SPM2 simulations compared to eddy flux measurements

Evaluation of modeled carbon fluxes for a slash pine ecosystem: SPM2 simulations compared to eddy flux measurements

Clark, Kenneth L

ABSTRACT. A process-based model (SPM2) was developed to simulate net hourly exchanges of carbon for slash pine (Pinus elliottii var. elliott#) ecosystems in north-central Florida, and simulated values were compared to those obtained using eddy covariance measurements above the canopy. The SPM2 simulated relatively low hourly rates of net carbon exchange at high irradiance during the day, especially during the month of May. This discrepancy contributed the most to differences between the two approaches, and suggests that the model underestimates the maximum rate of leaf-level net photosynthesis during the period of rapid growth in the spring. The SPM2 also simulated lower net carbon release during the nighttime across the range of air temperatures, relative to eddy covariance measurements. As a result of these discrepancies, accumulated daytime and nighttime sums were quite different between the two approaches. However, once results were aggregated over 24 hr, differences between simulations and measurements were insignificant at all but the highest and lowest daily net carbon gain values. Annual estimates for 1996 were similar: 693, 719, and 745 g C m-2 yr-1 for the SPM2 model, eddy covariance measurements and mass balance measurements, respectively. This model evaluation has lead to specific research questions for guiding further model development. FOR. Sci. 47(1):52-59.

Key Words: Simulation models, carbon dynamics, eddy covariance, slash pine, Florida.

ECOSYSTEM-LEVEL MODELS ARE ESSENTIAL TO the estimation of forest carbon balances for three major reasons. First, truly continuous direct measurements of net ecosystem C02 fluxes using micrometeorological techniques (i.e., eddy covariance) are not possible, so that extrapolation to periods of nonmeasurement is always necessary. Second, disaggregation of micrometeorological measurements at the ecosystem level into component processes for further analysis can only be accomplished using models. Third, hypotheses about the effects of future climate scenarios on the carbon dynamics of forests can only be evaluated using models.

Interpretation of carbon dioxide flux measurements made using micrometeorological techniques can be problematic. For example, under low wind conditions, increased CO2 concentrations below the sensor can diffuse upwards or “drain” along concentration gradients created by local topography and are typically not accounted for without additional measurements. Other problems arise during wet conditions, because signal transmission can be altered when the sensor heads on sonic anemometers are wet. Thus, much nighttime eddy covariance data is routinely discarded. Since annual flux estimates require continuous data, methods for “filling in” missing data are required. What the most appropriate approach should be is debated extensively within the scientific community (Ruimy et al. 1995, Goulden et al. 1996, Lavigne et al. 1997).

Process-level models can be used both for filling in missing eddy flux data, as well as to disaggregate such measurements into their component processes for further analysis. Micrometeorological techniques measure net ecosystem fluxes of C02 at a plane above the canopy, thus cannot be used alone to evaluate how environmental conditions might drive, for example, rates of canopy vs. understory photosynthesis, or autotrophic vs. heterotrophic respiration. C02 flux measurements can also be used in the development of empirical estimates of gross ecosystem production and ecosystem respiration, whose evaluation is then possible using independent model simulations (Waring et al. 1995, Law et al. 1998). Finally, models can be used to test hypotheses about the effects of future climate scenarios on the carbon dynamics of forests, such as the extensive pine plantations we have studied in the southeastern United States (Cropper and Gholz 1994, Cropper 1998).

Before models can be used in conjunction with eddy covariance data, however, it is important to understand how these two approaches compare across a range of time scales (e.g., hourly, daily, annually). This type of evaluation can reveal components or processes that may not be simulated correctly, or where there may be sampling biases in the eddy covariance data (Lavigne et al. 1997). At annual to longer time scales, both of these approaches can be evaluated against independent field measurements of biomass increment, decomposition, and litterfall (i.e., mass balance), although this latter information can be time-consuming and difficult to obtain accurately for forests with complex structure or slow tree growth.

Our objective was to compare rates of net ecosystem C02 exchange and net C gain simulated using the Slash Pine Model version 2 (SPM2, Cropper 1999) to those obtained using eddy covariance and mass balance for a managed slash pine (Pines elliottii var elliottii Engelm.) ecosystem in Florida.

