Early vs. asymptotic growth responses of herbaceous plants to elevated CO2.
Atmospheric carbon dioxide levels continue to increase at a rate of [approximately]0.45% per year, with the current global average exceeding 360 [[micro]liter]/L (IPCC 1996). The most immediate and well-characterized physiological effects of elevated C[O.sub.2] on plants are increased carbon uptake and reduced leaf-level transpiration (e.g., Bazzaz 1990, Bowes 1991, Curtis 1996). Physiological enhancement of carbon gain generally results in increased biomass accumulation over the short term. Accordingly, plant growth enhancements under elevated C[O.sub.2] have been extensively documented (Kimball 1983, Cure and Acock 1986, Poorter 1993, Poorter et al. 1996, Curtis and Wang 1998). The most commonly cited generalization to emerge from this work is that plants grown in a doubled C[O.sub.2] atmosphere (i.e., 700 vs. 350 [[micro]liter]/L) exhibit, on average, a 33-37% increase in biomass (e.g., Kimball 1983, Wullschleger et al. 1995).
Several reviews have recently argued that this 3337% figure, and enhancement ratios in general, do not adequately summarize C[O.sub.2] effects on plant growth (Loehle 1995, Gifford et al. 1996, Korner 1996, Thomas and Jasienski 1996). The essence of this argument, as we see it, is as follows: It is axiomatic that plant growth curves are generally sigmoidal in form, with an early exponential phase followed by a leveling off later in ontogeny (Causton and Venus 1981). A sigmoid growth function (in contrast to, for example, exponential or linear growth functions) must be described by at least two parameters. It follows that effects of C[O.sub.2] on plant "growth" must also be described by at least two parameters. Most commonly, sigmoid growth functions, such as the logistic equation, are described by an initial exponential parameter describing early growth, and an asymptotic maximal size parameter describing final plant size. Analyses of C[O.sub.2] effects on plant growth should thus also explicitly examine effects on these parameters. Enhancement ratios, as conventionally employed, thus obscure two distinct kinds of effects, both of which are of fundamental importance and interest [ILLUSTRATION FOR FIGURE 1 OMITTED].
Numerous studies examining plant growth responses to elevated C[O.sub.2] have employed one form or another of growth analysis (e.g., Rogers et al. 1984, Tolley and Strain 1984, Bazzaz et al. 1989, Coleman and Bazzaz 1992, Walters et al. 1993, Tremmel and Patterson 1994). There has also been a long-standing debate as to whether C[O.sub.2] effects on tree seedling growth will persist in larger trees and whole forests (e.g., Norby et al. 1992, 1996, Bazzaz et al. 1993, Johnson et al. 1996, Hattenschwiler et al. 1997, Idso and Kimball 1997). However, we are aware of no previous attempt to explicitly quantify C[O.sub.2] effects on asymptotic plant size per se. Nor has there been any empirical examination of issues that immediately arise from this approach, such as potential relationships between early and asymptotic growth responses. It seems likely that the physiological mechanisms responsible for differences in asymptotic size responses to C[O.sub.2] are fundamentally distinct from those responsible for early growth enhancements. For example, the physiological determinants of maximum plant size likely include ontogenetic shifts in allocation to reproductive structures (e.g., Chiariello and Roughgarden 1984, Thomas 1996a), biomechanical limitations (e.g., McMahon 1973, Givnish 1988), ontogenetic changes in photosynthetic physiology and water relations (Friend 1993, Yoder et al. 1994, Thomas and Bazzaz 1999), and, in at least some cases, meristem limitation (e.g., Geber 1990). Potential effects of elevated C[O.sub.2] on these and other processes may be due to direct or indirect interactions with hormonal regulation of plant growth (cf. Reekie 1996). Hormonal interactions may, for example, help explain accelerated rates of maturation under elevated C[O.sub.2] observed in some species, including effects on heteroblastic leaf development (Thomas and Bazzaz 1996), reproductive onset (Farnsworth et al. 1996), and whole-plant senescence (St. Omer and Horvath 1983).
The present study investigates early vs. asymptotic plant growth responses on the basis of a controlled-environment experiment examining C[O.sub.2] responses of eight herbaceous plants characteristic of recently disturbed sites on the Atlantic seaboard of North America. One of these species is replicated at the genotype level, allowing comparisons of intra- vs. interspecific patterns of variation in these variables. Each species was grown individually and in dense monocultures, enabling comparisons of individual plant vs. stand-level responses. The primary questions we address are: (1) Do the plant species in question exhibit increased asymptotic size under elevated C[O.sub.2]? (2) If so, is there a relationship between early growth responses and asymptotic size responses? (3) How do asymptotic size responses compare among growth metrics, such as biomass components and leaf area? (4) How do early and asymptotic responses of dense monocultures compare to responses of individually grown plants? and (5) Are early and asymptotic growth response patterns predictable in terms of growth form or other plant characteristics?
MATERIALS AND METHODS
Plant species and growth conditions. - Species chosen for study include widespread annuals and short-lived perennials characteristic of waste areas in the eastern United States. These include two species of erect annual plants (Abutilon theophrasti Medic. and Cassia obtusifolia L.), three short-lived perennial rosette species (Plantago major L., Rumex crispus L., and Taraxacum officinale Weber), and three grass species (Dactylis glomerata L., Lolium multiflorum Lam., and Panicum dichotomoflorum Michx.). A third erect annual species, Ambrosia artemesiifolia L., was also initially included in the experimental design, but performed very poorly under all treatments, and was therefore excluded from the final analyses. Plantago and Rumex seeds were collected locally in Cambridge, Massachusetts. Seed of Cassia, Taraxacum, Dactylis, Lolium, and Panicum was obtained through a commercial source (F and J Weed Seed Suppliers, Woodstock, Illinois), and originally collected from naturally occurring populations in Illinois (with the exception of Lolium, which was commercial grass seed originally cultivated in Oregon). Abutilon seed was obtained from the selfed offspring of eight cloned genotypes originally sampled in an abandoned agricultural field near Urbana, Illinois. Abutilon genotypes were a subset of those used in a previous experiment in which plants were cloned by apical meristem enhancement, and grown under a variety of light, water, and nutrient conditions (for further details see: Bazzaz et al. 1996, Jasienski et al. 1997). For each genotype seed was pooled from several individuals grown under high nutrient, high light conditions. Abutilon seed was stored in a dry room at 35 [degrees] C until use.
