Tree Physiology 00, 1–19doi:10.1093/treephys/tpr048
Paired comparison of water, energy and carbon exchanges over two young maritime pine stands (Pinus pinaster Ait.): effects of thinning and weeding in the early stage of tree growth
Virginie Moreaux1,3, Éric Lamaud1, Alexandre Bosc1, Jean-Marc Bonnefond1, Belinda E. Medlyn2 and Denis Loustau1
Received December 18, 2010; accepted May 10, 2011; handling Editor David Tissue
The effects of management practices on energy, water and carbon exchanges were investigated in a young pine plantation in south-west France. In 2009–10, carbon dioxide (CO2), H2O and heat fluxes were monitored using the eddy covariance and sap flow techniques in a control plot (C) with a developed gorse layer, and an adjacent plot that was mechanically weeded and thinned (W). Despite large differences in the total leaf area index and canopy structure, the annual net radiation absorbed was only 4% lower in plot W. We showed that higher albedo in this plot was offset by lower emitted long-wave radiation. Annual evapotranspiration (ET) from plot W was 15% lower, due to lower rainfall interception and transpiration by the tree canopy, partly counterbalanced by the larger evaporation from both soil and regrowing weedy vegetation. The drainage belowground from plot W was larger by 113 mm annually. The seasonal variability of ET was driven by the dynamics of the soil and weed layers, which was more severely affected by drought in plot C. Conversely, the temporal changes in pine tran-spiration and stem diameter growth were synchronous between sites despite higher soil water content in the weeded plot. At the annual scale, both plots were carbon sinks, but thinning and weeding reduced the carbon uptake by 73%: annual carbon uptake was 243 and 65 g C m−2 on plots C and W, respectively. Summer drought dramatically impacted the net eco-system exchange: plot C became a carbon source as the gross primary production (GPP) severely decreased. However, plot W remained a carbon sink during drought, as a result of decreases in both GPP and ecosystem respiration (RE). In winter, both plots were carbon sources, plots C and W emitting 67.5 and 32.4 g C m−2, respectively. Overall, this study highlighted the significant contribution of the gorse layer to mass and energy exchange in young pine plantations.
Keywords: carbon exchanges, energy balance, gorse, water budget, young forest stand.
Introduction
Forests play an important role in the climate system, providing important feedback to the level of carbon dioxide (CO2) in the atmosphere. Temperate forests in Europe are currently net car-bon sinks (Nabuurs et al. 2003, Loustau 2010), due to several factors: juvenile forest age structure; an increase in nitrogen (N) availability, atmospheric N deposition; and rising atmo-spheric CO2 concentration (Luyssaert et al. 2010). Several
modelling studies suggest that these factors will also have a positive effect on future forest production (Loustau et al. 2005, Ciais et al. 2008, Dezi et al. 2010).
However, on a long temporal scale, this positive net carbon uptake could be significantly compromised (Ciais et al. 2008), due to changes in the forest disturbance regimes, including nat-ural events such as fire, disease, drought or heat waves (Granier et al. 2007, Sohngen 2008), but also changes in management
Research paper
Tree Physiology Advance Access published June 30, 2011 at IN
practice such as site preparation, thinning and clear-cutting (Kowalski et al. 2003, Chen et al. 2004, Misson et al. 2005, Dore et al. 2010).
In European forests, management (e.g. clear-cutting, thin-ning) is the main disturbance during the forest life cycle (Janssens et al. 2001, Kowalski et al. 2003). In response to the demand for wood-energy, forest crop systems are being devel-oped, particularly in the Landes region in south-western France, which is climatically well suited to such systems. Future man-agement scenarios for this region include plantation of stands for intensive production of bio-energy. Mixed species planta-tions, particularly mixtures with pioneer or weedy N-fixing spe-cies, are under consideration. How to make optimal use of the forest for which biomass would be used in the bio-energy sec-tor is of particular interest. Such forests are not managed for carbon sequestration, but nonetheless contribute to climate mitigation efforts by reducing carbon emissions from the energy sector (Nabuurs et al. 2008). Therefore, management strategies such as thinning and weed control, play an impor-tant role when considering whole forest rotation and magni-tude of the sink. At the same time, N-fixing species growing between trees play a potentially useful ecological role in young forest plantations which suffer from soil N deficiency. However, because they may compete with growing trees for other resources such as water, it remains an open question whether they provide a net benefit to forest growth. Moreover, these forest systems are characterized by the importance of the period preceding canopy closure, during which the forest has a highly heterogeneous spatial structure. In France and particu-larly in the Landes region, this stage of the rotation is particu-larly important since the stand is characterized by the coexistence of trees growing with a developed herbaceous and weedy layer covering 70% of the ground surface. In this region, removal of weedy layers usually occurs 5 years after plantation or sowing (Lesgourgues et al. 1997), so that this stage corresponds to a critical point when the carbon balance is annually close to neutral or when the first centimetres of the soil become stabilized a few years after harvesting and planta-tion. During this period leading up to canopy closure, the respective contributions of soil and vegetation to atmospheric mass and energy exchanges change rapidly in an interdepen-dent way. The contribution of trees is negligible at the sowing or planting stage, but gradually becomes larger than the soil and weedy components and subsequently becomes the domi-nant stratum at ages ranging from 6 to 10 years. Thus, the nature of the herbaceous layer is important, particularly if it presents advantages such as natural N-fixing ability. However, the environmental responses of this young stage, with or with-out N-fixing species, are not well understood, making their management highly uncertain.
In the forest in the Landes region, many studies have been performed to determine energy, water and carbon exchanges,
but relatively few have been carried out in young stands (Kowalski et al. 2004, Stella et al. 2009). The majority were conducted in adult stands (Berbigier et al. 1991, Lamaud et al. 2001, Ogée et al. 2001, Medlyn et al. 2002, Bosc et al. 2003, Delzon and Loustau 2005, Jarosz et al. 2008, 2009). In other countries, there have been a number of studies on responses to disturbances such as fire (Dore et al. 2010) and thinning and clear-cutting (Kowalski et al. 2003, 2004, Misson et al. 2005, 2007, Dore et al. 2010, Sun et al. 2010). However, in general there are few studies on young stands and their response to management practices such as weeding and early thinning. The change in vegetation cover during this stage is likely to have a large effect on atmospheric and soil exchanges of carbon and water (Simonin et al. 2007). Comparing adult, clear-cut and young plantations, Kowalski et al. (2003, 2004) showed that the first phase after a disturbance (e.g. soil preparation) is par-ticularly important in the carbon cycle, the energy balance and greenhouse gas exchanges. Between plantation establishment and canopy closure, the net absorbed radiation increases by 50%, the albedo decreases from 50 to 25% and the total evapotranspiration (ET) increases from 20 to 50%. The net carbon exchange reverses: a young stand of 1–2 years old acts as a source of CO2, but rapidly becomes a carbon sink as the forest grows. To assess the contribution of the understorey, ver-tical partitioning between understorey vegetation and oversto-rey pines has also been evaluated in adult stands (Berbigier et al. 1991, Lamaud et al. 2001, Ogée et al. 2001, Misson et al. 2007, Jarosz et al. 2008). These studies concluded that differ-ences in the structural organization of pines and understorey influenced the partitioning of all the components involved in biophysical processes, such as heat fluxes or carbon uptake.
This study aimed to investigate the effects of thinning and weeding on energy, water and carbon ecosystem exchanges during the juvenile stage of tree growth. The presence of a layer of dwarf gorse and other weedy plants is also studied and its effects on these exchanges assessed. To achieve these objectives, we compared mass and energy exchanges in two contiguous young forest stands during one year (June 2009 to May 2010). The first plot was covered by a continuous layer of gorse between rows of maritime pine. In the second plot, the gorse layer was mechanically weeded and the maritime pines thinned. We analysed the variations in the biophysical charac-teristics and carbon exchange at seasonal and annual scales in response to the climate conditions. Several hypotheses were assessed: (i) the change in tree and gorse cover due to weed-ing and thinning would decrease available energy; (ii) the lower total leaf area would increase water availability for trees in the thinned and weeded plot, and thus mitigate the impacts of summer water deficit; and (iii) given that the young stands have a carbon balance that is close to neutral, weeding and thinning would change the ecosystem from a small carbon sink to a source.
Tree Physiology Online at http://www.treephys.oxfordjournals.org
Materials and methods
Experimental sites
The study was carried out at the Bilos site, located in the Landes forest ~50 km south-west of Bordeaux, France (44°29′37.99″N; 0°57′21.9″W). The climate is temperate with a maritime influence. The 1950–2000 mean annual tempera-ture and precipitation are respectively 13 °C and 977 mm. The site covers a 1 × 0.6 km area and is managed according to standard management practices (Lesgourgues et al. 1997). The soil is a sandy podzol with a discontinuous layer of iron hard pan at 75 cm depth that limits root extension. The charac-teristics of the site and prevailing climate features are summa-rized in Table 1.
