Seasonal variations in photosynthesis, intrinsic water-use efficiency and stable isotope composition of poplar leaves in a short-rotation plantation.

Photosynthetic carbon assimilation and transpirational water loss play an important role in the yield and the carbon sequestration potential of bioenergy-devoted cultures of fast-growing trees. For six poplar (Populus) genotypes in a short-rotation plantation, we observed significant seasonal and genotypic variation in photosynthetic parameters, intrinsic water-use efficiency (WUEi) and leaf stable isotope composition (δ13C and δ18O). The poplars maintained high photosynthetic rates (between 17.8 and 26.9 μmol m(-2) s(-1) depending on genotypes) until late in the season, in line with their fast-growth habit. Seasonal fluctuations were mainly explained by variations in soil water availability and by stomatal limitation upon photosynthesis. Stomatal rather than biochemical limitation was confirmed by the constant intrinsic photosynthetic capacity (Vcmax) during the growing season, closely related to leaf nitrogen (N) content. Intrinsic water-use efficiency scaled negatively with carbon isotope discrimination (Δ13Cbl) and positively with the ratio between mesophyll diffusion conductance (gm) and stomatal conductance. The WUEi-Δ13Cbl relationship was partly influenced by gm. There was a trade-off between WUEi and photosynthetic N-use efficiency, but only when soil water availability was limiting. Our results suggest that seasonal fluctuations in relation to soil water availability should be accounted for in future modelling studies assessing the carbon sequestration potential and the water-use efficiency of woody energy crops.

Introduction

Fast-growing tree species, such as poplar and willow, implemented in short-rotation bioenergy cultures (SRC), represent a promising renewable energy source (AEBIOM 2012). The success of this renewable bioenergy largely depends on the yields that can be achieved. The large genetic variability found within the Populus genus (Dunlap and Stettler 1998, Al Afas et al. 2005, Paris et al. 2011, Broeckx et al. 2012, 2012) offers the possibility to select highly productive genotypes. The high productivity of poplar has been associated with its high water use (water consumption) (Zsuffa et al. 1996, Allen et al. 1999, Meiresonne et al. 1999) and with its sensitivity to drought (Lindroth et al. 1994, Liang et al. 2006, Monclus et al. 2009). The increasing probability of seasonal droughts (Easterling et al. 2000, Seneviratne et al. 2010) and the prospects of freshwater scarcity (Berndes 2002) emphasize the importance of traits such as water-use efficiency (WUE) and drought tolerance as the selection criteria for biomass production under future climate conditions (King et al. 2013).

At the whole-plant level, WUE is defined as plant dry matter production per unit of water loss via transpiration. Substantial species and genotypic variation in whole-plant WUE have been reported (Cernusak et al. 2007, Linderson et al. 2007, Rasheed et al. 2013). At the leaf level, intrinsic water-use efficiency (WUEi) is defined as the instantaneous ratio between net CO2 assimilation rate (A) and stomatal conductance to water vapour (gs). The carbon isotope discrimination (Δ13C) is expected to scale negatively with WUEi (Farquhar and Richards 1984), and has been commonly used as an indicator of WUEi in poplar (Ripullone et al. 2004, Monclus et al. 2006, Bonhomme et al. 2008, Dillen et al. 2008, Fichot et al. 2011, Rasheed et al. 2013). However, the relationship between Δ13C and WUEi can be disturbed because of differences in the respective time of integration (Ponton et al. 2002, Ripullone et al. 2004) or of variable mesophyll diffusion conductances (gm) (Warren and Adams 2006, Soolanayakanahally et al. 2009). Selecting genotypes for low Δ13C—and assumingly for high WUEi—may not necessarily result in a selection towards higher productivity, as this depends on the main source of variation driving WUEi (Gilbert et al. 2011). As such, Δ13C does not allow distinguishing between the effects of A and gs. On the contrary, the oxygen composition of organic matter (δ18O) may be used to independently estimate variations in WUEi originating from variations in gs (Scheidegger et al. 2000, Barbour 2007). As water in the leaf is the most important source of oxygen, bulk leaf oxygen isotope composition (δ18Obl) integrates gs over the leaf life span. It thus combines source water oxygen isotope composition and leaf water enrichment, partly affected by evaporative processes. When combined with Δ13C data, δ18O is a means to distinguish between the different sources of variation in WUEi.