Plants were grown in a glasshouse at Harvard University over a period of 118-172 d from germination to final harvest (over the period August 1993-February 1994). Six glasshouse modules were maintained at a constant temperature regime (26 [degrees]/18 [degrees] C) and two C[O.sub.2] levels (nominally 350 and 700 [[micro]liter]/L, with daytime levels maintained to within 20 [[micro]liter]/L). A 14-h day length was maintained throughout the experiment, with light levels at bench height maintained at a minimum photosynthetic photon flux density (PPFD) of [approximately]400 [[micro]mol] [multiplied by] [m.sup.-2], [s.sup.-1] using metal halide lamps. There were three replicate glasshouse modules per C[O.sub.2] level, with treatment levels of each module assigned randomly (in a staggered array) prior to the start of the experiment. Seed of all species were initially allocated among C[O.sub.2] treatments and blocks, and germinated in plugs of the same soil as used in the experiment. Randomly selected seedlings of healthy appearance were then transplanted into the two density treatments, with timing varying according to the duration necessary for a given species to produce viable transplants. Individually grown plants were maintained in 15 cm internal diameter x 20 cm deep pots (3.5 L soil volume), with 12 replicate individuals per species per module (four in the case of Abutilon genotypes). High-density monocultures were planted in a hexagonal array at a density of 540 plants/[m.sup.2] in 38 x 48 x 20 cm deep plastic tubs ([approximately]35 L total soil volume). The planting design included two rows of "border plants" on each side of the tubs not included in the measurements, with a total of 48 "target" and 48 "border" plants/tub, and two replicate tubs per glasshouse module. Abutilon tubs were planted with an equal mixture of the eight genotypes (six replicates/genotype/tub) randomly allocated within the central target area of the hexagonal array. The total sample size initially planted was thus 576 plants/species at high density, and 72 plants/species for individually grown plants (192 in the case of Abutilon), giving a grand total of 5304 plants used in the main experiment.
Soil was obtained from a construction site in eastern Massachusetts, and was homogenized, but not sterilized, before use. The soil was texturally a sandy loam, with a pH of 5.2, organic content of 4.4%, and P, K, and Mg levels of 0.027, 0.22, and 0.099%, respectively (i.e., very low P and K, but moderately rich Mg content). (Soil analyses were conducted by the Schweizer Labor fur Bodenanalytik und Umwelttechnik, Zurich, Switzerland.) Growth containers were sealed in order to eliminate nutrient loss, and watering was conducted by hand on a daily basis to ensure that soil was moist, but below "field capacity." Measurements of gravimetric soil water content were made at the end of experiment for a subset of growth containers. The average was 14.6 % (dry mass basis), with differences no greater than 4% found among treatments or species (indicating that we were successful in providing similar water levels). Following seedling transplantation, a slow-release fertilizer was added as a top dressing to growth containers (Osmocote R 15:10:10 6-mo release; Sierra Chemical Company, Milpitas, California) at a rate corresponding to 150 kg N [multiplied by] [ha.sup.-1] [multiplied by] [yr.sup.-1], calculated on a soil surface area basis. The N dosage used is at the high end of estimated rates of net mineralization plus N deposition in midlatitude forests (e.g., Nadelhoffer et al. 1983, Aber et al. 1993), so as to approximate soil conditions likely to obtain in abandoned agricultural fields and other "waste areas" in the region. Growth containers for both density treatments had the same depth, so that N additions were the same on both an area and a soil volume basis. N addition rates on a per-plant basis naturally differ between the density treatments.
Harvests were conducted after a period of 118-172 d following germination (88-151 d after transplanting); [TABULAR DATA FOR TABLE 1 OMITTED] timing varied between species so as to insure that plants had reached a stable or declining leaf area, estimated on the basis of nondestructive measurements (see below). At harvest, plants were divided into above- and belowground parts, and reproductive parts (including fruits, seeds, and flowers), dried to constant dry mass in a forced-air oven, and weighed to the nearest 0.001 g. In the case of the three grass species it was not possible to extract roots separately for each individual plant within dense monocultures. For grasses the central target area of the growth containers was excised and washed, and the total root mass allocated in proportion to measured aboveground biomass for each individual.
Nondestructive allometric estimates of leaf area and plant biomass. - Growth and development of aboveground parts of plants were monitored using periodic nondestructive measurements. At each measurement period the length of every leaf on a given plant was estimated to the nearest centimeter. For most species, this could be accomplished by visual estimates, which greatly reduced physical contact with plants, particularly early in development. These leaf length records were kept for all individually grown plants, and for a subset of 12 plants/tub for plants located in central part of the tub of the high-density treatment. Three to six such sets of measurements were made for each species in each treatment. In addition, frequent notes on canopy height and developmental stage were also maintained to ensure that each species had reached a steady or declining leaf area prior to final harvest.
Several sources of data were used to establish allometric relationships between leaf length and area, and between total plant leaf area and biomass. For each of the study species, an additional set of plants, transplanted into the 3.5-L pots described above, were grown during the course of the experiment, and harvested after 30-50 d of growth. For each species there were six high-density pots (seven plants/pot), and six individually grown plants, giving 48 plants/species at ambient C[O.sub.2]. Total biomass and leaf area were determined for this set of plants (the latter using a LI-COR 1200 leaf area meter: LI-COR, Lincoln, Nebraska). In addition, relationships between leaf length and leaf area were established for each species using a subset of individuals and at least 50 leaves per species. This data was supplemented by similar measurements made on plants from the final harvest of the main experiment (thus including both C[O.sub.2] treatments) for species that did not senesce and drop leaves. For the species that did show marked leaf drop and senescence (Abutilon theophrasti and Cassia obtusifolia), allometric data for early growth at both C[O.sub.2] treatments from other experiments was utilized (S. C. Thomas, unpublished data; Farnsworth and Bazzaz, 1995). Species-specific allometric data on leaf size measures, total leaf area and biomass included plants at a wide range of sizes and ontogenetic states in all four density and C[O.sub.2] treatment combinations, but excluded senescing plants. The same allometric relationships were applied in all sequential measurements (although such relationships may change with plant growth, particularly in competing populations: e.g., Weiner and Thomas 1992). We acknowledge that the allometric approach used could thus potentially yield biased estimates of early relative growth rate (RGR) enhancements, particularly if there are large transient C[O.sub.2] or density effects on root:shoot relationships or leaf mass ratio. However, the early growth enhancements reported here for individually grown plants closely match previously published values on a species-by-species basis (comparing to values summarized by Poorter et al. 1996).
Most species showed no pronounced effects of density or C[O.sub.2] on leaf shape. A single allometric equation of the form A = a[L.sup.b] (where A is leaf area, L is leaf length, and a and b are constants, fit using model I regression of log-log transformed data) was therefore used to estimate leaf areas from leaf lengths (data not shown). The three rosette species, Plantago, Rumex, and Taraxacum, did show marked C[O.sub.2] effects on leaf shape (see Thomas and Bazzaz 1996), and for these species treatment-specific allometric relationships were used. Allometric equations used to estimate total biomass as a function of leaf area are listed in Table 1.