Following clear-cutting in 1999, the site was ploughed to 30 cm depth and fertilized with 60 kg P per ha in 2001. Measurements of turbulent fluxes of momentum, sensible and latent heat, and CO2 at this site have been made using the eddy covariance technique since 2000. In November 2004, the site was divided into two halves, which were seeded with maritime pine (Pinus pinaster Ait.) with a 1-year lag, in 2004 and 2005, respectively, tree rows being spaced at 4 m. In November 2008, the older stand (plot W) was thinned and weedy vegetation cleared while the younger stand (plot C) was left unmanaged. In the first stand, the soil surface was, subsequent to the thinning and weeding, covered by woody debris. Weedy vegetation, including gorse and other herba-ceous plants, regrew spontaneously and was mechanically destroyed for a second time in November 2009. In the second stand, plot C, a well-developed weed layer composed of gorse (Ulex minor, Roth), heather (Calluna vulgaris L.) and other her-baceous plants (Molinia coeruela M., Phytolacca americana L.), covered the soil surface. The weeding and thinning induced a difference in basal area of 2.92 m2 ha−1 between plot C (5.26) and plot W (2.34). The treatments also reduced the leaf area index (LAI) of the stand by ~1.9 m2 m−2 in summer and by roughly 2.4 m2 m−2 in winter 2009.
In 2009, all measurements were duplicated in plot C, giving fluxes in each plot, and the two plots were further equipped with sap flow, rainfall gauges beneath the canopy and auto-mated tree stem dendrometers. Tree diameter, height, crown height and maximal crown extension were inventoried annually (2009, 2010).
Meteorological measurements
At plot W, the incident short-wave radiation (SW↓) was mea-sured with a CE180 pyranometer (Cimel Electronique, Paris, France). A CGR2 pyrgeometer (Kipp & Zonen, Delft, The Netherlands) was also used to measure the incident and reflected long-wave radiation fluxes (respectively LW↓ and LW↑). The incident (PAR) and diffuse (PARd) photosynthetic photon flux densities were measured using a sunshine sensor
BF3H (Delta-T Devices, Cambridge, UK), while the reflected PAR was measured with an SKP 215 PAR sensor (Skye Instruments, Powys, UK). Plot C was equipped with a pyranom-eter to measure the upward short-wave radiation (SW↑) and an SKP 215 sensor for the outgoing PAR. On each plot, net radiation (Rn) was measured with an NrLite pyrradiometer (Kipp & Zonen). A CNR4 net radiometer (Kipp & Zonen) was used a posteriori to calibrate our data. The four components of the energy balance were measured with this sensor.
Atmospheric pressure was measured on plot W with a CS105 barometric pressure sensor (model PTB101B; Vaisala, Helsinki,
Paired comparison of water, energy and carbon exchanges 3
Table 1. Description of the selected plots and mean environmental conditions.
Plot C Plot W
Site characteristics Geographical coordinates 44°29′36.79″N 44°29′37.99″N
0°56′57.43″W 0°57′21.9″W Altitude (m) 40 40 Surface (ha) 30 30Vegetation characteristics Tree species P. pinaster Ait. Pinus pinaster Ait. Age (July 2009) 4 years 5 yearsInter-row vegetation dry weight (g m−2) Ulex minor 682.8 ± 108.3 90.8 ± 15.3 Calluna vulgaris 171.2 ± 59.8 6.4 ± 4.0 Heather 107.3 ± 20.1 9.3 ± 8.1 Diverse herbaceous 39.9 ± 11.8 27.9 ± 8.1 Diverse mosses 30.2 ± 6.4 0.2 ± 0.2 Pteridium aquilinium 3.0 ± 2.9 1.8 ± 1.8 Stocking (tree ha−1) 144,00 1803 Mean tree LAI (m2 m−2) 1.051 0.531
Mean inter-row vegeta-tion LAI (m2 m−2)
1.87 0.492
Mean height, Ht (m) (January 2010)
1.89 2.23
Min–Max Ht (m) 0.58–3.49 0.59–3.65 Mean DBH (mm)
(January 2010)21.9 39.9
Min–Max DBH (mm) 8.8–50.4 12.1–92.7Forest practices 2008 Weed layer control None First removal Trees None Early thinning 2009 Weed layer control None Second removal Trees None NoneMean meteorological data Annual precipitation (mm) 926.2 939.0 Mean temperature (°C) 12.46 12.37 Relative humidity (%) 77.95 78.14 Vapour presssure
deficit (kPa)0.39 0.39
Wind speed (m s−1)—mean direction (°)
2.0–173.3 1.8–194.5
1Experimental data and calculations from Shaiek et al. (2011).2This value was obtained in the summer period when the herbaceous layer was regrowing.
Finland). Air temperature and humidity were measured at 4 and 6 m heights using a CS215 probe (Campbell Scientific, Logan, UT, USA) on the two plots, respectively ~1.50 and 3.50 m above tree height. Precipitation above the canopy (Pi) was measured with rain gauges: ARG100 (Campbell Scientific) on plot W and Munroe on plot C. The rainfall fraction reaching the soil, called throughfall, was determined using home-made zinc V-shaped gutters of 2 m in length, with a V-angle of 60° and a width of 15 cm (plot W) or 5 cm (plot C) set up below the canopy and directed into automated rain gauges (ARG100). On plot W, two systems composed of four crossed troughs were installed in the inter-row. Two other linear systems with four troughs were installed along tree rows. On plot C, where the canopy is continuous, we installed two systems composed of six gutters (four gutters along the tree rows + two gutters per-pendicular to the rows). For both plots, the troughs were tilted at an angle of 6°. This device has a high receiving surface so that the risk of water loss by evaporation in the gutter needs to be taken into account. This loss was estimated with a reference system installed in an open area close to our laboratory and corrections were applied to throughfall measurements. The dif-ference between Pi and throughfall gives the rainfall intercep-tion, I. All meteorological data were measured every 10 s and half-hourly averages were stored by dataloggers (CR1000, CR23X and CR10X; Campbell Scientific).
Soil measurements
In plot W, soil water content (SWC) was determined using CS615 probes (Campbell Scientific) in four pits with six depths each: two pits at 0.15, 0.15, 0.30, 0.45, 0.60 and 0.80 m depth and two pits at 0.15, 0.15, 0.30, 0.30, 0.45 and 0.60 m because of the presence of a layer of hard iron pan below. In plot C, SWC was determined using CS615 probes (Campbell Scientific) in two pits with four depths: 0.10, 0.20, 0.45 and
0.80 m. We decomposed the soil volume into five volumes of 1 m2 of surface and five different thicknesses, determined by a piecewise integration. The SWC of the total volume represents the sum of SWC of each subvolume.
The two plots were equipped with four HFP01SC plates (Hukseflux Thermal Sensors, Delft, The Netherlands) to mea-sure soil heat fluxes at 0.05 m depth.
Sap flow measurements and stem diameter growth
We used the thermal dissipation method (Granier 1985, 1987) to measure sap flow. This method uses a pair of probes con-taining a copper–constantan (Cu–Cn) thermocouple and sur-rounded by a glass-coated constantan wire. The probes were designed for small trees and were 1.5 mm in diameter and 10 mm in length. They were inserted in aluminium tubes, and installed in the stem, vertically separated by 120 mm. The upper probe was heated continuously with a constant power of 0.140 A, whereas the lower probe was unheated and mea-sured wood temperature. Sensors were covered with an alu-minium sheet to prevent exposure to rain and direct sunlight. This system was installed on six representative trees in each plot, chosen to represent the frequency distribution of stem basal area and tree height (Table 2). The original equation relating the difference in temperature between probes to sap flux was applied (Granier 1985). Sensors were not heated dur-ing several periods to estimate the potential influence of the natural thermal gradients so that data could be corrected.
The mean sap flow density was then converted into transpi-ration (Tp in kg m dH O ground( ) ( )2
2 1− − ) using the stocking (tree m−2
(ground)) and cross-sectional area of sapwood at 1.3 m height (m2
(sapwood) tree−1). In our study, Tp refers to pine transpi-ration. The assumption of a constant radial profile of sap flow can introduce systematic bias into estimates of both tree and stand water use, particularly in older coniferous stands
4 Moreaux et al.
Table 2. Characteristics of trees selected for measurements of sap flow measurements and stem diameter growth. DBH represents the stem diam-eter at 1.30 m and Ht represents the total height of the tree. Hm and Hs refer to the micro-dendrometer and sap flow probe insertion heights, respectively, and Dm and Ds to the stem diameters at these heights.
Tree Physiology Online at http://www.treephys.oxfordjournals.org
(Clearwater et al. 1999, Delzon et al. 2005). However, we assumed that in small stems, all sapwood is active without any radial gradient in sap flux. Therefore, no correction for radial gradients was applied.
In each plot, the radial growth of tree stem was monitored using micro-dendrometers installed on the six trees sampled for the sap flow measurements. These home-made micro-den-drometers use a high-resolution linear position resistive sensor (RS317-780 Radiospares, Beauvais, France) mounted on a rigid frame to the trunk. These measurements started at the end of March 2010. Measurement heights are given in Table 2.
Turbulent flux measurement and data processing
Wind velocity, temperature and CO2/water vapour fluctuations were measured on both sites with, respectively, a sonic ane-mometer (model 1210R3; Gill Instruments, Lymington, UK) and an open-path dual CO2/H2O infrared gas analyser (IRGA; model Li7500; LiCor Inc., Lincoln, NE, USA) at the top of a 6 m tower, ~3.5 m above the canopy. The two towers were separated by a distance of 580 m. We calculated the footprints for each tower following the method used in Kljun et al. (2004) taking into account the mean direction of the wind on both sites (Table 1) and turbulence parameters. We determined that the two footprints took the shape of an ellipse around their mast with a main axis length of 200 m, and they did not overlap. Tourbillon Software (INRA Ephyse, France) was used for the acquisition of data obtained from the anemometer and the analyser with a 20 Hz frequency. The EdiRe software (R. Clement, 1999, University of Edinburgh, UK) was used for the determination of turbulent scalar eddy fluxes, every half hour calculated from the covariance between the vertical com-ponent of the wind velocity and the concentrations of H2O and CO2 and air temperature (Ta).
