coefficient (r) between the response and explanatory valuables in the generalized linear model. To increase the generality of our results, we performed phylogenetic analyses using phylogenetic independent contrasts (PICs; Felsenstein 1985), which reduce the risk of spurious correlations arising from evolutionary conservation and phylogenetically biased species selection. The phylogenetic tree for the 24 species used in the PIC calculations was constructed using the Web-based application Phylomatic (Webb & Donoghue 2009; Fig. S2). All branch lengths were set at 1, and polytomy was resolved by inserting very short branches to calculate the PICs. The PICs were calculated using the ape package for the r software (Paradis et al. 2011). All statistical analyses were performed using version 2.12.2 of the r software (R Development Core Team 2011).
Results
Size Dependence of Resprouting Ability for Each Species
All studied species had the ability to resprout, at least when they were small, although three distinct patterns of resprouting ability (BAr) in relation to stump size (BAs) were apparent. Simple linear increase models, in which the resprouting ability increased continuously with increasing stump size, were fit for seven of the 24 species (Figs 3a and S1). This group comprised six shrub species and one sub-canopy tree (Table 1). Piecewise models with a single breakpoint were fit for the other species, but two different patterns appeared. For two sub-canopy species, the piecewise without a decrease model showed that resprouting ability increased with increasing stump size and did not decrease after stump size reached the breakpoint, although the rate of increase in the resprouting ability slowed (Figs 3b and S1). For the other 15 species, most of which were sub-canopy and canopy trees, the piecewise with a decrease model showed that resprouting ability increased with stump size, peaked at the estimated breakpoint (Rmax), and then decreased as stump size increased (i.e. the species lost the ability to resprout at larger stump sizes; Figs 3c and S1).
We used the best-fit linear or piecewise models to estimate the parameters of resprouting ability (Rjuv, Rmax and Rrate; Fig. 2), which differed widely among the studied species (Table 1). For example, Corylus sieboldiana, which is a spontaneously multi-stemmed shrub species that lives mainly at the edges or on the floor of secondary forest, had the highest juvenile resprouting ability (Rjuv) and shared the lowest rate of increase in the resprouting ability with increasing stump size (Rrate), indicating a disproportionately large allocation to resprouting during the juvenile stage. Carpinus tschonoskii, which is a canopy tree in the same family as C. sieboldiana, had the second-smallest juvenile resprouting ability and lost the ability to resprout at a relatively small stump size (i.e. small Rmax), indicating a consistently poor resprouting ability throughout its lifetime; however, it had the highest Rrate.
Relationships Between Resprouting Ability and Other Life-History Traits
The maximum basal area (BAmax), minimum basal area at first reproduction (BArepro) and RGR95 were significantly positively correlated with each other across all species (Table 2; above the diagonal), indicating that larger species tended to grow faster during their juvenile stage and started to reproduce when they reached a relatively large size compared to smaller species. Only one of the functional traits was significantly correlated with the other functional traits across species; leaf toughness was significantly negatively correlated with foliar nitrogen concentration, suggesting that most of the functional traits measured in this study provided at least some ecological information independent of the others. These correlations were mostly unchanged when the phylogeny among species was taken into account using PICs (Table 2; below the diagonal). Seed mass was significantly positively correlated with SDI, although the significance of the correlation disappeared when phylogeny was considered.
The maximum basal area was significantly negatively correlated with juvenile resprouting ability (Table 3); that is, smaller species (i.e. shrub species) showed better resprouting ability during the juvenile stage (Fig. 4a). In the multiple-regression analysis, neither RGR95, minimum basal area at first reproduction, nor the two surrogates of photosynthetic rate (LMA and foliar nitrogen) was significantly correlated with the juvenile resprouting ability (Table 4). Rather, two attributes of mechanical strength (i.e. leaf toughness and wood density) were strongly and significantly negatively correlated with juvenile resprouting ability (Table 4 and Fig. 4), even though the wood density was not significantly correlated with the juvenile resprouting ability in the correlation analysis (Table 3). Wood density was significantly related to the juvenile resprouting ability only when the maximum basal area of the species was considered simultaneously (Tables 3 and 4; Fig. 4). That is, juvenile resprouting ability increased with decreasing leaf toughness and decreasing wood density in both shrub species and tree species, and the regression slopes did not differ significantly between the two life-forms (leaf toughness, F = 0.02, P = 0.89; wood density, F = 0.37, P = 0.55). However, the regression intercepts differed significantly (leaf toughness, F = 27.76, P < 0.01; wood density, F = 21.41, P < 0.01) between shrub and tree species, suggesting that shrub species had better juvenile resprouting ability when shrub species were compared with canopy species with similar values of leaf toughness and wood density (Fig. 4b,c). SDI was significantly positively correlated with juvenile resprouting ability, indicating that species with better juvenile resprouting ability tended to have fewer individuals in small diameter classes than in large diameter classes (Tables 3 and 4).