The earliest stages of postgermination plant growth are often characterized by a stable or even declining temporal pattern of plant biomass when plant growth is dependent mainly on stored seed reserves (cf. Causton and Venus 1981). The period of truly exponential plant growth is also very short, commonly [less than]3 d for agricultural species (e.g., Causton and Venus 1981, Gifford et al. 1996). These facts complicate the interpretation of "initial" relative growth rate in plants, and also present methodological difficulties, since plants must reach some appreciable size before nondestructive estimates of biomass are feasible. The estimates of "early" growth enhancements presented here are therefore based on a back-calculation procedure (see Poorter et al. 1996). Under the assumption of exponential growth, the difference in RGR between environments may be calculated as:
[RGR.sub.e] - [RGR.sub.a] = ln([W.sub.e]/[W.sub.a])/([t.sub.1] - [t.sub.2]) (1)
where [RGR.sub.e] is the relative growth under elevated C[O.sub.2], [RGR.sub.a] is the relative growth under ambient C[O.sub.2] (in g[multiplied by][g.sup.-1][multiplied by][d.sup.-1]), [W.sub.e] and [W.sub.a] are estimated plant masses at elevated and ambient C[O.sub.2], respectively, and [t.sub.1] - [t.sub.2] is the duration of growth, here taken as the time from seedling transplantation to the first measurement. Similar calculations were made for initial relative growth rate on a leaf area basis (leaf area relative growth rate [LARGR], expressed in square centimeters per square centimeter per day).
Statistical analysis. - The overall experiment employed a split-plot design in which C[O.sub.2] was the main plot effect, species/genotype and density were subplot effects, and pairs of glasshouse modules were the block (main plot) effect (e.g., Underwood 1997). Block was considered a random effect, and other effects were considered fixed; the denominator mean square for F tests of the C[O.sub.2] main effect was thus the C[O.sub.2] x block interaction term (i.e., the main plot error term), while the subplot error was used for all other tests. In order to compute a MANOVA for the experiment (and thus present global probability levels that preserve a correct experiment-wide error rate), it was necessary to omit the C[O.sub.2] x block interaction, and test all effects using the subplot error term.
Preliminary analyses included examination of residuals for both untransformed and log-transformed data. In both cases there were significant deviations from normality (Levene's test; Levene 1960) for some variables. Also, variances were generally not equal between density treatments (as is generally the case, Weiner and Thomas 1986; analysis based on Lilliefors' test, Lilliefors 1967). This source of heteroscedasticity was not readily eliminated by data transformation. Probability levels should be judged accordingly. For ease of interpretation, we present results based on untransformed data in all cases. Statistical analyses made use of several software packages (Data Desk 4.2 [Velleman 1992]; S-plus [Statistical Sciences 1992]; Systat [Wilkinson et al. 1992]). Hierarchical cluster analysis was [TABULAR DATA FOR TABLE 2 OMITTED] performed using Systat (version 5.2, Wilkinson et al. 1992). The cluster analysis presented used a single-linkage algorithm, and employed Mojena's "stopping rule" to estimate group number (as modified by Milligan and Cooper 1985; see Everitt 1993).
Overall patterns of growth and mortality. - Plants of all species established well under the experimental conditions, and closed canopies (leaf area index [LAI] [greater than] 1) were formed in the high-density treatments within 30-60 d [ILLUSTRATION FOR FIGURE 2 OMITTED]. High-density tubs ultimately formed stands with an LAI of 1.8-3.0. Several experimental units were excluded from analyses on an a priori basis: specifically, one block of Plantago suffered high mortality and poor establishment at both densities, and several growth containers for other species were severely flooded by a glasshouse roof leak. Excluding these experimental units, mortality was otherwise low for all species, averaging 0.3% for individually grown plants, and 2.2% for the high-density monocultures. Additionally, the seed stock for Panicum was contaminated by several other grass species (22% of the total number, mostly of the species Digitaria sanguinalis [L.] Scop.), and these plants were also excluded from all analyses.
Pooled statistical analyses. - Multivariate analysis of variance for the entire pooled data set indicates highly significant main effect terms for C[O.sub.2], density, and species, as well as significant C[O.sub.2] x density, C[O.sub.2] x species, and C[O.sub.2] x density x species interactions for the matrix of growth metrics examined (Table 2). C[O.sub.2], density, and genotype terms, and their interactions, were also significant in analyses using the eight sampled genotypes of Abutilon (Table 2; see also Bazzaz et al. 1995, Thomas and Jasienski 1996). In addition, pooled ANOVAs were conducted separately for each of the parameters describing early growth (leaf area and estimated biomass at 17-26 d) and asymptotic growth (maximal leaf area, and final biomass of roots, shoots, total vegetative mass, and vegetative plus reproductive mass). These ANOVA results also indicate significant (P [less than] 0.05) main effect and/or interaction terms in most cases (analyses not shown). Both sets of analyses support the general conclusion that growth responses to C[O.sub.2] and density treatments were species(and genotype-) specific, and that C[O.sub.2] and density effects were not simply additive.
[TABULAR DATA FOR TABLE 3 OMITTED]
Early growth parameters. - Among individually grown plants, there was a strong C[O.sub.2] x species interaction, with enhancements in initial RGR ranging from 0.011 to 0.074 g[multiplied by][g.sup.-1][multiplied by][d.sup.-1] (Table 3). Considering species individually, early growth enhancements were significant only in the case of Dactylis and Panicum; however, all species showed a qualitative trend toward increased RGR and LARGR under elevated C[O.sub.2]. Individual genotypes of Abutilon also showed consistently positive early growth effects, with enhancements in LARGR ranging from 0.012 to 0.039 [cm.sup.2][multiplied by][cm.sup.-2][multiplied by][d.sup.-1] and in RGR ranging from 0.037 to 0.063 g[multiplied by][g.sup.-1][multiplied by][d.sup.-1] among the eight genotypes. There was a significant C[O.sub.2] main effect term, though not a significant C[O.sub.2] x genotype interaction term, for individually grown Abutilon genotypes.
Positive effects of C[O.sub.2] on early growth were also generally found for plants grown in high-density monocultures, but with some exceptions (Table 4). Four of the eight species examined, and five of eight Abutilon genotypes, showed significantly enhanced LARGR, RGR, or both, under elevated C[O.sub.2]. Surprisingly, Plantago major showed significantly reduced LARGR and RGR under elevated C[O.sub.2], contributing to a significant C[O.sub.2] x species interaction for early growth responses under high density. There was a significant main effect term for early RGR responses of Abutilon genotypes to C[O.sub.2] at high density, but no corresponding C[O.sub.2] x genotype interaction.