Methods and Materials

Site Description

The study site was established in 1994, 15 km northeast of Gainesville, Alachua Co., FL (29 deg 44’N, 82″9’30″W). Long– term (1961-1990) mean annual air temperature was 21.7 deg C, and mean air temperature in January and July was WCC and 27 deg C, respectively (NOAA 1996). Mean monthly minimum and maximum temperatures throughout the study period (January 1995 to December 1998) ranged from 5.9 deg C to 19.5 deg C and 21.5 deg C to 32.9 deg C, respectively. Long-term mean annual rainfall for Gainesville was 1,342 mm, and annual precipitation in 1996, 1997, and 1998 was 1,391, 1,585, and 1,425 mm, respectively.

The site was level and was dominated by a 24-yr-old (in 1996) plantation of slash pine established on industry land and managed on a 25 yr rotation for the production of pulpwood. Soils of the site are ultic alaquods (sandy, siliceous, thermic), poorly drained and low in organic matter and available nutrients. The distributions of discontinuous subsurface spodic (organic) and argillic (clay) horizons range between 30-70 cm and 100-150 cm depth, respectively (Gaston et al. 1990). The surface soil water table fluctuated from ground level to over 1.5 m depth during the measurement period. Mixed genotype seedlings were planted at harvest density following stem-only harvest of the previous stand, chopping and broadcast burning of residues, and bedding. The stand had not been thinned or fertilized since establishment. Understory vegetation consisted of native species reestablished naturally after site preparation, primarily Serenoa repens (W. Bartram) Small, Ilex glabra (L.) A. Gray and Myrica cerifera L. The planting block used for this study was 100 ha, with a minimum fetch from our measurement tower of approximately 800 m, and was surrounded by other slash pine plantations ranging from 9 to 20 yr old.

Tree densities, heights, and stem diameters at 1.3 m height (dbh, cm) were measured annually on four 25 x 25 m plots (Clark et al. 1999). Tree biomass and growth increments were estimated from allometric relationships based on dbh (Gholz et al. 1991). Understory biomass was estimated from census data in each plot using allometric relationships based on various plant dimensions (Gholz et al. 1999, and personal observations). Fine litterfall was collected biweekly from ten 1 x 1 m traps at random locations in two of the four measurement plots. Forest floor mass was sampled in subplots throughout the duration of the study, and coarse woody debris was sampled in each plot at the end of the study.

The Slash Pine Model

The SPM2 model is an extension of a previously described model (SPM) developed to simulate the C dynamics of slash pine plantations in northern Florida (Cropper and Gholz 1993a,b, 1994). The original model was evaluated by comparing model output to independent data that was not used for calibration, including daily soil CO2 evolution, monthly labile carbon pool (starch and sugars) mass, annual stem growth, and stem biomass accumulation over 20 yr. Although the model showed reasonable agreement with observations (Cropper and Gholz 1993a), it was not tested against hourly observations or at sites other than those close to the main calibration plots.

The slash pine canopy is quite dynamic seasonally (Gholz et al. 1991), and has significant departures from the assumptions of the Beer-Lambert Law for light attenuation through plant canopies. The SPM2 uses logistic equations to model relative needle elongation and cumulative litterfall of two age classes of foliage (old and new) as a function of time, starting on March 1 of each year (typically just before budbreak; Cropper and Gholz 1993a). In order to correct for departures from the assumption of a uniform canopy, the simulated canopy is divided into nine layers each 1 in thick, a correction is made for relative gap frequency (Sinclair and Knoerr 1982), and a correction for the hourly solar elevation angle is applied (Brock 1981).

where PPFD is the photon flux density ((mu)mol m-2s-1) of photosynthetically active radiation, ap is the quantum efficiency (mol mol-1), kc is mesophyll conductance ((mu)mol m^sup -2^s^sup -1^ (mu)bar-1) and C^sub i^ is the mean internal leaf CO2 concentration ((mu)bar) for the mean ambient CO2. The asymptotic value of simulated leaf-level slash pine assimilation (A^sub max^) is determined as the limit of Equation (1) as PPFD approaches infinity (1.715 (mu)mol m-2s-1 for old foliage and 2.38 (mu)mol m-2s-1 for new foliage). The calculated Amax values are similar to the value of 2.24 (mu)mol m2s-1 reported by Teskey et al. (1994a), based on cuvette measurements in similar nearby stands.

where g^sub s^ is the stomatal resistance and C^sub a^ is the ambient concentration of CO2. Using this formulation, the k^sub c^C^sub i^ term of Equation (1) can be thought of as a new constant with units of (mu)mol m-2s-1. The resulting equation is then similar in form to Equation (1) in Teskey et al. (1994) or the assimilation term of Equation (7) below.

