Variability in the carbon isotopic composition of foliage carbon pools (soluble carbohydrates, waxes) and respiration fluxes in southeastern U.S. pine forests



[1] We measured the δ13C of assimilated carbon (foliage organic matter (δCOM), soluble carbohydrates (δCSC), and waxes (δCW)) and respiratory carbon (foliage (δCFR), soil (δCSR) and ecosystem 13CO2 (δCER)) for two years at adjacent ecosystems in the southeastern U.S.: a regenerated 32 m tall mature Pinus palustrisforest, and a mid-rotation 13 m tallPinus elliottii stand. Carbon pools and foliage respiration in P. palustris were isotopically enriched by 2‰ relative to P. elliottii. Despite this enrichment, mean δCER values of the two sites were nearly identical. No temporal trends were apparent in δCSC, δCFR, δCSR and δCER. In contrast, δCOM and δCW at both sites declined by approximately 2‰ over the study. This appears to reflect the adjustment in the δ13C of carbon storage reserves used for biosynthesis as the trees recovered from a severe drought prior to our study. Unexpectedly, the rate of δ13C decrease in the secondary C32–36 n-alkanoic acid wax molecular cluster was twice that observed forδCOM and the predominant C22–26 compound cluster, and provides new evidence for parallel but separate wax chain elongation systems utilizing different carbon precursor pools in these species. δCFR and δCER were consistently enriched relative to assimilated carbon but, in contrast to previous studies, showed limited variations in response to changes in vapor pressure deficit (D). This limited variability in respiratory fluxes and δCSC may be due to the shallow water table as well as the deep taproots of pines, which limit fluctuations in photosynthetic discrimination arising from changes in D.

1. Introduction

[2] A key variable fundamental for using δ13CO2 for investigating carbon transformations in the biosphere is photosynthetic carbon isotope discrimination (Δ) [Farquhar et al., 1989]. At the leaf scale, Δ can be calculated from mass balance of δ13C and CO2 concentration of air entering and leaving a leaf chamber [Evans et al., 1986] or from measurements of the ratio of intercellular to ambient air CO2 concentrations (Ci/Ca) that can then be related to Δ [Farquhar et al., 1989]. Measuring Δ at scales larger than the leaf or branches [Wingate et al., 2007] remains challenging [Wingate et al., 2010]. For scales larger than the leaf, proxies are used to infer Δ. At the ecosystem scale, for example, the δ13C of products of photosynthesis such as leaf carbohydrates (δCSC) [Brugnoli et al., 1988], phloem sugars [Scartazza et al., 2004; Barbour et al., 2005], leaf water soluble organics [Brandes et al., 2006], or bulk leaf organic matter (δCOM) [Farquhar et al., 1989] have been used to infer recent and seasonally integrated Δ. Additionally, the δ13C of ablated leaf waxes (δCW) have been used to infer Δ at the regional [Conte et al., 2003] to the sub-continental scale [Conte and Weber, 2002]. Photosynthetic discrimination has also been estimated from eddy covariance measurements of water flux and a Penman-Monteith inversion [Ogée et al., 2003] to calculate canopy conductance, or from biophysical models with varying degrees of complexity [Chen and Chen, 2007; Aranibar et al., 2006; Baldocchi and Bowling, 2003].

[3] Bowling et al. [2002] demonstrated that the δ13C of ecosystem respired CO2 (δCER) responded to changes in vapor pressure deficit (D) in a nonlinear fashion and suggested that this relationship was consistent with changes in Δ resulting from adjustments in stomatal conductance to moisture deficit. Based on the analysis of 11 ecosystems, Alstad et al. [2007] concluded that the observed relationships between δCER and environmental variables was indeed consistent with changes in Δ through physiological control on Ci/Ca by adjustment of stomatal conductance and photosynthesis [Ehleringer and Cerling, 1995].

[4] When the δ13C of carbon pools or respired CO2 are used as proxies to infer Δ it is typically assumed that they reflect 13C fractionation during photosynthesis [Farquhar et al., 1982], as modified by a relatively fixed biosynthetic offset. However, recent findings on fractionation during post photosynthetic carbon transformations as well as respiration call into question the validity of this assumption [Rascher et al., 2010; Hobbie and Werner, 2004; Tcherkez et al., 2011; Brüggemann et al., 2011]. For example, as a result of fractionation steps in the Calvin cycle, phloem sugars that result from transitory starch degradation at nighttime are δ13C enriched, while the daytime phloem sugars that originate from triose phosphates following the aldolase/transkeolase reactions are δ13C depleted [Gessler et al., 2008; Tcherkez et al., 2004; Tcherkez et al., 2011]. The δ13C of phloem sugars has therefore a diel variability that is superimposed on Δ.

[5] Isotopic labeling experiments furthermore have shown that soluble carbohydrates are partitioned into fast and slow turnover pools and that the relative size of these pools is species specific [Mortazavi et al., 2009]. Thus, the larger and/or slower turnover pool will buffer the δ13C signal of recent Δ imprinted on the plant's carbon pools. For example, leaf wax biosynthesis in the pine Pinus elliottii (slash pine) and in the deciduous tree Liquidambar styraciflua (sweetgum) draw significantly on carbon reserves as opposed to recently assimilated photosynthates, whereas in corn (Zea mays), which does not have significant reserves, wax δ13C more closely reflects recently assimilated carbon [Mortazavi et al., 2009]. Therefore, δCSC and δCW yield information on a temporally integrated Δ, but with an integration time that is species specific and will more than likely vary with the growth dynamics of the plant.

[6] Because substrates supporting autotrophic respiration are influenced by post photosynthetic fractionation [Tcherkez et al., 2003] and supported by long turnover carbon pools [Mortazavi et al., 2009; Nogués et al., 2004)] the degree to which respired δ13CO2 reflects Δ remains unclear. Foliage respiration, for example, can be enriched by several ‰ relative to δCSC [Ghashghaie et al., 2003]. In cases where aboveground respiration contributes significantly to nighttime ecosystem respiration [Lavigne et al., 1997; Mortazavi et al., 2006], estimate of Δ based upon δCER could be biased. In addition, δCERrepresents the flux-weighted contribution of heterotrophic and autotrophic respiration to ecosystem respiration. Because heterotrophic respiration is fueled by organic matter of different ages ranging from recent (e.g., litter) to decades old [Trumbore, 2000], the δCCR contribution to δCER will reflect a temporally integrated δ13C of carbon used for respiration that may differ significantly from recent Δ. Bowling et al. [2008] reported that with the exception of root respired CO2, δ13C fluxes in terrestrial ecosystems were enriched relative to δCOM.

[7] Despite these limitations, numerous investigations have used the δ13C of respiratory fluxes or carbon pools as proxies to estimate Δ at scales ranging from the leaf [Brugnoli et al., 1988] to the ecosystem [Wingate et al., 2010; Buchmann et al., 1998; Lai et al., 2005], and to the continental [Conte and Weber, 2002], and have related the temporal variability in the proxies of Δ to environmental variability [Bowling et al., 2002; Conte et al., 2003; Knohl et al., 2005; McDowell et al., 2004a, 2004b]. These investigations have yielded patterns that are consistent with the physiological understanding at the leaf level scale of assimilated carbon δ13C [Farquhar et al., 1989; Dawson et al., 2002].