EdiRe was also used for the data processing which was done in two steps. The first step consisted of applying several different functions to correct the data on plots C and W using a general standardized method of processing (Aubinet et al. 2000). Linear detrending was applied to scalar time series, to remove linear trends in the signals. Two-dimensional rotation was applied on plots W and C to align the streamwise wind velocity component with the direction of the mean velocity vector. The fluxes were corrected for spectral high-frequency losses using the approach of Moore (1986). CO2 and H2O fluxes were corrected from air density variations due to the use of an open-path system (Webb et al. 1980).
The second step consisted of filtering the data to remove points corresponding to technical problems, meteorological conditions not satisfying eddy correlation theory or data out of realistic bounds. Different statistical tests were applied for this filtering. Stationarity and turbulent conditions were tested with the steady state test and the turbulence characteristic test rec-ommended by Foken and Wichura (1996) and Kaimal and
Finnigan (1994). According to the different tests, only the values of the latent heat (LE), sensible heat (H) and carbon (NEE) fluxes that pass all the filters mentioned above and with a quality flag of <2 were retained. In what follows, negative values of NEE represent an uptake of atmospheric CO2 by the vegetation, whereas positive values correspond to a loss of carbon by the vegetation.
Energy balance closure
Evaluating energy balance closure is one of the methods used to check the scalar fluxes estimated by eddy covariance mea-surements (Aubinet et al. 2000). We estimated the sum of the measured latent and sensible heat fluxes, LE and H, at the half-hour scale, and compared it with the sum of the available energy, following the simplified energy balance equation:
H G+ = −LE Rn (1)
where Rn is the net radiation (W m−2) and G the soil heat flux (W m−2). We also used the method of the energy balance ratio, EBR (Wilson et al. 2002, Tanaka et al. 2008), by summing half-hourly values of (Rn − G) and (H + LE):
EBRLE
n
=+
−=
=
∑∑
( )
( )
H
R Gi
j
i
j1
1
(2)
where j indicates the number of half hours. In our case, other energy sources and sinks (Q) were ignored and, for short cano-pies, storage of energy in air and biomass (S) is expected to be small so that both terms were omitted in Eqs. (1) and (2) (McCaughey 1985). Energy balance ratio was plotted as a func-tion of the friction velocity u* at a half-hourly step to determine a threshold below which data are considered to be invalid.
Water balance
The uncorrected latent and sensible heat fluxes, LEu and Hu, were adjusted for energy balance closure following Twine et al. (2000). Briefly, the gap in energy balance between (Rn − G) and (Hu + LEu), which is likely due to underestimation of eddy fluxes, was filled by multiplying LEu and Hu by the correction factor Rn − G/Hu + LEu. On average, this correction augmented Hu and LEu by a factor of ~1.2 (see the section on energy bal-ance closure in Results). The adjusted fluxes are denoted as LE and H in what follows.
Lastly, missing values of fluxes were gap-filled following the method developed by Falge et al. (2001) and improved by Reichstein et al. (2005). This method combines the cor-relations of the fluxes with meteorological variables and their temporal auto-correlation. More information can be found in Reichstein et al. (2005). The series of corrected and gap-filled data were then used to calculate monthly or annual
Paired comparison of water, energy and carbon exchanges 5
budgets. For water balance analysis, LE was converted into total ET and summed on a monthly scale. The difference between ET and (I + Tp) provides an estimate of the soil evaporation and weed layer transpiration, Eh. Soil water bal-ance, ΔSWC, was calculated monthly as the net change in SWC for the 0–80 cm layer. Finally, water balance closure (WBC) was calculated as the difference between Pi and (ET + ΔSWC).
Net ecosystem exchange of carbon
To understand the response of forest processes to environ-mental changes observed in the studied year, we separated the NEE into gross primary production (GPP) and ecosystem respiration (RE) following Kowalski et al. (2003). Thus, NEE was partitioned into GPP and RE and modelled using a rectan-gular hyperbolic response to incident PAR as follows:
NEE
PARPAR ref
Ts= − + + ⋅ −aa
Q R1
210
15 10( / )
(3)
where Rref refers to the ecosystem respiration for a reference surface temperature of 15 °C, Q10 describes the respiration sensitivity to the surface temperature Ts, and a1 and a2 are parameters describing the maximum photosynthetic uptake and the light at half of the maximum photosynthetic rate, respectively. The mean surface temperature, Ts, was derived from the sensible heat flux and air temperature. Parameters Rref, Q10, a1 and a2 were fitted by a two-step non-linear regres-sion using a 15-day period moved by a 5-day increment. First, at each time step, Rref and Q10 were calculated using only night-time values of NEE (SW↓ < 20 W m−2) under sufficient turbu-lence (u* > 0.2 m s−1) during rain-free periods. We substituted surface temperature with soil temperature at 5 cm depth, but no difference was observed. Second, a1 and a2 values were calculated using the same conditions on u* and rainfall. Other approaches have been developed since, in which the depen-dence of GPP on vapour pressure deficit (VPD) is included (Reichstein et al. 2005, van Gorsel et al. 2009, Lasslop et al. 2010). However, we continued to use the earlier approach in order to maintain consistency with previous data from this study site. Note that our calculation implicitly ignores the puta-tive difference in respiration metabolism between night and day shown by Kok (1949).
Understorey biomass measurements
Understorey composition and biomass were measured destruc-tively on 20 plots of 1 m2 randomly distributed within a 30 m × 30 m area of each of the two study sites (Table 1). After cutting, plant material was separated according to spe-cies, dried at constant temperature (65 °C) and then weighed. Green and non-green components were separated to estimate the distribution of dry biomass weight between components.
The ratio of dry mass to projected area was estimated from sub-samples of green material by species and the LAI was calculated using green biomass values and the corresponding dry mass to area ratio.
Results
Meteorological conditions
The period extending from June 2009 to May 2010 had a mean temperature of 12.41 ± 0.06 °C and a total precipita-tion of 933 ± 9 mm, close to the 1950–2000 averages for this region (Table 1). At the measurement height (6.6 m for plot C and 6.7 m for plot W), air temperature Ta, wind speed U and VPD were nearly the same for both plots (Figure 1). Unusual climatic conditions occurred during several periods of the year. Low temperatures, below −5 °C, were recorded on several days in December 2009 and January 2010. Also, the first half of October 2009 was unusually cold with daily mean temperature values around 5 °C (Figure 1). A warm summer period occurred with daily maximum temperatures reaching over >30 °C at the end of June and during July. In August, the daily maximum temperatures reached >35 °C, coinciding with a first period of soil drought extending from the end of August to September. April 2010 was particularly warm. Exceptionally high precipitation occurred in November 2009 (223 mm compared with an average rainfall of 78 mm for the 2002–08 period). From February 2010, the monthly precipitation was 35% lower than the 2002–08 mean values (data from a Météo-France station at Biscarosse, France), so that from the beginning of March 2010, the two sites experi-enced a second drought period, which is rather unusual in this area.
Energy balance
Annually, plot W absorbed a marginally lower (4%) net radia-tion than plot C (Figure 2a) with a total net energy of 2880 MJ m−2, compared with 2942 MJ m−2 in plot C. This C–W difference varied seasonally, taking positive values in summer and negative values in winter. Since both canopies received the same incident short-wave and long-wave radiation, the differ-ence in net radiation was explained by differences in albedo and upward long-wave radiation (LW↑) (Figure 3). The albedo of plot W varied between 0.10 and 0.35 during the study year, while it varied from 0.08 to 0.15 on plot C, with average values of 0.19 and 0.12, respectively (Figure 3a). The greater sea-sonal variation in albedo on plot W was due to the regrowth of the weed layer, which progressively covered the soil surface. Its phenological development also induced changes in surface reflectance properties. In contrast, the canopy albedo of plot C evolved mainly owing to the changing pigmentation of the gorse layer, which turned green in summer, yellow in fall and dark brown in winter. In addition, the canopy wetness after
Tree Physiology Online at http://www.treephys.oxfordjournals.org
rainfall may also explain part of the daily change in the differ-ence in albedo between plots, since the canopy of plot C retained a greater amount of rainfall than plot W (see the results on water balance below).
Remarkably, the annual difference in albedo between plots was largely offset by a compensating difference in the upward radiation flux density, LW ↑ , which also had some temporal variation according to the season (Figure 3b). The largest dif-
ferences in Rn occurred during summer 2009 and spring 2010, reaching a difference of 6 and 8%, respectively (Figure 2b and d). These differences are explained by the larger LW↑ from plot W. The difference in Rn decreased in autumn, and was fol-lowed by an opposite difference in January 2010 (Figure 2c). During these periods, LW↑ was higher on plot C than on plot W (Figure 3b), which might have been due to a higher surface temperature and a change in the surface emissivity of plot C.
Paired comparison of water, energy and carbon exchanges 7
Figure 1. Time-course of the daily values of maximum global radiation (SW↓), mean air temperature (Ta), mean wind speed (U) and mean VPD at plot C (grey lines) and plot W (black lines). Rainfall and cumulative rainfall are also shown.