Seasonal variations in photosynthetic parameters and resource-use efficiency largely affect the modelling of ecosystem carbon uptake (Wilson et al. 2001, Wang et al. 2004, Kosugi and Matsuo 2006, Zhu et al. 2011), determining the efficiency of bioenergy cultures. Strong seasonal variations in photosynthetic parameters have been reported for deciduous species, but mostly under Mediterranean climate conditions in relation to water availability (Wilson et al. 2000, 2001, Xu and Baldocchi 2003, Limousin et al. 2010, Misson et al. 2010). Stomatal closure is generally the primary diffusive limitation to carbon assimilation rate and one of the earliest responses to drought during the growing season (Wilson et al. 2000, Chaves et al. 2002, Flexas and Medrano 2002). Mesophyll conductance to CO2 (gm) also decreases in response to decreasing soil water availability, adding an additional resistance to CO2 diffusion to the chloroplasts (Roupsard et al. 1996, Grassi and Magnani 2005, Limousin et al. 2010, Misson et al. 2010). Photosynthetic limitations because of biochemical impairments are generally observed under severe water stress (Bota et al. 2004, Flexas et al. 2004). Besides increasing WUE, stomatal closure decreases photosynthetic nitrogen-use efficiency (PNUE), defined as the ratio between A and leaf nitrogen (N) concentration (Warren and Adams 2006). The trade-off between WUE and PNUE arises from the generally observed relationship between light-saturated photosynthesis (Asat) and leaf N (Xu and Baldocchi 2003). Stomatal closure has a smaller effect on photosynthesis when compared with the direct impact on transpiration, and has no effect on leaf N. However, gm influences the variation in PNUE, and hence its relationship with WUE (Warren and Adams 2006). The role of gm and its relationship to gs is important for a better understanding of the economics of photosynthetic and N use in a changing climate (Buckley and Warren 2014).

For SRC plantations under temperate climate conditions, the seasonal evolution of photosynthesis, transpirational water loss and WUE are of utmost importance for their productivity and biomass yield. This is especially true when one considers that (i) seasonal variability in photosynthesis is a strong determinant of carbon balance and therefore of the environmental benefit of bioenergy-devoted plantations (Zona et al. 2012); (ii) SRC plantations devoted to biomass production rely on high-yielding species which are generally very sensitive to fluctuations in water availability, such as poplars (Lindroth et al. 1994, Liang et al. 2006, Monclus et al. 2009); (iii) high planting densities (6000–20,000) are likely to exacerbate competition for water acquisition and lead to faster water depletion (Toillon et al. 2013); and (iv) climate change might result in increased frequency and duration of abnormal drought episodes (IPCC 2007, Seneviratne et al. 2010). In view of the above, the seasonal evolution in Δ13C and δ18O as a potential indicator of WUEi in combination with seasonal changes in photosynthesis provides a more detailed study, in particular under field conditions of changing soil water availability. The rationale of the present study is also to quantify genotypic variation in the aforementioned seasonal evolution and in the WUEi relationships for poplar.

The objectives of this study were (i) to investigate and characterize seasonal and genotypic variation in photosynthesis, WUEi and leaf stable isotope composition (13C and 18O); and (ii) to examine how genotypes and timing throughout the growing season affect the relationships between the aforementioned leaf traits. Measurements were performed in a young SRC plantation on six genotypes throughout the growing season (from early May to the end of September 2011). An atypical dry spring to summer period allowed studying the effect of soil water availability. We hypothesized that decreased soil water availability would lead to an increased WUEi and to a decreased PNUE, mainly due to diffusional (stomatal and mesophyll) limitation of assimilation. We investigated the relationship between WUEi, Δ13C and δ18O, as well as between PNUE and WUEi, including the potential effect of gm. Based on the theory, we expected an inverse relationship between WUEi and Δ13C, between Δ13C and δ18O and between WUEi and PNUE. We hypothesized significant genotypic variation in the studied parameters and in their response to varying environmental parameters.