[8] Young pine forests and plantations of the southeastern U.S. appear to be among the largest terrestrial carbon sinks [Lai et al., 2002; Valentini et al., 2000; Wofsy et al., 1993; Clark et al., 1999; Hollinger et al., 1994]. These ecosystems are the dominant vegetation type in a mosaic that also includes mature second growth deciduous and pine forests. In this study we examined the δ13C of assimilated and respired carbon in a naturally regenerated Pinus palustris(Longleaf pine) forest and in a mid-rotationPinus elliottii(Slash pine) plantation that are located 6 km apart in central Florida. Our goals were to examine the temporal variability in (i) foliage-, soil-, and ecosystem-respiredδ13CO2 reflected the δ13C of foliage carbon pools (soluble carbohydrate and leaf waxes) and (ii) to determine to what extent the δ13C of these carbon pools and of respired CO2 reflected variability in environmental factors, for example, changes in D.

2. Methods

2.1. Study Sites

[9] Our study was conducted from June 2004 to June 2006 at two experimental forest stands, separated by ∼6 km, that are managed by the School of Forest Resources and Conservation at the University of Florida. Soils at both sites are ultic alaquods with a shallow water table that fluctuates between the surface and 130 cm [Clark et al., 2004]. Hurricane landfalls or severe droughts occur approximately every 15 years [Gholz and Boring, 1991], with droughts occasionally leading to wildfires.

[10] The first site, in Austin Cary Memorial Forest (AC) (29°44′N, 82°13′W), is an 85-year-old naturally regenerated forest with an open, 32 m tall overstory dominated by longleaf pine (Pinus palustris, 72% of basal area) and slash pine (Pinus elliottii var. elliottii, 28% of basal area) [Gonzalez-Benecke et al., 2011; Powell et al., 2008]. Prescribed fire is applied every 3–5 years to encourage natural regeneration and to control hardwood species. The stand was burned in January of 2003 prior to our study, and then again in January 2006.

[11] The second site, the Donaldson Tract (DT) (29°45′N, 82°10′W), is a mid-rotationPinus elliottii var. elliottii tree plantation. The stand was planted in 1990 at a density of 2000 trees ha−1following a clear-cut. Understory biomass at DT was about 5% of total ecosystem biomass [Clark et al., 1999], and is estimated to contribute 7%–9% of total ecosystem net primary production [Gholz and Cropper, 1991].

[12] Powell et al. [2008] measured carbon exchange at these sites for four years between 2000 and 2002 and in 2004–2005, overlapping the beginning of our study. The 2000–2002 period corresponded to a period of record drought for the region, with extended periods with a water table depth of >2 m at both sites. Extended spring droughts also occurred in both 2003 and 2004 and resulted in a large precipitation deficit that persisted until autumn 2004, just as our study began.

2.2. Meteorological Measurements

[13] Continuous measurements of photosynthetic photon flux density (PPFD) (LI-190, LI-COR, Inc.), incoming shortwave radiation (LI-200, LI-COR, Inc.), net radiation (Q7, Radiation and Energy Balance Systems, Inc.), air temperature and relative humidity (HMP35D-A, Vaisala, Inc.), wind speed and direction (R3, Gill Instruments, Ltd.), precipitation (TE525, Texas Instruments, Inc.) were collected at the top of walk-up towers at both sites. Additionally, water table depth was measured next to each tower with a Steven's water gauge (F-68, Leupold and Stevens, Inc.).D was calculated from measured temperature and humidity values.

2.3. Leaf to Bulk Air Vapor Pressure Gradient

[14] We calculated the leaf to bulk air vapor pressure gradient (Dsurf-air), which reflects the evaporative demand as sensed at the leaf surface, incorporating departures of leaf temperature from air temperature. Average leaf temperature was approximated as surface temperature, which was estimated from weather station and eddy covariance data [Bracho et al., 2012]:

display math

where Ts is surface temperature (°C), Tair is measured air temperature (°C), H is sensible heat flux (W m−2), ρa is air density (kg m−3), Cp is specific heat capacity of air (J kg−1 °C−1), and ga is aerodynamic conductance (m s−1). At each site Dsurf-air was then calculated as the difference between saturation vapor pressure at Ts and air vapor pressure measured at the weather station.

2.4. Foliage and Soil Collections

[15] Foliage from P. palustris at AC and P. elliottiiat DT was collected approximately every three weeks at sunset using a shotgun to obtain needles from mid-crown branches of several trees. Foliage collections were made on the same trees from which needles were used for foliage respiration (δCFR) measurements (See Section 2.5.1). Needles collected in 2004 were analyzed for bulk organic matter (δCOM) only. Needles collected in 2005–2006 were divided into three replicates for determination of δCOM, δCSC and δCW and immediately placed in liquid nitrogen. In addition, foliage from the top of the canopy was also collected from a second group of P. elliottii trees at DT between February and August 2005. Needles from terminal branches of three trees were pooled and measured for total carbon and wax concentrations and isotopic compositions.

[16] Soil sample profiles were collected on 13 July 2004 at two locations at each site. Duplicate soils cores were collected and soil samples from 0 to 2, 4–5, 9–10, 14–15, 20–21 and 40–41 cm below the soil surface were collected. Roots were removed, samples were dried and ground to a fine power for the determination of carbon content and the δ13C of soil organic matter.

2.5. Respiration Measurements

2.5.1. Foliage

[17] Gas samples of foliage respiration for determination of δCFRwere collected predawn on dates when foliage was collected for total carbon and wax analyses. Needles from mid-canopy branches of two trees at each site were measured. Needles from the same year cohort were excised and pooled to generate one sampling run. The respired CO2from the excised needles was measured using a sequenced-air-sampling system previously described in detail [Prater et al., 2006].

2.5.2. Ecosystem

[18] δCER was calculated from measurements of CO2 concentration and isotopic composition using the Keeling plot method, as previously described [Mortazavi et al., 2005, 2006]. δCER was determined on or near the same dates as samples were collected for foliage respiration and carbon analyses. For each measurement, air samples were collected predawn at eight different heights (0.01, 1, 2, 10, 16, 20, 26 and 30 m) within the canopy at AC and at seven different heights (0.01, 0.5, 2, 6, 10, 12 and 18 m) at DT; total collection time was <1 h.