Pairwise tests indicate that differences in responses of LARGR and RGR to elevated C[O.sub.2] were significantly greater in magnitude for individually grown plants than for the high-density monocultures [ILLUSTRATION FOR FIGURE 3 OMITTED]. This pattern is supported for comparisons among species, among Abutilon genotypes, and for the combined data set.
Asymptotic growth parameters. - There were significant C[O.sub.2] x species interactions detected for most asymptotic growth parameters, with both positive and significant negative effects of C[O.sub.2] detected (Tables 3 and 4). Negative effects were commonly observed for maximum leaf area, with the majority of species and genotypes showing such a trend under both density conditions. Surprisingly, such declines were larger and more consistent for plants grown individually. In contrast, vegetative mass at final harvest generally increased under elevated C[O.sub.2] in both density treatments. In some cases, such as Rumex crispus and Abutilon genotype E, taxa with negative C[O.sub.2] effects on maximal leaf area showed substantial positive effects on final vegetative biomass. All plant species with significant C[O.sub.2] effects on final vegetative mass also showed substantial C[O.sub.2] effects on final root mass. For some species, such as Rumex and Taraxacum, biomass enhancements were largely or entirely due to enhanced root growth.
Only four of the species reproduced during the course of the experiment. Cassia obtusifolia showed a trend toward declining reproductive mass under elevated C[O.sub.2] in both density treatments (Tables 3 and 4; see also Farnsworth and Bazzaz 1995). Abutilon genotypes showed both increases and decreases in reproductive mass, with a significant C[O.sub.2] x genotype interaction at high density (Tables 2-4; see also Bazzaz et al. 1995, Thomas and Jasienski 1996). C[O.sub.2] effects on reproduction did not reach significance for the two grass species that produced seed during the experiment (Lolium and Panicum).
Relationships among response variables. - Table 5 lists the correlation matrices among early and asymptotic growth response measures. In all cases significant correlations are positive in sign for both among-species and among-genotype relationships (note that in three cases pairs of variables are not statistically independent: namely, total vegetative mass vs. root or shoot mass, and early LARGR vs. early RGR). However, there were also numerous low negative correlations between response variables, accounting for one-third of the total number (excluding the six autocorrelated variable pairs). For example, species and genotypes exhibiting greater early growth responses tended to show greater reductions in maximal leaf area under elevated C[O.sub.2]. The overall correspondence between patterns found among species and among genotypes (i.e., the correlation between elements of the species matrix vs. the genotype matrix in Table 5) was low (r = 0.216; P [greater than] 0.05). Among genotypes, early growth responses were correlated with final reproductive responses. Differences between the interspecific and genetic correlation matrices are likely due, at least in part, to the fact that genotypes at high density were grown in competition (see Thomas and Jasienski 1996).
[TABULAR DATA FOR TABLE 4 OMITTED]
Relationships were also examined between RGR under ambient C[O.sub.2] and observed responses among species and genotypes [ILLUSTRATION FOR FIGURES 4 AND 5 OMITTED]. There was no significant correlation between these measures for individually grown plants (r = -0.089; P [greater than] 0.644: [ILLUSTRATION FOR FIGURE 4 OMITTED]). However, lack of correlation is largely determined by outlying values for a few species that showed quite high RGR values, but very low RGR responses: namely, Taraxacum and Plantago, as well as the grass Lolium. Excluding these species there is a positive correlation between RGR and the observed RGR response, similar to that reported by Poorter et al. (1996) (r = 0.706; Model I regression equation: [Delta]RGR = -0.014 + 0.524[RGR]). In contrast, there is no evidence for any relationship between RGR and the C[O.sub.2] response of final plant biomass [ILLUSTRATION FOR FIGURE 5 OMITTED].
A hierarchical cluster analysis was conducted to describe patterns of variation in the varying aspects of C[O.sub.2] response observed among plant species and Abutilon genotypes [ILLUSTRATION FOR FIGURE 6 OMITTED]. Two trends of interest are apparent. First, plant growth forms (namely erect annuals, rosette species, and grasses) were not consistently grouped, but rather interspersed throughout the phenetic tree. Second, although five of the eight Abutilon genotypes were closely grouped, response patterns shown by other genotypes are interspersed among the broader sample of species. Genotype H, in particular, was an extreme genetic variant, showing a remarkably high C[O.sub.2] response at high density, with a disproportionate root response (Table 3). Genotype H was nearly as divergent in its multivariate response pattern as was the species showing the greatest deviation, Panicum dichotomofolium. Application of a standard "cutoff rule" segregates three groups: Panicum, genotype H, and the residual group of all other species and genotypes [ILLUSTRATION FOR FIGURE 6 OMITTED]. The cluster analysis thus indicates that the multivariate patterns of response among genotypes are similarly variable to those among species, and also fails to detect discrete groups among the taxa considered.
[TABULAR DATA FOR TABLE 5 OMITTED]
A primary finding of this study is that enhancements in early plant growth under elevated C[O.sub.2] do not necessarily correspond to similar effects on plant size late in ontogeny. In fact, each of the theoretical possibilities illustrated in Fig. 1 was actually found empirically among the species examined [ILLUSTRATION FOR FIGURE 7 OMITTED]. For example, Rumex crispus showed substantial enhancements of both early relative growth and maximum vegetative mass under elevated C[O.sub.2]. In contrast, Lolium multiflorum, while exhibiting significant early growth enhancements, showed almost no C[O.sub.2] effect on asymptotic biomass. Finally, Taraxacum officinale showed little or no effect of elevated C[O.sub.2] on early relative growth, but a positive effect on final biomass. In accordance with this range of patterns, among-species correlations between early vs. asymptotic growth enhancements were quite weak, ranging from -0.23 to +0.48 (Table 5).
The absence of significant correlations between early and asymptotic enhancements does not imply that such effects are unpredictable. Rather, C[O.sub.2] effects on asymptotic plant size varied in a consistent manner among plant size metrics, and in response to plant density. Most species exhibited reductions in maximum leaf area under elevated C[O.sub.2], while also showing substantial increases in final biomass (Table 2, [ILLUSTRATION FOR FIGURE 3 OMITTED]). In addition, species exhibiting large increases in final total biomass also tended to show a disproportionate effect of C[O.sub.2] on root growth (Tables 3-5). This pattern was especially pronounced in species forming large taproots (i.e., Rumex and Taraxacum). Early growth responses were lower for high-density monocultures than for individually grown plants, with a threefold difference in average LARGR responses observed between density treatments. Similarly, responses for asymptotic values of biomass components were also much lower at high density than for individually grown plants. The most dramatic interactive effect of density and C[O.sub.2] involved reproductive responses, which were positive for individually grown plants, but were, on average, negative for plants at high density [ILLUSTRATION FOR FIGURE 3 OMITTED].