Materials and methods

Experimental site and plant material

The experimental site was located in Lochristi, East-Flanders, Belgium (51°06′44″N, 3°51′02″E; 6.25 m above sea level). The poplar bioenergy plantation (http://uahost.uantwerpen.be/popfull) was established in April 2010 on 18.4 ha of former agricultural land. The long-term average annual temperature at the site is 9.5 °C and the average annual precipitation is 726 mm, equally distributed over the year. A detailed soil analysis prior to planting characterized the soil type as sandy in texture, with clay-enriched deeper soil layers. After site preparation, 25-cm-long dormant and unrooted hardwood cuttings were planted at a density of 8000 ha−1 in a double-row design, with alternating distances of 0.75 and 1.50 m between the rows and 1.1 m between the individuals within each row. A total area of 14.5 ha was planted with 12 selected poplar genotypes representing different species and hybrids of Populus deltoides Bartr. (ex Marsh.), Populus maximowiczii Henry, Populus nigra L. and Populus trichocarpa Torr & Gray (ex Hook), arranged in large monoclonal blocks. Neither irrigation nor fertilization was applied. Additional information on the site, the soil characteristics and the plantation layout can be found in Broeckx et al. (2012. For the present study, six out of the 12 poplar genotypes were retained, covering the different parentages present in the plantation: Koster and Oudenberg (P. deltoides × P. nigra), Bakan and Skado (P. trichocarpa × P. maximowiczii), Grimminge (P. deltoides × (P. trichocarpa × P. deltoides)) and Wolterson (P. nigra) (see Table S1A available as Supplementary Data at ; Broeckx et al. 2012). All measurements reported in this contribution were performed between May and September 2011, i.e., during the second growing season of the SRC plantation.

Meteorological parameters

Meteorological parameters were recorded half-hourly using a meteorological mast installed within the SRC plantation at the experimental site. Air temperature and relative humidity data, recorded using Vaisala probes (Model HMP45C, Vaisala, Helsinki, Finland), were used to calculate vapour pressure deficit (VPD). The incoming short-wave radiation (SWR, 0.3–3 μm) was measured using a pyranometer (Model CNR1, Kipp & Zonen, Delft, The Netherlands). The amount of precipitation was measured with a tipping bucket rain gauge (Model 3665R, Spectrum Technologies, Inc., Plainfield, IL, USA). Moisture probes (TDR Model CS616, Campbell Scientific, Logan, UT, USA) placed at depths of 20 and 40 cm close to the mast were used to measure soil water content (SWC, m3 m−3). As a complement, soil water potential (ΨS) was measured from June to November 2011 using calibrated equitensiometer probes (Type EQ-2, Delta-T Devices Ltd, Cambridge, UK) installed at depths of 20 and 40 cm at four locations around the mast. We chose to characterize soil water availability along the growing season through the time course of ΨS values, averaged among the four locations. Therefore, ΨS values were extrapolated for the missing period (May–early June) based on the relationship observed between SWC and ΨS measurements at each measuring depth. A feed-forward Neural Network (Matlab R2012a, Mathworks, Natick, MA, USA) was used to interpolate missing values. The correlation between predicted and measured values ranged between 0.83 and 0.98 for different soil depths (see Broeckx et al. 2013).