2.5.3. Soil

[19] At each site, soil CO2 fluxes were measured using two replicate static chambers (65 × 65 cm base and a volume of 102 L, including the collar base) placed within 5 m of each other. The chamber collars were permanently inserted into the soil to a depth of 2 cm prior to the initiation of the study. A fan inside each chamber gently circulated the air. Replicate gas samples (125 ml) from the chamber headspace were drawn with a syringe from a sample port, while a port at the other end of the chamber was left open to the atmosphere to minimize variations in pressure. Gas samples were stored and processed according to Mortazavi et al. [2005, 2006]. Previous comparison of soil δ13CO2efflux determined with the chambers and open-topsoil-mini towers as well as soil probes have shown no bias in the chamber measurements in the sandy soils in Florida [Mortazavi et al., 2004].

2.6. Laboratory Analyses

[20] The δ13CO2of ecosystem respiration, foliage respiration, and soil respiration samples was determined using a gas chromatograph isotope ratio mass spectrometer (GC-IRMS) (GC: Hewlett Packard 5890 Series II, IRMS: Finnigan Delta S operated in continuous flow mode (Mortazavi and Chanton, 2002). The CO2 concentrations of the foliage and soil respiration samples were calculated from the CO2voltages obtained by the GC-IRMS [Mortazavi and Chanton, 2002; Mortazavi et al., 2005]. CO2concentrations of the ecosystem respiration samples were determined with a LI-COR 6200 portable photosynthesis system (LI-COR, Inc.) calibrated with a NOAA reference gas (363.5 ppm CO2). Isotopic ratios are presented in the δnotation and are reported relative to VPDB. Analytical precision for isotopic measurements was ±0.2‰ based on repeated measurements of a laboratory-working standard. The intercept of the Keeling plots for ecosystem-, foliage-, and soil-respired CO2 was then determined with ordinary least square regressions [Zobitz et al., 2006] applied to the CO2 isotopic ratios and associated 1/CO2 concentrations of the samples.

[21] δCSCwas determined on freeze-dried ground needles. The soluble carbohydrates were purified using an ion-exchange technique [Göttlicher et al., 2006] as described in detail in Mortazavi et al. [2009]. Briefly, freeze-dried foliage samples (100 mg) were ground to a fine powder with a ball mill and then extracted with 1.5 ml of methanol\chloroform\water (12:5:3, v\v\v) for 30 min at 70°C. The samples were cooled and then centrifuged (10,000 g, 3 min). The supernatant (0.8 ml) was removed and mixed with water (0.8 ml) and chloroform (0.25 ml). Samples were centrifuged (10,000 g, 2 min) to separate the chloroform and aqueous phases. The upper phase (1.2 ml) was removed and 0.5 ml of chloroform added. The sample was then mixed and centrifuged to separate the phases. The upper phase was transferred to a vial, evaporated to dryness, and then redissolved in DI water. The extracted samples were next run through a combination of ion exchange resins (cation-exchange: DOWEX 50 W × 8, 50–100 mesh prepared in the H+-form, anion-exchange DOWEX 1 × 8, 50–100 mesh prepared in the formate-form). The eluate was collected, dried in vacuo and redissolved in 0.2 ml ethanol. An aliquot was removed to a tin capsule and dried forδ13C determination at UC Davis Stable Isotope Facility. A working standard (freeze-dried foliage) was processed for every four experimental samples.

[22] Waxes were measured in both the 2004 and 2005 cohorts of needles obtained from mid-canopy branches ofP. elliottii and of P. palustris and from the same branch as needles analyzed for δCOM and δCSC. Needles from two trees were pooled at each site. In addition, we also measured waxes in P. elliottii in needles from terminal branches at the top of the canopy (canopy top). Needles from three trees were pooled for the canopy data.

[23] Wax isotopic analytical procedures have been detailed in Conte et al. [2003] and Mortazavi et al. [2009]. Briefly, waxes were extracted in dichloromethane after addition of an internal standard mixture (21:0 n-alkanol, 23:0n-alkanoic acid, 36:0n-alkane, 5Α cholestane). The total extract was evaporated to dryness, resuspended in 4:1 dichloromethane:methanol and passed through a minicolumn containing anhydrous Na2SO4 to remove residual water. The extract was evaporated just to dryness, resuspended in toluene and transesterified using 5% methanolic HCl (55°C, 12 h). The transesterified products were extracted into hexane, dried and then trimethylsylilated under N2 using BSTFA + 1% TMCS in pyridine (55°C, 1 h). The individual compounds were quantified using a Fisons 8000 series GC fitted with a Chrompack CPSil5CB column (60 m × 0.25 mm, 0.25 μm film thickness, H2 carrier gas). The GC programming was 50 to 150°C at 10°C min−1 and from 150 to 320°C at 4°C min−1 with a 30 min isothermal hold at 320°C.

[24] For isotopic analyses of the n-alkanoic acids, the samples were saponified, urea adducted to isolate the n-alkyl chain compounds and then purified by solid phase extraction (SPE) using the methods detailed inConte and Weber [2002]. The δ13C of the individual compounds was determined using a GV IsoPrime GC-mass spectrometer at the MBL Ecosystems Center Mass Spectrometry Facility. A standard mixture containingn-alkanoic acids of known isotopic composition was derivatized with the samples and was used to correct for theδ13C of the derivatization group (CH3). The median measurement error was 0.32‰ and did not differ significantly for the individual compounds.

[25] Bulk organic matter δ13C was determined on freeze-dried ground needles at the MBL isotopic facility, using a Europa ANCA-SL elemental analyzer gas chromatograph preparation system attached to a continuous-flow Europa 20–20 gas source stable isotope ratio mass spectrometer.

3. Results

3.1. Environmental Conditions

[26] Mean daytime temperatures varied between 3 and 29°C (Figure 1a). Mean daytime D at DT and AC were 0.95 kPa and 0.97 kPa, respectively (Figure 2a and 2c). Maximum values of daytime D at DT and AC were 2.67 kPa and 2.59 kPa, respectively (Figure 2a and 2c). The mean values for water table depth were 93.0 and 69.7 cm at DT and AC, respectively (Figures 2b and 2d). Water table depth was at its deepest levels in June 2004 (Figure 2b and 2d) and precipitation deficit for 2004 was not relieved until the three tropical storms in later summer and early fall (Figure 1c). Annual precipitation in 2004 was 1496 mm, which exceeded the long-term (1975–2008) mean of 1226 mm (National Climatic Data Center,, 2009). Annual precipitation in 2005 (1182 mm) was close to the long-term mean.

Figure 1.

(a) Mean daily temperature and (b) photosynthetically active radiation (PPFD) at Donaldson Tract and Austin Cary, Fla.; (c) daily precipitation; and (d) cumulative precipitation. Vertical lines represent sampling events. Precipitation data is from Bracho et al. [2012].

Figure 2.

(a) Mean daily vapor pressure deficit and (b) mean monthly water table depth at Donaldson Tract; (c) mean daily vapor pressure deficit; and (d) mean monthly water table depth at Austin Cary.