The SPM simulation model was originally designed, in part, to attempt a realistic simulation of slash pine carbon dynamics without including physiological details that might not be necessary to successfully scale up to the stand level. Analysis of the cuvette data (Teskey et al. 1994) indicated that PPFD explained 69% of the variation in assimilation, with vapor pressure deficit and air temperature improving the fit by 7 and 2%, respectively. Although it is simple to attribute all variation in assimilation to environmental factors, it should be remembered that biological factors such as tree genetics and health, and needle phenology and health, may also contribute to this variation. The simulated assimilation function represents a mean response to light, and therefore does not capture the observed extreme values. We made the assumption that given a large enough population of needles (stand level), the mean light response would adequately simulate aggregate behavior.

where KM is the Michaelis-Menten constant (the PPFD value leading to an ASSIM value of half of the Amax value). Palmetto A^sub max^ was set at 3.499 (mu)mol m-2s-1 and gallberry A^sub max^ at 3.30 (mu)mol m-2s-1, based on a more limited set of physiological measurements (Gholz and Cropper unpubl. data).

where for each tissue (i), the base rate (BR(i)) is the respiration rate at 20 deg C (g CO2 g-1 h-1 ), XW is the tissue dry mass (g), TEMP is the ambient temperature (OC), and Q^sub 10^ is the rate of change in respiration per 10 deg C change in temperature. A similar temperature dependence was used to simulate hourly decomposition of detritus (Cropper 1999). Growth respiration was simulated as a linear function of tissue growth, with the growth respiration coefficient (C(i), g COZ g-1 new tissue growth), based on estimates for slash pine from Chung and Barnes (1977).

Hourly meteorological variables (above-canopy PPFD, air temperature and relative humidity) measured at the site were used to drive the model. Additional details on model structure, performance, sensitivity analyses and evaluation can be found in Cropper and Gholz (1993a) and Cropper (1999).

Eddy Covariance Measurements

where F^sub CO2^ is the net ecosystem flux of CO2 and F^sub DeltaS^ is the flux associated with the change in storage of CO2.

Net fluxes were measured using a closed-path eddy covariance system (EdiSol; Moncrieff et al. 1997). The system was composed of (1) a 3-dimensional sonic anemometer (A 1002R, Gill Instruments Ltd., Lymington, UK) mounted at the top of a 24 rn antenna tower; (2) a fast-response, infrared gas analyzer (LI-6262, Li-Cor Inc., Lincoln, NB); (3) a 30 m long, 0.4 cm ID nylon tube, a flow controller and a small air pump; and (4) a laptop PC running EdiSol software. The inlet of the tube was placed between the upper and lower sensors of the sonic anemometer, and air was drawn through the LI-6262 ata rate of 6.0 1 min-1. Raw data were collected at 21 Hz. The EdiSol software carries out coordinate rotation of the raw sonic anemometer signals to obtain turbulence statistics perpendicular to the local streamline. The maximum values for the covariance between turbulence and CO2 concentrations were compared to a 200 s running mean to calculate instantaneous fluxes. Average net CO2 fluxes were then calculated at hourly intervals. Finally, fluxes were corrected for the frequency attenuation of the gas concentration down the sampling tube, sensor separation loss and the nonideal frequency response of the LI-6262 using transfer functions (Moncrieff et al 1997).

Hourly changes in the mean CO2 concentration at the height of the sampling inlet were used to estimate F^sub DeltaS^ in the volume of air beneath the inlet (24 m^sup 3^; cf. Hollinger et al. 1994). For the analyses presented here, only daytime and nighttime data collected during turbulent air periods, when the canopy was well-coupled with the atmosphere (e.g., the friction velocity, u*, was >0.2 m s-1), were used (Goulden et al. 1996, 1997). During these periods, the sum of sensible and latent heat accounted for >80% of net radiation, and the balance was assumed to be primarily heat exchange by the biomass and soil.

