Competition in the regeneration niche between conifers and angiosperms: Bond's slow seedling hypothesis

What features of angiosperms have enabled them to dominate the biomass of much temperate and tropical vegetation despite the high productivity of many adult gymnosperms, especially conifers? Bond (1989) argued that slow-growing gymnosperm juveniles are poor competitors with angiosperms regenerating in forest gaps and other well lit, well watered habitats. Consequently, gymnosperms are restricted to areas where growth of angiosperm competitors is reduced, for example by cold or nutrient shortages. A historical implication of Bond's hypothesis is that angiosperms overran gymnosperms during the Cretaceous and Tertiary by invading their regeneration niche. Bond's (1989)‘slow seedling’ hypothesis has considerable appeal because it seeks to explain major phytogeographical patterns and evolutionary trends in terms of observable vegetative traits. Data suitable for testing Bond's plausible arguments were then quite limited, and what was proposed as hypothesis has often been cited as fact in recent reviews of conifer biology (Enright & Hill 1995; Richardson 1998; Smith & Hinckley 1995). Studies of large-scale trends may be exploratory (generating hypotheses) or sceptical (testing hypotheses); both modes of inquiry are essential and indeed complementary (McShea 1998). This review emphasizes the sceptical mode, and aims to test some of the main premises and generalizations of Bond's (1989) slow seedling hypothesis. Although Bond's arguments embraced the polyphyletic gymnosperms, relevant data are mostly limited to the monophyletic conifers, also constraining this review. A key element of Bond's (1989) hypothesis is that conifer seedlings cannot match the fastest growth rates of angiosperms. This is supported here for early-succession angiosperms, but not for late-succession, north-temperate trees. Few comparative data are available from regions where both angiosperms and conifers have long-lived leaves, a trait that is correlated with reduced seedling growth rate (Reich 1998). Whether Bond's hypothesis can explain the rarity of conifers and gymnosperms in lowland tropical forests therefore remains uncertain. Bond (1989) attributed the slow growth of juvenile conifers to an inefficient transport system and a low maximum leaf photosynthetic rate, which together prevent conifers from achieving high productivity until sufficient leaf area accumulates. Here a review of hydraulic conductances of temperate and tropical trees suggests that conifers and angiosperms have similarly adequate transport capabilities at the whole-plant level, despite the simple leaf venation and narrow xylem conduits of conifers. Data for temperate trees show that both deciduous and evergreen angiosperms have higher maximum photosynthetic rates (mass basis) than conifers. This difference disappears after accounting for leaf longevity, suggesting that conifers and angiosperms do not differ in their biochemical efficiency of energy conversion, and leaving uncertain the relative photosynthetic capacities of long-lived, tropical angiosperm and gymnosperm leaves. An oft-repeated generalization is that conifer-dominated vegetation is mainly restricted to nutrient-poor or wet soils, or cold areas, which Bond (1989) interpreted as the outcome of the competitive inferiority of gymnosperms in the regeneration niche. It is useful to distinguish between situations where a species is dominant because of traits for tolerating the environment as opposed to traits for suppressing neighbours; the former type of dominance could occur in the absence of competition (Keddy 1989). The tendency of conifers to dominate at high altitudes and latitudes has numerous exceptions (Sprugel 1989) and could be explained by tracheids being more resistant than vessels to freezing-induced embolization (Sperry & Sullivan 1992; Wang, Ives & Lechowicz 1992). Occupation of wet soils by trees is strongly linked to tolerance of anoxia (Kramer & Boyer 1995). A comprehensive understanding of the biogeography of angiosperms and gymnosperms will need to consider their entire distributions. In the absence of experiments to separate the effects of tolerance and competition, however, this review is restricted to biogeographical patterns on uplands at moderate elevations and latitudes where Bond's (1989) arguments are most likely to apply. The supposed association of conifers with poor (sandy) soils is shown to be an oversimplification, in at least one case, by quantitative analysis of presettlement forests and soil fertility in a northern cold-temperate region. Although the impression of angiosperm dominance on productive sites still requires confirmation, it is likely that there is a probability gradient of angiosperm dominance paralleling gradients of site productivity. Bond (1989) emphasized that the angiosperm innovation of the herbaceous habit enabled more effective exploitation of ephemeral habitats, including regeneration gaps in forests, than the woody habit which all