Leaf gas exchange, chlorophyll content and PNUE

Leaf gas exchange measurements were performed repeatedly on the same trees during the 2011 growing season in seven measurement campaigns (MCs): 4–6 May (MC1), 18–20 May (MC2), 4–8 July (MC3), 27–29 July (MC4), 16–19 August (MC5), 5–9 September (MC6) and 26–30 September 2011 (MC7). For the six genotypes, measurements were done on four replicate trees located close to the mast with a LI-6400 open path photosynthesis system (Li-Cor, Lincoln, NE, USA) equipped with a leaf chamber fluorometer (LI-6400-40, Li-Cor). Measurements were taken in the upper canopy, on the first fully mature sunlit leaf of the current-year main axis. To minimize differences in leaf age across MCs, we sampled leaves of the same leaf rank. Leaves were first acclimated for 10 min in the chamber at a CO2 concentration of 400 ppm and under a photosynthetic photon flux density (PPFD) of 1500 μmol m−2 s−1. Preliminary test experiments had shown that this PPFD was enough to ensure saturating light conditions for all genotypes. Afterwards, light-saturated assimilation rate at atmospheric CO2 concentration (Asat, μmol m−2 s−1) and stomatal conductance (gs-sat, mol m−2 s−1) were recorded before establishing the response of the net assimilation rate (A) to varying intercellular CO2 concentrations (Ci), i.e., the A–Ci curve. Each curve consisted of 10 steps of external CO2 concentrations set in succession to 400, 300, 250, 150, 100, 50, 500, 750, 1000 and 1250 ppm (Monclus et al. 2006). Leaves were allowed to equilibrate at least 3 min at each step before data were logged. Net assimilation rates were corrected for the effect of CO2 diffusion, according to the instrument manual (LI-6400XT Version 6), using a diffusion correction term of 0.46 μmol s−1. Before logging at each step of the A–Ci curves, steady state (F) and maximum fluorescence (F′) were measured during a light-saturating pulse (7 mmol m−2 s−1) and the efficiency of Photosystem II (ΦPSII) was determined as:

Then, the CO2 concentration in the chamber was set back to 400 ppm. Once the net assimilation rate had stabilized, the response to varying light intensities was recorded (A-light curve). Leaf photosynthesis was measured at eight PPFD intensities in the following order: 1500, 1000, 800, 600, 400, 200, 100, 0 μmol m−2 s−1. A minimum of 2 min of leaf equilibration was set at each step before data were logged. Dark respiration was defined as the absolute CO2 exchange rate measured during the last step of the A-light curve. All measurements were done at a constant block temperature (25 °C) and at a controlled VPD close to 1 kPa (1.2 ± 0.04, mean ± SE). Intrinsic water-use efficiency under saturating conditions () was calculated as the ratio between the values of Asat and gs-sat obtained from the A–Ci and A-light curves under reference conditions (PPFD of 1500 μmol m−2 s−1 and CO2 concentration of 400 ppm).

Once gas exchange measurements were completed, a minimum of six chlorophyll readings was taken on the same leaf with a portable chlorophyll content meter (CCM-200, Opti-Sciences, Inc., Hudson, NH, USA). Total chlorophyll content (Chl) was estimated from the CCM values according to the equations reported in Richardson et al. (2002). The leaf sampled was then harvested and the individual leaf area (LA) was measured using a LI-3000 leaf area meter (Li-Cor). A subsample was punched out of the leaf lamina to determine leaf mass per area (LMA, g m−2) after drying at 70 °C; LMA was only available from MC3 onwards. The dried leaf material was then ground and used for the assessment of the leaf N content per unit mass (NM, mg g−1) with an elemental analyser (Carlo Erba, NA 1500-NC, Milan, Italy). The values of NM were converted to N content per unit area (NA, mg cm−2) using LMA values. Photosynthetic nitrogen-use efficiency (μmol mg−1 s−1) was calculated as the ratio of Asat (μmol m−2 s−1) to NA.

Estimation of mesophyll conductance and photosynthetic parameters

The mesophyll diffusion conductance to CO2 from the sub-stomatal cavities to the chloroplast (gm) was estimated by combining gas exchange and chlorophyll fluorescence measurements (Pons et al. 2009). The rate of photosynthetic electron transport (JETR) was calculated as: where α is the leaf absorptance and 0.5 is the fraction of photons absorbed by Photosystem II. Absorptance was derived from the CCM readings according to Bauerle et al. (2004) after conversion of the CCM readings to soil plant analysis development (SPAD) values (Richardson et al. 2002). Mesophyll conductance was then estimated following the equation of Harley et al. (1992): where 42.7 is the CO2 compensation point in the absence of dark respiration, as taken from Bernacchi et al. (2001), and Rd is the mitochondrial respiration in the light, taken as half of the dark respiration obtained from the A-light curves (Piel et al. 2002, Niinemets et al. 2005). The values of gm were then used to convert A–Ci curves to A–Cc curves, with Cc being the CO2 concentration in the chloroplast stroma calculated as (Limousin et al. 2010, Misson et al. 2010):

The maximum carboxylation rate (Vcmax) and the maximum rate of electron transport (Jmax) were estimated by fitting the A–Cc curves to the biochemical photosynthesis model of Farquhar et al. (1980) using the routine developed by Sharkey et al. (2007). The Michaelis constant of Rubisco for carbon dioxide (Kc), the inhibition constant of Rubisco for oxygen (Ko) and the photocompensation point (Γ*) used for fitting were taken from Sharkey et al. (2007).