3.2. Carbon Isotope Ratios of Foliar Pools

[27] Bulk organic matter: Average δCOM for the combined 2004 and 2005 flush data for P. elliottii at DT (−29.68 ± 0.23‰, mean ±1 SE, n = 50) was 13C depleted on average by roughly 2‰ relative to P. palustris at AC (−27.33 ± 0.17‰, n = 49) (Figure 3a). The P. elliottiifoliage collected at mid-canopy height at DT was13C depleted by roughly 1‰ in comparison to foliage collected at canopy height (Figure 3a). Additionally, the newly emergent 2005 flush of needles was δ13C enriched by ∼1‰ relative to needles from the 2004 flush (Figure 3a).

Figure 3.

Seasonal trends in the δ13C of (a) foliage organic matter (mean of 2–3 replicate ±1 SD), δCOM, (b) foliage soluble carbohydrates (mean of 2–6 replicates ±1 SD, except in 03/06 at DT where n = 1), δCSC at Donaldson Tract (P. elliottii) and Austin Cary (P. palustris).

[28] Soluble carbohydrates: δCSC was δ13C depleted on average by approximately 1‰ relative to δCOM at both sites. Consistent with δCOM, the δCSC of P. elliottii at DT (−30.38 ± 0.10‰, n = 70) was also depleted by roughly 2‰ relative to P. palustris at AC (−28.38 ± 0.18‰, n = 28) (Figure 3b). No temporal trend in δCSC was apparent over the study period.

[29] Leaf waxes: Temporal trends in needle wax n-alkanoic acid concentration are shown in Figure 4. Total needle wax concentration (sum of the C20–36 n-alkanoic acids) at the beginning of the study period was 40% higher in P. elliottii at the DT site than in P. palustris at the AC site (700 versus 500 μg gdw−1, respectively) (Figure 4a). Wax concentrations decreased during the study at both sites, but the decrease at the DT site (−0.37 μg/gdw d−1), although more variable, was 80% faster than the AC site (−0.21 μg/gdw day−1). Thus, by the end of the study wax concentrations in trees at both sites were nearly comparable.

Figure 4.

Seasonal trends in wax concentration of 2004 and 2005 needle flushes in P. palustris at Austin Carey (top of canopy needles) and in P. elliottiiat Donaldson Tract (top of canopy and mid-canopy needles). (a) Total needle wax (sum of C20–36 n-acids), (b) total C20–22 n-acids, and (c) total C32–36 n-acids.

[30] A striking difference was observed in the concentrations of the individual wax n-alkanoic acids between the sites, specifically the relative differences in the short-chain C20 and C22 compounds and in the long chain C32–36 compounds in the 2004 and 2005 needle flushes (Figures 4b and 4c). P. elliottii needles at the top of the canopy from the 2004 flush, but not the 2005 flush, were strongly enriched in C20–22 n-alkanoic acids and depleted in C32–36 n-alkanoic acids. Interestingly, the P. elliottiineedles from the mid-canopy had similar trends but those from the 2004 flush were less enriched in C20–22 n-alkanoic acids and similarly less depleted in C32–36 n-alkanoic acids. InP. palustrisat AC, mid-canopy needles from the 2004 flush were also notably depleted in C32–36 n-alkanoic acids relative to the 2005 flush but only slightly enriched in C20–22 n-alkanoic acids.

[31] At both sites, the concentrations of the C32–36 n-alkanoic acids were maximal in newly emerged needles of the 2005 flush (May–June samples). In bothP. palustris and P. elliotti, concentrations in mid-canopy needles then decreased by >50% before rising again in August (Figure 4c). This swing in concentration was not observed in needles from the top of the canopy. For the remainder of the study, concentrations of the C32–36 n-alkanoic acids steadily decreased at a rate of ∼0.2 μg/gdw d−1. In contrast, no temporal trend in C32–36 n-alkanoic acid concentration was apparent in needles from the 2004 flushes. Neither was there any temporal trend in the C20–22 n-alkanoic acid concentrations in needles from the 2005 flush at either site.

[32] The within class distribution of the individual compounds is very similar in P. palustris and P. elliottii and is characterized by a primary peak at the C24 n-alkanoic acid and and secondary peak at the C34 n-alkanoic acid (Figures 5a–5c). The greater spread seen in the percent contribution of C20–22 n-alkanoic acids and of C32–36 n-alkanoic acids in these graphs reflects the switchover previously described between concentrations of the C20–22 n-alkanoic acids versus the C32–36 n-alkanoic acids (Figure 4). In contrast to these compounds, the relative concentrations of the C24–32 n-alkanoic acids are far less variable.

Figure 5.

Within-class molecular distribution of waxn-alkanoic acids. The ID numbers are the sequence day of the individual sampling dates, starting 1 January 2005. (a) Top of canopy needles ofP. palustris at Austin Carey, (b) top of canopy needles of P. elliottiiat Donaldson Tract, and (c) mid-canopy needles ofP. elliottii at Donaldson Tract.

[33] Averaged over the entire study, the δCW of the wax n-acids (the concentration-weighted average of the C22–36 compounds) for P. palustris at AC and for P. elliottii at DT was 13C depleted relative to δCOM (Figure 3a) by 3.4 ± 0.3‰ and 3.2 ± 0.7‰, respectively. As observed for δCOM and δCSC, the needle wax of P. elliottii at DT was depleted by 2.3 ± 0.3‰ relative to the mature P. palustris at AC. This difference in δCW between the two species mirrored that observed in δCOM and δCSC (Figure 3) and indicates that waxes are similarly recording the broad isotopic shifts that are associated with different Δ in these two species.

[34] As observed in wax concentration and composition, there are clear differences in temporal trends in the δ13C of individual wax compounds in P. palustris and P. elliottii and, for P. elliottii, in the 2004 and 2005 flushes (Figure 6). In P. palustris at AC, there was a decrease in δ13C over the study period for all compounds, with no distinct discontinuity in this trend apparent between the 2004 and 2005 flushes. The decline in δ13C of the primary molecular cluster of C22, C24 and C26 n-alkanoic acids during the study period (−0.002‰ to −0.004‰ d−1), although variable, was comparable to the decline in δ13COM (−0.004‰ d−1) (Figure 6, Table 1). Quite unexpectedly, the rate of decline in δ13C of the secondary molecular cluster of C32, C34 and C36 compounds (−0.008‰ to −0.009‰ d−1) was twice that of the C22–26 compounds in the primary molecular cluster (Figure 6, Table 1). Additionally, the temporal trend in the C32–36 n-alkanoic acids was far less variable than observed for the shorter chain length compounds. For the C28 acid, a minor wax component (Figure 5), the rate of decline (−0.006‰ d−1) fell between that of the primary and secondary molecular clusters (Table 1).

Figure 6.

(a) The δ13C of bulk carbon in needles shown for comparison with temporal trends in (b–h) the individual C22 through C36 n-alkanoic acids and in (i) the mass weighted average of the C32–36 n-alkanoic acids.

