• Atmospheric change;
  • carbon dioxide enrichment;
  • forest dynamics;
  • growth;
  • Panama;
  • solar radiation


  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

1. Plants growing in deep shade and high temperature, such as in the understorey of humid tropical forests, have been predicted to be particularly sensitive to rising atmospheric CO2. We tested this hypothesis in five species whose microhabitat quantum flux density (QFD) was documented as a covariable. After 7 (tree seedlings of Tachigalia versicolor and Beilschmiedia pendula) and 18 months (shrubs Piper cordulatum and Psychotria limonensis, and grass Pharus latifolius) of elevated CO2 treatment (c. 700 μl litre–1) under mean QFD of less than 11 μmol m–2 s–1, all species produced more biomass (25–76%) under elevated CO2.

2. Total plant biomass tended to increase with microhabitat QFD (daytime means varying from 5 to 11μmol m–2 s–1) but the relative stimulation by elevated CO2 was higher at low QFD except in Pharus.

3. Non-structural carbohydrate concentrations in leaves increased significantly in Pharus (+ 27%) and Tachigalia (+ 40%).

4. The data support the hypothesis that tropical plants growing near the photosynthetic light compensation point are responsive to elevated CO2. An improved plant carbon balance in deep shade is likely to influence understorey plant recruitment and competition as atmospheric CO2 continues to rise.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Tropical forests, like all other vegetation on earth, existed at atmospheric CO2 concentrations as low as 180–220 μl litre–1 only 20 000 years ago (Neftel et al. 1988). Now plants experience almost twice that concentration. Research is devoted worldwide to study the potential quadrupling-effects of atmospheric CO2 expected to occur by the end of the next century. Because almost half of the carbon in global biomass is found in the tropics and subtropics (Brown & Lugo 1982), understanding the CO2 response of these biomes deserves high priority, in particular because predictions of plant responses to elevated CO2 commonly assume that responses are more pronounced the warmer the climate (Drake & Leadley 1991; Mooney et al. 1991; Gifford 1992; Rawson 1992; Koch & Mooney 1996).

Biomass responses in tropical plants growing in pots or model ecosystems (to date seven projects) were found to vary with experimental conditions (see review by Arnone 1996). Reekie & Bazzaz (1989) and Arnone & Körner (1993, 1995) found no significant overall biomass response in experimental communities exposed to elevated CO2. Körner & Arnone (1992) working with model communities in a greenhouse on more fertile substrate, found a moderate stimulation of plant community biomass (significant below ground, little above ground). In contrast Ziska et al. (1991, shade-house in Panama), Lovelock, Kyllo & Winter (1996, open top chambers in Panama) and Oberbauer, Strain & Fetcher (1985, growth chambers) observed significant CO2 stimulation of biomass in potted tropical plants. It appears that responses are species specific, related to plant morphology, and are dependent on soil fertility, plant competition and experimental duration. A problem with these previous experimental systems using tropical plants is that they were artificial to variable degrees, lacking one most fundamental prerequisite for assessing realistic CO2 responses, namely a natural plant–soil association, a feature that was present in the current study with plants growing in the understorey of an undisturbed tropical forest.

Seedlings below a tree canopy and understorey plants in general are constrained by a lack of solar radiation. Is this low level of quantum supply likely to preclude growth stimulation by CO2? It is important here to distinguish between relative and absolute effects of elevated CO2. Simply because most plants grow faster in light compared to shade, absolute gains in biomass owing to CO2 fertilization over similar time intervals can be expected to be higher in full sunlight. However, relative effects are likely to be more pronounced in deep shade because higher concentrations of CO2 reduce photorespiration, increase quantum yield and, thus, decrease the light compensation point of photosynthesis (e.g. Valle et al. 1985; Wong & Dunin 1987; Long & Drake 1991). Hence, the leaf carbon balance should be improved in very low light under elevated CO2 and the largest relative effects should occur in plants whose leaves are operating close to the light compensation point under current ambient CO2 concentrations. Although small in absolute terms, such large relative effects could induce far-ranging changes in community dynamics in a forest (Bazzaz & Miao 1993; Hättenschwiler & Körner 1996), the motive for the current investigation.

In situ CO2 enrichment in the understorey of a natural forest is challenging. The first problem is the extremely patchy light climate. Without carefully accounting for this, it obviates any meaningful interpretation of growth responses to CO2. We met this challenge by installing 64 light sensors next to our experimental plants in the experimental tents, so that we could treat microhabitat QFD as a measured covariable. The second potential problem relates to the frequent assumption that CO2 concentrations are naturally high in the forest’s understorey, and that any further increase in CO2 concentrations would have minimal effects. However, there is no substance to this belief. A substantial body of evidence, particularly in the less known older forest meteorology literature, shows that natural CO2 enrichment during the day in forests really is confined to the immediate soil boundary layer (e.g. Medina et al. 1986) and to calm night-time hours. Daytime CO2 concentrations at heights between 0·1 m and the top of the canopy rarely deviate by more than 30 μl litre–1 from concentrations measured in the free atmosphere (Schimper & Von Faber 1935; Mitscherlich, Kern & Künstle 1963; Baumgartner 1967; Elias et al. 1989; Bazzaz & Williams 1991; Buchmann, Kao & Ehleringer 1996; for tropical forests see also Allen & Lemon 1976; Kira & Yoda 1989). We will present data for our experimental site which support these earlier findings and justify experimental CO2 enrichment in the understorey.