gymnosperms possess. Being short-lived, such habitats account for only a small proportion of extant vegetation. In all studies, tree seedlings have slower net assimilation rates and relative growth rates (RGR) than non-woody species (Grime & Hunt 1975; Hunt & Cornelissen 1997a; Poorter 1989). With time, however, perennials become more effective competitors as they accumulate reserves and a deeper root system (Poorter 1989; Schulze 1982). If disturbance does not eliminate competition from trees, whether angiosperms or conifers, their longevity and height enable them to shade and eventually kill herbs and shrubs (Bazzaz 1996; Scholes & Archer 1997; Schulze 1982). Plants with more access to light also construct more roots and are better able to forage for nutrients, so simple traits such as height and crown diameter substantially account for measured competitive ability (Keddy 1989). For these ecological reasons, and because it is uncertain whether the absence of gymnosperm herbs represents a chance failure to evolve or constraints imposed by the gymnosperm bauplan, comparison of growth rates between conifers and angiosperms is here restricted to shared life forms. The largest suitable data set (Cornelissen, Castro Diez & Hunt 1996) showed that RGRs of seedlings of four conifer genera (and some angiosperms) were less than the lower regression confidence limit estimated for 16 angiosperm tree genera (Fig. 1). Correction for initial seed mass by regression analysis is justified by its inverse relationship to RGR (Grubb 1998; Reich et al. 1998b), but the outcome of competitive interactions will be determined by the RGRs of the actual contestants, regardless of their seed size. There was no significant difference between angiosperms and conifers in a straightforward comparison of their RGRs (pooled variance t = 1·41, P = 0·18). Relationship between mean relative growth rate (sensu stricto) of temperate angiosperm (○) and conifer (▴) tree seedlings and (ln-transformed) mean seed dry mass. Both variables were averaged by genus when more than one species was measured. A least-squares, linear regression (solid line) and 95% confidence interval (dashed lines) were fitted to the angiosperm data only. Most angiosperm species were deciduous and most conifers were evergreen. The slowest-growing angiosperms (RGR < 0·1 day−1) were mid- and late-succession species (Andrew Cameron, personal communication). Data from Appendix 1 of Cornelissen et al. (1996). Consistent with Bond's (1989) generalizations, only angiosperms achieved the fastest RGRs (Fig. 1). However, all seven trees (Fraxinus excelsior, Malus sylvestris, Alnus glutinosa, Sorbus aucuparia, Ulnus glabra, Betula pendula and Salix caprea) that exceeded the maximum RGR of conifers are species of early succession (Andrew Cameron, personal communication). If these results are representative, then Bond's (1989) arguments could explain the dominance of angiosperms in early-succession vegetation. For example, Shainsky & Radosevich (1992) demonstrated experimentally that the initial height advantage of the early succession species Alnus rubra (Red Alder), sustained by its faster height growth rate, was probably the most important factor determining its competitive superiority to Pseudotsuga menziesii (Douglas Fir). However, Bond's arguments might be less relevant to late-succession stages, even on productive sites, where canopy species of both angiosperms and conifers have slower seedling RGRs (Fig. 1). Bond's (1989) arguments comparing the competitive abilities and ecophysiological characteristics of conifers and angiosperms parallel those developed previously for deciduous and evergreen angiosperms (Goldberg 1982 and references therein). He would contend that deciduousness is difficult for gymnosperms to evolve and maintain because of their poor transport capacity (Bond 1989 and personal communication). The ecophysiological and morphological traits that affect RGR are strongly associated with leaf life span, which in turn is partly controlled by the length of the growing season (Eamus 1999; Kikuzawa 1995; Reich 1998). Incorporation of seasonal effects into a cost–benefit model of leaf longevity predicted a bimodal geographical (latitude and altitude) distribution of evergreenness (Kikuzawa 1995). This matches the distribution of conifers, except at low latitude where evergreen angiosperms strongly dominate (Reich et al. 1995) – a biogeographical pattern which it is important to explain. Deciduous woody species grow more quickly than evergreen ones (Cornelissen et al. 1996; Reich 1998), but the effect of leaf habit could not be tested with the data in Fig. 1 because nearly all angiosperms were deciduous and nearly all conifers were evergreen. This partly indicates sampling bias, but is more a reflection of real phenological differences between conifers and angiosperms in cold-temperate regions. After accounting for differences in leaf life span by regression analysis, height-growth rates of temperate conifer and angiosperm saplings were similar (Reich et al. 1995). The