| Juvenile resprouting ability (Rjuv) | Stump size of maximum resprouting ability (Rmax) | Rate of increase in resprouting ability (Rrate) | ||||
|---|---|---|---|---|---|---|
| r | r-PICs | r | r-PICs | r | r-PICs | |
| BAmaxa | −0.66 (<0.001) | −0.76 (<0.001) | 0.66 (0.008) | 0.71 (0.005) | −0.12 (0.582) | 0.08 (0.722) |
| BAreproa | −0.49 (0.015) | −0.49 (0.017) | 0.14 (0.618) | 0.42 (0.136) | −0.03 (0.903) | 0.06 (0.794) |
| RGR95 | 0.00 (0.991) | 0.05 (0.845) | 0.19 (0.516) | 0.13 (0.669) | −0.02 (0.946) | 0.22 (0.357) |
| LT | −0.65 (<0.001) | −0.62 (0.002) | 0.01 (0.983) | −0.17 (0.561) | 0.31 (0.140) | 0.45 (0.029) |
| LMA | −0.14 (0.515) | −0.08 (0.733) | 0.59 (0.021) | 0.29 (0.314) | −0.40 (0.052) | −0.35 (0.100) |
| N | 0.42 (0.041) | 0.39 (0.066) | 0.10 (0.723) | −0.03 (0.919) | 0.15 (0.497) | 0.13 (0.551) |
| WD | −0.37 (0.071) | −0.41 (0.051) | 0.17 (0.545) | 0.00 (0.990) | 0.11 (0.601) | 0.11 (0.603) |
| SMa | −0.21 (0.327) | −0.21 (0.327) | 0.84 (<0.001) | 0.81 (<0.001) | −0.21 (0.321) | −0.14 (0.510) |
| SDI | 0.51 (0.044) | 0.49 (0.044) | 0.39 (0.169) | 0.24 (0.439) | −0.16 (0.533) | −0.30 (0.261) |
- Values are Pearson's correlation coefficients for untransformed values (r) and for phylogenetic independent contrasts (r-PICs), and boldfaced values represent significant correlations (P < 0.05). P-values are given in brackets. Abbreviations are same as shown in Table 2.
- a Log10-transformed.
| Juvenile resprouting ability (Rjuv) | Stump size of maximum resprouting ability (Rmax) | Rate of increase in resprouting ability (Rrate) | |||||||
|---|---|---|---|---|---|---|---|---|---|
| W i | Coefficients | P | W i | Coefficients | P | W i | Coefficients | P | |
| BAmax (cm2)a | 0.97 | −0.188 (0.048) | 0.002 | 0.77 | 0.634 (0.136) | 0.002 | 0.31 | −0.055 (0.023) | 0.026 |
| BArepro (cm2)a | 0.30 | 0.80 | 0.36 | ||||||
| RGR95 | 0.45 | 0.47 | 0.17 | ||||||
| LT (MN m−2) | 0.98 | −0.198 (0.051) | 0.002 | 0.93 | −0.347 (0.093) | 0.006 | 0.98 | 0.097 (0.063) | 0.141 |
| LMA (g m−2) | 0.66 | 0.005 (0.003) | 0.102 | 0.87 | 0.020 (0.006) | 0.013 | 0.97 | −0.009 (0.003) | 0.011 |
| N (%) | 0.34 | 0.44 | 0.226 (0.191) | 0.271 | 0.70 | 0.166 (0.075) | 0.040 | ||
| WD (g cm−3) | 0.84 | −1.386 (0.404) | 0.006 | 0.74 | −1.519 (0.903) | 0.131 | 0.32 | ||
| SM (g)a | 0.43 | 0.81 | 0.177 (0.079) | 0.055 | 0.37 | ||||
| SDI | 0.48 | 4.220 (1.061) | 0.002 | 0.42 | 0.06 | ||||
| AIC | −22.13 | AIC | 3.26 | AIC | −16.875 | ||||
| R 2 | 0.911 | R 2 | 0.988 | R 2 | 0.585 | ||||
| P | <0.001 | P | <0.001 | P | 0.002 | ||||
- Bold values represent significant variables (P < 0.05).
- a Log10-transformed.