Thus, there are good reasons to assume that growth of plants in the understorey of a tropical forest should be stimulated by elevated CO2. We tested this hypothesis by exposing populations of understorey plants in situ to CO2-enriched air while leaving all other environmental conditions largely unchanged. A great advantage to working in the understorey is that CO2 enrichment can be achieved in transparent enclosures without risking any of the well-known greenhouse effects. Plant responses were assessed as biomass accumulation, leaf number, leaf area, specific leaf area and non-structural leaf carbohydrate pools.

Materials and methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References


The experiment was conducted at the Barro Colorado Island Nature Monument in the Canal Zone of Panama (9° 09 ′N, 79° 51 ′W, altitude 20 m). Holdridge et al. (1971) classified the vegetation of this biological reserve as a tropical moist forest. Approximately 90% of the mean annual precipitation of 2600 mm falls in the rainy season which is interrupted by a drier season extending from late December to early May (Wright 1991). The annual mean temperature is 27 °C. Soils are slightly acidic oxisols (pH 6·3) derived from basaltic bedrock (Dietrich, Windsor & Dunne 1992), are silty in texture and low in extractable nutrients (Table 1). The surface in the experimental plots remained undisturbed and covered with natural litter (300–400 g m–2 at the beginning of the experiment). Our experimental plants shared the soil with the surrounding forest vegetation. Within our plots, the abundance of fine roots in the top 10 cm of the soil profile varied between 120 and 150 g m–2. The canopy leaf area index (above the understorey) was about 6 during the wet and about 5·5 during the dry season (synchronized pair of ‘Plant Canopy Analyzers’, LAI-2000, LI-COR Inc., Lincoln, NE, USA; the reference sensor was mounted on a 40 m tower at 200 m distance).

Table 1.  . Soil analysis in experimental plots. Means ± SD for eight soil samples, four from the upper end and four from the lower end of the four tents. Mineral nutrient concentration in mg kg–1 dry mass of the 2 mm soil fraction (dried at 40 °C) Thumbnail image of


We selected five species commonly found in the immediate vicinity of our experimental plots. Two of them are tree seedlings [Beilschmiedia pendula (Sw.) Hemsl., Lauraceae; Tachigalia versicolor Standl. & L.O. Wms., Caesalpinioideae], two are evergreen shrubs (Piper cordulatum C. DC, Piperaceae; Psychotria limonensis Krause, Rubiaceae) and one is a shade tolerant grass (Pharus latifolius L., Poaceae). We subsequently refer to species by genus only.

Beilschmiedia and Pharus seeds were collected in the forest and germinated in an outdoor shade-house. Immediately after germination, one to three seedlings per species were planted at the same position within each of the 32 plots (see Experimental Design) with minimum soil disturbance. Tachigalia seedlings, which had produced one small leaf with three to five leaflets, and clonal cuttings (with a piece of rooted rhizome attached) of Piper and Psychotria (mostly with two to three leaves), were collected in the surrounding forest and transplanted directly into the experimental plots. At planting, leaf area per individual for Pharus, Piper and Psychotria was c. 10, 140 and 190cm2, respectively. Beilschmiedia and Tachigalia were added to the experiment 11 months later, when it became clear that the initial set of the experimental plants would not cover more than 10% of the ground area, owing to slower than expected growth. In each tent we planted 24 Pharus individuals, 16 Tachigalia, and eight individuals each of Psychotria, Piper and Beilschmiedia. Only a few individuals were lost during the experiment. Each of the 32 plots contained individuals from all species, with 20–30 cm distances between individuals. Hence, spacing of experimental plants was so wide that any below-ground interaction among them was highly unlikely and above-ground interaction did not occur. The few other seedlings present in the plots at the beginning were removed but otherwise all vegetation surrounding the tents remained undisturbed.


Thirty-two plots (0·75 m–2) were arranged in four tents (6 m × 1 m × 1 m high; 25 cm of vertical side walls topped by a triangular roof) that were placed across a narrow light gradient (mean daytime QFD 5–11 μmol m–2 s–1) in the understorey of the closed forest. The tent construction was a light frame from 20 mm electrical insulation tubes coated with clear polyethylene film (Hummert, MO, USA), absorbing c. 9% of incoming radiation. Sixteen plots (two tents with eight plots each) received ambient CO2 (daytime mean of 414 μl litre–1, see below) and two tents received elevated CO2 (daytime mean of 713 μl litre–1). Each of the 32 plots contained seven to eight plants: three Pharus, two Tachigalia, one Beilschmiedia (not in all plots), one Piper and one Psychotria at similar positions and two light sensors.


