Declining net primary production (NPP) with forest age is often attributed to a corresponding decline in gross primary production (GPP). We tested two hypotheses explaining the decline of GPP in ageing stands (14–115 years old) of Pinus taeda L.: (1) increasing N limitation limits photosynthetic capacity and thus decreases GPP with increasing age; and (2) hydraulic limitations increasingly induce stomatal closure, reducing GPP with increasing age. We tested these hypotheses using measurements of foliar nitrogen, photosynthesis, sap-flow and dendroclimatological techniques. Hypothesis (1) was not supported; foliar N retranslocation did not increase and declines were not observed in foliar N, leaf area per tree or photosynthetic capacity. Hypothesis (2) was supported; declines were observed in light-saturated photosynthesis, leaf- and canopy-level stomatal conductance, concentration of CO2 inside leaf air-spaces (corroborated by an increase in wood δ13C) and specific leaf area (SLA), while stomatal limitation and the ratio of sapwood area (SA) to leaf area increased. The sensitivity of radial growth to inter-annual variation in temperature and drought decreased with age, suggesting that tree water use becomes increasingly conservative with age. We conclude that hydraulic limitation increasingly limits the photosynthetic rates of ageing loblolly pine trees, possibly explaining the observed reduction of NPP.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Previously, we demonstrated that NPP declines strongly with age in loblolly pine (Pinus taeda L.) forests, driven exclusively by an 80% decline in pine wood production (Drake et al. 2010). Sap-flow-derived estimates of GPP and measurements of stem respiration indicated that the decline in pine NPP was driven by a decline in pine GPP, not an increase in Ra (Drake et al. 2010).
Here, we tested two hypotheses regarding an age-related decline in GPP: (1) increasing N limitation increasingly limits photosynthetic capacity and thus decreases GPP with increasing age; and (2) increasing hydraulic limitation increasingly induces stomatal closure and thus decreases GPP with increasing age. If hypothesis (1) was true, we expected to observe declining foliar N%, declining leaf area per tree, increased translocation of N before leaf abscission and declining photosynthetic capacity as defined by the maximum rate of carboxylation (Vc,max) and the maximum rate of electron transport (Jmax; Long & Bernacchi 2003). If hypothesis (2) was true, then we expected to observe declines in stomatal conductance, reduced specific leaf area (SLA) driven by reduced turgor-induced cellular expansion, decreased concentrations of CO2 inside leaves (ci) relative to the CO2 concentrations in the atmosphere around leaves (c.), reduced canopy conductance and compensation of hydraulic limitation through an increasing ratio of sapwood area (SA) to leaf area (Ryan et al. 2006). Many of these predictions were tested with leaf-level gas exchange and stem sap-flow measurements. We also used a dendroclimatological approach to investigate correlations between radial growth and climate, with the prediction that increased hydraulic limitation would make radial growth more strongly correlated with inter-annual variation in temperature and drought with increasing age.
We identified a chronosequence of 12 forest stands that were similar in all respects except the age of the dominant loblolly pine (P. taeda) trees. All sites were within the Korstian and Durham divisions of the Duke Forest, located in Durham County, NC, USA (centred near 36°00′N 78°58′W). Sites were chosen to have similar land use, initial stocking densities, management history and soil types (J. Edeburn, personal communication), but the age of the pines varied from 14 to 115 years, including the oldest loblolly pines present in this region (Table 1). The Helena soil series was chosen for site selection because it closely matches the soil at the nearby Duke Free-air CO2 Enrichment (FACE) site (Hamilton et al. 2002), enabling comparisons between data presented here and this 26-year-old forest that lies ∼5 km to the west. The maximum distance between stands was 6 km. Detailed descriptions of the site selection criteria, stand characteristics and production measurements are provided in Drake et al. (2010).
