Results and Discussion Model Validation

Before we discuss the sensitivity of 'Keeling plot' intercepts to variations in factors such as time, soil respiration, turbulent mixing, leaf area, and photo-synthetic capacity, we examine how well the model reproduces Keeling plot intercepts under field conditions. Figure 7.1 shows that the CANISOTOPE model is able to compute a Keeling plot intercept, for the summer growing season, that is indistinguishable from the one measured in the field; the measured intercept was —26.13 ± 0.56%o; and the computed intercept was —26.26 ± 0.01%o. We also report that the model was able to reproduce measured carbon isotope discrimination value in leaves and that computed values of isotopic flux densities agreed within 30% of measured values (Baldocchi and Bowling, 2003). With an acceptable level of agreement between measured and computed Keeling plot intercepts, we next use the model to diagnose the behavior of Keeling plot intercepts with respect to variations in controlling variables.

Model Sensitivity Studies

Two lines of evidence exist regarding the temporal variation of Keeling plot intercepts for 13C02. One body of data shows that the intercept does not change from night to day (Buchmann and Ehleringer, 1998;

0.0020 0.0021 0.0022 0.0023 0.0024 0.0025 0.0026 1/C02

Figure 7.1 A test of the ability of the CANISOTOPE model to compute Keeling plot intercepts within a broadleaved deciduous forest. The measured intercept was — 26.13 ± 0.56%o; the computed intercept was —26.26 ± 0.01 %o (after Baldocchi and Bowling, 2003). The geometric mean regression method was used to fit the data.

Mortazavi and Chanton, 2002), and another set of evidence suggests that it does (Bowling et al, 1999; Pataki et al., 2003).

Two theoretical Keeling plots are compared in Fig. 7.2. One is computed for daytime conditions and the other for nighttime conditions. The linear regressions through the two sets of data are offset from one another since

0.0020 0.0022 0.0024 0.0026 0.0026 0.0030 1/COg

Figure 7.2 Computed Keeling plots for a daytime and nighttime period on day 194 during the 1998 growing season; the data were fit with geometric mean regressions.

Day 194.1998 0200-0400 hours

• measured v computed

Figure 7.2 Computed Keeling plots for a daytime and nighttime period on day 194 during the 1998 growing season; the data were fit with geometric mean regressions.

Figure 7,3 Diurnal pattern of Keeling plot intercepts computed with the CANISOTOPE model. These results are for a typical summer day, D194,1998. The geometric mean regression technique was used to evaluate the regression intercepts and the error bars represent the statistical error associated with that estimate.

Time

Figure 7,3 Diurnal pattern of Keeling plot intercepts computed with the CANISOTOPE model. These results are for a typical summer day, D194,1998. The geometric mean regression technique was used to evaluate the regression intercepts and the error bars represent the statistical error associated with that estimate.

daytime CO2 concentrations are lower than nighttime values. The nighttime regression produces an intercept of —26.26 ±0.01 %o and the regression intercept during the daytime is —25.85 ± 0.34%o, values that are significantly different from one another.

Computing Keeling plot intercepts over the course of a typical summer day reveals that a diurnal pattern exists (Fig. 7.3). In general, Keeling plot intercepts are on the order of —26.3%o during the night; during the day the intercepts fluctuate between —25.4%o and —25.9%o. An important feature to notice in Fig. 7.3 is that the standard deviation of the calculated intercept is negligible at night (<0.01%o) but relatively large during the day (~0.3%o).

Figure 7.4 provides an explanation why there is a diurnal pattern in the Keeling plot intercept and why the standard error of the regression intercept is large during the day. A daytime Keeling plot is composed of two intersecting lines, rather than one continuous line. One line is associated with air above the canopy, the other line corresponding with air inside the canopy. The Keeling plot intercept derived for inside the canopy equals —26.2 ±0.01 %o, a value close to the intercept value derived from nighttime measurements that represents the respiring ecosystem. The Keeling plot intercept derived from air measured above the canopy yields an intercept of — 30.36 ± 0.01 %o. The isotope discrimination value associated with air above the upper canopy is lower than the value that is associated with the nighttime Keeling plot because the photosynthetic sink in the upper portion of the canopy is probably respired air recycling from the lower levels

Keeling intercept: -25.98 ±0.28%. • Below canopy; b0 =-26.2 %o » Above canopy: b0 =-30.36%o

Keeling intercept: -25.98 ±0.28%. • Below canopy; b0 =-26.2 %o » Above canopy: b0 =-30.36%o

1/C02

Figure 7.4 Keeling plots computed on separate profiles of air, from above and within the canopy. These theoretical estimates are derived from the CANISOTOPE model. D194, 1200 hours.