Tents received air collected at 2 m height by fans from one end. For the high CO2 treatment, pure CO2 was added at a constant rate by manually adjusting a needle valve. Regular measurements at two positions (at one-third and two-thirds of the length of the tent, twice a week) with a portable infrared gas analyser (EGM-1, PP Systems, Stotfold, Hitchin, Herts., UK) during daylight hours indicated a mean (± SD) CO2 concentration of 713 ± 54 μl litre–1 in the high and 414 ± 16 μl litre–1 CO2 in the low CO2 tents. Because of respiratory CO2 from the soil, mean CO2 concentrations in control tents were about 40 μl litre–1 above the ambient mean of 373 ± 13 μl litre–1 (see Results) and there was a gradient of about 10 μl litre–1 between inlet and outlet positions of each tent. CO2 concentrations were always higher at the outlet, so net plant CO2 uptake never exceeded soil CO2 evolution. The CO2 concentrations in the tents followed ambient trends with the CO2 addition rate remaining constant. We also measured the diurnal variation of ambient CO2 concentrations 0·5 m above ground with the same gas analyser on selected days during the experiment.

Temperatures were the same inside and outside the tents (a consequence of the very low radiation levels; repeated hand measurements during the dry and the wet season). The minimum and maximum temperatures inside the tents were 24·7 °C and 30·8 °C and outside the tents 23·6 °C and 30·1 °C (minimum/maximum thermometer). Inside the tents, midday humidity averaged 4% higher than in the forest during the dry season (tent: 82%; forest: 79%) with no measurable differences within or between tents. During the rainy season, humidity was generally higher and close to 100% for most of the day both, inside and outside the tents. Plots were watered with rain-water according to ambient precipitation patterns.


Within the tents QFD was measured 15 cm above the ground by 64 sensors (two in each of the 32 plots; GaAsP photo diodes, spectral response range 300–680 nm, G1115, Hamamtsu Photonics, Hamamtsu City, Japan), each sealed in a waterproof tube covered with a white diffuser cap. Each sensor was calibrated twice, once at the start of the experiment and once after 6 months, with a quantum-sensor (LI-189, LI-COR Inc., Lincoln, NE, USA). Readings were logged at 1 min intervals (CR 10 with relay multiplexer, Campbell Scientific Ltd, Loughborough, UK). For high frequency records (6 days in total) means for 3 min were stored, (about 14 000 values per 12 h day light period). In the longer term (15 months) hourly means of 1 min data were logged (= low frequency; 860 values per day). Of the high frequency data, 4 days fall in the wet season and 2 days in the dry season. High and low frequency data for the 32 microhabitat sets of two sensors correlated closely (r2 = 0·86), indicating that the differences captured over the short period of high frequency records are adequately reflecting the longer term trends. We used the high frequency data set to analyse the microhabitat specific frequency distribution of QFD.

Because the response of photosynthesis to QFD is non-linear, differences in QFD cannot a priori be expected to have a proportional effect on CO2 uptake (Fig. 3). Therefore we had accounted for this non-linearity by calculating relative rates of net photosynthesis (A, i.e. a biologically weighted QFD), with 100% equal to A at saturating QFD. For this procedure we used an equation from Küppers & Schulze (1985) which describes the curvature of the light response and we assumed a light compensation point of 3 μmol m–2 s–1, 50% of maximum A at 40 μmol m–2 s–1 and light saturation at 250 μmol m–2 s–1, typical values for such understorey plants (cf. Huber 1978; Mulkey 1986; K. Winter, unpublished data; the latter two references for Psychotria, Piper and Beilschmiedia of this study). However, QFD values falling in the non-linear part of the light response curve of A were so rare that a linear (initial slope only) and the non-linear procedure (whole curve) for describing the photosynthetically relevant microhabitat light regime produced the same results (hence the dashed line in Fig. 3 remained unaffected).


Figure 3. . The relative frequency of QFD (3 min logging interval for 64 sensors and six different days) and the estimated fraction of potential net photosynthesis (A; light compensation point, LCP, 3 μmol m–2 s–1; 50% of maximum A at 40 μmol m–2 s–1). Arrows indicate the daily means for all plots between 06.00 and 18.00 h.

Download figure to PowerPoint


We counted the number of leaves on each plant biweekly at first and then monthly. At the end of the experiment (in May 1995) all plants were harvested. The below-ground harvest required first soaking the soil with water. As in any field study, fine roots could not be recovered to 100% (although this was almost the case for the shallow rooted Psychotria, Piper and Pharus) and some of the finest roots were lost in the extremely solid subsoil. Rated by where the recovered roots broke in Tachigalia and Beilschmiedia, this was a rather minor dry-matter fraction and the error associated with this should be equally distributed across treatments. Also root mass typically represents a small fraction of total plant mass in such understorey species (< 10%). Each plant was divided into roots, stem and leaves, with petioles treated as stems. Leaf area (fresh) was measured with a portable area meter (conveyer belt LI-3050 A, LI-COR Inc., Lincoln, NE, USA) and all samples were dried at 65 °C beginning not later than 3 h after harvest. Specific leaf area (SLA) was calculated as projected leaf area per leaf dry matter.


At harvest we measured total non-structural carbohydrates (TNC) in leaves, as a measure of the plant’s carbon supply status. Increasing TNC can be used as an indicator of enhanced carbon uptake. After grinding, samples were first boiled for 40 min in distilled water and the soluble fraction analysed for sugars (see below). Aliquots of the solution were treated with isomerase and invertase and analysed for glucose using a Hexosekinase reaction kit (Sigma Diagnostics, St. Louis, MO, USA). In a second step the insoluble material including starch was incubated for 20 h at 40 °C with the dialyzed crude enzyme Clarase (a fungal α-amylase from Aspergillus oryzae var.; Miles Laboratory Inc., Elkhart, IN, USA). Starch and sugar standards as well as a laboratory standard of plant powder were used as controls for all analyses. Carbohydrates other than starch, sucrose, fructose and glucose are not covered by this assay (for more detail see Körner & Miglietta 1994).