Table 1. Stand characteristics of 12 Pinus taeda forests of varying age in North Carolina
Stand maximum LAI (m2 m−2)
Pine LAI (m2 m−2)
Pine SA (cm2 m−2)
Pine SA/LAI (cm2 m−2)
Values are the mean (±1 SE) of 30–150 measurements for DBH and 10 measurements for height and LAI. LAI refers to projected leaf area index, and SA refers to sapwood area per unit ground area. SA and SA/LAI were calculated at the stand level, and thus have no subreplicate measurements.
Pine SA was measured in all stands using an allometric relationship between diameter at breast height (DBH) and SA; the allometry was derived by coring 41 trees of varying size, with two perpendicular cores per tree, and differentiating sapwood with a 10% ferric chloride solution [SA (cm2) = 18.24*DBH (cm) − 258.05; r2 = 0.81, P < 0.001, n = 41).
Ecosystem leaf area index (LAI, m2 leaf area m−2 ground area) was measured at 10 locations within each stand with a canopy analyser (LAI 2000, Li-Cor, Lincoln, NE, USA) at approximately monthly intervals. Measurements of LAI were corrected for foliar clumping and woody interception (Chen 1996) as in Drake et al. (2010). The relative contribution of pines and hardwoods to total LAI was estimated from litterfall collections (Drake et al. 2010). Pine LAI was estimated as the difference between summer maximum ecosystem LAI (optical method) and hardwood LAI (litterfall method). The ratio of pine SA to pine leaf area was calculated at the stand level by summing the SA of all pines and dividing by pine LAI (Table 1).
Foliage gas exchange was measured with cross-calibrated, open gas exchange systems (model 6400, Li-Cor Biosciences) under saturating irradiance (1500 µmol PAR m−2 s−1, 6400-02B red/blue light source, 2 × 3 cm chamber, Li-Cor Biosciences). The midpoint of three fascicles (i.e. nine needles) was enclosed in the chamber; leaks formed by clamping the flat chamber gaskets onto irregular needle surfaces were sealed with Permagum (Virginia KMP, Broadview, IL, USA), and each sample was verified to be leak tight by exhaling on the chamber. Relationships between net photosynthesis (Anet) and the CO2 partial pressure (ci) inside needle air-spaces (i.e. A:ci curves) were measured by recording Anet at the following ambient concentrations of CO2 (c.) in the reference cell of the gas exchange system: 400, 250, 150, 50, 400, 550, 750, 1000, 1200, 1400 and 1800 ppm. All A:ci curves were analysed using the PS-fit program (version 7.3, Bernacchi et al. 2002; Bernacchi, Pimentel & Long 2003; Long & Bernacchi 2003). Parameters derived from this fitting procedure (e.g. Vc,max, and Jmax) were reported at 25 °C.
Loblolly pine trees carry needles for 18 months (Zhang & Allen 1996), and thus have cohorts of needles that were produced in the current year and the previous year. The previous year's foliage was measured in all cases, as these needles were more likely to have experienced conditions of water stress and would be more likely to display evidence of hydraulic limitation. Additionally, old needle cohorts often maintain similar photosynthetic activity relative to new foliage (Greenway, Macdonald & Lieffers 1992; Crous & Ellsworth 2004), and thus contribute substantially to whole stand photosynthesis.
As these sites lacked canopy access, we sampled canopy foliage with a shotgun (Liu, Ellsworth & Tyree 1997; Thomas et al. 2008). To investigate potential artefacts of measuring gas exchange on detached branches, especially the possibility of reduced stomatal conductance (gs), we conducted two experiments using the canopy access towers at the Duke FACE site where gas exchange of attached and detached branches was compared; trees at this site averaged 22 cm DBH and 20 m tall in 2008. In the first experiment, needles were placed in gas exchange chambers while maintaining the connection to the branch and until light-saturated photosynthesis (Asat) reached steady state, after which the branch was detached from the tree and Asat was measured every minute over the next hour. We performed these measurements throughout the course of 2 d to identify the optimal time of sampling. In the second experiment, A:ci curves were measured on attached needles in the mid-morning; the branch was then detached and a second A:ci curve was measured after 10 min. Five pairs of A:ci curves were measured in this manner to investigate the potential bias of branch sampling on gas exchange measurements.