Seed mass and maximum basal area were significantly positively correlated with Rmax, suggesting that species with a larger maximum size and larger seeds tended to maintain their ability to resprout until a relatively large stump size (Table 3). In the multiple-regression analysis (Table 4), leaf toughness was significantly negatively correlated with Rmax, even though it was not significantly correlated with Rmax in the correlation analysis (Table 3). This suggested that species with higher leaf mechanical strength lost their ability to resprout at a relatively small stump size only when the maximum basal area was considered. Although LMA also showed a significant correlation with Rmax, this relationship weakened and became non-significant when phylogeny (PIC) was considered (Table 3). Note that shrub species did not have an Rmax (i.e. they did not lose their ability to resprout throughout their lifetimes, at least within the range of sizes that we examined), so those species were not included in this analysis.
Although none of the species traits was significantly correlated with the rate of increase in the resprouting ability along stump size (Rrate; Table 3), multiple regression showed that Rrate was significantly positively correlated with foliar nitrogen and significantly negatively correlated with BAmax and LMA (Table 4). This suggested that foliar nitrogen was positively correlated with Rrate only when differences in BAmax and LMA were considered. Thus, Rrate was generally higher in shorter species with lower LMA, and within the shorter species with lower LMA, foliar nitrogen was positively related to Rrate.
Discussion
In the Ogawa Forest Reserve, catastrophic disturbances such as wildfire and logging have been eliminated for 100–200 years, and the stands are now shifting towards a stable mature forest. However, the current tree species composition still includes species that are adapted to the past large-scale disturbances. In such an environment, all the studied species had the ability to resprout, at least in the juvenile stage, although this ability varied among species (Table 1). This suggests that resprouting is important to enhance their survival even in the absence of catastrophic disturbance (Hara 1987; Paciorek et al. 2000; Del Tredici 2001). In the following sections, we discuss the size dependence of resprouting ability to relate the ability with other life-history traits in different disturbance regimes and along with light gradients in a stable forest.
Size Dependence of Resprouting Ability
In this study, we successfully characterized the size dependence of the resprouting ability of a species after loss of its above-ground biomass by cutting, and we found that this relationship differed greatly among the coexisting species in the studied forest. For all the studied species, the resprouting ability (BAr) increased in small stems with increasing stump size (BAs) (Figs 3 and S1). Larger stumps should have larger root systems with a large amount of the carbohydrate and nutrient pool (Kabeya & Sakai 2005), which would support the initial growth of the resprouts until photosynthates come from their new leaves. Most of the studied canopy and sub-canopy tree species then exhibited drastically reduced resprouting ability after they attained a specific stump size, and the maximum stump size at which they retained their resprouting ability (Rmax) varied widely among the species (Table 1). The cost of maintaining viable buds and the cost of respiration of large established roots may explain this reduction, but the causal mechanisms are not yet fully understood (Vesk & Westoby 2004; Clarke et al. 2013). Among those species, some species had large individuals with no resprouting, but the others had no such individuals. For example, C. laxiflora lost its resprouting ability at a relatively small size, and we therefore observed many individuals that had grown beyond the size at which they lost their resprouting ability. Some other species such as Castanea crenata, Quercus crispula and Prunus verecunda can retain their resprouting ability at a relatively large size, so we therefore observed few individuals of these species with no resprouting. The relationship between the maximum size (or age) and the size (or age) at which resprouting ability was lost is a key factor in determining the response of a species to disturbance frequency (Kamitani 1986). Trees can regenerate by resprouting if the return interval of the catastrophic disturbance is smaller than the age (the time to reach a given size) at which the species lost their resprouting ability. Thus, species with high Rmax seem to have adapted to less-frequent disturbances. Trees that have grown beyond this size are likely to reproduce by dispersing seeds or other propagules. On the other hand, most shrub species did not exhibit decreased resprouting ability even when they attained their maximum size (Table 1). We also observed no individual shrub without resprouts in this study (Figs 3a and S1). Thus, shrub species tend to die or their main stem is replaced by resprouts before they reach a size at which they lose their resprouting ability.