Comparisons with previous research. - As noted in the introduction, previous studies examining C[O.sub.2] effects on plant growth have not explicitly sought to quantify asymptotic size. However, a closely related issue that has received attention is the response of steady-state leaf area index in single and multispecies stands (Oechel and Riechers 1986, Korner and Arnone 1992, Arnone and Korner 1995, Schappi and Korner 1995, Hattenschwiler and Korner 1996, Hirose et al. 1996). In general, little or no response in steady-state LAI to elevated C[O.sub.2] has been found in nonagricultural systems (cf. Korner 1996). The present study supports this general conclusion, but also indicates substantial variation in potential LAI responses among species. LAI responses averaged only +6%; however, observed values varied from -21% in Rumex crispus monocultures, to +48% in Lolium multiflorum monocultures (Table 4).
It has often been assumed that changes in LAI under rising C[O.sub.2] are driven by decreases in the leaf-level light compensation point (e.g., Valle et al. 1985, Long and Drake 1992, Hirose et al. 1996). However, observations of negative effects of elevated C[O.sub.2] on leaf area are not expected under this mechanism. An additional unexpected result in the present study was that individually grown plants showed more consistent effects of C[O.sub.2] on leaf area than did high-density monocultures, with a strong trend toward reduced maximum leaf area under elevated C[O.sub.2] [ILLUSTRATION FOR FIGURE 3 OMITTED]. The species with the largest observed reduction in maximal leaf area under elevated C[O.sub.2] was Taraxacum officinale, which produces an exaggerated "adult" leaf morphology under elevated C[O.sub.2] (Thomas and Bazzaz 1996). These observations suggest that C[O.sub.2] effects on LAI may commonly be determined by species-specific changes in leaf development and morphology, rather than by changes in photosynthetic physiology.
A number of reviews have previously stressed the importance of belowground storage organs and other morphological carbon sinks as a determinant of C[O.sub.2] growth responses (e.g., Cure 1985, Korner 1993). Because such structures do not generally develop until late in plant ontogeny, we suggest that this sink strength factor may primarily determine plant responses to C[O.sub.2] late in ontogeny, but have little predictive value for early growth responses. This hypothesis is supported by our data: strong responses of final biomass were in every case accompanied by disproportionate effects on root mass at harvest, most notably in species producing taproots. However, such species did not necessarily show strong C[O.sub.2] enhancements of growth early in ontogeny (Tables 3-5). A similar result has also recently been reported by Reekie (1996), in a study of agricultural species and varieties within the genus Brassica. Taxa that developed large reproductive structures late in ontogeny, such as broccoli and cauliflower, showed the strongest C[O.sub.2] responses late in ontogeny, whereas taxa that did not produce such structures, such as mustard and rape, showed little or no size increase at final harvest.
Only a few previous comparisons have been made between growth responses to elevated C[O.sub.2] of individually grown plants vs. dense monocultures. Wayne and Bazzaz (1996) have recently reported data for yellow birch (Betula alleghaniensis) seedlings, finding that growth enhancements after one season were +49% for individually grown seedlings vs. + 16% for dense populations. These figures are not directly comparable to the calculated RGR enhancements presented here; however, both sets of results suggest that C[O.sub.2] enhancements of early plant growth are reduced by local crowding. This generalization thus appears to hold for both comparisons between individuals vs. plant monocultures, as studied here, and also for comparisons between individuals vs. mixed-species stands (Ackerly and Bazzaz 1995).
Functional types and correlates of C[O.sub.2] responses among and within species. - Much recent attention has been given to the issue of elucidating functional types among plants that show similar responses to rising C[O.sub.2] and other aspects of global change (e.g., Smith et al. 1997). We agree that it is absolutely necessary to aggregate responses in some manner in order to formulate tractable models. However, this goal may be accomplished in many ways. The idea that discrete functional types exist at all should be treated as a hypothesis, and not as an a priori assumption. The comparative analyses presented here (cf. [ILLUSTRATION FOR FIGURES 4-6 OMITTED]), suggest that readily recognizable growth forms among herbaceous plant species do not correspond closely to patterns of growth response to elevated C[O.sub.2]. Moreover, genetic variation within a single species was found to be remarkably large, with, for example, the most variant genotype examined being nearly as divergent in response pattern as the most variant species [ILLUSTRATION FOR FIGURE 6 OMITTED]. This result agrees with other recent analyses suggesting that intraspecific variation in responses to C[O.sub.2] rivals the variation observed among species (cf. Curtis et al. 1996, Thomas and Jasienski 1996). Particularly in light of this result, we suggest that the "problem of aggregation" would be better approached by using methods analogous to those employed in quantitative genetics (e.g., Lynch and Lande 1993). Ecologists interested in vegetation responses to global change should be quantifying parameters that describe distributions of response and patterns of covariation among response components within and among species, rather than assuming uniform types on an a priori basis.
Comparative studies and reviews have previously noted that species with higher intrinsic growth rates tend to show greater responsiveness to elevated C[O.sub.2] (e.g., Cure 1985, Poorter 1993). This pattern has recently been expressed as a relationship between the absolute enhancement in RGR, and RGR under controlled conditions (Poorter et al. 1996). It has also been suggested that differences in intrinsic RGR may be the primary determinant of interspecific variation in C[O.sub.2] responses (Poorter 1993, Poorter et al. 1996). Our results do not consistently support this latter hypothesis. Rather, we found a substantial proportion of species with high RGR that do not show large early growth responses to elevated C[O.sub.2]. Notable examples of this pattern include Lolium multiflorum and Taraxacum officinale [ILLUSTRATION FOR FIGURE 3 OMITTED]. As illustrated in Fig. 3, variation in RGR (due to either genetic or environmental effects) appears to impose an upper bound to possible RGR enhancement under elevated C[O.sub.2], with some species consistently falling well below this upper bound. This pattern suggests that aspects of physiology and morphological development other than RGR can substantially constrain early growth responses to C[O.sub.2] in some taxa. We also examined possible relationships between RGR and C[O.sub.2] responses of asymptotic plant size [ILLUSTRATION FOR FIGURE 4 OMITTED]. In contrast to the pattern found for early growth, there was no evidence for any relationship. This result, as well as the lack of a close correspondence between early and asymptotic growth responses, supports the general conclusion that physiological determinants differ greatly for early vs. asymptotic growth responses.