Each plot with its plot-specific QFD was used as an independent replicate in the statistical analysis (n mostly between 14 and 16 for each CO2 treatment). CO2 and QFD effects and their interaction were tested using analyses of variance (type III ANOVA JMP version 3·1 released by SAS Institute, Cary, NC, USA). The dependent variable leaf number was analysed with repeated measure analyses using the MANOVA procedure of JMP. Each species was tested separately.

We tested effects of (1) CO2 as the controlled factor with two levels (low and high CO2), (2) light as a covariable being quantified as the mean for the two sensors within each of the 32 plots over the measurement period, (3) position of tent as a factor nested in CO2 with four levels, (4) individual plant height at start of the experiment as a covariable and (5) the interaction between CO2 and light. The factor position of tent was considered to be fixed. Starting with the model including all five factors we stepwise eliminated effects and interactions, if a particular error term was P > 0·25. This procedure allowed us to test and estimate the main effects more powerfully and precisely. P values presented in Table 3 are the result of this procedure.

Table 3.  . Summary of statistical analysis (ANOVA) Thumbnail image of


  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References


The mean CO2 concentration at the experimental site measured about 50 cm above the ground was 373 μl litre–1 (Fig. 1; data collected between 08.00 and 17.00 h). The highest concentrations in Fig. 1 are from early morning and late afternoon hours. During high QFD hours at midday and under calm weather ambient CO2 concentrations were as low as 320 μl litre–1 owing to photosynthetic CO2 depletion in the forest. Night-time concentrations and concentrations within the litter layer, temporarily exceeded 500 μl litre–1. It usually took less than 2 h during dawn and early morning for the higher night-time CO2 concentrations to approach normal ambient concentrations. During these transition periods, understorey QFD was less than 5 μmol m–2 s–1.


Figure 1. . Frequency distribution of ambient CO2 concentration at the experimental site within the closed forest. One-hundred and forty-four randomly collected hand measurements with a portable infrared gas analyser between 08.00 and 17.00 h on 2 days in early May (1995) about 50 cm above ground. No difference was found between these data and measurements at 2 m height at the fan inlet of the experimental tents (for long-term means in the experimental tent see the text).

Download figure to PowerPoint


The mean QFD of the low frequency data for 15 months of the experiment, was 8·3 μmol m–2 s–1 (hourly means for 64 sensors, Fig. 2). This is a value at which potential A (according to our model) reaches only about 7% of a theoretical maximum achieved during a full 12 h day.


Figure 2. . Daily means of QFD in the understorey (μmol m–2 s–1, 06.00–18.00 h, upper line, each point is the mean of 64 sensors) and daily sums of precipitation and QFD (dotted line) outside the forest about 200 m distant from the experimental area (these latter data by the Environmental Science Program of the Smithsonian Tropical Research Institute). Gaps in the data are owing to instrumental failures.

Download figure to PowerPoint

For a more specific analysis of the light regime at the experimental site we considered the high frequency data set for 6 days (Fig. 3; note that low and high frequency data are closely correlated; see Materials and methods). It is very important that such an analysis accounts for the biological effectiveness of QFD. However, we recall that the exact in situ light compensation point of the plant populations studied is not known but the physiological data used are the best available. With this caveat, almost half (45%) of all 3 min means of QFD between 06.00 and 18.00 h were below the assumed light compensation point of 3 μmol m–2 s–1. These accounted for 5% of the daily sum of QFD which we assume did not contribute to a positive A but reduced respiratory losses. About one third (37%) of all 3 min means were between 3 and 10 μmol m–2 s–1 representing 32% of the total quantum influx per day, (accounting for approximately one-third of the estimated daily CO2 uptake). QFDs between 10 and 20 μmol m–2 s–1 occurred during 13% of the daylight hours and represented 24% of the daily QFD sum (contributing a further third to the daily CO2 fixation). Although intensities higher than 20 μmol m–2 s–1 were infrequent (5% of all means), they represented 40% of the daily quantum supply to the understorey and were responsible for the final third of the estimated daily A. All of these estimates do not account for suboptimal induction of photosynthesis in very short sunflecks (Pearcy 1987). This would further reduce the photosynthetic gain. Figure 3 illustrates the interplay between frequency and potential photosynthetic effectiveness of QFD in this understorey habitat. Most of the predicted cumulative carbon gain (in relative terms) was achieved during periods when QFD was around 10 μmol m–2 s–1. The overall mean QFD for each plot covering various periods of both wet and dry seasons was used as a covariable in the analysis of CO2 responses (Fig. 4).