Meteorological Measurements

Incoming PPFD (LI-190, Li-Cor, Inc.), air temperature, and relative humidity (#ES- 110, Omnidata, Inc., Ogden, UT; mounted in a Stevens enclosure) and precipitation (TI-525, Texas Instruments, Inc., Dallas, TX) were measured within 1 m of the sonic at the top of the tower. Soil temperature was measured at 5 cm depth (# ES-060, Omnidata, Inc.). Meteorological data were recorded with automated data loggers (Easy Logger #EL824-GP, Omnidata, Inc.).

Data Analyses

where alpha is the apparent quantum yield (dF^sub CO2^/dPPFD at PPFD = 0), F^sub sat^ is the net CO2 exchange at light saturation and R is the mean net CO2 exchange at PPFD = 0.

where a and b are regression coefficients and T is the mean hourly air or soil temperature. SigmaPlot 5.0 Regression Wizard software (SPSS, Inc., Chicago, IL) was used to estimate parameters in Equations (6) and (7). Both types of data were then compared at daily, monthly, and annual time scales. Annual values were also compared to a mass balance of tree woody biomass increment (stems + branches + coarse roots) plus litterfall, corrected for decomposition, using data collected at the site.

Results and Discussion

Daytime Carbon Exchange

Daytime hourly net CO2 exchange rates simulated using the SPM2 were significantly different than rates measured using eddy covariance (Figure 1; t277 = 9.915, P

The apparent “compensation points” for net CO2 exchange occurred at similar PPFD levels for both approaches (ca. 100 and 125 (mu)mol m-2 s-1 for SPM2 simulations and eddy covariance measurements, respectively). However, simulated daytime hourly net CO2 exchange saturated at lower light levels when compared to eddy covariance values. For example, at a PPFD of 1500 (mu)mol m-2 s-1, the mean simulated net CO2 exchange was ca. 12 (mu)mol m-2 s-1, compared to 18 (mu)mol CO2 m-2 s-1 for the eddy covariance data (Figure 2, closed circles). The discrepancy was particularly marked during May, when the model simulated maximum fluxes of only 8 lmol m-2 s-1 at the highest light levels (Figure 2, open triangles).

The relatively large difference between the two approaches in the spring may be a result of several factors. First, the SPM2 likely underestimates the maximum rate of leaf-level photosynthesis, Amax, during this period when growth is at a maximum and at least the new foliage age class is a large internal sink for carbon (Hendry and Gholz 1986, Gholz and Cropper 1991). Leaf-level net photosynthesis in the SPM2 is based on an average annual value for Amax, developed from extensive leaf-level diurnal cuvette data collected biweekly over 2 yr (Teskey et al. 1994a). However, springtime conditions for the new foliage were underrepresented in the leaf– level cuvette data, because we were unable to use the cuvette system on the small needles emerging between March and June. As a result, complete diurnals on new foliage were not obtained until mid-June, when these needles were almost fully expanded. Therefore, seasonal changes in Amax of new foliage due, for example, to increased sink strength during periods of active tissue growth, are not represented by the model. Seasonal changes in Amax have been documented for other pines, including the related Pinus taeda (Teskey et al. 1994b). A possible solution is to simulate seasonal variation in the Amax of new foliage explicitly, using data from a different cuvette system obtained through further research. This is not an issue for old foliage, since fully expanded needles were sampled year round by Teskey et al. (1994a).