Resprouting Strategies in Relation to Disturbance Regimes
In support of our hypothesis, we found that species with a smaller maximum size had better resprouting ability than species with larger maximum size during their juvenile stage (Tables 3 and 4, Fig. 4). Our results also suggested that species with good resprouting ability had limited seedling recruitment (i.e. high SDI; Tables 3 and 4), which agrees with the results of Nzunda & Lawes (2011). In addition, most shrub species did not exhibit decreased resprouting ability throughout the size range we observed (Table 1). Thus, the good resprouting ability of shrub species throughout their lifetimes would be advantageous under an environment with frequent small disturbance, but would occur at the expense of maximum height. This is because the resources allocated to many stems so as to maintain a bank of viable buds and sufficient resource storage to permit resprouting would be unavailable to support height growth (Midgley 1996). Many shrub species resprout without a disturbance, and naturally become multi-stemmed (Hara et al. 2004; Fujiki & Kikuzawa 2006). Indeed, most of the studied shrub species developed a multi-stemmed architecture even without a disturbance. This architecture, with good resprouting ability after the loss of above-ground plant parts, is likely an adaptation by understorey shrubs to increase survival and enhances their lifetime fitness under a shaded forest floor (Tanentzap et al. 2012), where they are frequently disturbed by falling debris (Clark & Clark 1991). Aralia elata, which is a single-stemmed light-demanding shrub species, was an exception, as it lives exclusively at open sites such as clearings. This species can regenerate after fire both by resprouting and by seed (Goto et al. 1996). In our study, A. elata had a good resprouting ability in the juvenile stage, but exhibited an exceptional reduction in its ability to resprout during the adult stage (Table 1). The combination of fast growth, a short life span, and good resprouting ability at an early life stage would be especially advantageous in a frequently disturbed environment.
In addition to the juvenile resprouting ability, we were able to relate the maximum size at which resprouting ability was retained to other life-history traits. Our results showed that the species with lower leaf toughness tended to retain their resprouting ability until a relatively large stump size (Tables 3 and 4). This suggested that this maximum size was generally higher in species with a larger maximum basal area, and that species with lower leaf toughness within a specific size class tended to retain their resprouting ability at a relatively large size. The lower leaf toughness could be associated with light-demanding species, and therefore, in consistent with our hypothesis, light-demanding species may retain their resprouting ability until a larger size than shade-tolerant species. For example, C. crenata and P. verecunda resprouted at stump diameters >20 cm, and Q. crispula resprouted at a stump diameter >30 cm (Table 1). These three species are light-demanding (but long-lived) species with limited seedling recruitment (i.e. large SDI), and are thought to be recruited after large-scale disturbances such as fire and logging (Masaki 2002). Clearly, resprouting of mature individuals is advantageous for a species to persist after a catastrophic disturbance. However, further investigation is needed to reveal how much resprouting of mature trees accounts for the filling of treefall gaps in a stable forest community.
Resprouting Strategy in Relation to Shade Tolerance
Theoretical studies predict a trade-off between relative growth rate and resprouting ability, because species with better resprouting ability allocate more biomass to root storage than to shoot growth (Bellingham & Sparrow 2000; Bond & Midgley 2001). In contrast to our hypothesis, we found no significant relationship between the RGR95 of juveniles and the resprouting parameters (Tables 3 and 4). Some studies in fire-prone ecosystems also showed that the growth rate did not differ between species with poor and good resprouting ability (Bell 2001; Knox & Clarke 2005). This could be because species with good resprouting ability had leaf traits that were optimized to enhance the photosynthetic rate (e.g. lower LMA or higher foliar nitrogen; Reich et al. 1998; Wright et al. 2004), which would allow them to allocate more photosynthate to both root storage and above-ground growth (Paula & Pausas 2006; Hernández, Pausas & Vilagrosa 2011). However, LMA and foliar nitrogen concentration were not strongly associated with the juvenile resprouting ability in this study (Tables 3 and 4). The independence of juvenile resprouting ability from LMA and foliar nitrogen suggests that variations in photosynthetic rate are not a likely explanation for the interspecific variation in resprouting ability in the juvenile stage. On the other hand, our study showed that species with better resprouting ability during the juvenile stage tended to have lower wood density and leaf toughness, indicating lower construction costs for their shoots and leaves, and this might allow higher allocation of photosynthates to root storage. The theoretical trade-off between growth and resprouting assumes that the carbon gain or the cost for resprouting does not differ between the species with good and poor resprouting ability. However, the carbon gain or the cost should actually differ between those species depending on other life-history strategies, which makes the relationship between growth rate and resprouting ability unclear.
Our results instead suggested a trade-off between resprouting ability and defences. The juvenile resprouting ability was generally poorer for species with higher leaf toughness and wood density (Fig. 4; Tables 3 and 4), which are associated with stronger physical defences, resulting in a higher survival rate of juveniles (Poorter et al. 2008; Kitajima & Poorter 2010). This indicated that resprouting would be less important for species with higher leaf toughness and wood density, because juveniles of these species can tolerate these kinds of damage due to their stronger physical defences. In contrast, species with lower leaf toughness and wood density had higher juvenile resprouting ability, possibly to compensate for their weaker physical defences. Thus, in contrast to our hypothesis, there could be a trade-off between juvenile resprouting ability and defensive investment, possibly because simultaneously increasing the resource allocation to both strategies would be difficult.