Carbon and oxygen stable isotope analyses

Isotopic analyses were performed at the Stable Isotope Laboratory of the James Hutton Institute (Invergowrie, Dundee, UK). Bulk leaf carbon isotope composition (δ13Cbl) was determined on the leaves used for gas exchange measurements. Subsamples of ground leaf material were enclosed and weighed in tin capsules and combusted in a continuous flow isotope ratio mass spectrometer (IRMS) (Delta V, Thermo Fisher Scientific, Bremen, Germany). The CO2 produced by combustion was purified and its 13CO2/12CO2 ratio was analysed by the IRMS. The δ13Cbl (‰) was expressed relative to the Pee Dee Belemnite standard (Craig 1957). The accuracy of measurements was assessed by repeated measures of laboratory standards and was ±0.08‰ (standard deviation). Carbon isotope discrimination between the atmosphere and the bulk leaf organic matter (Δ13Cbl, ‰) was then calculated as in Farquhar et al. (1989): with δ13Cair assumed to equal −8‰.

The same leaf powder used for δ13C analyses was used to measure the 18O composition of bulk leaf matter (δ18Obl). Leaf material was enclosed and weighed in silver capsules. Analyses were conducted with a continuous flow IRMS (Delta Plus XP, Thermo Fisher Scientific) interfaced with a high temperature elemental analyser. Bulk leaf oxygen isotope composition was expressed relative to the sea mean ocean water standard and the analytical precision for repeated measurements was ±0.09‰ (standard deviation).

Statistical analysis

Some data were missing at random (MAR) and could be ignored for parameters and genotypes that were measured from the third MC onwards (Verbeke and Molenberghs 2000). The mixed procedure for repeated measurements was used to analyse the effects of genotype and seasonality on the above-mentioned parameters. Measurements were performed seven times during the growing season (repeated variable ‘MC’) on the same four replicate trees of each genotype (subject variable ‘Tree’). A linear mixed model with fixed effects genotype, MC and their interaction—indicating the genotype-specific behaviour in time—was used. The unstructured repeated covariance structure was chosen and the variance component was estimated using restricted maximum likelihood (REML). When significant genotype or MC effects were found, pairwise comparisons of the means were performed using the Bonferroni adjustment (see Table S2A available as Supplementary Data at ). Relationships between photosynthetic, WUE and isotopic parameters were examined using linear and non-linear regression analyses and coefficients of determination (R2) as well as genotype-specific Spearman's rank correlation coefficients. All statistical tests were considered significant when P < 0.05. All statistical analyses were performed in SPSS 20.0 (IBM Corp., SPSS Statistics for Windows, Armonk, NY, USA).

Results

The seasonal course of precipitation, air temperature, daytime (SWR >20 W m−2) maximum VPD (VPDmax) and soil water availability at the site during the 2011 growing season (May–September) is presented in Figure 1. Although the 2011 growing season showed a normal pattern of temperature and rainfall throughout the season, there were a few periods with dry conditions. The first drop in soil water potential (Ψs) was observed at the end of May in the upper soil layer (20 cm depth). Close to MC2 Ψs peaked at approximately −1.8 MPa, while there was no apparent response in the 40-cm depth layer. The precipitation and lower VPD that occurred afterwards led to a progressive recovery. The second drop in Ψs to −1.5 MPa was observed in mid-July (close to MC3) for both soil layers, concomitantly with high air temperatures and high VPD. The high amount of precipitation after mid-July combined with a progressive decrease in VPD resulted in the recovery of Ψs close to zero for the rest of the growing season (MC4–MC7).

Figure 1.

Seasonal time course of the main meteorological parameters during the period of this study (May–September 2011): (a) precipitation; (b) daily minimum (Tair min) and maximum (Tair max) air temperature; (c) daytime maximum vapour pressure deficit (VPDmax); and (d) soil water potential at 20 cm (solid line) and 40 cm (dotted line) soil depths. Grey bars indicate the timing of gas exchange MCs: 4–6 May (MC1), 18–20 May (MC2), 4–8 July (MC3), 27–29 July (MC4), 16–19 August (MC5), 5–9 September (MC6), 26–30 September 2011 (MC7).