Having confirmed the efficacy of shotgun sampling (see Results), 8–10 upper canopy branches were shot down from 9 of the 12 sites in the midmorning through the early afternoon (1000 to 1300 EST). Three sites were excluded because their proximity to roads or public dwellings made shotgun sampling prohibitively dangerous. Gas exchange was measured immediately following shotgun sampling. One site was measured per day from July 12th to July 23rd, 2008. We controlled the block temperature of the gas exchange systems at 30 °C at all sites; average leaf temperature was between 29.9 to 32 °C and did not vary significantly with age (regression, P > 0.1). We did not control water vapour pressure in the cuvette which varied from 1.29 to 1.99 kPa, but did not vary systematically with stand age (regression, P > 0.1). These results indicate that day-to-day variation in climate conditions was small. The needle area inside the chamber was photographed, and projected leaf area was measured using the ImageJ program (Abramoff, Magelhaes & Ram 2004). Afterward, needles were dried for 2 d at 60 °C, ground and analysed for carbon and nitrogen concentrations using micro-Dumas combustion elemental analysis (ECS 4010, Costech Analytical, Valencia, CA, USA). Gas exchange rates were reported on a projected area basis, and SLA was measured as projected needle area per dry mass (cm2 g−1).
Nitrogen translocation and leaf area
Fresh litterfall was collected from 10 0.22 m2 litter traps per plot 3 d after the traps were emptied to ensure that litter had not experienced precipitation. Percent N retranslocation was calculated as (Nfresh − Nlitter)/Nfresh as in Finzi et al. (2001), where Nfresh was the N concentration in fresh needles obtained by shotgun, and Nlitter was the N concentration in freshly abscised needle litterfall.
In addition to affecting foliar N concentrations, increasing N limitation has the potential to reduce the leaf area of individual trees (Ryan et al. 2004). Pine LAI at the ecosystem level declined from 4.1 to 2.2 m2 m−2 from 15 to 115 years old, but total ecosystem LAI remained constant at ∼4.5 for all ages because of the accumulation of later successional hardwood foliage (Table 1, Drake et al. 2010). To distinguish whether increasing N limitation reduced the leaf area supported by each individual tree, or if pine LAI declined because self-thinning reduced pine stem density, we quantified pine leaf area per individual by dividing summer pine LAI by stem density and compared these values to those predicted by an equation relating DBH to leaf area derived from destructive sampling at the Duke FACE site (Naidu, DeLucia & Thomas 1998).
We instrumented eight pine trees in five of the 12 stands (14, 19, 36, 70 and 97 years of age) with thermal dissipation sap-flow probes (Granier 1987). We estimated radial and circumferential variation in Js (Phillips, Oren & Zimmerman 1996; Ewers & Oren 2000; Wullschleger & King 2000) with additional measurements on the north and south sides (0–20 mm depth) and at 20–40 mm depth (north side) on two pines per stand. There was no difference between Js measured in the outer 20 mm on the north-and south-facing sides (P > 0.1), so we did not apply circumferential corrections. We accounted for a measured decline in Js with increasing sapwood depth as in Phillips et al. (1996) and Schäfer et al. (2002). Atmospheric humidity and temperature were measured with a capacitive relative humidity sensor and a thermistor (Vaisala HMP 35C; Campbell Scientific, Logan, UT, USA) suspended in the upper third of the canopy in each stand, and vapour pressure deficit (VPD) (D) was calculated from relative humidity and air temperature. Leaf transpiration (El) and canopy conductance (Gs) were calculated using monthly measurements of LAI and a site-specific relationship between DBH and SA (Drake et al. 2010) as in Ewers et al. (2001).