Table 1. Regression Parameters (a + bx) of Temporal Trend Lines for δ13C of Bulk Leaf Carbon, Individual n-Alkanoic Acids and Mean of C32–36 Compounds (Mass Weighted Average) Shown in Figure 6a
n-Alkanoic Acid
ParameterBulk Leaf CarbonC22C24C26C28C32C34C36Weight Average C32–36
  • a

    The Austin Carey regressions are for combined data for both 2004 and 2005 flushes. The Donaldson Tract regressions are given for combined data and for the 2005 flush only.

Austin Carey
Donaldson Tract (All)
aNo TrendNo TrendNo TrendNo Trend−30.30−31.09−30.88−30.55−30.44
Donaldson Tract (2005 Flush Only)

[35] In contrast to P. palustris, needles of the 2004 and 2005 flushes of P. elliottii at DT showed large isotopic differences in temporal trends of the individual compounds (Figure 6). For needles from the 2004 flush, there was no temporal trend in the δ13C of the primary cluster of C22–26 n-alkanoic acids. Additionally, the C22–26 n-alkanoic acids were also isotopically depleted relative to the new flush of needles in 2005. However, needles from the 2005 flush clearly showed a decreasing trend, albeit noisy, in theδ13C of C22–26 n-alkanoic acids. These differences in the primary molecular cluster can be contrasted with those of the C28 to C36 compounds, in which both the 2004 and 2005 flushes showed a similar isotopic trend (Figure 6).

[36] A second striking observation was that the rates of decline in δ13C of the secondary C32–36 molecular cluster in P. elliottii at the DT site and in P. palustris at the AC site were indistinguishable (Table 1), although the data were much noisier at the DT site (Figure 6). The mass weighted average δ13C of the C32–36 molecular cluster in P. palustris at the AC and DT sites decreased at similar rates (Table 1).

3.3. Carbon Isotope Ratios of Foliar, Ecosystem, Soil Respired CO2 and Soil Organic Matter

[37] Foliar respiration exhibited the greatest overall temporal variability in the δ13C of respired CO2 (Figure 7a). Consistent with differences observed in carbon pools, δCFR of P. elliottii at DT (−28.49 ± 0.29‰, n = 35) was depleted by ∼2‰ relative to P. palustrisat AC (−26.34 ± 0.22‰, n = 34). No seasonal or long-term trend was apparent inδCFR at either site. In contrast to δCFR, δCER was not significantly different at the two sites (Figure 7b). Mean δCER at DT was −27.16 ± 0.20‰ (n = 18) and −26.94 ± 0.14‰ (n = 20) at AC. All δCER values at AC, and all but one (January 2006) at DT, were within 1‰ of the site average.

Figure 7.

Temporal trends in the δ13C of (a) foliage (δCFR), (b) ecosystem δCER, and (c) soil-respired (δCSR) CO2 at the Donaldson Tract (P. elliotii) and Austin Cary (P. palustris) sites. The error bars indicate the standard error of Keeling plots intercepts.

[38] On average, δCSR at DT (−27.07 ± 0.13‰, n = 40) was only 0.7 ± 0.2‰ more depleted than at AC (−25.89 ± 0.55‰, n = 38). Minimum values of δCSR were observed in the fall of 2004 at both sites (Figure 7c). Leaf temperature differed from air temperature by less than 0.55°C for 95% of daylight hours, resulting in little difference between Dsurf-air and D of bulk air (data not shown). The shortest time lag that yielded a significant correlation between δCSR and D was 11 days at AC and 5 days at DT. However, no significant correlations of δCFR or δCER were observed with temperature, PPFD or D for time lags of 0–21 days (data not shown).

[39] Soil: The average δ13C of soil organic matter in the 0–40 cm soil interval (n = 7) was similar at both the AC and DT sites (−25.72 ± 0.37‰ and −25.32 ± 0.61‰, respectively) (Figure 8).

Figure 8.

Synthesis showing the δ13C of the various pools and CO2 fluxes measured for (a) P. elliottii at the Donaldson Tract and (b) P. palustris at Austin Cary. The median, 25% upper/lower quartiles as well as the 5% and 95% limits of the data are shown.

4. Discussion

4.1. Temporal Trends in Wax Isotopic Composition

[40] Significant temporal trends in wax concentrations and in molecular and isotopic composition were found in both P. palustris at AC and P. elliottii at DT over the course of this study. Additionally, there were clear differences between the 2004 and 2005 needle flushes, and in P. elliottii, between top of canopy and mid-canopy needles. These observations provide further evidence for rapid wax turnover [Sachse et al., 2009; Neinhuis and Barthlott, 1998; Hauke and Schreiber, 1998] and for the dynamic adjustments in wax biosynthesis in response to environmental stressors such as drought [Shepherd and Griffiths, 2006; Kosma et al., 2009; Leide et al., 2011; Percy et al., 2009; Yang et al., 2011].

[41] The thickness and microstructure of the wax coating of leaves is critical to leaf photoprotection, to control cuticular respiration and to regulate other interactions with the environment [reviewed in Shepherd and Griffiths, 2006]. We found significantly higher wax loadings in needles of the 2004 flush which were formed during a very dry spring (Figure 2) [Powell et al., 2008]. In particular, concentrations of shorter chain length C20 and C22 n-alkanoic acids were much higher than in needles of the 2005 flush that were formed under more typical spring conditions. This increase in wax loading in a drought year is consistent with other studies that have observed increased wax deposition in response to water stress and low humidity to reduce cuticular respiration [reviewed inShepherd and Griffiths, 2006].

[42] Light and temperature have also been observed to influence wax chain length in several species. For example, higher levels of UV-B were associated with an increase in the proportion of shorter (C < 30) to longer (C > 30) chainn-alkanes inNicotiana tabacum [Barnes et al., 1996]. Thus, higher light levels and leaf temperature at the top of the canopy may have contributed to the wax compositional differences and higher levels of shorter chain length compounds in fully sunlit needles of P. elliottiias compared to the more shaded needles in the mid-canopy (Figure 5).

[43] Corresponding with the increase in C20–22 n-alkanoic acid concentration was a decrease in C32–36 n-alkanoic acid concentration. Previous studies have attributed changes in wax chain length composition in response to its growth conditions to an adjustment in channeling wax precursors into free fatty acids and the reductive and decarbonylation pathways [Giese, 1975; Shepherd et al., 1995, 1997]. Our data suggests that a similar environmentally induced switchover in channeling carbon into biosynthetic pathways may be occurring here in both P. palustris at AC and P. elliottii at DT, as they show remarkably similar trends and changes in wax molecular distributions.

[44] Waxes of differing chain lengths are biosynthesized via chain elongation within fatty acid elongase (FAE) complexes in the endoplasmic reticulum. In a simple model of wax biosynthesis [e.g., Kolattukudy et al., 1976], the longer-chain length compounds are thought to be biosynthesized in the same FAE complexes from elongation of waxes of shorter-chain lengths, so both are sourced from the same wax precursors. This simple model predicts that the <C30 n-alkanoic acids provide the substrate for elongation to produce the C32, C34 and C36 n-alkanoic acids. Therefore, the average δ13C of the C1 to C28 carbons in the alkyl chain of the >C28 n-alkanoic acids should be identical to that of the C28 n-alkanoic acids.