RGRs of temperate conifer seedlings tended to be greater than those of tropical evergreen angiosperms (Reich 1998). In a comparison of evergreen seedlings of the main canopy species of a Tasmanian cool-temperate rain forest, the three gymnosperms had slower RGRs than the three angiosperms (Read 1985). Existing data do not permit generalization about growth-rate differences between gymnosperm and angiosperm seedlings in the tropics, where both taxa are evergreen. RGR is the product of a physiological component, net assimilation rate, and a morphological component, leaf area ratio (LAR = total leaf area/total plant mass). Various studies have indicated that allocation is substantially more important than physiology in explaining variation in growth rate on an area basis, but this was reversed in mass-based growth analyses (Hunt & Cornelissen 1997b; Reich 1998; Reich et al. 1998c). Specific leaf area (SLA = total leaf area/total leaf dry mass) was implicated as the principal means for adaptive changes in RGR (Hunt & Cornelissen 1997b). The trade-off for increasing RGR by increasing SLA is reduced leaf longevity and increased leaf palatability, which may be tolerable only in productive environments. Leaf life span tends to be longer and to control seedling RGR more strongly in unproductive habitats (Aerts 1995; Reich 1998). The RGRs of seedlings of deciduous angiosperms were significantly greater than those of conifers, even at comparable LAR or its component, SLA, both of which tended to be smaller in conifers (Reich 1998). The maximum leaf photosynthetic rate of temperate conifers was less than that of angiosperms on an area and a mass basis for deciduous species, and on a mass basis for evergreen species (Fig. 2). Generally, the ranking of mass-based net photosynthesis among functional groups was consistent for seedlings and mature plants in the field (Reich 1998), so this large database for saplings and adults may be indicative of trends for seedlings. Greenhouse-grown seedlings of temperate deciduous angiosperms (three species) had higher photosynthetic rates than evergreen conifers (five species) on a mass basis, but not on an area basis (Reich et al. 1998c). Glasshouse seedlings of four angiosperms had mass-based, maximum photosynthetic rates double those of three codominant gymnosperms from Tasmanian temperate rain forest (Read & Busby 1990). There is a consistently stronger relationship between RGR and maximum photosynthetic rate on a mass basis than on an area basis (Reich et al. 1998c), suggesting that mass-based gas-exchange rates are more relevant to the competitive ability of seedlings in the context of Bond's (1989) arguments. Maximum net photosynthetic rate of temperate angiosperms and conifers on projected leaf-area and leaf-mass bases for deciduous (leaf life span < 1 year) and evergreen (leaf life span ≥ 1 year) trees. Fully expanded, young- to medium-aged ‘sun’ leaves growing on juvenile and adult plants (2–15 m tall) were measured. The number of species in each sample is indicated below the lower axes. Box plots show median (central horizontal line), interquartile range (box) and range (whiskers) with outlying values depicted as asterisks. Probability values are for separate variance t-tests of ln-transformed photosynthetic rates. Data from Reich et al. (1998a) and Peter Reich (personal communication), averaged when measurements were made on the same species at different sites. The age classes in Fig. 2 are probably still too coarse to fully account for effects of leaf longevity because (i) so many temperate angiosperms have very short leaf life spans and so many conifers have very long leaf life spans; and (ii) maximum photosynthetic rate declines with leaf life span (data not shown), partly because long-lived leaves tend to have smaller SLA and percentage N (Reich et al. 1999). For the range of leaf life spans (15–30 months) having a similar and substantial number (10) of both conifer and angiosperm species, there was no significant difference (pooled variance t-tests, P > 0·85) between taxa in maximum photosynthetic rate on either area or mass bases (Reich, Ellsworth & Walters 1998a, data not shown). This suggests that conifer and angiosperm leaves of similar longevity do not differ in their biochemical efficiency of energy conversion. As in the case of RGR, available data do not permit extrapolation to tropical systems, where both angiosperms and gymnosperms have long-lived leaves. Bond (1989) argued that anastamosing vein systems enable faster solute transport and thus more rapid expansion of angiosperm leaves compared with gymnosperm leaves which have open venation. Hydraulic conductance is the ratio of water-flow rate to the pressure differential causing flow, and is a measure of transport sufficiency (when normalized by leaf area). Computer simulations have estimated lower overall conductance in leaves with open, dichotomous venation compared with those having closed, anastamosing venation but an identical volume ratio of parenchyma and xylem (Anita Roth-Nebelsick, Dieter