Overall, nearly all of the photosynthetic leaf traits differed significantly among genotypes and fluctuated during the growing season, i.e., along the MCs (Figure 2, Table 1; see Table S2A available as Supplementary Data at ). Considering the average pattern across the six genotypes, Asat, gs-sat and Δ13Cbl exhibited a similar time course with a pronounced decrease in mid-July (MC3), when the soil water potential was low at both 20 and 40 cm depths, and a progressive increase towards the end of the growing season (Figure 2a, b and d). The values of WUEi followed an opposite trend (Figure 2c). The time course observed for Vcmax, Jmax and gm was slightly different. Vcmax and Jmax increased progressively during the growing season and gm decreased, especially in July (MC3; Figure 2f and g; Jmax data not shown). Mesophyll conductance showed overall a similar seasonal evolution to that of gsat for the dry period May–July but remained higher than gs-sat (Figure 2b and g). From August onwards, however, gs-sat and gm followed an opposite pattern; gm decreased progressively (Figure 2b and g). The overall means of Vcmax and Jmax were 125.2 and 172.1 μmol m−2 s−1, respectively (Figure 2f), both parameters being strongly and linearly correlated across genotypes and MCs (R2 = 0.79; P < 0.0001). A decrease throughout the growing season was also observed in δ18Obl (Figure 2e). The seasonal trends in NA and in LMA were less obvious (despite a significant time effect; Table 1), potentially due to some missing data points in the beginning of the growing season (Figure 2h and i). Photosynthetic nitrogen-use efficiency was lower in July at low soil water availability (MC3; Figure 2j), but—as for NA and LMA—the response to water availability was less clear. Genotypes were not significantly different in their seasonal evolution of NA and PNUE (Table 1).

Figure 2.

Seasonal evolution of (a) net assimilation rate (Asat); (b) stomatal conductance (gs-sat); (c) intrinsic water-use efficiency (WUEi); (d) bulk leaf carbon isotope discrimination (Δ13Cbl); (e) bulk leaf oxygen isotope composition (δ18Obl); (f) maximum carboxylation rate (Vcmax); (g) mesophyll conductance (gm); (h) area-based leaf N content (NA); (i) leaf mass per area (LMA); and (j) photosynthetic nitrogen-use efficiency (PNUE). Data points represent genotypic means (±SE) for six poplar genotypes of different parentages: T × M (Bakan, Skado), D × N (Koster, Oudenberg), D × (T × D) (Grimminge) and N (Wolterson).

Table 1.

Output of the mixed model analysis (REML) showing the effects of genotype and time in the season (MC) on photosynthetic and related parameters. The different parameters have been identified and described in the text. *, 0.01 < P ≤ 0.05; **, 0.001 < P ≤ 0.01; ***, P ≤ 0.001; MC, measurement campaign.

Parameters	df	Genotype	Df	MC	df	Genotype × MC
Asat	5	***	6	***	25	***
gs-sat	5	***	6	***	25	***
WUEi	5	**	6	***	25	*
Δ13Cbl	5	***	6	***	26	*
δ18Obl	5	**	6	***	26	***
Vcmax	5	**	6	***	25	***
Jmax	5	**	6	***	25	***
gm[1]	5		6	***	24	**
NA	5	***	4	**	20
LMA	5		4	***	20	***
PNUE	5	**	4	***	19

[1]Heterogeneous Toeplitz covariance structure was used.