Wood δ13C composition
We measured the stable carbon isotope composition (δ13C) of wood to investigate if ci/c. measured by gas exchange was characteristic of the long-term behaviour of these trees. Biomass δ13C values are often used as long-term integrators of ci/c. (Farquhar, Oleary & Berry 1982; Katul, Ellsworth & Lai 2000; Dawson et al. 2002) because the relative importance of isotopic fractionation by ribulose 1·5-bisphosphate carboxylase/oxygenase (Rubisco) carboxylation (fractionation of ∼27‰) versus diffusion through the stomatal pore (fractionation of 4.4‰) varies with ci/c. Previously, we reported wood δ13C values of the last 5 years of growth of eight pines per site (Drake et al. 2010); we include these data as an independent reference for the measurements of ci/c. derived from gas exchange. Wood was sampled with tree cores, dried, ground and analysed for δ13C at the University of Illinois using an elemental analyser (ECS 4010; Costech Analytical) coupled to a Conflo IV interface (Thermo, Bremen, Germany) and a Delta-V advantage isotope ratio mass spectrometer (Thermo). The precision calculated from 12 samples run in duplicate was 0.09‰, and the average measured deviation of an in-house isotopic reference material was <0.05‰ of its known value (n = 5).
Tree ring analysis
Twenty canopy trees were cored per site, with two cores at perpendicular angles per tree. The cores were sanded, imaged using a flatbed scanner (Epson Perfection V700 Photo; Epson, Rockford, IL, USA) and ring widths for the past 30 years were quantified using an image analysis program (Windendro; Regent Instruments Inc, Nepean, Ontario, Canada) and averaged by tree. Ring widths were detrended and built into site-level chronologies using standard dendrochronology techniques and the dplR package (R Development Core Team 2007; Bunn 2008). Detrended ring widths were then regressed against climate variables using DENDROCLIM (Franco & Kishor 2004), which determines the strength of the relationship between climate variables and annual ring width, and evaluates the significance of these correlations with bootstrap resampling techniques. Monthly average temperature and Palmer drought severity index (PDSI) for the past 30 years in Durham County were obtained from the National Climatic Data Center (http://www1.ncdc.noaa.gov/pub/data/cirs/).
Statistical analyses were performed using the MIXED and REG procedures of the SAS system (SAS 9.1; SAS Institute, Cary, NC, USA). The assumptions of homoskedasticity and normality of residuals were checked using the UNIVARIATE and REG procedures. Non-linear regressions were fit in Sigmaplot (version 10.0; Systat Software, San Jose, CA, USA). Stands were treated as the unit of replication in all cases (n = 9 for gas exchange, fluorescence and SLA; n = 12 for wood isotopes and dendroclimatology; n = 5 for sap-flow).
Validation of gas exchange measurements
Detached branches sampled in the late afternoon exhibited a strong decline in gs once harvested, while branches sampled in the mid-morning showed no change in gs for >40 min (Fig. 1a), which was sufficient to measure an A:ci curve. A:ci curves measured on attached and detached foliage showed a strong correlation between corresponding rates of photosynthesis (Fig. 1b, detached Anet = 0.2 + 0.95* attached Anet, P < 0.001, r2 = 0.99, slope not significantly different than 1, P > 0.5). Additionally, Vc,max and Jmax derived from these measurements did not differ between attached or detached foliage (paired t-test, P > 0.3 for Vc,max, P > 0.1 for Jmax). Thus, we concluded that A:ci curves could be measured on detached foliage without measurement artefacts as long as the branches were detached before 1400 h EST.
N limitation hypothesis
While ecosystem-level pine LAI declined with age (Table 1; Drake et al. 2010), leaf area per individual tree increased with age along the trajectory predicted by an equation derived from destructive measurements at the Duke FACE site (Fig. 2a; Naidu et al. 1998). There was a marginally significant increase in foliar N (Fig. 2b, P = 0.08) and no evidence of a systematic change in photosynthetic capacity with increasing age (Fig. 2c,d) with no detectable change in Vc,max or Jmax. There was no detectable change in the retranslocation of N (Table 2).