Figure 2. Comparison of the net radiation (Rn) between the two plots. Yearly (a) and monthly (b, c, d) data are shown. The red line indicates the linear regression.
Figure 3. Comparison of (a) daily albedo and (b) daily long-wave radiation emitted from the surface (LW↑) between the two plots. LW↑ is shown for spring/summer (grey) and fall/winter (black).
Tree Physiology Online at http://www.treephys.oxfordjournals.org
Energy balance closure
The sum of the uncorrected turbulent fluxes Hu and LEu con-stituted 78 and 85% of the available energy (Rn minus the soil heat flux G) on plots W and C, respectively (Figure 4a and b). The intercept values were not significant. Plots of half-hourly EBR as a function of friction velocity u* (Figure 4c and d) showed that EBR increased with friction velocity, consis-tent with the results of Wilson et al. (2002). We observed that EBR was <0.60 when u* was <0.20 m s−1, and increased as u* increased. Therefore following Falge et al. (2001), we rejected and gap-filled turbulent flux values for u* values <0.20 m s−1.
Partitioning the energy fluxes into H and LE
For the peak of the growing season (June 2009, Figure 5a), most of the available energy was dissipated into latent heat flux LE on plot C, while on plot W the sensible heat flux H was slightly higher than the latent heat flux. Hence, the mean diur-nal ratio of sensible heat flux to latent heat flux, the Bowen ratio β, was 0.50 on plot C compared with 1.55 on plot W. Daytime LE was twice as high in the control plot as in the thinned and weeded plot. Conversely, H was 50% higher on
plot W than on plot C. Soil heat fluxes were similar between sites despite the difference in soil cover.
During the dry period (end of August, beginning of September 2009; Figure 5b), the available energy and its components were similar in the two plots, with Rn mostly dissipated into sensible heat, so that the diurnal β value increased consider-ably in both plots, reaching 2.21 on plot C and 2.01 on plot W. Midday values of LE were reduced by a factor of 2 in plot W and by a factor of 4 in plot C compared with the values of LE in the previous period, showing that the latent heat flux of plot C was severely affected by the drought.
No difference in Rn was observed during winter (January 2010; Figure 5c). The dissipation of the available energy was characterized by similar proportions of H and LE, with the β value close to 1. During cold weather, the soil heat flux was negative in the two plots, indicating that the soil was a source of energy for the atmosphere.
Water balance: annual and seasonal differences in the stand water budget
Annually, the partitioning of ET among stand layers was modified by both weeding and thinning (Table 3). Rainfall interception I,
Paired comparison of water, energy and carbon exchanges 9
Figure 4. (a, b) Relationships between H + LE and net available energy Rn − G and between the energy balance ratio (H + LE)/(Rn − G) and friction velocity u* (c, d) for plot W (a, c) and plot C (b, d). The red line is the regression line.
Figure 5. Mean daily pattern of energy fluxes Rn, H, LE and G for the three periods: (a) growing season, (b) drought and (c) winter. Fluxes on plot W are represented by solid lines and fluxes on plot C are represented by dotted lines. Rn (black line), LE (blue line), H (red line) and G (green line) are shown. The mean values of the Bowen ratio (β) are also given for each plot.
Table 3. Components of the water balance on plot C (a) and on plot W (b). Seasonal and annual values are given. The values shown in grey for plot C correspond to adjusted data from plot W, since June and July ΔSWC data were missing.
Pi (mm) ET (mm) I (mm) Tp (mm) Eh (mm) ΔSWC (mm) WBC (mm)
(a) Plot CGrowing season June 2009 82.6 132.3 14.0 34.2 84.0 −61.8 12.1 July 55 113.4 9.4 30.9 73.1 −18.9 −39.5 August 75.8 78.8 12.9 23.4 42.5 −2.5 −0.5Drought September 82.0 38.7 13.9 20.1 4.7 10.5 32.8 October 67.2 53.6 11.4 20.1 22.1 10.2 3.4Winter (dormant season) November 221.8 69.9 37.7 15.9 16.3 55.2 96.7 December 85.2 28.7 14.5 9.8 4.4 60.6 −4.1 January 2010 92.4 29.5 15.7 14.3 0.0 17.8 45.1 February 46.8 30.9 8.0 17.1 5.8 −34.6 50.5Growing season March 54.5 47.8 9.3 20.5 18.0 −36.9 43.6 April 23.3 65.6 4.0 25.4 36.2 −21.4 −20.9 May 39.6 80.6 6.7 25.3 48.5 −45.4 4.4Annual 926.2 769.6 157.4 257.2 355.6 −67.2 223.7(b) Plot WGrowing season June 2009 81.2 88.6 8.4 20.0 60.1 −65.7 58.3 July 58.1 101.9 6.0 21.1 74.8 −20.1 −23.7 August 69 72.9 7.2 17.3 48.5 −1.9 −2.0Drought September 84 41.3 8.7 12.3 20.2 13.8 28.9 October 68.5 53.4 7.1 13.1 33.2 14.7 0.4Winter (dormant season) November 223.1 51.5 23.2 11.9 16.4 112.8 58.8 December 90.9 25.8 9.5 9.6 6.7 22 43.1 January 2010 91.4 24.0 9.5 8.8 5.8 17.5 49.9 February 53 33.5 5.5 8.5 19.5 −40 59.5Growing season March 55.4 40.9 5.8 10.3 24.9 −33.7 48.2 April 24 58.8 2.5 12.6 43.6 −65.1 30.3 May 40.4 79.6 4.2 15.3 60.1 −24.8 −14.4Annual 939.0 672.3 97.7 160.8 413.9 −70.5 337.2
Tree Physiology Online at http://www.treephys.oxfordjournals.org
pine transpiration Tp and evaporation from the soil + weed layer Eh accounted for 14, 24 and 62% on plot W of the annual ET respectively, so that soil evaporation and weed layer transpira-tion were considerably larger than pine transpiration. On plot C, this partitioning was quite balanced between pine contribution (I + Tp = 54%) and weed layer contribution (46%).
The annual total ET was 15% higher in plot C. The ET differ-ence between the plots resulted mainly from a larger ET from the pine trees on plot C. This difference was partly offset by an opposite difference in ET from the soil + regrowing weed veg-etation on plot W. Indeed, I and Tp were 61 and 60% higher in the control plot, respectively. This difference was roughly simi-lar to the difference in pine LAI between the two plots: the LAI of plot W was 50% lower in January 2010. Conversely, its Eh was higher by 58.3 mm year−1. The residual term of the WBC was assigned to belowground drainage; surface runoff was largely absent because of the high soil permeability and hori-zontal topography. Annual belowground drainage was 50.7% (+113.5 mm year−1) higher in the weeded plot.
The difference between the plots in ET and its partitioning between layers changed according to the seasons, in response to summer drought and to the regrowth of herbaceous plants in the weeded plot. In particular, we found that:
(1) During the peak of the 2009 growing season (JJA), the dif-ference in ET was highest and caused mainly by higher pine transpiration in plot C.
(2) During the drought of September 2009, there was a severe reduction in ET, which equalled 71 and 53% in plots C and W, respectively. Despite a higher SWC (0–80 cm) in plot W (+30 mm, Figure 6), it is noteworthy that pine transpiration was reduced similarly and simulta-neously in both plots. Consistently, in a second period of severe drought (July 2010, data not shown), fluctuations in tree diameter were remarkably synchronous across the treatments, supporting the hypothesis that the trees were exposed to a simultaneous water stress in both plots:
growth on both sites stopped simultaneously in July 2010 (Figure 7). Conversely, in September 2009, Eh diverged dramatically between the two plots: it was reduced almost to zero in the control plot but maintained at a substantial rate, 20.2 mm month−1, in the weeded plot. Hence, Eh on plot C was 40% higher in June than on plot W, but 77% lower in September, probably caused by gorse stomatal closure.
(3) During the 2009–10 winter (October to February), ET from plot C exceeded that from plot W by 13%. This differ-ence was associated with Tp and I being higher than on plot W by ~55%, whereas Eh was halved. In wintertime, Tp and Eh on plot C contributed 36 and 23% of ET, respec-tively, whereas on plot W it was 28 and 43%, respectively. Simultaneously, the change in SWC on plot W was only 16% higher than on plot C, despite the variability between the months.
(4) Lastly, during spring 2010 (MAM), the difference in ET between plots was reduced to 8% but the difference in transpiration of pines between sites grew and conversely the deviation in Eh was lower. This period was character-ized by an early drought with SWC at 0–80 cm decreas-ing from mid-April on both sites (compared with mid-May in 2009, data not shown), in response to a 77 (60) mm imbalance between ET and Pi in plot C (W). However, contrasting with the pattern shown during September 2009, we did not find any evidence of a reduction in Eh at either site. Despite its lower ET, the soil water was depleted more severely in the weeded plot (Figure 6) because of both larger Eh and belowground drainage compared with plot C.
Carbon exchanges
Annually, the two plots were net carbon sinks, but plot W stored 73% less carbon than plot C (NEE in Table 4). This dif-ference was mostly attributed to the difference in the GPP
Paired comparison of water, energy and carbon exchanges 11
Figure 6. Evolution of the SWC of plot W (black line) and plot C (grey line) for the 0–80 cm layer.
between the two plots, with a 53% higher carbon uptake on plot C. The ecosystem respiration RE on plot C was 40% higher than on plot W.