In light of the great interest in long-term C[O.sub.2] responses of trees (e.g., Norby et al. 1992, 1996, Bazzaz et al. 1993, Curtis and Wang 1998), it is worth considering the potential implications of our results for forest responses to rising C[O.sub.2]. Woody plants tend to have a lower initial RGR than do herbaceous species, and comparative analyses also indicate that trees display relatively low stimulations of RGR under elevated C[O.sub.2] (Poorter
et al. 1996). However, our results suggest that asymptotic growth responses are not related to RGR at all, but rather are determined, at least in part, by the degree to which large carbon sinks are formed late in plant ontogeny. Some aspects of tree growth, such as height extension, may be asymptotic (Thomas 1996b); however, the limited functional life span of xylem tissue predisposes woody plants to a non-asymptotic pattern of diameter growth and wood production (Zimmerman 1983, Thomas 1996b). This growth pattern could also predispose trees to show increasing sink potential, and thus relatively large effects of elevated C[O.sub.2] on biomass, late in ontogeny. The longest term data available examining C[O.sub.2] effects on tree growth extend to only slightly beyond the inflection point of estimated growth curves (cf. Hattenschwiler et al. 1997, Idso and Kimball 1997). Forest modelers are thus presently compelled either to assume that asymptotic growth responses to rising C[O.sub.2] are zero (e.g., Bolker et al. 1995), or to apply the same physiological or growth responses observed for seedlings and saplings to large trees and/or whole forests (e.g., Luxmoore et al. 1990, Kirschbaum et al. 1994). In the absence of direct data, values for asymptotic responses of herbaceous species may provide some indication of possible asymptotic forest growth responses; however, there is a clear and pressing need for experiments examining asymptotic size responses for short-lived woody plants.
Given the lack of correlation between early vs. asymptotic growth responses, the question also arises as to which parameter is of greater ecological significance. This will depend to a large extent on the specific ecological system and time frame in mind; however, we suggest that both kinds of effect are generally of importance and interest. For example, consider the case of forest plantations managed as a carbon sink (e.g., Row 1996). Assuming that harvests occur near the growth curve inflection point, species or genotypes that show higher initial RGR should generally be preferred. A positive correlation between RGR and early growth response simply reinforces this conclusion. On the other hand, initial RGR responses may be of little value in predicting C[O.sub.2] effects on carbon storage by older forests. From an evolutionary perspective, we expect that asymptotic growth responses may be more closely related to reproductive output, and would thus be favored by natural selection under rising C[O.sub.2]. However, in order to predict the consequences of selection to future growth patterns, analyses should examine genetic correlation structure through the entire course of ontogeny (cf. Thomas and Jasienski 1996; M. Jasienski, S. C. Thomas, and F. A. Bazzaz, unpublished manuscript). Finally, it should also be emphasized that more than two growth parameters may be necessary to understand many important ecological impacts of rising C[O.sub.2]. Of particular interest are potential C[O.sub.2] effects on reproductive timing, seed germination, and plant senescence (e.g., Farnsworth and Bazzaz 1995, Farnsworth et al. 1996).
In conclusion, fundamental principles and empirical data both point to the importance of distinguishing between what we have termed early vs. asymptotic components of plant growth responses to elevated C[O.sub.2]. Distinguishing between these two types of response is a necessary step in understanding physiological mechanisms that account for variation in responses among and within plant species. This distinction is also essential in formulating models aimed at predicting vegetation responses to global change. A complete understanding of the multidimensional effects of C[O.sub.2] through plant ontogeny, including effects on leaf area, biomass components, leaf and root loss, and allometric relationships among these components, is ultimately desirable. However, we suggest that even for the modest goal of characterizing effects of a stepchange in C[O.sub.2] on plant growth for a given species in a given environment, one parameter is not enough.
We thank D. Ramseier for assistance with soils analyses, E. Farnsworth for sharing Cassia data, B. Traw for help in setting up the experiment, and P. Wayne and J. Weiner for insightful discussions. C. Potvin, P. Reich, W. Winner, and an anonymous reviewer provided helpful input on the penultimate manuscript. This research was supported by grants from the Department of Energy and the National Science Foundation.
Aber, J. D., A. Magill, R. Boone, J. M. Melillo, P. Steudler, and R. Bowden. 1993. Plant and soil responses to chronic nitrogen additions at the Harvard forest, Massachusetts. Ecological Applications 3:156-166.
Ackerly, D. D., and E A. Bazzaz. 1995. Plant growth and reproduction along C[O.sub.2] gradients: non-linear responses and implications for community change. Global Change Biology 1:199-207.
Arnone, J. A., and C. Korner. 1995. Soil and biomass carbon pools in model communities of tropical plants under elevated C[O.sub.2]. Oecologia 104:61-71.
Bazzaz, F. A. 1990. The response of natural ecosystems to the rising global C[O.sub.2] levels. Annual Review of Ecology and Systematics 21:167-196.
Bazzaz, F. A., S. L. Bassow, G. M. Berntson, and S. C. Thomas. 1996. Elevated C[O.sub.2] and terrestrial vegetation: implications for and beyond the global carbon budget. Pages 43-76 in B. Walker and W. Steffen, editors. Global change and terrestrial ecosystems. Cambridge University Press, Cambridge, UK.
Bazzaz, F. A., K. Garbutt, E. G. Reekie, and W. E. Williams. 1989. Using growth analysis to interpret competition between a [C.sub.3] and a [C.sub.4] annual under ambient and elevated C[O.sub.2]. Oecologia 79:223-235.
Bazzaz, F. A., M. Jasienski, S.C. Thomas, and E Wayne. 1995. Microevolutionary responses in experimental populations of plants to C[O.sub.2]-enriched environments: parallel results from two model systems. Proceedings of the National Academy of Sciences, USA 92:8161-8165.
Bazzaz, F. A., S. L. Miao, and P. Wayne. 1993. C[O.sub.2]-induced growth enhancements of co-occurring tree species decline at different rates. Oecologia 96:478-482.
Bolker, B. M., S. W. Pacala, F. A. Bazzaz, C. Canham, and S. A. Levin. 1995. Species diversity and carbon dioxide fertilization of temperate forests. Global Change Biology 1:373-381.
Bowes, G. 1991. Growth at elevated C[O.sub.2]: photosynthetic responses mediated through Rubisco. Plant, Cell and Environment 14:795-806.
Causton, D. R., and J. C. Venus. 1981. The biometry of plant growth. Arnold Press, London, UK.
Chiariello, N., and J. Roughgarden. 1984. Storage allocation in seasonal races of an annual plant: Optimal versus actual allocation. Ecology 65:1290-1301.
Coleman, J. S., and F. A. Bazzaz. 1992. Effects of C[O.sub.2] and temperature on growth and resource use of co-occurring [C.sub.3] and [C.sub.4] annuals. Ecology 73:1244-1259.
Cure, J. D. 1985. Carbon dioxide doubling responses: A crop survey. Pages 99-116 in B. R. Strain and J. D. Cure, editors. Direct effects of increasing carbon dioxide on vegetation. United States Department of Energy Report DOE/ER-0238. National Technical Information Service, Springfield, Virginia, USA.