[45] Assuming this model, the average δ13C of the terminal (i.e., >C28) carbons of the C32, C34 and C36 n-alkanoic acids can be calculated from a simple mass balance. The results (Figure 9) indicate that the δ13C of the terminal carbons of the C32 to C36 n-acids declined by nearly 10‰ over the course of our study. The rate of decline is virtually indistinguishable betweenP. palustris at AC and P. elliottii at DT. For the combined data, the fitted regression for the calculated trend line is δ13C = −28.63 − 0.017 day−1 (r2 = 0.56). However, the magnitude of the decline that is predicted using this mass balance approach seems unlikely given the minimal change in δCSC and δCOM over the study period (Figure 3) as it would require large internal isotopic fractionations during carbon remobilization and/or utilization.

Figure 9.

Estimated mean δ13C of the C32–36terminal carbons in the long-chain waxes for a hypothetical model that assumes these are synthesized via chain elongation of C28 precursor n-acids having the sameδ13C as that measured for the C28 n-acid. The regression for the combined data (solid line):δ13C = −28.63–0.017 month−1, r2 = 0.56. The short and long dashed lines show the separate regressions for P. palustris and P. elliottii, respectively.

[46] An alternative model of wax biosynthesis indicates that waxes of different chain lengths are synthesized independently [Samuels et al., 2008; Kunst and Samuels, 2009]. For example, comparative studies of genetic mutants known to be defective in specific points in the elongation pathway have indicated that rather than a single pathway, there are sequential elongation systems with multiple isoforms of elongases or subunits, such as the condensing enzyme β-ketoacyl-CoA synthase (KCS). The KCS(s) in each of these systems appears to have a strict substrate specificity and catalyze only a few successive elongation steps.

[47] The isotopic differences observed among the waxes of different chain lengths (Figure 6), as well as the switchover pattern observed in their production (Figure 4), are consistent with this alternative model, i.e. there are parallel but separate elongation systems that regulate the biosynthesis of the primary C20–28 molecular cluster and C32–36 molecular cluster. Our results furthermore suggest that these parallel elongation systems are supplied by distinct pools of wax precursors. This hypothesis is supported by evidence in some species for separate but parallel elongation systems linked with the different associated biosynthetic pathways that do not share a common pool of precursors [reviewed in Shepherd and Griffiths, 2006]. Furthermore, there may be changes in the relative activities of these parallel systems during plant growth and/or in response to environmental variables. For example, in corn the ED-I elongation system is more active in seedlings and produces very long-chain alcohols, aldehydes, acids and alkanes while a second parallel system, ED-II, is active over all stages of growth and produces shorter acyl chains primarily for ester formation [Avato et al., 1984; Avato et al., 1990; Beattie and Marcell, 2002]. In barley and wheat, ED-1 is also paralleled by a 3-ketoacyl elongation system. ED-I is more active in seedlings, and the 3-ketoacyl elongation system more active in mature plants [Wettstein-Knowles, 1995]. Physiological switching between parallel elongation systems that is induced by environmental conditions has been observed in several species. For example, a switch between parallel pathways of biosynthesis of alkanes and β-diketones appears to occur in some species in response to water or salinity stress [Wettstein-Knowles, 1972; Bengtson et al., 1978; Rao et al., 1981] and has been suggested to explain light-induced changes in wax chain length in barley [Shepherd and Griffiths, 2006].

[48] No matter what the biochemical explanation for the differences observed between the primary and secondary molecular clusters, the decline in δ13CW during the study (Figure 6) indicates a slow decline in the δ13C of carbon precursors(s) used for wax biosynthesis. As storage reserves provide the primary carbon source for wax biosynthesis in P. elliottii [Mortazavi et al., 2009], the decline strongly suggests that the storage carbon reserves of both P. palustris at AC and P. elliottii at DT were becoming more 13C depleted over the study period.

[49] The remarkably similar rate of decline in δ13CW at both sites (Figures 6i, Table 1) strongly suggests a uniform physiological response to a strong environmental driver rather than any site- or species-specific variable. An explanation consistent with environmental data is that the trees were recovering from the extreme drought of 2000–2002. This “drought of the century” depressed the water table to >2 m and resulted in severe water stress, as indicated by premature needle fall and reduced daytime carbon assimilation and net ecosystem exchange [Powell et al., 2008] Extended spring droughts in both 2003 and 2004 also resulted in a large precipitation deficit that persisted until autumn 2004 (Figure 1).

[50] Although no isotopic data were collected during the drought of 2000–2002, it can be speculated that the water stress depleted tree reserves and resulted in a build-up ofδ13C enriched carbon that was assimilated during the drought. As water stress was subsequently alleviated in late 2004, Δ would have reverted back to typical values as the trees would assimilate more 13C depleted photosynthates compared to those assimilated during the drought. This addition of more δ13C depleted carbon would slowly decrease the δ13C of the carbon storage reserve(s) that would then supply the wax biosynthetic precursors.

[51] Our interpretation that the isotopic trend in δ13CW reflects a slow change in the δ13C of the tree's carbon reserves is supported in other studies of conifers and deciduous trees that have found a large influence of storage reserves on the δ13C of carbon pools. The labeling experiment of Kagawa et al. [2006] found that larch (Larix gmelinii) needles drew nearly half of their carbon from material assimilated during the prior year and stored within the branch. Label was still detectable within the needle starch pool three years post-labeling, suggesting the presence of significant long-term metabolic reserves in this species. The isotopic labeling experiments ofMortazavi et al. [2009] found that in addition to P. elliottii, the deciduous sweetgum tree L. styraciflua also heavily utilized stored carbon for biosynthesis. Labeled carbon from storage reserves was still detectable in waxes of new leaves that were produced the year after the trees were removed from the labeled CO2. Similarly, in a whole-tree labeling experiment in a natural temperate hardwood forest,Körner et al. [2005] observed that carbon reserves from the prior year accounted for ∼60% of the foliage organic matter in Quercus petraea, 30%–35% in Fagus sylvatica and Acer campestre and 20% in Carpinus betulus.A small (∼1%), but still measurable, fraction of carbon from long-term storage reserves was detectable 4 years post-labeling.

[52] The dichotomy between the temporal trends in the isotopic composition of the C22–26 and C32–36 molecular clusters of wax n-alkanoic acids (Figure 6) highlights the gaps in our knowledge of how assimilated carbon is subsequently redistributed within the plant tissues to support biosynthesis. Although current knowledge sheds no light on reasons why the δ13C of the carbon precursor pools for the primary C20–28 and secondary C32–36 molecular cluster may differ, we speculate that newly assimilated carbon may be channeled more strongly into the precursor pool supplying the wax biosynthetic pathway for the secondary C32–36 molecular cluster. In contrast, the precursor pools that supply the pathway for the primary C20–28 molecular cluster appear to be sourced mainly from the tree's carbon storage reserves, a speculation that is supported by the similarity between the isotopic trend in the C20 to C26 n-alkanoic acids and bulk leaf carbon (Figure 6, Table 1). The physiological basis for different isotopic fluxes and turnover rates in wax precursor pools is currently an enigma and a ripe topic for future studies.