In the two plots, GPP and RE exhibited a seasonal variability during 2009–10, with significant range for GPP (Figure 8, Table 4). Canopy photosynthesis was maximal in June. While GPP on plot C started to decrease at the beginning of July, the onset of the decrease in GPP occurred later in plot W (Figure 8). Both GPP and RE were severely affected by the September soil drought. This pattern in GPP paralleled the evolution of ET dur-ing the same period on the two plots (Tables 3 and 4), support-ing the hypothesis of canopy-scale stomatal closure. Plot C even became a carbon source in September. During winter, GPP and RE also showed dramatic reductions during cold events.
The partitioning of NEE into GPP and RE (Eq. (3)) allows us to investigate the between-site differences in the asymptotic
maximal GPP, a1; light-use efficiency (LUE), a1/a2; and night respiration, RE(PAR=0). These differences are illustrated by the response of NEE to available light at two critical periods: the peak of the growing season and the dry period of September (Figure 9). As expected, the light-saturated NEE value was three times higher in plot C than in plot W, with respective val-ues of −22 and −7 µmol m−2 s−1 for the growing season. This difference almost vanished in the dry period where NEE at light saturation showed similar values of 8.4 and 6.1 µmol m−2 s−1 on plots C and W, respectively. The carbon released during the night-time was also higher in plot C (RE(PAR=0) = 5.7 µmol m−2 s−1) than in plot W (RE(PAR=0) = 4.3 µmol m−2 s−1) during the grow-ing season but was similar at the end of the drought (RE(PAR=0) = 3.8 µmol m−2 s−1).
The derivative of the first term in Eq. (3) at PAR = 0, ~a1/a2, interpreted as the apparent ecosystem LUE (µ µmol molCO
1PAR2( ) ( )
− ),
12 Moreaux et al.
Table 4. Annual and seasonal values of carbon exchange on the two plots.
Plot C Plot W
NEE (g C m−2) GPP (g C m−2) RE (g C m−2) NEE (g C m−2) GPP (g C m−2) RE (g C m−2)
Growing season June 2009 −74.6 283.3 208.6 −2.4 133.8 131.4 July −64.8 277.8 213.0 −18.4 166.6 148.1 August −37.8 175.4 137.6 −10.9 134.3 123.4Drought September 13.1 75.9 89.0 −2.6 73.9 71.4 October −0.5 130.4 129.9 −8.5 96.6 88.1Winter (dormant season) November 55.8 84.4 140.2 31.9 64.5 96.4 December 20.8 57.4 78.2 13.0 40.3 53.3 January 2010 −2.5 58.0 55.4 −1.6 34.4 32.8 February −6.2 74.6 68.4 −2.4 40.1 37.7Growing season March −28.2 113.2 85.0 −10.2 71.9 61.7 April −45.4 169.6 124.2 −15.7 112.1 96.5 May −73.0 219.5 146.5 −37.4 152.4 115.0Annual −243.3 1719.4 1476.1 −65.2 1121.0 1055.8
Figure 7. Mean stem diameter growth of six trees from April 2010 to July 2010. Plot W is shown in black and plot C in grey. The vertical grey lines indicate the period of the cessation of stem growth due to dry conditions.
Tree Physiology Online at http://www.treephys.oxfordjournals.org
was higher in plot C during the growing and dormant seasons (Figure 10). This trend seems to reverse in the dry period dur-ing which plot W became more light-use efficient than C, as the gorse layer on plot C started its senescence. However, given the standard deviations during these 2 months, this trend should be interpreted with caution.
Discussion
Energy balance closure
The gap in the energy balance obtained for the two plots is comparable to that found in other forest sites (Berbigier et al.
2001, Wilson et al. 2002, Foken 2008, Jarosz et al. 2008, Tanaka et al. 2008). A number of authors have proposed con-vergent explanations for the gap in energy balance closure (Massman and Lee 2002, Foken 2008, Kidston et al. 2010). In our study, two explanations are possible. First, the LEu + Hu term might have been underestimated, due to a possible loss of low- or high-frequency contributions to the turbulent fluxes, and/or to advection. Advection might have been more accentu-ated in plot W where the canopy was discontinuous, with a large free space between trees that can drive development of low- or high-frequency turbulences. Second, we neglected the heat storage in the upper soil and dead organic matter accu-mulated at the soil surface following thinning and weeding. This factor might have also led to a larger underestimate of the available energy in both plots.
Changes in radiative transfer
The annual available energy was in the same range on both plots, with small seasonal variations which contradict our first hypothesis. Several studies conducted on stands of different ages and structures (clear-cut versus old stands) or disturbed (thinned, burnt) versus undisturbed stands showed more sig-nificant effects on annual Rn (Gholz and Clark 2002, Kowalski et al. 2003, Sun et al. 2010). These studies also showed that Rn was reduced when the LAI was lower (clear-cut, burnt) and attributed this reduction to an increase in albedo. Here we found that changes in albedo alone could not explain the small difference in Rn that we observed, and that the difference in upward long-wave radiation was offsetting most of the albedo effect. We assume that surface temperature and probably sur-face emissivity are sensitive to changes in stand structure and canopy wetness, which differed between plots. Similarly, Amiro et al. (1999) conducted a paired comparison between an old forest surface and a 1-year-old burnt surface and showed that the lower albedo on the burnt area was compensated for by higher temperatures, yielding similar values of Rn on the two
Paired comparison of water, energy and carbon exchanges 13
Figure 8. Annual time-courses of ecosystem respiration (RE) and gross primary production (GPP). GPP values are presented here as negative values to express the uptake of carbon in contrast to the release of carbon by the vegetation in the RE term. Plot W is represented in black and plot C in grey.
Figure 9. Response of net ecosystem exchange, NEE, to incident light, PAR, for two critical periods: (a) growing season (June) and (b) dry period (September). Symbols show measured values. Modelled NEE is shown as a polynomial adjustment of the model (lines, R2 > 0.85 in each case). Plot W is represented in black and plot C in grey.
surfaces. In south-western France, Jarosz et al. (2009) showed a larger difference in Rn during May and June, mainly lower val-ues in a clear-cut stand versus winter crops and in a clear-cut stand versus mature stands of maritime pines. These differ-ences reached 20 and 35%, respectively, and were explained mainly by a higher emission of long-wave radiation LW↑ in the clear-cut stand. During the same period our values were lower and ranged from 6 to 15%. Contrary to the findings of Sun et al. (2010), the difference we observed in Rn between plots was larger during the growing season than during winter. This is explained by a compensatory effect between albedo and LW↑ in the dormant season and a reinforcing effect of the two terms in the growing season: in spring and summer, both albedo and upward long-wave radiation were lower in the control plot.
The specific pattern of energy partitioning on each plot and its seasonal change show how environmental drivers and can-opy structure and composition controlled this partitioning (Wilson and Baldocchi 2000). We found contrasting patterns in the energy partitioning between the two plots, with a higher Bowen ratio on the weeded plot. The difference in LAI is likely to be the main cause of this difference in Bowen ratio. The canopy of the control plot is a continuous mixed layer of pine crowns and gorse, with a low albedo and high stomatal sensi-tivity to soil water deficit. In contrast, the weeded plot had a large fraction of bare soil that was progressively covered by the regrowing vegetation and the expansion of pine crowns, which explains its lower sensitivity to soil water availability. We saw that during the dry period (end of August, beginning of September), the net radiation was preferentially dissipated into sensible heat on the two sites, and the transpiration of the canopy and understorey decreased as a result of water stress and presumably stomatal closure (see the section on water balance below). Consistently, Jarosz et al. (2008, 2009) observed a similar evolution in Bowen ratio of a mature mari-time pine stand during drought. Baldocchi et al. (2000) also showed that the Bowen ratio increased with drier environmen-tal conditions in a Pinus ponderosa stand. This pattern was also
found in a grassland and an adjacent Scots pine forest, where under low water supply, Rn was mostly converted into H (Rost and Mayer 2006).
The presence of the soil + weed layer appeared to play a major role in the partitioning of energy fluxes in plot C. Unfortunately, in our study, no measurement of LE and H on each layer allows us to differentiate the fluxes from the weed vegetation from those coming from the soil surface. Previous studies have shown that the understorey layer contribution is controlled by the overstorey LAI in mature stands with closed canopies. For example, Baldocchi et al. (1997) found that the understorey contributed between 20 and 40% of the total energy exchange in a jack pine forest. Jarosz et al. (2008) assessed this contribution to be 32 and 38% for the sensible and latent heat fluxes, respectively, in a mature maritime pine stand. In young, open stands with a vigorous weed layer of similar height to the trees, it should be expected that the weed layer would play a major role in the partitioning of the energy.
Placing our results in the perspective of the forest life cycle, the partitioning between sensible heat and latent heat fluxes appears to be controlled during the juvenile stage, first by the share of canopy LAI between the weed layer and tree canopy and secondly by the available soil water (see the next section). Then, following overstorey canopy closure, the available energy is mostly controlled by the overstorey LAI alone.
Water balance
As a major component of the water balance in both plots, ET on the weeded plot was only reduced by 15% compared with the control plot. In the context of the Landes forest, little litera-ture is available on water exchanges comparing adjacent stands with contrasting canopy structures induced by forest manage-ment and evolving under an identical climate. However, other studies have been conducted in adjacent plots comparing dif-ferent types of ecosystem. In south-west Germany, Wicke and Bernhofer (1996) showed ET from 50 to 80% higher in a Scots pine forest compared with an adjacent grassland in a
14 Moreaux et al.
Figure 10. Annual time-course of the maximum light use efficiency a1/a2 (µ µmol molCO1
PAR2( ) ( )− ). Plot W is represented in black and plot C in grey.