Even if this could account for the discrepancies in May, there appears to still be a more general problem of agreement between the SPM2 and eddy covariance measurements at high irradiance (Figures 1 and 2). It may be that the SPM2 misrepresents the foliage distribution and/or the relative gap frequency of the slash pine canopy at the field site, so that light conditions deeper in the canopy are inadequately simulated. Although seasonal canopy light attenuation and pine litterfall in the SPM2 model were similar to field measurements, foliage distribution and relative gap frequency in the model were estimated from research at previous research sites and in different years. Therefore, the layered canopy submodel of SPM2 may not adequately simulate light interception by the canopy at the current site during high light conditions. At the previous stands, PPFD under the pine canopy, as predicted with the Beer-Lambert equation using seasonal canopy LAI, a cosine-corrected k value and incident PPFD, showed good agreement with measured values, with no apparent seasonal bias (Gholz et al. 1991). This suggests that LAI may be underestimated in the model, or more likely that the rate of LAI accumulation in the spring of these particular simulation years was underestimated. Simulated values match eddy covariance measurements later in the summer because needles are fully expanded and haze and clouds are more common, reducing incident levels at the canopy level compared to those in the spring (Ewel and Gholz 1986).

Finally, it is also possible that the SPM2 underestimates net CO2 gain by the understory in the spring and early summer. Direct beam light penetration through the pine canopy is greatest in May, because new needles in the canopy have not yet fully expanded (i.e., LAI is not yet at a seasonal maximum), the solar zenith angle is at a minimum, and clear sky conditions prevail. Since the SPM2 uses a layered canopy model, it cannot simulate local areas of direct beam radiation reaching the understory, so that high spring rates of understory carbon gain may be not be simulated and understory carbon gain during these periods underestimated. Physiological data for the understory is less abundant than for the pine canopy, which may also lead to incorrect parameterization. Independent estimates using destructive sampling and mass balance approaches for these sites support a low annual contribution to ecosystem carbon gain (e.g., an understory contribution equal to 7-9% of overstory aboveground net primary production, Gholz et al. 1999). This suggests that whatever misrepresentations of understory dynamics there might be in the model, the impact on ecosystem carbon balances is relatively small.

Nighttime Carbon Exchange

Nighttime hourly net C02 exchange rates (ecosystem respiration) simulated using the SPM2 were also significantly different than rates measured using eddy covariance during turbulent conditions when the canopy was well– coupled to the atmosphere (t^sub 261^ = 6.175, P

Underestimation of ecosystem respiration by the SPM2 could result from underestimation of pool sizes or basal rates of litter and/or soil organic matter decomposition and/or needle and foliage respiration. We used aboveground tree, understory, and litter biomass measured at the 24-yr-old site to parameterize the SPM2 for these simulations. Decomposition of needle and fine root litter has now been measured independently in two studies (Gholz et al. 1985, 2000), and results indicate a slightly higher decomposition rate than currently used in the model. For example, a decomposition rate of 15% yr-1 for all needle and fine roots litter was used in these simulations (Gholz et al. 1985), while the more recent study suggests a rate of 18-20% yr-1 may be more appropriate. As a further test, we are currently comparing simulated soil CO2 emission with independent field measurements. In addition, as is the case with leaf-level photosynthesis, there may be biases in the previous in situ small chamber measurements of needle and fine root respiration, which can be determined by further measurements. Understory and bole respiration are only small components of total ecosystem respiration, so are not likely to be major contributing factors (Ryan et al. 1995, Cropper and Gholz 1994).

Daily, Monthly, and Annual Carbon Balances

Because the relationships between hourly fluxes in night or day and PPFD or air temperature were significantly different between SPM2 simulations and eddy covariance measurements, accumulated daytime and nighttime sums were also often quite different (Table 1). For example, when mean daytime and nighttime net COZ exchanges were compared by month, differences in daytime C gain exceeded 2 g m-2 day-1 during the month of April, and differences in nighttime C release exceeded -0.7 g m-2 day-1 from December through April.

However, once results were aggregated over 24 hr, differences between simulated and measured net C gain were insignificant at all but the highest (> 3.5 g C m-2 day-1) and lowest (

Conclusions

The evaluation of whole-ecosystem carbon balances simulated by “process models” using independent measurements has only become possible with the advent of eddy covariance techniques. However, neither approach is without potential error or bias. Difficulties are most often observed in “real time” comparisons (at half-hourly or hourly time step). Our results suggest that some process-level measurements used in constructing the SPM2 model may contain biases that when scaled to the ecosystem level lead to the underestimation of both daytime net carbon gain at relatively high light levels and nighttime release across the range of air temperatures. This resulted in a substantial underestimation of carbon gain by the model during the late spring, while simulated values were similar to eddy covariance measurements during other times of the year.