4.2. Controls on the δ13C Respired CO2

[53] For both species δCFR measured prior to dawn remained on average ∼2‰ enriched relative to δCSC and by a ∼1‰ relative to δCOM (Figure 3). Enrichment in δCFR relative to δCSC has been observed in sweetgum [Mortazavi et al., 2009], oak [Hymus et al., 2005], in beans, tobacco, sunflower [Tcherkez et al., 2003; Xu et al., 2004; Ghashghaie et al., 2003] and as well as herbaceous species [Werner et al., 2009]. The enrichment in δCFR and δCER relative to δCSC, the carbon substrate presumably used for autotrophic respiration is consistent with other studies conducted at the leaf [Ghashghaie et al., 2003] and ecosystem [Bowling et al., 2008] scale. Isotopic enrichment in δCFR has been explained by post photosynthetic fractionation that results in enriched δ13CO2 during foliage respiration [Ghashghaie et al., 2003]. In P. elliottii [Mortazavi et al., 2006; Prater et al., 2006] as well as other species [Barbour et al., 2007]. Tcherkez et al. [2003] suggested that the enrichment in δCFRresults from the non-random distribution of13C in glucose and molecular fragmentation that occurs during respiration. δCFR is highly enriched immediately upon darkening and this enrichment declines exponentially to stable values within a few hours [Mortazavi et al., 2006]. Gessler et al. [2008] demonstrated that the δ13C enrichment observed in dark respired CO2 for light acclimated foliage arises from excess malate previously synthesized by the leaf in the light.

[54] Variability in δCFR (Figure 7a) was significantly greater than that in δCSC (Figure 3b), the presumed carbon source used during foliage respiration. Predawn δCSC in P. elliotiiconsists of fast and slow turnover substrate pools, with the fast turnover pool (∼3–4 d half-life) accounting for only 1%–2% of the total [Mortazavi et al., 2009]. Therefore, δCsc reflects temporally integrated values of Δ, with the isotopic values of recently assimilated carbon overshadowed by the contribution of a large carbon pool with a long turnover rate [Mortazavi et al., 2009].

[55] The greater variability in δCFR relative to that for δCSC is consistent with other observations [Hymus et al., 2005; Mortazavi et al., 2009]. Tcherkez et al. [2003] related the changes in δCFR to switching of carbon substrates and metabolic pathways as foliage respiration proceeds. The pool of carbohydrates declines with respiratory losses, and the fraction of 13C-enriched CO2 that is produced declines. With the reduction in carbohydrates there is a greater rate of fatty acid oxidation coupled to the Krebs cycle, and subsequently the fraction of δ13C-depleted CO2produced increases. This mechanism can account for 8‰–9‰ short-term variability (hours) observed inδCFR [Sun et al., 2009; Hymus et al., 2005; Mortazavi et al., 2006; Prater et al., 2006]. Therefore, the greater variability in δCFR compared to δCSC reflects the contribution of different metabolic pools to predawn respiration rather than changes in Δ.

[56] In contrast to the isotopic enrichment of foliar carbon pools and foliar respiration in P. palustris at AC in comparison to P. elliottii at DT, δCSR was only slightly more enriched at AC site (−25.89 ± 0.55‰ at versus −27.07 ± 0.13‰, respectively) (Figure 7c). Interestingly, there was an increase in δCSR from June to December 2004. This increase coincided with a steep rise in the water table (Figures 2b and 2d) and declining respiration rates (data not shown). The increase in δCSR could reflect changes in autotrophic/heterotrophic respiration or the contribution of anoxic respiration as soil pore spaces became saturated following the recovery from the drought.

[57] The shortest time lags that yielded any correlation between δCSR and D were 11 and 5 days at AC and DT sites, respectively (data not shown). Our results are in line with other recent studies that indicate multiple controls on δCSR, in addition to change in the δ13C of substrates supplied by autotrophic respiration [e.g., Ekblad and Högberg, 2001]. Wingate et al. [2010] observed in a maritime pine forest that δCSRwas at times decoupled from short-term variability in canopy dynamics. For example, immediately following rainy days soil respired CO2 flux increased and became δ13C enriched while all other ecosystem components became δ13C depleted. Moyes et al. [2010] showed that much of the variability in δCSR was due to change in heterotrophic processes and mechanisms that alter the ratio of soil autotrophic to heterotrophic respiration. Similarly, Powers et al. [2010] measured a 5‰ depletion in δCSRin a semi-arid woodland after water addition to the soil and suggested that changes in the relative contribution of heterotrophic and autotrophic respiration were responsible for the large changes in soil CO2 flux and δCSR. In addition to short-term changes in soil respiratory components, diel variability inδCSR can also arise due to transient diffusive isotopic fractionations [Moyes et al., 2010; Maseyk et al., 2009; Kayler et al., 2010]. In the absence of high frequency measurements such as those by Wingate et al. [2010], the mechanisms leading to variability in δCSR at these sites remains unclear.

[58] The relative contribution of soil and folaige to nighttime ecosystem respiration influnces δCER. Despite the higher mean δCSC and δCFR values for P. palustris at AC compared to P. elliottii at DT, mean δCER values were nearly identical (Figure 8). δCER at AC is nearly identical to δCSR (−25.98 ± 0.55‰) and δCFR (−26.34 ± 0.22‰). However, δCER at DT is more similar to δCSR (−27.07 ± 0.13‰) than it is to δCFR (−28.49 ± 0.29‰), potentially indicating a greater influence of soil respiration to nighttime ecosystem respiration. This interpretation is consistent with results in a Loblolly pine plantation, where soil respiration was 4.6 times higher than plant respiration [Lai et al., 2002].

[59] The relation between δCER and Dsurf-air at the sites are shown in Figure 10. The close coupling between conditions at the leaf surface and bulk air results from high boundary layer conductance in our systems at both the leaf and canopy scales resulted in Dsurf-air and D of bulk air being similar. At the leaf scale, small needle dimensions result in large boundary layer conductances to heat transfer, causing conifer needle temperatures to seldom, if ever, differ from air temperature by more than one degree [e.g., Jarvis et al., 1976; Angell and Miller, 1994]. At the canopy scale, pine forests have high aerodynamic conductance due to their height and low leaf area index, resulting in minimal gradients in temperature and humidity from bulk air through the canopy space, and dominance of convective versus latent heat transfer [Baldocchi and Vogel, 1996; Ewers and Oren, 2000; Powell et al., 2005].

Figure 10.

The 13C of ecosystem respired CO2 (δCER) at Donaldson Tract (P. elliotii) and Austin Cary (P. palustris) as a function of daytime average site Dsurf-air for days 1–4 prior to sampling (correlations were not significant, P > 0.05). The error bars indicate the standard error of Keeling plots intercepts.