Figure 4. . Mean QFD for the 32 experimental plots (eight plots in a row per tent presented in the actual sequence of plot positions, data for 06.00–18.00 h). Examples are provided for the rainy seasons (October 1993 and 1994) and the dry season (January 1994). Solid line with symbols for the two rainy seasons, without symbols for the 15 months of available data irrespective of season, dotted line for the dry season. The wet and dry season data differ, reflecting both the difference in cloudiness and the higher leaf area index of the canopy during the wet season. Except for one tent, the ranking of plots (with respect to QFD) remained the same between seasons. The higher QFD at the lower end of one of the low CO2 tents in the dry season is probably owing to a gap in the canopy during this period.

Download figure to PowerPoint


Mean total leaf area per plant roughly quadrupled over the course of the experiment and positive CO2 effects were observed in all five species at the time of harvest (Pharus + 25%, Tachigalia + 33%, Psychotria + 45%, Beilschmiedia + 50% and Piper + 61%; Table 2). The difference as a result of elevated CO2 was significant in all species except Psychotria in which the naturally common leaf infections and subsequent leaf deformations caused greater leaf area variation among individuals.

Table 2.  . Responses of biomass (g), leaf area (dm2), SLA (dm2 g–1) and TNC (% dry matter) for all species (mean ± SE) Thumbnail image of

Repeatedly counted numbers of leaves (Fig. 5) are not directly convertible to biomass but illustrate at least three different types of temporal growth response. (1) Grass individuals (Pharus) produced the same number of leaves under both ambient and elevated CO2 but leaves grown in elevated CO2 had significantly (P < 0·04) more area. It took Pharus more than a year for a significant leaf area production to become visible during the second dry season. Similarly, Psychotria showed no difference in leaf production (number of leaves) over year one. Once plants started to flush in the second dry season, high CO2 plants actually produced fewer leaves than ambient CO2 plants but high CO2 leaves were much larger (P < 0·001). (2) Piper was stimulated continuously (although more during the second dry season) and, at harvest, plants in elevated CO2 had 14 leaves, compared to nine at ambient CO2, and the area of individual high CO2 leaves was significantly larger (P < 0·05). (3) Seedlings of the two species of trees, were unaffected by CO2 for the first 2 months after germination and planting. However, in the following 120 days, both species produced more leaves under elevated CO2.


Figure 5. . Repeated counts of leaf numbers per individual (leaflets in the dissected leaves in Tachigalia). Note that for the sake of graphical resolution y-axes do not start at zero and have different scales.

Download figure to PowerPoint


In elevated CO2, mean total plant biomass at harvest was 25–76% higher than in ambient CO2 depending on the species (Table 2). The analysis of variance, including initial leaf number and plant height as covariables, showed a significant CO2 effect on total biomass for all species (Table 3). The increase in biomass under elevated CO2 was mainly owing to increased above-ground biomass. Leaf mass per total plant mass (LMR, data not shown) was not influenced by elevated CO2 in any of the species. The biomass response was strongest in the shrubs of Piper and Psychotria. With microhabitat QFD increasing from about 5 to 11 μmol m–2 s–1 the absolute stimulation of total biomass became greater but the relative effect of CO2 on biomass was higher in the plots receiving lowest QFD, except for Pharus. In Piper, the relative stimulation by elevated CO2 was so pronounced that it removed the QFD sensitivity in this range of QFD (Fig. 6). Because of the large scatter the slopes of the two regressions did not differ significantly, except for Piper.


Figure 6. . The response of plant biomass to microhabitat QFD at ambient (open symbols) and elevated CO2 (closed symbols). QFD represents the 125 day mean for each of the 32 plots (two sensors per plot). Biomass includes leaves, stems and roots after 7 (tree seedlings) and 18 months (shrubs and grass) of treatment.

Download figure to PowerPoint


At harvest, specific leaf area (SLA, dm2 g–1) remained unaffected by CO2 in all species except Psychotria (– 14%). Despite the very low TNC concentrations in leaves (2·3–6·3%), CO2 effects were significant in Pharus (+ 27%) and Tachigalia (+ 40%) but not in the other species (Table 2).


  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References


Our analysis of the natural understorey CO2 regime during daylight periods is in line with all earlier observations: understorey plants experience atmospheric CO2 concentrations which differ little from those outside the forest at times when positive A is possible. The spatial distribution of QFD across our plots represents a twofold range in the most sensitive, linear part of the photosynthetic light response. QFDs greater than 40 μmol m–2 s–1 (less than 2% of full sunlight) apparently play a negligible role for carbon gain in this environment.


Plant growth data support our hypothesis that plants living in the deep shade of a humid tropical forest would show a marked positive growth response to elevated CO2. The prevailing QFDs of less than 10 μmol m–2 s–1 (Fig. 2), i.e. far below the light compensation point of plants adapted to full sunlight, did not preclude a growth stimulation by CO2 enrichment. However, the stimulation developed very slowly. For a valid comparison with the temperate zone one needs to account for the fact that the 18 experimental months in this tropical climate correspond to the accumulated duration of three full seasons in a temperate climate. According to the leaf number data for the shrubs and the grass, it took these plants a full year of growth to enter a phase during which a significant above-ground effect became measurable. Earlier in the experiment, assimilates may have been allocated to roots. Tree seedlings appeared to be more responsive (within 7 months) to elevated CO2 than the shrubs and the grass. However, leafing in most tropical plants is not a continuous process (e.g. Kursar & Coley 1991; Van Schaik, Terborgh & Wright 1993) but occurs periodically and is related to the plant’s assimilate reserves (besides other controls). We have captured just one pulse of leafing.