Vertical lines refer to standard deviations for each plot.
Tree Physiology Online at http://www.treephys.oxfordjournals.org
sunny spring. Sun et al. (2010) also found a 16–40% higher value of ET in a 16-year-old loblolly pine plantation compared with a nearby 4-year-old plantation.
Intuitively, the lower values of ET on plot W were associated with its lower Tp due to a reduction in both basal area and LAI. However, the reduction in leaf area also allowed greater amounts of energy and water to reach the soil surface (Figure 6), allow-ing higher surface evaporation just after weeding and transpi-ration when the weed layer regrew. For example, in November 2009, the herbaceous layer was totally removed on plot W, which could explain the higher contribution of Eh on plot W than on plot C, by the presence of a large litter layer, which likely had a high moisture-holding content. The reduction in leaf area also led to an additional annual loss of water by deep drainage (Table 3). Therefore, soil surface evaporation might have increased, so that Eh on plot W was higher than that on plot C, mostly gorse transpiration. Indeed, a lower pine LAI increased radiation load to the ground, inducing higher soil surface temperature and therefore soil evaporation, particularly in summer. This response is in accordance with Simonin et al. (2007), who found that soil evaporation was greater in a thinned stand of ponderosa pines. Thus, our second hypothe-sis, that lower total leaf area would enhance resource availabil-ity (i.e. soil water) for trees in the weeded plot, and thus mitigate the impacts of summer water deficit, was not con-firmed. To insist on that result, despite the weeding and thin-ning, the pines in both plots seem to have responded similarly to drought conditions, contrary to what we would have expected. The parallel decrease in pine transpiration during August and September 2009 was surprising when considering the stronger decrease in ET on plot C than on plot W. Contrary to expectations, the weeding and thinning did not enhance transpiration per unit of needle area in plot W by increasing light availability and exposure of needles to wind and high VPD. Moreover, stem diameter growth showed a parallel decrease during the 2010 drought. For these reasons, we suggest that the entire soil volume may not have been fully explored by the pine root systems following thinning and weeding.
Our results clearly demonstrate that the pine and weed lay-ers exhibited differential sensitivity to soil conditions and par-ticularly to soil drought, which impacted annual and seasonal partitioning of evaporation from the total canopy. Partitioning of evaporation between overstorey and understorey or surface has already been studied in a number of mature stands. Law et al. (2000) measured the contribution of the surface to total ET in a mature P. ponderosa stand. Soil evaporation was 44% of the total evaporation at the beginning of the growing season, decreasing to 33% during summer. Wilson et al. (2000) also reported that soil evaporation can account for 30–50% of total ET in temperate pine forests, particularly in spring and summer. Our values were higher because of higher throughfall in a young and open stand compared with a mature and closed
canopy. They reached 68 and 66% of ET on plot W in June and August, respectively. On plot C, those values were 64 and 54% of ET. Moreover, as soil evaporation and weed layer transpira-tion could not be separated in terms of contribution, these per-centages included both layers’ contributions. Herbst et al. (2008) reported understorey transpiration values representing 9 and 18% of the total ET in two broad-leaved woodlands, with a sparsely developed understorey and a well-developed under-storey, respectively. Iida et al. (2009) also noticed the impor-tance of the understorey in a larch forest where its contribution was 51% of the total ET. In the same region as our study, Loustau and Cochard (1991) assessed a non-negligible contri-bution of the understorey, which represented annually 20% of the total evaporation in a 19-year-old stand. Finally, in the same stand, Berbigier et al. (1991) emphasized the fact that this understorey was not affected by drought, taking advantage of the rain events in the summer, contrary to the adult pines. Indeed, the contribution of the understorey represented 45% of the total evaporation in the growing season and 54% during the drought, while transpiration of the pines represented 45% and then 12% of the total evaporation, in response to stomatal closure for the trees. Moreover, Simonin et al. (2007) found that the herbaceous layer contributed 92 and 75% of the total ET after extreme drought in thinned and unthinned plots, respectively. During the drought event, transpiration of our gorse layer and soil evaporation was clearly reduced and con-tributed 12% of ET on the control plot. These contrasting results show the importance of considering the nature of the weed layer but also the structure of the stand. In other studies this has comprised small understorey plants growing below mature trees, for example M. coerulea M. growing beneath mature P. pinaster (Berbigier et al. 1991), and Festuca arizonica and Elymus alymoides growing beneath mature P. ponderosa (Simonin et al. 2007). In our case, the weedy species domi-nated by gorse appear to be more sensitive to drought than the pines in this stand configuration. The drought sensitivity we observed in Eh on plot C might be explained by stomatal clo-sure, followed by leaf mortality.
Carbon exchanges
The weeding and thinning in plot W reduced its annual carbon uptake by up to 73%. Despite the reduction of 65% of the total LAI, the weeded plot remained a carbon sink annually. We sup-posed that the rapid regrowth of the herbaceous layer in plot W contributed to this pattern. This suggestion could explain the time lag in the summer decrease in ET and GPP between the two plots, so that this rapid regrowth compensated for a pos-sible decrease in GPP of the pines, sustaining constant values of GPP in July. This explanation was supported by biomass analysis of the weed layer, which showed that its biomass growth was much faster in plot W than in plot C: the biomass of ground vegetation rose from 136 ± 46 to 282 ± 99 g C m−2
Paired comparison of water, energy and carbon exchanges 15
in plot W, a net annual gain of 146 g C m−2, whereas it did not vary significantly in plot C from 1034 ± 232 to 906 ± 214 g C m−2, possibly because of increased shading by tree crowns and a high proportion of leaf mortality after the 2009 drought. As more rainfall and light reached the soil of the open canopy, the water use of the weed layer could increase. This observa-tion is consistent with the results from Moore et al. (2006) on ponderosa pine forests, where the thinning of the trees stimu-lated the rapid regrowth of the understorey.
From the successive studies at this site by Kowalski et al. (2003, 2004) and Stella et al. (2009) and data provided for mature sites by Berbigier et al. (2001) and Jarosz et al. (2009), we can conclude that the regrowth of vegetation (weeds + trees) has turned a clear-cut site into a carbon sink after only 3 years.
The annual values of NEE and its different components on plot C were comparable to the results of Stella et al. (2009) for plot W in 2007 before it was thinned and weeded, at which time it had a similar structure and composition to plot C, includ-ing a well-developed gorse layer. In 2007, NEE was −335 g C m−2 and RE reached 1650 g C m−2. These values are slightly higher than our observations, but were obtained during a wet-ter year, with more favourable soil conditions, particularly dur-ing the spring and summer of 2007 compared with summer 2009 and spring 2010. The annual carbon balance of plot W was also studied at an earlier stage, immediately after harvest-ing, by Kowalski et al. (2003). At that time, the site was a net source of carbon with an annual NEE value of 276 g C m−2 with a GPP of 602 g C m−2 and an RE of 878 g C m−2. Misson et al. (2005) also studied the effect of thinning on carbon balance in a young ponderosa pine plantation. In their study, ecosystem respiration was largely unaffected by thinning, with an increase of only 1%, in contrast to canopy photosynthesis which decreased by 14%, so that their site became a net source of carbon the year following thinning. Recently, Dore et al. (2010) found that thinning of an adult ponderosa pine stand reduced the ecosystem carbon uptake by 30% compared with an undis-turbed stand. The change in LAI is the main explanation for the decrease in carbon uptake observed in each study. In our study, in comparison, the total LAI was reduced by 65%, including the removal of the weed layer, against a reduction of 35% in Dore et al. (2010), which accounts for the larger decrease in carbon uptake that we observed. One limitation of our study is the fact that GPP could not be partitioned between the pines and the weed layer, so that the contribution of the gorse layer on plot C and the regrowing vegetation on plot W during summertime could not be assessed. However, given the strong parallel in the dynamic of ET and GPP during the grow-ing and dry periods in both plots, we can infer that the dynamic of GPP was also driven by the dynamic of the gorse layer. Therefore, it seems that the drought strongly impacted the gorse layer. This conclusion is consistent with observations at
the same site directly after harvesting (Jarosz et al. 2009), which showed that the weed layer was as sensitive as the pines to summer drought.
The apparent LUE values obtained in summer had a mean value of 0.041 and 0 0. ( ) ( )38 mol molCO
1PAR2
µ µ − for plot W and C, respectively. These values are lower than the mean value of 0.064 found for mature pine stands in Finland (Pinus sylvestris and Grasses/dwarf shrubs), Britain (Picea sitchensis and Graminae) and France (P. pinaster and Graminae/gorse), but within the range of the values found in clear-cuts in the same countries: from 0.028 to 0.057 (Kowalski et al. 2004). We saw that the weed layer was more sensitive to dry conditions. Therefore, a hypothesis to understand the inversion observed during drought is that the gorse layer would become less light efficient under those conditions compared with the pines. As a dominant species on plot C compared with plot W, it would tend to decrease the ratio a1/a2.