Our result was inconsistent with Poorter et al. (2010), which illustrated that slow-growing shade-tolerant species with higher wood density had better resprouting ability than fast-growing light-demanding species particularly in tropical moist forest. Because Poorter et al. (2010) measured the survival and growth of the resprouts after the stem clipping, their resprouting ability included ‘the ability to resprout’ and ‘the survival of the resprouts’ (Moreira, Tormo & Pausas 2012). The former can be affected by the resource storage; on the other hand, the latter by physical defences with tougher leaves or denser wood. Thus, their result may reflect higher survival of the resprouts in shade-tolerant species with denser wood than the light-demanding species in a shaded understorey.
On the other hand, we measured the ability to resprout after the stem clipping in an open area, and showed that species with weaker physical defences had better resprouting ability. Such species could have greater storage of resources due to a higher photosynthetic rate or could produce resprouts at a lower cost. The latter explanation appears to be more plausible because of the lack of relationship between resource storage and physical defences for our studied species (R. Shibata, H. Kurokawa, T. Nakashizuka, unpubl. data). In our study, Acanthopanax sciadophylloides had the lowest wood density, but their SDI was small. This indicated that this species has stable recruitment in the shaded understorey, and there were actually many juveniles of this species with resprouts (Hara 1987; R. Shibata, pers. obs.). Individuals of this species therefore can prolong their life span by repeated resprouting in the shaded understorey, like the shrub species (Hara et al. 2004; Fujiki & Kikuzawa 2006). Previous studies in temperate forests showed that large parts of the canopy gaps created by windthrow were filled by resprouting of damaged individuals as well as by seedling recruitment (Cooper-Ellis et al. 1999; Dietze & Clark 2008). Rapid regrowth of damaged individuals by means of vigorous resprouting is a strategy that would enable species with a small maximum size or low mechanical strength to increase their survival under canopy and chance of filling the canopy gaps. The light-demanding fast-growing species with lower physical defences would therefore potentially have better resprouting ability than shade-tolerant slow-growing species, but the resprouting ability measured in a shaded understorey often seems to be higher in shade-tolerant species because light-demanding species cannot persist due to lower physical defences in a shaded condition. Thus, the shade tolerance of a species can be achieved by both a combination of higher persistence as a result of higher physical defences and recovery from damage by a better resprouting ability. Some species could increase their shade tolerance by means of a high resprouting ability even if they have low leaf toughness and wood density.
Furthermore, the inconsistency between our results and those of Poorter et al. (2010) suggests that the relationship between resprouting ability and other strategies along a shade-tolerance gradient may be somewhat ecosystem-dependent. As suggested by Poorter et al. (2010), the moist evergreen forest can cast a deep and more persistent shade than the deciduous temperate forest, or the tropical dry forest, which could provide a steep growth-survival trade-off along with the gradient of shade tolerance. Compared to the moist forest, Poorter et al. (2010) found a weaker positive relationship between resprouting ability and wood density in the dry tropical forests of their study. Moreover, being inconsistent with the result from tropical forests, our study species in temperate forest did not show the significant relationship between wood density and shade tolerance. Thus, wood density may be more related to other strategies, such as drought tolerance, than the shade tolerance in our systems. Furthermore, although we did not consider differences in site productivity in the present study, this could also affect the resprouting ability of a species (Clarke et al. 2005). Further research is needed to determine how various environmental factors influence the relationship between resprouting and other traits (Salk & McMahon 2011).
Acknowledgements
We thank Dr. S. Abe and other members of the Forestry and Forest Products Research Institute (FFPRI) for the opportunity to conduct research in the Ogawa Forest Reserve and for their advice in the field; Mr. S. Ishida and M. Uno from Tohoku University for helping to collect the resprouting data; Prof. K. Hikosaka from Tohoku University for detailed comments and suggestions on an early version of this article; Drs. S. Sakai, T. Sasaki, M. Aiba and T. Itagaki from Tohoku University for providing valuable suggestion throughout our study; and all researchers who were studying in the Ogawa Forest Reserve for helping in the field and in the laboratory work. We also thank an associate editor, Dr. Peter Klinkhamer, and two anonymous reviewers for providing valuable comments to improve our manuscript. This study was supported by a Grant-in-Aid for Scientific Research (B) (2337007), by the Environment Research and Technology Development Fund (S-9-3) of the Ministry of the Environment, Japan, by Research Funds for the National Organizations for Pollution Prevention of the Ministry of the Environment, Japan, and by Research grant of the FFPRI, Japan.