Results also suggest that the use of the simple empirical relationships based on PPFD and temperature may be more appropriate for “filling in” missing short-term (i.e., less than 24 hr) eddy covariance values than the more complicated process model, at least as currently formulated. However, for longer periods, either approach appears appropriate. Finally, as a result of model evaluation, we have identified specific research questions that can be answered through field research and then used to guide further model development.

Literature Cited

BROCK, T.D. 1981. Calculating solar radiation for ecological studies. Ecol. Model. 14:1-19.

CHUNG, H.H., AND R.L. BARNES. 1977. Photosynthate allocation in Pines taeda. I. Substrate requirements for synthesis of shoot biomass. Can. J. For. Res. 7:106-111.

CLARK, K.L., H.L. GHOLZ, J.B. MONCRIEFF, F. CROPLEY, AND H.W. LOESCHER. 1999. Environmental controls over net exchanges of carbon dioxide from contrasting Florida ecosystems. Ecol. Appl. 9:936-948.

CROPPER, W.P., JR. 1998. Modeling the potential sensitivity of slash pine (Pines elliottii) stem growh to increasing temperature and CO2. P. 353– 366 in The productivity and sustainability of southern forest ecosystems in a changing environment, Mickler, R.A., and S. Fox (eds.). Springer– Verlag, New York.

CROPPER, W.P., JR. 1999. SPM2: A simulation model for slash pine (Pines elliottii) forests. Ecol. Model. (in press).

CROPPER, W.P., JR., AND H.L. GHOLZ. 1993a. Simulation of the carbon dynamics of a Florida slash pine plantation. Ecol. Model. 66:231-249.

CROPPER, W.P., JR., AND H.L. GHOLZ. 1993b. Constructing a seasonal carbon balance for a forest ecosystem. Clim. Res. 3:7-12.

CROPPER, W.P., JR., AND H.L. GHOLz. 1994. Evaluating the potential response mechanisms of a forest stand to fertilization and night temperature: A case study using Pines elliottii. Ecol. Bull. (Copenhagen) 43:154-160.

EWEL, K.C., AND H.L. GHOLZ. 1991. A simulation model of the role of belowground dynamics in a Florida pine plantation. For. Sci. 37:397– 438.

GASTON, L., P. NKEDI-KizzA, G. SAWKA, AND P.S. RAO. 1990. Spatial variability of morphological properties at a Florida flatwoods site. Soil Sci. Soc. Am. J. 54:527-533.

GHOLZ, H.L., AND W.P. CROPPER, JR. 1991. Carbohydrate dynamics in mature Pines elliottii var. elliottii trees. Can. J. For. Res. 21:1742-1747.

GHOLZ, H.L., D.N. GuERIN, AND W.P. CROPPER, JR. 1999. Phenology and productivity of saw palmetto (Serenoa repens) in a north Florida slash pine plantation. Can. J. For. Res. 29:1248-1253.

GHOLZ, H.L., C.S. PERRY, W.P. CROPPER, JR., AND L.C. HENDRY. 1985. Litterfall, decomposition, and nitrogen and phosphorus dynamics in a chronosequence of slash pine (Pines elliottii) plantations. For. Sci. 31:463-478.

GHOLZ, H.L., D.A. WEDIN, S.M. SMITHERMAN, M.E. HARMON AND W.J. PARTON. 2000. Long-term dynamics of pine and hardwood litter in contrasting environments: toward a global model of decomposition. Global Change Biol. (in press).

GHOLZ, H.L., Ex AL. 1991. Dynamics of canopy structure and light interception in Pines elliottii stands, North Florida. Ecol. Monogr. 61:33-51.

GouLDEN, M.L., J.W. MUNGER, S.-M. FAN, B.C. DAUBE, AND S.C. WOFSY. 1996. Measurements of carbon sequestration by long-term eddy covariance: methods and a critical evaluation of accuracy. Glob. Change Biol. 2:169-182.

GOULDEN, M.L., B.C. DAUBE, S.-M. FAN, D.J. SuTTON, A. BAZZAZ, J.W. MUNGER AND S.C. WOFSY. 1997. Physiological response of a black spruce forest to weather. J. Geophys. Res. 102:28,987-28,996.