1/C02

Figure 7.4 Keeling plots computed on separate profiles of air, from above and within the canopy. These theoretical estimates are derived from the CANISOTOPE model. D194, 1200 hours.

of the canopy depleted in 13CC>2; the isotopic content of the recycled air has a more negative isotopic ratio (~ — 25.0%o) than that of air originating from above the canopy (~ — 8.0%o) (e.g., Sternberg, 1989). Ironically, the linear regression applied to both sets of data produces a less-negative Keeling plot intercept.

The split in the two regression lines is either due to the differential carbon sources and sinks, as discussed earlier, or is an artefact of counter-gradient transfer that is occurring at a zone within the canopy, as is demonstrated by the calculations presented in Fig. 7.5. For example, between the canopy-atmosphere interface (~24m) and 21 m, CO2 concentrations decrease as one moves into the canopy, reflecting down-gradient diffusion and the uptake of carbon dioxide by the canopy. Below 21m the CO2 concentration gradient changes sign and CO2 concentrations increase with further depth into the canopy (Fig. 7.5A) even though the canopy is a strong sink for CO2 down to 16 m (Fig. 7.5B). At the lowest levels, the source strength of CO2 released by the soil outpaces the vegetated sink (not shown); so, in this region, down-gradient transfer is occurring.

We emphasize that the subtle differences in the dependency of 513C on 1 / C for air inside and above the canopy may be overlooked when examining field data. First, sampling errors can be large (10-40%) due to the interrelation between the timescales of turbulence and the frequency of sampling

A D194 1200 hours

Temperate forest

A D194 1200 hours

Temperate forest

Source/sink strength, dF/dz ((.irnol m"3 s_1)

Figure 7.5 Computed vertical profiles of CO2 concentrations (A) and source sink strength (B) for Day 194 1200. Counter-gradient transfer is identified in the region between 16 and 21 m.

Source/sink strength, dF/dz ((.irnol m"3 s_1)

Figure 7.5 Computed vertical profiles of CO2 concentrations (A) and source sink strength (B) for Day 194 1200. Counter-gradient transfer is identified in the region between 16 and 21 m.

air at a given level (Baldocchi and Bowling, 2003). Second, the finite sampling period required at a given level in the atmosphere and the cost and labor requirements of analyzing such air with a mass spectrometer restricts the number of levels that can be sampled for a representative period of time. Model calculations, on the other hand, have the luxury of producing information for many discrete levels throughout the canopy.

Next we focus on soil, plant, and atmospheric factors that can theoretically modulate the Keeling plot intercept. First, let us examine the impact of changing rates of soil respiration on Keeling plot intercepts. By turning off soil respiration, theoretically, we observe a narrower range of CO<2

D194-195, 1998

D194-195, 1998

1/COz

Figure 7.6 Theoretical computations of Keeling plot intercepts for cases with and without soil respiration. The arrow direction and magnitude indicates the tendency of change.

1/COz

Figure 7.6 Theoretical computations of Keeling plot intercepts for cases with and without soil respiration. The arrow direction and magnitude indicates the tendency of change.

at night and a Keeling plot intercept that is 2.37%o lower than the reference case, a difference of 9% (Fig. 7.6). This lower (more negative) value reflects leaf respiration of the previous day's photosynthesis, which has a lower isotopic discrimination value than that of carbon decomposing in the soil; the isotopic signal of soil respiration can reflect isotopic enrichment by Basidiomycete fungi (Henn and Chapela, 2001) and the isotopic signal of last year's leaves and carbon in older soil pools, in addition to root respiration that is reflecting recent photosynthesis (Hogberg et al., 2001; Bowling et al., 2002).

Doubling the base rate of soil respiration produces greater concentrations of CO2 near the soil (Fig. 7.7). But this modification has a negligible effect on the Keeling plot intercept. In this circumstance, doubling the base rate of soil respiration causes the Keeling plot intercept to decrease by only 0.15 %o. So we conclude that Keeling plot intercepts are independent of the rate of respiration.

How variations in leaf area index affect isotopic discrimination is a question of much interest to isotope biogeochemists (Buchmann et al, 1997). In principle, variations in leaf-area index will have direct and indirect effects on the Keeling plot by changing the sink and source strengths for leaf photosynthesis and respiration, by altering the fraction of sunlit and shaded leaves, by altering turbulent mixing in the canopy, and by changing the pool size of decomposing litter and the amount of respiring roots. Data in Fig. 7.8 represent results from a numerical experiment where we doubled leaf-area index of the forest. The influence this change had on the Keeling plot

Figure 7,7 Theoretical Keeling plots for conditions of 2 times the base respiration case, which equalled 0.61 pmol m~2 s^1. D194, 0300 to 0400 hours.

1/C02

Figure 7,7 Theoretical Keeling plots for conditions of 2 times the base respiration case, which equalled 0.61 pmol m~2 s^1. D194, 0300 to 0400 hours.

intercept was minor (a decrease of 0.12%o), a result that is consistent with the experimental findings of Buchmann et al. (1997). In general, calculated values of <513C and C02 moved up and down the regression line without greatly altering the slope or intercept.