Table 2. Foliar N and C concentrations, nitrogen retranslocation and the ratio of sapwood area (SA) to leaf area of Pinus taeda at 12 forests of varying age in North Carolina
% Pine N retrans location
Values are the mean (±1 SE) of 8–10 replicates. The collection of fresh foliage by shotgun was not feasible in three of the 12 sites; these values are indicated as NA. There were no significant relationships between measured variables and stand age (regression, P > 0.05).
Hydraulic limitation hypothesis
Light-saturated photosynthesis declined by ∼25% with increasing age, from 12 to 9 µmol CO2 m−2 s−1 at 15 and 115 years of age, respectively (Fig. 3a). There was a corresponding age-related decline of ∼40% in stomatal conductance, from 0.15 to 0.09 mol H2O m−2 s−1 (Fig. 3b). The decline in conductance led to a decline in ci (Fig. 3c) and an increase in the stomatal limitation to photosynthesis (Fig. 3d) with increasing age. When combined with the absence of an age-related change in Vc,max or Jmax, these results suggest that the shape of A:ci curve did not vary with age, but rather that older trees were operating at a lower ci (Fig. 4). In addition, needle SLA declined significantly with age (Fig. 5) from 47 to 40 cm2 g−1 at 15 and 115 years of age, respectively, and the ratio of SA to leaf area increased with age from 3.1 cm2 m−2 at age 15 to ∼6 cm2 m−2 at age 115 (Table 1; y = 4.01 + 0.017x, P < 0.05, r2 = 0.34).
The decline in ci with increasing age was corroborated by wood C isotopic measurements, which reflect time-integrated ci/c. values (Katul et al. 2000). Wood δ13C increased from −27.8‰ at age 14 to −26.3‰ at age 115 (y = −27.29 + 0.0125x, r2 = 0.45, P < 0.05), which represents a decline in ci/c. from 0.65 to 0.58 (y = 0.65 − 0.0005x, r2 = 0.45, P < 0.05). Estimates of mean ci/c. from gas exchange and wood δ13C were correlated [isotopic ci/c. = 0.321 * (gas exchange ci/c.) + 0.442, r2 = 0.39, P < 0.05].
Sap-flow measurements indicated that day-time leaf transpiration rates declined by ∼60% with increasing age (Fig. 6a). The youngest forests had the highest day-time canopy-weighted stomatal conductance (Gc) at low VPD, but these forests rapidly reduced Gc as VPD increased (Fig. 6b). In contrast, the oldest forests always had low Gc, and were thus less sensitive to increases in VPD. Age-related differences in Gc were most pronounced at low VPD values.
The dendroclimatologial analysis demonstrated that radial growth of all forest ages was greatest in years with cool summers and high water availability (i.e. high values of PDSI; Fig. 7). However, the strength of the correlation between inter-annual variation in climate and radial growth declines with age. Thus, the youngest pines were more sensitive to hot and dry summers relative to older forests. Summer mean temperature and PDSI were not correlated (P > 0.5).
Light-saturated photosynthesis declined by ∼25% with increasing age, and the corresponding decline in stomatal conductance and decrease in ci suggest that this was driven by increasing hydraulic limitation. Results were not consistent with any of the predictions regarding the hypothesis of increasing N limitation, while all but one of the predictions regarding the hypothesis of increasing hydraulic limitation were supported. Leaf- and canopy-level stomatal conductance declined, ci/c. declined, stomatal limitation increased, SLA decreased and the ratio of SA to leaf area increased. We conclude that hydraulic limitation increasingly limits photosynthesis in ageing loblolly pines. The decreasing sensitivity of radial growth to adverse inter-annual climate variation with increasing age, the opposite of our prediction, suggests that hydraulic limitation may reflect an increasingly conservative growth strategy of pine trees as they age.