[60] The lack of a clear relation between δCER and Dsurf-air (Figure 10) is in contrast to those observed for a plantation of the pine P. taeda at Duke Forest, North Carolina [Mortazavi et al., 2005], and in a coniferous forest in the NW Pacific [Bowling et al., 2002]. Alstad et al. [2007] compared the relationship between δCER and D in several forest ecosystems. The authors noted that while the overall patterns were consistent with a physiological response of lowered stomatal conductance (and Ci/Ca) to increasing D, there were important differences in the strength of the relationship among plant functional groups. For example, coastal sites exhibited little variability in comparison with continental boreal sites, which they attributed to the more moderate maritime climate of coastal regions that minimizes the seasonal variations in temperature and D.

[61] With increasing D, Ci/Ca is expected to decline due to reduction in stomatal conductance, resulting in assimilates becoming δ13C enriched. The magnitude of the trend between stomatal conductance and D, however, exhibits a high degree of variability among species and for individuals within the same species located at different sites [Whitehead et al., 1981; Pataki et al., 1998; McNaughton and Jarvis, 1991]. The empirical analysis of Oren et al. [1999] shows a consistent and robust relationship between stomatal conductance at low D and the sensitivity of the stomatal closure response to increasing D. This analysis indicates that individuals, species, and stands with low stomatal conductance at low D exhibit a lower sensitivity to D [Oren et al., 1999]. Thus, the lack of a clear response of δCER to D in this study is consistent with physiological studies in seedlings [Johnson et al., 1995] and mature trees [Teskey et al., 1994] that find a limited influence of D on net photosynthesis and stomatal conductance in P. elliottii and P. palustris [Gonzalez-Benecke et al., 2011]. This limited influence of D is further supported by a limited variability in δCSC (Figure 3b) despite large variations in D during the study period (Figures 2a and 2c). δCER is reflecting the δ13C of assimilated photosynthates but with an offset (Figure 8).

[62] Much of the insensitivity of δCSC and δCER to D at the Florida sites compared to that observed in a Loblolly pine at Duke Forest [Mortazavi et al., 2005] is likely related to the differences in the soil water availability patterns. In the flat spodosols common in the lower coastal plain of Florida, the surficial water table is generally within 1 m of the surface except during episodic drought [Abrahamson and Hartnett, 1990; Powell et al., 2008]. During the study period, depth to the water table varied from 22 to 224 cm (Figures 2b and 2d) and averaged 101 and 74 cm at DT and AC, respectively. The fine roots of P. palustris reach 2 m [Addington et al., 2006] and in the Florida sandy soil have taproots that can extend at least 4 m [Heyward, 1933]. The pine tree taproots are usually therefore in contact with saturated soils, mediating any effects of D or soil drying in upper layers. Integrative modeling studies [McMurtrie et al., 1994;Cropper and Gholz, 1993] and production ecology investigations [Martin and Jokela, 2004; Neary et al., 1990] find that P. elliottii productivity, for example, at this site is seldom limited by atmospheric or soil water stress except during periods of extreme drought. The consequences of extreme droughts are severe, however, in these ecosystems and appear to leave a long lasting metabolic impact as evidenced by δCW (Figures 4 and 6). For example, the drought of 2000–2002 which resulted in the water table being depressed to >2 m for extended time periods caused premature needle fall in May and June of 2001 that was equivalent to 50% of the annual needlefall [Powell et al., 2005]. Access to groundwater at a Douglas-fir forest was also suggested byMcDowell et al. [2004a] to have buffered variability in canopy gas exchange from drought.

[63] The low ranges of variability in δ13C of the foliage metabolic pools and in the respiratory fluxes observed here (Figures 3b, 8 and 10) are consistent with studies that have shown a lower sensitivity in stomatal conductance to D for deep rooting species [Bréda et al., 1993; Bréda et al., 1995], but a high sensitivity in stomatal conductance to D for species in areas where rooting depth is limited [Pataki et al., 1998; Bréda et al., 1993]. The contrasting results between this study in the coastal plain Florida and the Duke forest in North Carolina supports this view, as the Duke forest has a clay pan at 35 cm depth that isolates a surficial soil layer from which 90% of water for vegetation transpiration is derived from [Oren et al., 1999]. Thus, canopy leaf conductance at Duke Forest is strongly related to soil moisture availability in the upper soil layer [Oren et al., 1999] and Ci/Ca and δCER vary predictably in response to D [Mortazavi et al., 2005].

[64] The range of δCER values for the identical D values (Figure 10) highlights the influence of additional factors besides stomatal conductance on δCER. Changes in the relative contribution of soil and aboveground biomass-to-ecosystem respiration, as well as variability in the ratio of autotrophic to heterotrophic respiration ultimately influenceδCER.Seasonal changes in source-to-sink [Kozlowski, 1992], for example, influence carbohydrate flow in the phloem [Peuke et al., 2006] will therefore influence the temporal coupling between assimilated and respired δ13CO2. As suggested by Wingate et al. [2010] understanding the temporal variability δCER and how it varies with environmental variables that influence Δ will require explicit accounting for how recently assimilated carbon is mixed into a larger carbon pool that serves as the substrate for respiration.

5. Conclusions

[65] The persistent availability of moisture due to a shallow water table and deep taproots dampen fluctuations in the δ13C of either the assimilated (and subsequently respired) CO2 at the study sites that might have been observed in areas with more limited moisture availability. We found a similar 2‰ enrichment in δCOM, δCSC and δCW in P. palustris relative to P. elliotii, indicating that these isotopic proxies, on average, similarly record species-level differences in Δ. However, theδ13C of total needle carbon and waxes δ13C in P. elliottii and P. palustrisappear to more strongly reflect long-term trends in theδ13C of storage reserves used for biosynthesis rather than any short-term variability in Δ, and thus provide a temporally integrated signal of photosynthetic discrimination. Differences in molecular and isotopic trends observed in the primary C20–26 and secondary C32–36 wax n-alkanoic acids provides evidence for the existence of parallel but separate chain elongation systems for compounds of differing chain lengths in these species, each utilizing different carbon precursor pools. Our results point to a need for targeted isotopic labeling experiments that focus on the internal carbon cycling of long-lived species with substantial storage reserves, as currently we have little understanding of the processes by which carbon is shunted internally to supply the cell's biosynthetic pathways and how this affects the isotopic signals that are recorded in different carbon pools and in respiration.


[66] We thank M. Smith and J. Malheur for field and laboratory assistance, and R. Mitchell for comments on an earlier version of this manuscript. We also thank L. Samuels for stimulating discussions on wax biosynthesis. This work was supported by a NSF grants DEB-0343604, DEB-0344562 and DEB-0552202, and DOE grant DE-FC02-06ER64156/06-SC-NICCR-1063. We would like to thank 3 anonymous referees, the associate editor and D. Baldocchi for their comments and suggestions on a previous version of this manuscript.