The Caesalpinioideae Tachigalia did not achieve a greater growth stimulation than the Lauraceae Beilschmiedia (as might have been expected for a legume). 15N values for all species were similar (δ15N 0·1‰ for Tachigalia vs a mean for all other species of – 0·3‰; unpublished data). However, we do not know how active the two to three tiny root nodules in Tachigalia actually were. Other observations suggest that plant growth in deep shade is commonly not nutrient limited (Denslow et al. 1990; Whitehead et al. 1995; Hättenschwiler & Körner 1996).

The mobile carbohydrate pools found here in the understorey are much smaller than those observed in fully sunlit or shaded leaves in the upper canopy (Würth & Körner 1998 and the relative increase in TNC owing to CO2 enrichment occurred in only two out of five species. Yet life in deep shade does not completely preclude the accumulation of starch in elevated CO2, confirming the observations by Körner & Arnone (1992).

The overall slow growth and delayed occurrence of a CO2 response in these late successional species corresponds to patterns found by Bazzaz & Miao (1993) for seedlings of temperate zone tree species grown in low light. In a parallel study in Panama, Piper cordulatum was grown in isolation in a controlled environment under QFD of 6·5μmol m–2 s–1 and showed a 54% increase in dry matter accumulation at elevated than at ambient levels of CO2 (K. Winter and A. Virgo, unpublished data). In another experiment, the growth responses of potted seedlings of Beilschmiedia to elevated CO2 were examined in open top chambers on Barro Colorado Island at 10 times higher QFD as in our experiment (Lovelock et al. 1996). Plants maintained for 140 days at elevated CO2 produced 30% more dry matter than plants at ambient CO2, with the presence of mycorrhiza (improved plant nutrition) further enhancing the response as compared to non-mycorrhizal controls.


Based on our results, it seems reasonable to assume that rising CO2 stimulates plant growth in the understorey of a closed tropical forest, at least during the early phase of recruitment. It is not possible to explain faster tree turnover among older age classes of trees, as implied by Phillips & Gentry (1994) and Phillips (1996). However, if seedlings of lianas were similarly advantaged by elevated CO2 as were the different plant types studied here (in the light of our data we see no reason for dismissing this possibility) the current CO2 enrichment could contribute to a greater survival or aggressiveness of these climbers during recruitment, which in turn could enhance forest dynamics. It is known that lianas invest very few nutrients in their early structures (Bigelow 1993) and are mainly carbon limited until they reach the canopy.

Furthermore, our experiment, like any other one in which responses of several species were compared (see Körner & Bazzaz 1996), indicates that individual species respond differently to elevated CO2. In the long term, this suggests that atmospheric CO2 enrichment is likely to influence species composition of tropical forests, regardless of the overall biomass responses (cf. Whitmore 1990; Körner 1995, 1998; Hättenschwiler & Körner 1996; Hättenschwiler et al. 1997).

However, from these results it cannot be concluded whether carbon sequestration in tropical forests will be enhanced. Although, a mature tropical forest in the Amazon basin has been reported to exhibit currently a positive carbon balance (Grace et al. 1995; Lloyd et al. 1995), growth stimulation alone, as reported here, is not sufficient to draw conclusions about carbon pools per unit land area (Körner 1998). Increased C-sequestration would require mean life span of trees to remain unaltered or become increased, or mean residence time of carbon in litter or humus to increase, for which we have no evidence (Hirschel, Körner & Arnone 1997; Körner 1998). Whatever the effects of elevated CO2 on the ecosystem carbon balance might be, our findings suggest that there is a realistic chance that forest dynamics are enhanced owing to CO2 fertilization.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

This project was supported by the A.W. Mellon Foundation through the Smithsonian Tropical Research Institute, Panama City and Washington DC, and by Freie Akademische Gesellschaft, Basel. We thank Anne Larigauderie and J.A. Arnone for their helpful comments. We are grateful to Monica Mejía, Milton García and Pilar Angúlo for field assistance and technical help, to Paul Jordan for statistical consultations and to Susanna Peláez-Riedl who helped with Figures and Tables, to Reto Stocker who designed and installed the light monitoring system and helped with tent construction.