Uhl & Volker Mosbrugger, unpublished results). However, there is no empirical evidence that the hydraulic conductance of leaves depends on the complexity of venation, including the presence of tertiary venation (Becker, Tyree & Tsuda 1999a; Tyree et al. 1999). In ferns, reticulate venation is generally viewed as a strengthening trait because it is not correlated with the transpiration demands of different habitats (Wagner 1979). Specialized transfusion tissue in conifer needles, and large flow resistances in the finest veins and non-vascular tissue of angiosperm leaves, may tend to equalize leaf hydraulic conductances in these taxa (Becker et al. 1999a). Furthermore, reduced sap-flow rates do not necessarily imply an insufficient supply of nutrients to the leaves because so-called ‘futile’ nutrient cycles, such as those involving N and Mg, deliver nutrients at rates far in excess of the demands for plant growth (Stitt & Schulze 1994). Finally, closed networks require more vascular tissue than open networks to adequately supply equivalent volumes of leaf parenchyma (Kull & Herbig 1994; Roth et al. 1995), and this extra energetic expense could negatively affect leaf expansion rates. Bond (1989) also supposed that the conducting system of conifers was less efficient than that of angiosperms due to the greater resistance of tracheids compared with vessels. This was supported by measurements on stem segments (Becker et al. 1999a; Wang et al. 1992), which unfortunately are not a surrogate for whole-plant measurements. Leaf-area normalized, whole-plant hydraulic conductances of conifers and angiosperms were statistically indistinguishable (Becker et al. 1999a) for both tropical sapling-sized plants and temperate adult trees (Fig. 3). Differences in xylem anatomy between conifers and angiosperms manifested at the branch level are apparently compensated to equalize the sufficiency of water supply to leaves. This is consistent with the observation that the maximum leaf diffusive conductances of temperate conifers and deciduous angiosperm trees did not differ significantly when compared on the same leaf-area basis (Becker et al. 1999a). Despite having the lowest vein density, the gymnosperm Gingko biloba had the second fastest maximum transpiration rate (leaf-area basis) compared with 11 deciduous angiosperms (Kull & Herbig 1994). Whole-plant hydraulic conductance (dots depict species means) of angiosperms and conifers, normalized by leaf area for (a) tropical sapling-sized plants growing on N-poor white sand and (b) temperate adult trees. The three largest values in (a) for evergreen angiosperm saplings correspond to colonizer species that did not compete with the broad-leaved conifers. Values in (a) were measured with a high-pressure flow meter; those in (b) were estimated from the relationship between transpiration rate and xylem water potential. Box plots as in Fig. 2. Data from Tables 3 and 4 (excluding the suspiciously large value for Betula alba) of Becker et al. (1999a). One obvious compensatory mechanism, incorporated in the formal plant hydraulic model of Whitehead, Edwards & Jarvis (1984), is to adjust the area ratio of leaf : sapwood cross-section. This was significantly larger in first-year shoots of deciduous temperate plants than in those of evergreens, but ratios of evergreen angiosperms and conifers were statistically indistinguishable (Fig. 4). The latter result strongly suggests that differences in water-transport efficiency do not entirely account for the differences in leaf : twig area ratios. Compared with deciduous leaves, evergreen leaves require more mechanical support because of their higher mass density, and greater redundancy of transport tissue to compensate embolism-induced loss of conductance capacity during their longer lifetimes. Both of these requirements could lead to increased wood cross-sectional area (Brouat et al. 1998). Leaf : twig area ratios for first-year shoots of temperate trees and shrubs from open, mesic habitats in the Great Smokey Mountains of Tennessee, USA, to compare deciduous and evergreen angiosperms and evergreen conifers. Twig cross-sectional area was measured over bark but otherwise represented mostly sapwood. Leaf area per shoot was estimated as the product of average leaf area and number of leaves per shoot. Plot conventions as in Fig. 2. Leaf : twig area ratios of evergreen species did not differ significantly between seven angiosperms and four conifers (Mann–Whitney U = 16, P = 0·71) but were significantly smaller (U = 209, P = 0·00001) than those of 19 deciduous angiosperm species. Data from White (1983a,b). Allometric analysis of Sitka Spruce (Picea sitchensis) branches suggested that their diameters were more related to the requirements for mechanical load than to hydraulic transport (Farnsworth & Van Gardingen 1995), but such relationships may be size-dependent (Bertram 1989). Additional allocation to sapwood by evergreens could reduce their potential growth rate, depending on the degree of C limitation. Quantitative studies are necessary