HENDRY, L.C., AND H.L. GHOLZ. 1986. Aboveground phenology in north Florida slash pine plantations. For. Sci. 32:779-778.

HOLLINGER, D.Y., F.M. KELLIHER, J.N. BYERS, J.E. HUNT, T.M. MCSEVENY, AND P.L. Were. 1994. Carbon dioxide exchange between an undisturbed old-growth temperate forest and the atmosphere. Ecology 75:134-150.

LAVIGNE, M.B., M.G. RYAN, D.E. ANDERSON,,D.E. ANDERSON, D.D. BALDOCCHI, P.M. CRux, D.R. FITZJARRALD, M.L. GOULDEN, S.T. GowER, J.M. MAsHEDER, J.H. MCCAUGHEY, M. RAYMENT, AND R.G. STRIEGL. 1997. Comparing nocturnal eddy covariance measurements to estimates of ecosystem respiration made by scaling chamber measurements at six coniferous boreal sites. J. Geophys. Res. 102:28,977-28,985.

LAW, B.E., M.G. RYAN, AND P.M. ANTHoNi. 1998. Seasonal and annual respiration of a ponderosa pine ecosystem. Glob. Change Biol. 5:169– 182.

MoNcRiEFF, J.B., J.M. MASSHEDER, A. VERHOEH, J. ELBERs, B. HEUSINKVELD, S. SCOTT, H. DE BRuiN, P. KABAT, AND PT JARVIS. 1997. A system to measure surface fluxes of energy, momentum and carbon dioxide. J. Hydrol. 188189:589-611.

NOAA. 1996. Climatological data, Florida. Nat. Ocean. Atmos. Admin., Nat. Climatic Center Environ. Data Service, Asheville, NC.

Rumy, A., P.G. JARVIS, D.D. BALDOCCHI, AND B. SAUGIER. 1995. COZ fluxes over plant canopies and solar radiation: A review. Adv. Ecol. Res. 26:1– 68.

RYAN, M.G., S.T. GOWER, R.M. HuBBARD, R.H. WARING, H.L. GHoLz, W.P. CROPPER, JR., AND S.W. RuNNING. 1995. Stem maintenance respiration of conifer stands in contrasting climates. Oecologia 101:133-140.

SINCLAIR, T.R., AND K.B. KNOERR. 9182. Distribution of photosynthetically active radiation in the canopy of a loblolly pine plantation. J. Appl. Ecol. 19:183-191.

TESKEY, R.O., H.L. GHoLz, AND W.P. CROPPER, JR. 1994a. Influence of climate and fertilization on net photosynthesis of mature slash pine. Tree Physiol. 14:1215-1227.

TESKEY, R.O., D. Wl=rE D, AND S. LINDER. 1994b. Photosynthesis and carbon gain by pines. Ecol. Bull. (Copenhagen) 43:35-49.

WARING, R.H., B.E. LAW, M.L. GOuLDEN, S.L. BAssow, R.W. McCREIGHT, S.C. WOFSY, AND F.A. BAzzAz. 1995. Scaling gross ecosystem production at Harvard Forest with remote sensing: A comparison of estimates from a constrained quantum-use efficiency model and eddy correlation. Plant, Cell Environ. 18:1201-1213.

Kenneth L. Clark, SFRC, University of Florida, Gainesville, FL 32611-E-mail: klcl@gnv.ifas.ufl.edu, Fax: (352) 846-1277; Wendell P. Cropper, Jr., University of Miami, Miami, FL 33149-E-mail: wcropper@rsmas.miami.edu; Henry L. Gholz, SFRC, University of Florida, Gainesville, FL 32611E-mail: hlg@nersp.nerdc.ufl.edu.

Acknowledgments: We thank John Moncrieff, Ford Cropley, and Henry Loescher for assistance with the eddy covariance measurements, and the Jefferson-Smurfit Corporation for providing access to the field site. This research was funded in part by Department of Energy, National Institute of Global Environmental Change (NIGEC), Southeast Regional Center. This is Florida Agricultural Experiment Station Journal Series No. R-07890.

Copyright Society of American Foresters Feb 2001

Provided by ProQuest Information and Learning Company. All rights Reserved