  1. To whom correspondence should be addressed. E-mail:


  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  • 1
    Allen, L.H. Jr & Lemon, E.R. (1976) Carbon dioxide exchange and turbulence in a Costa Rican tropical rain forest. Vegetation and the Atmosphere, vol. 2 (ed. J. L. Monteith), pp. 265–308. Academic Press, London.
  • 2
    Arnone III, J.A. (1996) Predicting responses of tropical plant communities to elevated CO2: lessons from experiments with model ecosystems. Carbon Dioxide, Populations, and Communities (eds Ch. Körner & F. A. Bazzaz), pp. 101–121. Academic Press, San Diego.
  • 3
    Arnone III, J.A. & Körner, Ch. (1993) Influence of elevated CO2 on canopy development and red:far-red ratios in two-storied stands of Ricinus communis. Oecologia 12, 510515.
  • 4
    Arnone III, J.A. & Körner, Ch. (1995) Soil and biomass carbon pools in model communities of tropical plants under elevated CO2. Oecologia 12, 6171.
  • 5
    Baumgartner, A. (1967) Ecological significance of the vertical energy distribution in plant stands. Resources Naturelles 12, 367374.
  • 6
    Bazzaz, F.A. & Miao, S.L. (1993) Successional status, seed size, and responses of tree seedlings to CO2, light, and nutrients. Ecology 12, 104112.
  • 7
    Bazzaz, F.A. & Williams, W.E. (1991) Atmospheric CO2 concentrations within a mixed forest: implications for seedling growth. Ecology 12, 1216.
  • 8
    Bigelow, S.W. (1993) Leaf nutrients in relation to stature and life form in tropical rain forest. Journal of Vegetation Science 12, 401408.
  • 9
    Brown, S. & Lugo, A.E. (1982) The storage and production of organic matter in tropical forests and their role in the global carbon cycle. Biotropica 12, 161187.
  • 10
    Buchmann, N., Kao, W.Y., Ehleringer, J.R. (1996) Carbon dioxide concentration within forest canopies: variation with time, stand structure, and vegetation type. Global Change Biology 12, 421432.
  • 11
    Denslow, J.S., Schultz, J.C., Vitousek, P.M., Strain, B.R. (1990) Growth responses of tropical shrubs to treefall gap environments. Ecology 12, 165179.
  • 12
    Dietrich, W.E., Windsor, D.M., Dunne, T. (1992) Geología, clima e hidrología de la isla Barro Colorado. Ecología de un Bosque Tropical (eds E. G. Leigh Jr, A. S. Rand & D. M. Windsor), pp. 25–51. Tropical Research Institute, Balboa, Panama.
  • 13
    Drake, B.G. & Leadley, P.W. (1991) Canopy photosynthesis of crops and native plant communities exposed to long-term elevated CO2. Plant, Cell and Environment 12, 853860.
  • 14
    Elias, P., Kratochvilova, I., Janous, D., Marek, M., Masarovicova, E. (1989) Stand microclimate and physiological activity of tree leaves in an oak-hornbeam forest. I. Stand microclimate. Trees 12, 227233.
  • 15
    Gifford, R.M. (1992) Interaction of carbon dioxide with growth-limiting environmental factors in vegetation productivity: implications for the global carbon cycle. Advances in Bioclimatology 12, 2458.
  • 16
    Grace, J., Lloyd. J., McIntyre, J., Miranda, A.C., Meir, P., Miranda, H.S., Nobre, C., Moncrieff, J., Massheder, J., Malhi, Y., Wright, I., Gash, J. (1995) Carbon dioxide uptake by an undisturbed tropical rain forest in southwest Amazonia, 1992–93. Science 12, 778780.
  • 17
    Hättenschwiler, S. & Körner, Ch. (1996) Effects of elevated CO2 and increased nitrogen deposition on photosynthesis and growth of understory plants in spruce model ecosystems. Oecologia 12, 172180.
  • 18
    Hättenschwiler, S., Miglietta, F., Raschi, A., Körner, Ch. (1997) Thirty years of in situ tree growth under elevated CO2: a model for future forest response? Global Change Biology 12, 463471.
  • 19
    Hirschel, G., Körner, Ch., Arnone III, J.A. (1997) Will rising CO2 affect leaf litter quality and in situ decomposition rates in native plant communities? Oecologia 12, 387392.
  • 20
    Holdridge, L.R., Grenke, W.C., Hatheway, W.H., Liang, T., Tosi, J.A. (1971) Forest Environments in Tropical Life Zones: a Pilot Study. Pergamon Press, Oxford.
  • 21
    Huber, O. (1978) Light compensation point of vascular plants of a tropical cloud forest and an ecological interpretation. Photosynthetica 12, 382390.
  • 22
    Kira, T. & Yoda, K. (1989) Vertical stratification in microclimate. Tropical Rain Forest Ecosystems (eds H. Lieth & M. J. A. Werger), pp. 55–71. Elsevier Science Publishing, Amsterdam.
  • 23
    Koch, G.W. & Mooney, H.A. (1996) Response of terrestrial ecosystems to elevated CO2: a synthesis and summary. Carbon Dioxide and Terrestrial Ecosystems (eds G. W. Koch & H. A. Mooney), pp. 415–428. Academic Press, New York.
  • 24
    Körner, Ch. (1995) Biodiversity and CO2: global change is under way. GAIA 12, 234243.
  • 25
    Körner, Ch. (1998) Tropical forests in a CO2-rich world. Climatic Change 12, 297315.
  • 26
    Körner, Ch. & Arnone III, J.A. (1992) Responses to elevated carbon dioxide in artificial tropical ecosystems. Science 12, 16721675.