Since we are using a Lagrangian-based turbulence scheme, these computations do not consider how changes in leaf area will alter turbulent mixing; for example, turbulent mixing deep within a canopy increases as leaf area

Figure 7.8 Impact of doubling leaf-area index on the Keeling plot for a nighttime period during day 194, 1998. The arrow directions and magnitudes identify the tendency of change.

1/COs

Figure 7.9 The impact of altering turbulent mixing in the canopy on the Keeling plot. The arrow directions and magnitudes identify the tendency of change.

1/COs

Figure 7.9 The impact of altering turbulent mixing in the canopy on the Keeling plot. The arrow directions and magnitudes identify the tendency of change.

decreases (Dwyer etal., 1997; Massman and Weil, 1999). In contrast, a higher order closure Eulerian model can accommodate this feature because it calculates explicitly the impact of canopy drag on the kinetic energy and momentum exchange budgets of the canopy (see Meyers and Paw U, 1987; Katul and Albertson, 1999). Despite this limitation, we can mimic the interaction between turbulence and leaf-area index by prescribing a change in turbulent mixing within the canopy. Doubling turbulent mixing in the understorey—as quantified by the standard deviation of vertical velocity fluctuations near the soil—increases the Keeling plot intercept by only 0.10%o (Fig. 7.9). The biggest change that occurs is a shifting of CO2 concentrations up and down the linear regression line; as turbulent mixing increases, the range of CO2 and 13C02 concentrations encountered diminishes.

Despite the results shown in Figs 7.8 and 7.9, we do not imply that leaf-area index has no impact on carbon isotope discrimination of a forest canopy, A canopy- Theoretical calculations, presented in Fig. 7.10, show that the canopy isotopic discrimination value is about 10% less negative with a 50% decrease in leaf-area index. Additional model calculations (not shown) also indicate that the average value of Q/Ca for sunlit leaves is lower than that for shaded leaves (~0.75 vs ~0.89). Reducing canopy leaf-area index increases the proportion of sunlit leaves and thereby forces the canopy isotopic discrimination value to be lower during daylight hours. It must also be recognized that changes in photosynthesis and isotopic discrimination, imposed by changes in leaf area, will affect the signature of carbon that is respired later, a feedback that is not satisfactorily simulated in our

D140-240, 1998

D140-240, 1998

Figure 7.10 The impact of leaf-area index on the canopy isotopic discrimination for a temperate forest over the course of the summer growing season. The arrows identify the tendency of the change.

"base

Figure 7.10 The impact of leaf-area index on the canopy isotopic discrimination for a temperate forest over the course of the summer growing season. The arrows identify the tendency of the change.

computations; at present we respire carbon from the litter and roots with the signal of the previous day's photosynthesis.

Next we examine how physiological capacity (Fig. 7.11) and stomatal conductance (Fig. 7.12) alters in the Keeling plot. The Keeling plot intercept is weakly associated with Vcmax and the stomatal conductance proportionality constant because changes in photosynthetic and stomatal capacity do not change Q/Ca appreciably. In practice, sunlit leaves tend to open and close their stomata in concert with changes in photosynthesis to keep Q/Ca rather conservative, near 0.7 (Wong et al, 1979; Baldocchi, 1994). In addition, because these Keeling plots are computed at night, they do not reflect the impact that long-term changes in physiological capacity may have on the isotopic signature of litter and roots; in these calculations only dark respiration scales directly with photosynthetic capacity.

We computed stomatal conductance by coupling it to photosynthesis {A), relative humidity (rh) and CO2, gs = k(A ■ rh/Ca). When plants experience soil water deficits, it is common to see a decline in the proportionality constant, k (Sala and Tenhunen, 1996). Our model calculations, shown in Fig. 7.12, indicate that an alteration in the stomatal conductance proportionality constant, k, may not alter the Keeling plot intercept by much, despite the fractionation of carbon that occurs when it diffuses through the stomata. On the other hand, a change in k does affect and lower canopy isotope discrimination (Fig. 7.13). On the basis of this finding, we could expect long-term drought to affect the isotopic signal of decomposed leaves and root respiration that would be measured later in the year.

0.0020 0.0022 0.0024 0.0026 0.0028

1/C02

Figure 7.11 Impact of changing photosynthetic capacity on Keeling plots. In these calculations dark respiration scales with maximum carboxylation velocity, Vrmax. The arrow direction and magnitude indicates the tendency of change.

0.0020 0.0022 0.0024 0.0026 0.0028 1/C02

Figure 7.12 Impact of the stomatal conductance proportionality constant (e.g., the Ball-Berry stomatal conductance parameter) on the 13C-C02 Keeling plot. The arrow indicates the tendency for change in direction and magnitude of the Keeling plot.

Was this article helpful?

0 0

Post a comment