Overall, weeding of the gorse layer associated with thinning of the pines seemed to have a beneficial effect on tree growth (Figure 7), in accordance with Juodvalkis et al. (2005). They showed that, in several young stands including pine species, thinning resulted in increased diameter at breast height (DBH) growth. The 1-year difference in age between our two plots was not considered to explain the differences observed in diameter growth. Indeed, physical properties of wood and tree structure are thought not to vary within this stage of growth, and photosynthetic parameters have been shown not to vary in a chronosequence of maritime pine stands in the Landes forest (Delzon and Loustau 2005). Therefore, in our study, the age difference is implicitly taken into account through the differ-ences in LAI and basal area of the stand.
Conclusion
In the context of climate change, important roles of forests are in carbon sequestration and intensive production of bio-energy. Management options need to be designed to maximize these roles. Management options available for pine stands in the south-west region of France include thinning, weeding during the early stage of growth and intersowing with coexisting N-fixing species. We examined the consequences of these management options for carbon, water and energy exchanges on two adjacent young maritime pine stands representative of both stages (before and after thinning and weeding) using a paired measurement system, including soil, plant and atmo-spheric measurements.
We first showed that at the annual scale, a small difference was observed in the net radiation between the two plots, which was explained by opposing effects on albedo and long-wave radiation emitted by the surface and their respective dynamics through the study year. The components of the energy balance showed contrasting trends at the seasonal scale. Latent heat
Tree Physiology Online at http://www.treephys.oxfordjournals.org
flux and GPP were larger during the growing season on the control plot, which could be attributed to a higher LAI on that plot. This difference was severely reduced, with the same dynamics for LE and GPP, in the drought event, during which flux values of the control plot reached those observed on the weeded plot. Plot W also underwent a decrease in LE and GPP. This decrease was probably linked to stomatal closure on both plots. Observations of tree transpiration suggest that the extreme decrease in LE and GPP on plot C was probably due to the effects of drought on gorse rather than on pines. It sug-gests that the pines on both plots had similar responses to drought, even though the stocking densities differed. This behaviour was also observed in tree growth: the relative growth in stem diameter stopped simultaneously on the two plots dur-ing the drought event. We attributed this pattern to the fact that trees on plot W did not take advantage of the higher soil water availability due to removal of the vegetation layer, potentially because the pine roots had not fully explored the soil.
The reduction in total LAI due to the removal of the N-fixing gorse layer and thinning did not cause the weeded plot to become a source of carbon as we could have expected. Both plots were carbon sinks at annual scale and during summer. However, both of them were small sources in winter. Despite the small difference in annual total ET, net carbon sequestra-tion was lower by 73% in the disturbed plot. These observa-tions can contribute to an understanding of the operational advantage of controlling the gorse layer during the early stage of maritime pine stands, although in this study we did not con-sider N input benefits. Watt et al. (2003) showed for broom growing with young Pinus radiata that the N released upon weeding could enhance tree growth in stands deficient in N. However, it is known that in dry conditions, water stress can limit nutrient uptake by trees (Kreuzwieser and Gessler 2010). Augusto et al. (2009) and Kreuzwieser and Gessler (2010) both documented significant levels of N fixation by gorse (Ulex europeaus L.) and broom, respectively, which might increase tree growth in the long term, but such increases in pine growth would also depend on water availability to the pine root sys-tem. Future studies of management benefits of these options to growth of young stands for wood-energy need to consider these aspects.
Acknowledgments
The Tranzfor project also contributed to the partnership with Australia. We thank Pierre Trichet, Michel Sartore, Didier Garrigou, Marc Irvine and Christophe Chipeaux, as well as the experimental unit of INRA (Pierroton) for their contribution to the technical part of the experiments and field assistance. The site belongs to the commune of Salles and is managed by the Office National des Forêts who kindly provided the support and authorizations for our research study. We finally thank the
IUFRO workshop that favoured multiple discussions during the Sir Oliphant conference (7–15 October 2010, Victoria and Tasmania, South East Australia).
Funding
This work was supported by the Tuck foundation and the Enerbio funds.
References
Amiro, B.D., J.I. MacPherson and R.L. Desjardins. 1999. BOREAS flight measurements of forest-fire effects on carbon dioxide and energy fluxes. Agric. For. Meteorol. 96:199–208.
Aubinet, M., A. Grelle, A. Ibrom, et al. 2000. Estimates of the annual net carbon and water exchange of forests: the EUROFLUX methodology. Adv. Ecol. Res. 30:113–175.
Augusto, L., M.R. Bakker, C. De Lavaissière, L. Jordan-Meille and E. Saur. 2009. Estimation of nutrient content of woody plants using allometric relationships: quantifying the difference between concen-tration values from the literature and actuals. Forestry 82:463–477.
Baldocchi, D.D., C.A. Vogel and B. Hall. 1997. Seasonal variation of carbon dioxide exchange rates above and below a boreal jack pine forest. Agric. For. Meteorol. 83:147–170.
Baldocchi, D.D., B.E. Law and P. Anthoni. 2000. On measuring and modeling energy fluxes above the floor of a homogeneous and het-erogeneous conifer forest. Agric. For. Meteorol. 102:187–206.
Berbigier, P., A. Diawara and D. Loustau. 1991. Etude microclimatique de l’effet de la sècheresse sur l’évaporation d’une plantation de pins maritimes et du sous-bois. Ann. Sci. For. 22:157–177.
Berbigier, P., J.-M. Bonnefond and P. Mellmann. 2001. CO2 and water vapour fluxes for 2 years above Euroflux forest sites. Agric. For. Meteorol. 108:183–197.
Bosc, A., A. de Grandcourt and D. Loustau. 2003. Variability of stem and branch maintenance respiration in a Pinus pinaster tree. Tree Physiol. 23:227–236.
Chen, J., K.T. Paw U, S.L. Ustin, T.H. Suchanek, B.J. Bond, K.D. Brosofske and M. Falk. 2004. Net ecosystem exchanges of carbon, water, and energy in young and old-growth Douglas-fir forests. Ecosystems 7:534–544.
Ciais, P., M.J. Schelhaas, S. Zaehle, et al. 2008. Carbon accumulation in European forests. Nature Geosc. 1:425–429.
Clearwater, M.J., F.C. Meinzer, J.L. Andrade, G. Goldstein and N.M. Holbrook. 1999. Potential errors in measurement of nonuniform sap flow using heat dissipation probes. Tree Physiol. 19:681–687.
Delzon, S. and D. Loustau. 2005. Age-related decline in stand water use: sap flow and transpiration in a pine forest chronosequence. Agric. For. Meteorol. 129:105–119.
Delzon, S., A. Bosc, L. Cantet and D. Loustau. 2005. Variation of the photosynthetic capacity across a chronosequence of maritime pine correlates with needle phosphorus concentration. Ann. For. Sci. 62:537–543.
Dezi, S., B.E. Medlyn, G. Tonon and F. Magnani. 2010. The effect of nitrogen deposition on forest carbon sequestration: a model-based analysis. Glob. Change Biol. 16:1470–1486.
Dore, S., T.E. Kolb, M.C. Montes-Helu, et al. 2010. Carbon and water fluxes from ponderosa pine forests disturbed by wildfire and thin-ning. Ecol. Appl. 20:663–683.
Falge, E., D. Baldocchi, R. Olson, et al. 2001. Gap filling strategies for defensible annual sums of net ecosystem exchange. Agric. For. Meteorol. 107:43–69.
Paired comparison of water, energy and carbon exchanges 17
Foken, T. 2008. The energy balance closure problem: an overview. Ecol. Appl. 18:1352–1367.
Foken, T. and B. Wichura. 1996. Tools for quality assessment of sur-face based flux measurements. Agric. For. Meteorol. 78:83–105.
Gholz, H.L. and K.L. Clark. 2002. Energy exchange across a chronose-quence of slash pine forests in Florida. Agric. For. Meteorol. 112:87–102.
Granier, A. 1985. Une nouvelle méthode pour la mesure du flux de sève brute dans le tronc des arbres. Ann. Sci. for. 42:81–88.
Granier, A. 1987. Sap flow measurement in Douglas fir stems using a new thermal method. Ann. For. Sci. 44:1–14.
Granier, A., M. Reichstein, N. Bréda, et al. 2007. Evidence for soil water control on carbon and water dynamics in European forests during the extremely dry year: 2003. Agric. For. Meteorol. 143:123–145.
Herbst, M., P.T.W. Rosier, M.D. Morecroft and D.J. Gowing. 2008. Comparative measurements of transpiration and canopy conduc-tance in two mixed deciduous woodlands differing in structure and species composition. Tree Physiol. 28:959–970.
Iida, S., T. Ohta, K. Matsumoto, et al. 2009. Evapotranspiration from understory vegetation in an eastern Siberian boreal larch forest. Agric. For. Meteorol. 149:1129–1139.
Janssens, I.A., H. Lankreijer, G. Matteucci, et al. 2001. Productivity overshadows temperature in determining soil and ecosystem respi-ration across European forests. Glob. Change Biol. 7:269–278.
Jarosz, N., Y. Brunet, E. Lamaud, M. Irvine, J.M. Bonnefond and D. Loustau. 2008. Carbon dioxide and energy flux partitioning between the understorey and the overstorey of a maritime pine for-est during a year with reduced soil water availability. Agric. For. Meteorol. 148:1508–1523.
Jarosz, N., P. Béziat, J.M. Bonnefond, Y. Brunet, J.C. Calvet, E. Ceschia, J.A. Elbers, R.W.A. Hutjes and O. Traullé. 2009. Effect of land use on carbon dioxide, water vapour and energy exchange over terrestrial ecosystems in Southwestern France during the CERES campaign. Biogeosci. Discuss. 6:2755–2784.