  • 27
    Körner, Ch. & Bazzaz, F.A. (1996) Carbon Dioxide, Populations, and Communities. Academic Press, San Diego.
  • 28
    Körner, Ch. & Miglietta, F. (1994) Long term effects of naturally elevated CO2 on Mediterranean grassland and forest trees. Oecologia 12, 343351.
  • 29
    Küppers, M. & Schulze, E.D. (1985) An empirical model of net photosynthesis and leaf conductance for the simulation of diurnal courses of CO2 and H2O exchange. Australian Journal of Plant Physiology 12, 513526.
  • 30
    Kursar, T.A. & Coley, P.D. (1991) Nitrogen content and expansion rate of young leaves of rain forest species: implications for herbivory. Biotropica 12, 141150.
  • 31
    Lloyd, J., Grace, J., Miranda, A.C., Meir, P., Wong, S.C., Miranda, H.S., Wright, I.R., Gash, J.H.C., McIntyre, J. (1995) A simple calibrated model of Amazon rainforest productivity based on leaf biochemical properties. Plant, Cell and Environment 12, 11291145.
  • 32
    Long, S.P. & Drake, B.G. (1991) Effect of the long-term elevation of CO2 concentration in the field on the quantum yield of photosynthesis of the C3 sedge, Scirpus olneyi. Plant Physiology 12, 221226.
  • 33
    Lovelock, C., Kyllo, D., Winter, K. (1996) Growth response to vesicular–arbuscular mycorrhizae and elevated CO2 in seedlings of a tropical tree, Beilschmiedia pendula. Functional Ecology 12, 662667.
  • 34
    Medina, E., Montes, G., Cuevas, E., Rokzandic, Z. (1986) Profiles of CO2 concentration and ∂13C values in tropical rain forests of the upper Rio Negro Basin, Venezuela. Journal of Tropical Ecology 12, 207217.
  • 35
    Mitscherlich, G., Kern, K.G., Künstle, E. (1963) Untersuchungen über den Kohlensäuregehalt der Waldluft in Plenterwald und Fichtenreinbestand. Allgemeine Forst- und Jagd-Zeitung 12, 281290.
  • 36
    Mooney, H.A., Drake, B.G., Luxmoore, R.J., Oechel, W.C., Pitelka, L.F. (1991) Predicting ecosystem responses to elevated CO2 concentrations. BioScience 12, 96104.
  • 37
    Mulkey, S.S. (1986) Photosynthetic acclimation and water-use efficiency of three species of understory herbaceous bamboo (Gramineae) in Panama. Oecologia 12, 514519.
  • 38
    Neftel, A., Oescher, H., Staffelbach, T., Stauffer, B. (1988) CO2 record in the Byrd ice core 50,000–5,000 years BP. Nature 12, 609611.
  • 39
    Oberbauer, S.F., Strain, B.R., Fetcher, N. (1985) Effect of CO2 enrichment on seedling physiology and growth of two tropical tree species. Physiologia Plantarum 12, 352356.
  • 40
    Pearcy, R.W. (1987) Photosynthetic gas exchange responses of Australian tropical forest trees in canopy, gap and understory micro-environment. Functional Ecology 12, 169178.
  • 41
    Phillips, O.L. (1996) Long-term environmental change in tropical forests: increasing tree turnover. Environmental Conservation 12, 235248.
  • 42
    Phillips, O.L. & Gentry, A.H. (1994) Increasing turnover through time in tropical forests. Science 12, 954958.
  • 43
    Rawson, H.M. (1992) Plant responses to temperature under conditions of elevated CO2. Australian Journal of Botany 12, 473490.
  • 44
    Reekie, E.G. & Bazzaz, F.A. (1989) Competition and patterns of resource use among seedlings of five tropical trees grown at ambient and elevated CO2. Oecologia 12, 212222.
  • 45
    Schimper, A.F.W. & Von Faber, F.C. (1935) Pflanzengeographie auf physiologischer Grundlage. Gustav Fischer, Jena.
  • 46
    Valle, R., Mishoe, J.W., Campbell, W.J., Jones, J.W., Allen, L.H. (1985) Photosynthetic responses of ‘bragg’ soybean leaves adapted to different CO2 environments. Crop Science 12, 333339.
  • 47
    Van Schaik, C.P., Terborgh, J.W., Wright, S.J. (1993) The phenology of tropical forests: adaptive significance and consequences for primary consumers. Annual Review of Ecology and Systematics 12, 353377.
  • 48
    Whitehead, D., Hogan, K.P., Rogers, G.N.D., Byers, J.N., Hunt, J.E., McSeveny, T.M., Hollinger, D.Y., Dungan, R.J., Earl, W.B., Bourke, M.P. (1995) Performance of large open-top chambers for long-term field investigations of tree response to elevated carbon dioxide concentration. Journal of Biogeography 12, 307313.
  • 49
    Whitmore, T.C. (1990) An Introduction to Tropical Rain Forests. Clarendon Press, Oxford.
  • 50
    Wong, S.C. & Dunin, F.X. (1987) Photosynthesis and transpiration of trees in a eucalypt forest stand: CO2, light and humidity responses. Australian Journal of Plant Physiology 12, 619632.
  • 51
    Wright, S.J. (1991) Seasonal drought and the phenology of understory shrubs in a tropical moist forest. Ecology 12, 16431657.
  • 52
    Würth, M.K.R., Winter, K., Körner, Ch. (1998) Leaf carbohydrate responses to CO2 enrichment at the top of a tropical forest. Oecologia 12, 1825.
  • 53
    Ziska, L.H., Hogan, K.P., Smith, A.P., Drake, B.G. (1991) Growth and photosynthetic response of nine tropical species with long-term exposure to elevated carbon dioxide. Oecologia 12, 383389.