The data presented here support five predictions from the hydraulic limitation hypothesis specified by Ryan et al. (2006): (1) reduced canopy Gs; (2) reduced leaf level gs; (3) increased δ13C and reduced ci/c.; and (4) reduced Asat. Additionally, (5) there was evidence of compensation for increasing hydraulic limitation, as the ratio of SA to leaf area (cm2 m−2) increased significantly with age; SA to leaf area was 3.1 cm2 m−2 at age 14, and 6 cm2 m−2 at age 115. DeLucia, Maherali & Carey (2000) demonstrated that published values of pine SA to leaf area were positively related to summer average daily maximum VPD; the youngest forest measured here corresponds to pines growing at a VPD of ∼1.5 kPa, while the oldest forests correspond to pines growing at ∼2.8 kPa. Thus, pines are known to vary the amount of sapwood per unit leaf area in response to evaporative demand, and pines observed here compensated across the full range of observed values in the DeLucia et al. (2000) data set.
While the dendroclimatological results (Fig. 7) were the opposite of predicted, they were consistent with the Gc measurements derived from sap-flow (Fig. 6b). Gc of the youngest trees was very sensitive to variation in evaporative demand, while the oldest trees had consistently low Gc that varied little with VPD. This is consistent with the correlations between radial growth and inter-annual climate variation (Fig. 7); growth by the youngest trees was most sensitive to environmental variation related to water availability, while growth by the oldest trees was less affected. Together, these results suggest that these trees increasingly adopt a stress-tolerant, conservative growth strategy with increasing age. Grime (1977) hypothesized that forest communities should change from ruderal to stress-tolerating species during the course of secondary succession. The same process of growth strategy acclimation may occur within individual species exhibiting state-dependent plasticity in life history strategies, as has been recognized in animals (McNamara & Houston 1996).
The results of this study are in general agreement with the literature, showing little support for increasing N limitation as a cause of the age-related decline in GPP, and hence NPP. While N limitation of NPP is widespread across ecosystem types and particularly strong in temperate forests (Elser et al. 2007; LeBauer & Treseder 2008), there is no established pattern of declining nutrient availability with increasing age commensurate with declining NPP (Ryan et al. 1997a; Olsson, Binkley & Smith 1998). Some ecosystem modelling efforts have suggested that nutrient availability should decline with age as N is immobilized in woody biomass (Murty & McMurtrie 2000), but observational support for this process is rare (Johnson 2006), and N availability can even increase as forests age (Smithwick et al. 2009).
In conclusion, leaf-level photosynthetic rates by loblolly pine declined with increasing age through increasing hydraulic limitation of stomatal conductance, not increasing nutrient limitation. Combined with previous measures of declining production in these forests (Drake et al. 2010), it appears that hydraulic limitation of GPP can explain the observed age-related decline in NPP for these loblolly pines. Thus, this study supports the hypothesis of Ryan et al. (1997a) that GPP is the driver of age-related changes in NPP, rather than Ra as was previously thought (e.g. Odum 1969). This conclusion has implications for the future of the C sink in aggrading temperate forests (Pacala et al. 2001), because elements of global change that primarily affect photosynthesis (i.e. CO2) may have a larger influence on the future productivity of these forests than elements primarily affecting Ra (i.e. temperature).
We gratefully acknowledge George Hendrey and Robert Nettles (Brookhaven National Laboratory) for the operation of the Duke FACE experiment. We thank Judson Edeburn and Marissa Hartzler (Duke Forest) for assistance identifying the chronosequence stands, providing historical management information and allowing the use of shotguns on University property. Richard Thomas, Andrew Leakey and Patrick Morgan gave helpful suggestions concerning the gas exchange measurements. We thank Ram Oren and Eric Ward for help designing and implementing the sap-flow instrumentation. The Duke FACE experiment was supported by the Office of Science (BER), US Department of Energy Grant No. DE-FG02-95ER62083 and through its Southeast Regional Center (SERC) of the National Institute for Global Environmental Change (NIGEC) under Cooperative Agreement No. DE-FC02-03ER63613. Additional support was provided by DOE (BER) Grant No. DE-FG02-04ERG384. J.D. gratefully acknowledges support through a University Distinguished Fellowship from the University of Illinois, and multiple travel grants from the Department of Plant Biology and the Program of Ecology, Evolution, and Conservation Biology at the University of Illinois.