Cure, J. D., and B. Acock. 1986. Crop response to carbon dioxide doubling: A literature survey. Agricultural and Forest Meteorology 38:127-145.
Curtis, P. S. 1996. A meta-analysis of leaf gas exchange and nitrogen in trees grown under elevated carbon dioxide. Plant, Cell and Environment 19:127-137.
Curtis, P. S., D. J. Klus, S. Kalisz, and S. J. Tonsor. 1996. Intraspecific variation in C[O.sub.2] responses in Raphanus raphanistrum and Plantago lanceolata: assessing the potential for evolutionary change with rising atmospheric C[O.sub.2]. Pages 13-29 in C. Korner and F. A. Bazzaz, editors. Carbon dioxide, populations and communities. Academic Press, San Diego, California, USA.
Curtis, P. S., and X. Wang. 1998. A meta-analysis of elevated C[O.sub.2] effects on woody plant mass, form, and physiology. Oecologia 113:299-313.
Everitt, B. S. 1993. Cluster analysis. Third edition. Arnold Press, London, UK.
Farnsworth, E. J., and F. A. Bazzaz. 1995. Inter- and intrageneric differences in growth, reproduction, and fitness of nine herbaceous annual species grown in elevated C[O.sub.2] environments. Oecologia 104:454-466.
Farnsworth, E. J., A. M. Ellison, and W. K. Gong. 1996. Elevated C[O.sub.2] alters anatomy, physiology, growth, and reproduction of red mangrove (Rhizophora mangle L.). Oecologia 108:599-609.
Friend, A. D. 1993. The prediction and physiological significance of tree height. Pages 101-115 in A. M. Solomon and H. H. Shugart, editors. Vegetation dynamics and global change. Chapman and Hall, New York, USA.
Geber, M. A. 1990. The cost of meristem limitation in Polygonum arenastrum: Negative genetic correlations between fecundity and growth. Evolution 44:799-819.
Gifford, R. M., D. J. Barrett, F. L. Lutze, and A. B. Samarakoon. 1996. Agriculture and global change: scaling direct carbon dioxide impacts and feedbacks through time. Pages 229-259 in B. Walker and W. Steffen, editors. Global change and terrestrial ecosystems. Cambridge University Press, Cambridge, UK.
Givnish, T. J. 1988. Adaptation to sun and shade: A whole plant perspective. Australian Journal of Plant Physiology 15:63-92.
Hattenschwiler, S., and C. Korner. 1996. System-level adjustments to elevated C[O.sub.2] in model spruce ecosystems. Global Change Biology 2:377-387.
Hattenschwiler, S., F. Miglietta, A. Raschi, and C. Korner. 1997. Thirty years of in situ tree growth under elevated C[O.sub.2]: a model for future forest responses? Global Change Biology 3:463-471.
Hirose, T., D. D. Ackerly, and F. A. Bazzaz. 1996. C[O.sub.2] elevation and canopy development in stands of herbaceous plants. Pages 413-430 in C. Korner and F. A. Bazzaz, editors. Carbon dioxide, populations and communities. Academic Press, San Diego, California, USA.
Idso, S. B., and B. A. Kimball. 1997. Effects of long-term atmospheric C[O.sub.2] enrichment on the growth and fruit production of sour orange trees. Global Change Biology 3:89-96.
IPCC. 1996. Climate change 1995. International Panel on Climate Change. Cambridge University Press, Cambridge, UK.
Jasienski, M., F. J. Ayala, and F. A. Bazzaz. 1997. Phenotype plasticity and similarity of DNA among genotypes of an annual plant. Heredity 78:176-181.
Johnson, D. W., P. H. Henderson, J. T. Ball, and R. F. Walker. 1996. Effects of C[O.sub.2] and N on growth and N dynamics in ponderosa pine: results from the first two growing seasons. Pages 23-40 in G. W. Koch and H. A. Mooney, editors. Carbon dioxide and terrestrial ecosystems. Academic Press, San Diego, California, USA.
Kimball, B. A. 1983. Carbon dioxide and agricultural yield: An assemblage and analysis of 430 prior observations. Agronomy Journal 75:779-788.
Kirschbaum, M. U. F., D. A. King, H. N. Comins, R. E. McMurtrie, B. E. Medlyn, S. Pongracic, D. Murty, H. Keith, R. J. Rason, P. K. Khanna, and D. W. Sheriff. 1994. Modelling forest response to increasing C[O.sub.2] concentration under nutrient-limited conditions. Plant, Cell and Environment 17:1081-1099.
Korner, C. 1993. C[O.sub.2] fertilization: the great uncertainty in future vegetation development. Pages 53-70 in A.M. Solomon and H. H. Shugart, editors. Vegetation dynamics and global change. Chapman and Hall, New York, New York, USA.
-----. 1996. The response of complex multispecies systems to elevated C[O.sub.2]. Pages 20-42 in B. Walker and W. Steffen, editors. Global change and terrestrial ecosystems. Cambridge University Press, Cambridge, UK.
Korner, C., and J. A. Arnone III. 1992. Responses to elevated carbon dioxide in artificial tropical ecosystems. Science 257:1672-1675.
Levene, H. 1960. Robust tests for equality of variance. Pages 278-292 in I. Olkin, editor. Contributions to probability and statistics. Princeton University Press, Palo Alto, California, USA.
Lilliefors, H. W. 1967. On the Kolmorogrov-Smirnov test for normality with mean and variance unknown. Journal of the American Statistical Association 64:399-402.
Loehle, C. 1995. Anomalous responses of plants to C[O.sub.2] enrichment. Oikos 73:181-187.
Long, S. P., and B. G. Drake. 1992. Photosynthetic C[O.sub.2] assimilation and rising atmospheric C[O.sub.2] concentrations. Pages 69-101 in N. R. Baker, and H. Thomas, editors. Crop photosynthesis: spatial and temporal dynamics. Elsevier, Amsterdam, The Netherlands.
Luxmoore, R. J., M. L. Tharp, and D. C. West. 1990. Simulating the physiological basis of tree-ring responses to environmental changes. Pages 393-401 in R. K. Dixon, R. S. Meldahl, G. A. Ruark, and W. G. Warren, editors. Process modeling of forest growth responses to environmental stress. Timber Press, Portland, Oregon, USA.
Lynch, M., and R. Lande. 1993. Evolution and extinction in response to environmental change. Pages 234-250 in P. M. Kareiva, J. G. Kingsolver, and R. B. Huey, editors. Biotic interactions and global change. Sinauer Associates, Sunderland, Massachusetts, USA.
McMahon, T. 1973. Size and shape in biology. Science 179:1201-1204.
Milligan, G. W., and M. C. Cooper. 1985. An examination of procedures for determining the number of clusters in a data set. Psychometrica 50:159-179.