Juodvalkis, A., L. Kairiukstis and R. Vasiliauskas. 2005. Effects of thin-ning on growth of six tree species in north-temperate forests of Lithuania. Eur. J. For. Res. 124:187–192.
Kaimal, J.C. and J.J. Finnigan. 1994. Atmospheric boundary layer flows, their structure and measurements. Oxford University Press, UK.
Kidston, J., C. Brümmer, T.A. Black, K. Morgenstern, Z. Nesic, J.H. McCaughey and A.G. Barr. 2010. Energy balance closure using eddy covariance above two different land surfaces and implications for CO2 flux measurements. Bound.-Lay. Meteorol. 136:193–218.
Kljun, N., P. Calanca, M.W. Rotach and H.P. Schmid. 2004. A simple parameterisation for flux footprint predictions. Bound.-Lay. Meteorol. 112:503–523.
Kok, B. 1949. On the interrelation of respiration and photosynthesis in green plants. Biochim. Biophys. Acta 3:625–631.
Kowalski, A.S., M. Sartore, R. Burlett, P. Berbigier and D. Loustau. 2003. The annual carbon budget of a French pine forest (Pinus pinaster) following harvest. Glob. Change Biol. 9:1051–1065.
Kowalski, A.S., D. Loustau, P. Berbigier, et al. 2004. Paired compari-sons of carbon exchange between undisturbed and regenerating stands in four managed forests in Europe. Glob. Change Biol. 10:1707–1723.
Kreuzwieser, J. and A. Gessler. 2010. Global climate change and tree nutrition: influence of water availability. Tree Physiol. 30:1221–1234.
Lamaud, E., J. Ogée, Y. Brunet and P. Berbigier. 2001. Validation of eddy flux measurements above the understorey of a pine forest. Agric. For. Meteorol. 106:187–203.
Lasslop, G., M. Reichstein, D. Papale, A.D. Richardson, A. Arneth, A. Barr, P. Stoy and G. Wohlfahrt. 2010. Separation of net ecosystem exchange into assimilation and respiration using a light response
curve approach: critical issues and global evaluation. Glob. Change Biol. 16:187–208.
Law, B.E., M. Williams, P. Anthoni, D.D. Baldocchi and M.H. Unsworth. 2000. Measuring and modeling seasonal variation of carbon dioxide and water vapor exchange of a Pinus ponderosa forest subject to soil water deficit. Glob. Change Biol. 6:613–630.
Lesgourgues Y., D. Merzeau, L. Crémière and V. Bailleres. 1997. Conduite des boisements de Pin maritime sur le plateau landais: iti-néraires techniques—pistes pour le futur. Conférence: De la forêt cultivée à l’indstrie de demain. De la gestion au développement durable, Bordeaux, 20–21 Novembre 1997, pp 207-222.
Loustau, D. 2010. Forests, carbon cycle and climate change, QUAE, Versailles, 350 pp.
Loustau, D. and H. Cochard. 1991. Utilisation d’une chambre de tran-spiration portable pour l’estimation de l’évapotranspiration du sous-bois de pin maritime. Ann. Sci. For. 48:29–45.
Loustau, D., A. Bosc, A. Colin, et al. 2005. Modeling climate change effects on the potential production of French plains forests at the sub-regional level. Tree Physiol. 25:813–823.
Luyssaert, S., P. Ciais, S.L. Piao, et al. 2010. The European carbon bal-ance. Part 3: forests. Glob. Change Biol. 16:1429–1450.
Massman, W.J. and X. Lee. 2002. Eddy covariance flux corrections and uncertainties in long-term studies of carbon and energy exchanges. Agric. For. Meteorol. 113:121–144.
McCaughey, J.L. 1985. Energy balance storage terms in a mature mixed forest at Petawawa, Ontario—a case study. Bound.-Lay. Meteorol. 31:89–101.
Medlyn, B.E., D. Loustau and S. Delzon. 2002. Temperature response of parameters of a biochemically based model of photosynthesis. I. Seasonal changes in mature maritime pine (Pinus pinaster Ait.). Plant, Cell Environ. 25:1155–1165.
Misson, L., J. Tang, M. Xu, M. McKay and A. Goldstein. 2005. Influences of recovery from clear-cut, climate variability, and thinning on the carbon balance of a young ponderosa pine plantation. Agric. For. Meteorol. 130:207–222.
Misson, L., D.D. Baldocchi, T.A. Black, et al. 2007. Partitioning forest carbon fluxes with overstory and understory eddy-covariance mea-surements: a synthesis based on FLUXNET data. Agric. For. Meteorol. 144:14–31.
Moore, C.J. 1986. Frequency response corrections for Eddy Correlation systems. Bound. -Lay. Meteorol. 37:17–35.
Moore, M.M., C.A. Casey, J.D. Bakker, J.D. Springer, P.Z. Fulé, W. Wallace Covington and D.C. Laughlin. 2006. Herbaceous vegetation responses (1992–2004) to restoration Treatments in a ponderosa pine forest. Rangeland Ecol. Manage. 59:135–144.
Nabuurs, G.-J., M.J. Schelhaas, G.M.J. Mohren and C.B. Field. 2003. Temporal evolution of the European forest sector carbon sink from 1950 to 1999. Glob. Change Biol. 9:152–160.
Nabuurs, G.J., E. Thurig, N. Heidema, et al. 2008. Hotspots of the car-bon cycle in European forests. For. Ecol. Manage. 256:194–200.
Ogée, J., E. Lamaud, Y. Brunet, P. Berbigier and J.M. Bonnefond. 2001. A long-term study of soil heat flux under a forest canopy. Agric. For. Meteorol. 106:173–186.
Reichstein, M., E. Falge, D. Baldocchi, et al. 2005. On the separation of net ecosystem exchange into assimilation and ecosystem respiration: review and improved algorithm. Glob. Change Biol. 11:1424–1439.
Rost, J. and H. Mayer. 2006. Comparative analysis of albedo and sur-face energy balance of grassland site and an adjacent Scots pine forest. Clim. Res. 30:227–237.
Shaiek, O., D. Loustau, P. Trichet, C. Méredieu, B. Bachtobji, S. Garchi and M.H. El Aouni. 2011. Generalized biomass equations for the main aboveground biomass components of maritime pine across contrasting environments. Ann. For. Sci. 68:443–452.
Tree Physiology Online at http://www.treephys.oxfordjournals.org
Simonin, K., T.E. Kolb, M.C. Montes-Helu and G.W. Koch. 2007. The influence of thinning on components of stand water balance in a ponderosa pine forest stand during and after extreme drought. Agri. For. Meteorol. 143:266–276.
Sohngen, B. 2008. Climate change, agriculture, forests, and biofuels. International Agricultural Trade Research Consortium. 7–9 December 2008, Scottsdale, Arizona.
Stella, P., E. Lamaud, Y. Brunet, J.-M. Bonnefond, D. Loustau and M. Irvine. 2009. Simultaneous measurements of CO2 and water exchanges over three agroecosystems in South-West France. Biogeosciences 6:2957–2971.
Sun, G., A. Noormets, M.J. Gavazzi, S.G. McNulty, J. Chen, J.C. Domec, J.S. King, D.M. Amatya and R.W. Skaggs. 2010. Energy and water bal-ance of two contrasting loblolly pine plantations on the lower coastal plain of North Carolina, USA. For. Ecol. Manage. 259:1299–1310.
Tanaka, H., T. Hiyama, N. Kobayashi, H. Yabuki, Y. Ishii, R.V. Desyatkin, T.C. Maximov and T. Ohta. 2008. Energy balance and its closure over a young larch forest in eastern Siberia. Agric. For. Meteorol. 148:1954–1967.
van Gorsel, E., N. Delpierre, R. Leuning, et al. 2009. Estimating noctur-nal ecosystem respiration from the vertical turbulent flux and change in storage of CO2. Agric. For. Meteorol. 149:1919–1930.
Watt, M.S., P.W. Clinton, D. Whitehead, B. Richardson, E.G. Mason and A.C. Leckie. 2003. Above-ground biomass accumulation and nitrogen fixation of broom (Cytisus scoparius L.) growing with juvenile Pinus radiata on a dryland site. For. Ecol. Manage. 184:93–104.
Webb, E.H., G.I. Pearman and R. Leuning. 1980. Correction of flux measurements for density effects due to heat and water transfer. Q. J. Roy. Meteorol. Soc. 106:85–100.
Wicke, W. and C. Bernhofer. 1996. Energy balance comparison of the Hartheim forest and an adjacent grassland site during the HartX experiment. Theor. Appl. Climatol. 53:49–58.
Wilson, K.B. and D.D. Baldocchi. 2000. Seasonal and interannual vari-ability of energy fluxes over a broadleaved temperate deciduous for-est in North America. Agric. For. Meteorol. 100:1–18.
Wilson, K.B., P. Hanson and D. Baldocchi. 2000. Factors controlling evaporation and energy partitioning beneath a decidous forest over an annual cycle. Agric. For. Meteorol. 102:83–103.
Wilson, K., A. Goldstein, E. Falge, et al. 2002. Energy balance closure at FLUXNET sites. Agric. For. Meteorol. 113:223–243.
Paired comparison of water, energy and carbon exchanges 19