to resolve this issue, but clearly a whole-plant perspective is essential (Becker et al. 1999a). The hypothesized inferior competitive ability of gymnosperms (Bond 1989) may have been exacerbated by atmospheric changes in geological time. A large decline in atmospheric CO2 concentration, [CO2], was probably a consequence of enhanced silicate-weathering rates associated with the evolution of land plants, with faster weathering assumed after the appearance of angiosperms (Berner 1994). Evidence is that gymnosperms (and other plant are to their of and poor that when atmospheric CO2 was more than at & argued that angiosperms & were better able than gymnosperms to adjust and physiology to the of [CO2], and that this is in the higher per water of angiosperms. most of the Cretaceous & as the number of angiosperm genera but the number of genera during this & the evidence is a competitive of conifers. However, Bond (1989) that is not with ecological Consistent with forests less land and herbaceous vegetation during the Maximum and the when was lower than during the et al. 1994; A recent of CO2 effects on net CO2 assimilation and biomass showed no significant difference between conifers and angiosperms, but the of these was and effects may have been with effects & Wang 1998). Bond (1989) suggested that conifers were restricted to sites, or that assumed to have less competition in the regeneration niche. Various that competition with remains as productivity or is on the ratio of supply to which may be to productivity & and references therein). to the effects of all the ability of a or an to a of limited (Keddy 1989). studies, mostly of non-woody plants, do not permit generalizations about the relationship of total and competition to productivity & of the of competition in woody vegetation productivity gradients are to competition species dominance on unproductive sites, and whether or growth is then the more important to measure (Goldberg & 1997; 1998; The traits that competitive superiority differ and conifers not so much where growth rates are slow as where they are less important in determining the outcome of competition (Aerts & & & Poorter 1992; et al. & assumed that species of had faster growth rates at low supply than species from environments. However, species from species from at both high and low & & 1995; & Poorter 1992; Walters & Ellsworth 1992; et al. but Walters & Reich In where nutrients limited growth rate, species achieved greater biomass and species with faster nutrient loss (Aerts 1995; & 1994). assumed a trade-off between plant traits associated with nutrient and productivity area or N depending on the is of evergreens with long-lived leaves having low nutrient and gas-exchange rates when is However, that plants that strongly reduce their rate when are in short supply might show some nutrient into empirical studies of competition requires of rates. This will be by the likely inverse relationship between the of loss and the actual gradient (Keddy 1989). Bond (1989) that shade would a to conifers by the growth-rate advantage of angiosperms in the same as nutrient differences in RGR do with both temperate and tropical species, there is an inverse between growth rate of well seedlings and their in shade 1996; et al. 1995). Bond (1989) predicted that conifers would angiosperm competitors in forests where gaps are small and saplings are In the forests of however, was to plant species because interactions among juveniles in shade long a et al. 1990). studies in the temperate rain forest of that do not to be a major evolutionary for conifers, because species maintain large seedling closed forest Bond (personal that to be a generalization and it to the of broad-leaved conifers in as demonstrated for species & Hill 1999). It remains to be whether conifers are generally as by Smith & There was no quantitative to support Bond's (1989) generalization and conifers occur on nutrient-poor sites, especially or can be conifers dominate or with angiosperms on soils in et al. 1998; and the & 1995; & & and as long-lived in the personal communication). exceptions to a hypothesis are to be in systems & Bond 1989) so their be indicates that the impression of on north-temperate sites & is an oversimplification, especially when forests conifers are also forest and of soils of of forest soils (Fig. and were probably for this association between vegetation and soils & & 1996; et al. 1999; & of (a) forest soils by lines) and (b) angiosperm forest and on soils in to mesic forest with angiosperms and most of the northern of the soils is not are vegetation in areas by lines) of soils which some For areas of water and are not The are After et al. and & In early deciduous trees conifers on soils in about the of soils over the et al. If vegetation generally so much more quickly than the soil it considerable in may be large areas of forest are not by nutrient-poor despite the of their association & 1998). The fertility of soils by conifers be than to be In USA, either of the species or on were associated with soils &