Nadelhoffer, K. J., J. D. Aber, and J. M. Melillo. 1983. Leaf-litter production and soil organic matter dynamics along a nitrogen mineralization gradient in Southern Wisconsin (USA). Canadian Journal of Forest Research 13:12-21.
Norby, R. J., S. D. Wullschleger, and C. A. Gunderson. 1996. Tree responses to elevated C[O.sub.2] and implications for forests. Pages 1-22 in G. W. Koch and H. A. Mooney, editors. Carbon dioxide and terrestrial ecosystems. Academic Press, San Diego, California, USA.
Norby, R. J., S. D. Wullschleger, C. A. Gunderson, E. G. O'Neill, and M. K. McCracken. 1992. Productivity and compensatory responses of yellow-poplar trees in elevated C[O.sub.2]. Nature 357:322-324.
Oechel, W. C., and G. H. Riechers. 1986. Impacts of increasing C[O.sub.2] on natural vegetation, particularly the tundra. Pages 36-42 in C. Rosenzweig and R. Dickinson, editors. Climate-vegetation interactions. Goddard Space Flight Center, Greenbelt, Maryland, USA.
Poorter, H. 1993. Interspecific variation in the growth response of plants to an elevated ambient C[O.sub.2] concentration. Vegetatio 104/105:77-97.
Poorter, H., C. Roumet, and B. D. Campbell. 1996. Interspecific variation in the growth response of plants to elevated C[O.sub.2]: a search for functional types. Pages 375-412 in C. Korner and F. A. Bazzaz, editors. Carbon dioxide, populations, and communities. Academic Press, San Diego, California, USA.
Reekie, E. G. 1996. The effect of elevated C[O.sub.2] on developmental processes and its implications for plant-plant interactions. Pages 333-346 in C. Korner and F. A. Bazzaz, editors. Carbon dioxide, populations, and communities. Academic Press, San Diego, California, USA.
Rogers, H. H., J. D. Cure, J. F. Thomas, and J. M. Smith. 1984. Influence of elevated C[O.sub.2] on growth of soybean plants. Crop Science 24:361-366.
Row, C. 1996. Effects of selected forest management options on carbon storage. Pages 59-90 in R. N. Sampson and D. Hair, editors. Forests and global change. Volume 2. American Forests, Washington, D.C. USA.
Schappi, B., and C. Korner. 1995. Growth responses of an alpine grassland to elevated C[O.sub.2]. Oecologia 105:43-52.
Smith, T. M., H. H. Shugart, and F. I. Woodward, editors. 1997. Plant functional types: their relevance to ecosystem properties and global change. Cambridge University Press, Cambridge, UK.
Statistical Sciences. 1992. S-plus for DOS reference manual. Statistical Sciences, Seattle, Washington, USA.
St. Omer, L., and S. M. Horvath. 1983. Elevated carbon dioxide concentrations and whole plant senescence. Ecology 64:1311-1314.
Thomas, S. C. 1996a. Reproductive allometry in Malaysian rain forest trees: biomechanics vs. optimal allocation. Evolutionary Ecology 10:517-530.
-----. 1996b. Asymptotic height as a predictor of growth and allometric characteristics in Malaysian rain forest trees. American Journal of Botany 83:556-566.
Thomas, S. C., and F. A. Bazzaz. 1996. Elevated C[O.sub.2] and leaf shape: are dandelions getting toothier? American Journal of Botany 83:106-111.
Thomas, S. C., and F. A. Bazzaz. 1999. Asymptotic height as a predictor of photosynthetic characteristics in Malaysian rain forest trees. Ecology 80:278-294.
Thomas, S. C., and M. Jasienski. 1996. Genetic variability and the nature of microevolutionary responses to elevated C[O.sub.2]. Pages 51-81 in C. Korner and F. A. Bazzaz, editors. Carbon dioxide, populations, and communities. Academic Press, San Diego, California, USA.
Tolley, L. C., and B. R. Strain. 1984. Effects of C[O.sub.2] enrichment on growth of Liquidambar styraciflua and Pinus taeda seedlings under different irradiance levels. Canadian Journal of Forest Research 14:343-350.
Tremmel, D. C., and D. T. Patterson. 1994. Effects of elevated C[O.sub.2] and temperature on development in soybean and five weeds. Canadian Journal of Plant Science 74:43-50.
Underwood, A. J. 1997. Experiments in ecology: their logical design and interpretation using analysis of variance. Cambridge University Press, Cambridge, UK.
Valle, R., J. W. Mishoe, W. J. Campbell, J. W. Jones, and L. H. Allen. 1985. Photosynthetic response of 'Bragg' soybean leaves adapted to different C[O.sub.2] environments. Crop Science 25:333-339.
Velleman, P. F. 1992. Data desk 4.2 handbook. Data Descriptions, Ithaca, New York, USA.
Walters, M. B., E. L. Kruger, and P. B. Reich. 1993. Growth, biomass distribution and C[O.sub.2] exchange of northern hardwood seedlings in high and low light: relationships with successional status and shade tolerance. Oecologia 94:7-16.
Wayne, P. M., and F. A. Bazzaz. 1996. Seedling density modifies the growth responses of yellow birch maternal families to elevated carbon dioxide. Global Change Biology 1:315-324.
Weiner, J., and S. C. Thomas. 1986. Size variability and competition in plant monocultures. Oikos 47:211-222.
Weiner, J., and S. C. Thomas. 1992. Competition and allometry in three species of annual plants. Ecology 73:648-656.
Wilkinson, L., A. Hill, J. P. Welna, and G. K. Birkenbeuel. 1992. SYSTAT for Windows: statistics, version 5. Systat, Evanston, Illinois, USA.
Wullschleger, S. D., W. M. Post, and A. W. King. 1995. On the potential for a C[O.sub.2] fertilization effect in forests: estimates of the biotic growth factor based on 58 controlled-exposure studies. Pages 85-107 in G. M. Woodwell and F. T. MacKenzie, editors. Biotic feedbacks in the global climatic system: will the warming feed the warming? Oxford University Press, New York, New York, USA.
Yoder, B. J., M. G. Ryan, R. H. Waring, A. W. Schoettle, and M. R. Kaufman. 1994. Evidence of reduced photosynthetic rates in old trees. Forest Science 40:513-527.
Zimmerman, M. 1983. Xylem structure and the ascent of sap. Springer-Verlag, Berlin, Germany.
|Printer friendly Cite/link Email Feedback|
|Author:||Thomas, S.C.; Jasienski, M.; Bazzaz, F.A.|
|Date:||Jul 1, 1999|
|Previous Article:||Detecting population-level consequences of ongoing environmental change without long-term monitoring.|
|Next Article:||Validity of extrapolating field CO2 experiments to predict carbon sequestration in natural ecosystem.|