A closer look at the observed results showed that the six genotypes did not respond in the same way or with the same amplitude with time in the growing season, as indicated by the significant genotype × MC interactions observed for most traits (Table 1; see also Figure 2). Differences among genotypes were particularly reinforced during the dry period around MC3 (Figure 2). Genotypes Wolterson and Oudenberg were clearly less responsive to the dry period than the other genotypes in terms of Asat, gs-sat, WUEi and PNUE (Figure 2a–c and j). Overall, genotype Wolterson showed the highest values of Asat, gs-sat, Vcmax and NA throughout most of the growing season (Figure 2a, b, f and h), while WUEi was at the lower end of the genotypic range (Figure 2c). The ranking of the other genotypes changed substantially during the growing season, although genotypes Bakan and Skado remained consistently at the lower end of the range for Asat and gs-sat (Figure 2a and b). Genotype Skado had the lowest Δ13Cbl and the highest WUEi throughout the entire growing season (Figure 2d and e). On the other end, the highest Δ13Cbl values were observed for genotype Grimminge, which also showed the lowest Vcmax and Jmax values with an early decrease from August onwards (MC5–7; Figure 2d and f). In contrast to other leaf traits, the genotypic ranking for NA did not significantly change throughout the growing season (no significant genotype × MC interaction, Table 1; Figure 2h). Genotypes Wolterson and Grimminge generally showed the highest and lowest NA values, respectively (Figure 2h).

The values of Asat and gs-sat were significantly, but non-linearly, related (Asat = 36.25 × gs-sat/(0.22 + gs-sat)), with Asat reaching saturation at high gs-sat (Figure 3). A similar but less significant pattern was found between Asat and gm , with Asat saturating at high gm. The relationship between gs-sat and gm was linear at low values, with a decoupling among both parameters at higher values . The linear part of the relationship was mainly determined by values recorded during MC3, when soil water availability was reduced.

Figure 3.

Curvilinear relationships between (a) stomatal conductance (gs-sat) and net assimilation rate (Asat); (b) mesophyll conductance (gm) and net assimilation rate (Asat); and (c) gs-sat and gm. Data points indicate the first letter of each genotype (B, Bakan; S, Skado; K, Koster; O, Oudenberg; G, Grimminge; W, Wolterson) followed by the number of MCs (1–7) and represent the mean of four individuals. For (a) the genotype-specific Spearman's correlation coefficients are presented (**, 0.001 < P ≤ 0.01; ***, P ≤ 0.001).

Net assimilation rate was linearly and negatively related to WUEi (Figure 4a). A stronger negative, but non-linear (WUEi = (0.15 × 0.31)/(0.31 + gs-sat) relationship was found between gs-sat and WUEi (Figure 4b). The ratio of gm to gs-sat was significantly and positively related to WUEi (Figure 4c). The highest values along the curve were identified as MC3 observations, suggesting a shift in the relative contribution of mesophyll vs stomatal limitation when soil water was limiting. As expected from the theory, Δ13Cbl and WUEi were linearly and negatively related although the relationship showed some scatter (Figure 5b). There was no significant correlation between δ18Obl and Δ13Cbl (Figure 5c). However, a significant and positive relationship was found between δ18Obl and WUEi (Figure 5a), while a significant and negative (non-linear; δ18Obl = (28.24 × 2.51)/(2.51 + gs-sat); R2 = 0.42, P < 0.0001) relationship was observed between δ18Obl and gs-sat.

Figure 4.

Relationships between (a) net assimilation rate (Asat) and intrinsic water-use efficiency (WUEi); (b) stomatal conductance (gs-sat) and WUEi; (c) the ratio of mesophyll conductance (gm) to stomatal conductance (gs-sat) and WUEi. Data points indicate the first letter of each genotype (B, Bakan; S, Skado; K, Koster; O, Oudenberg; G, Grimminge; W, Wolterson) followed by the number of MCs (1–7) and represent the mean of four individuals. For (b) the genotype-specific Spearman's correlation coefficients are presented (*, 0.01 < P ≤ 0.05; **, 0.001 < P ≤ 0.01; ***, P ≤ 0.001).

Figure 5.

Relationships between (a) intrinsic water-use efficiency (WUEi) and bulk leaf oxygen isotope composition (δ18Obl); (b) WUEi and bulk leaf carbon isotope discrimination (Δ13Cbl); (c) bulk leaf oxygen isotope composition (δ18Obl) and bulk leaf carbon isotope discrimination (Δ13Cbl). Data points indicate the first letter of each genotype (B, Bakan; S, Skado; K, Koster; O, Oudenberg; G, Grimminge; W, Wolterson) followed by the number of MCs (1–7) and represent the mean of four individuals. For (b) the genotype-specific Spearman's correlation coefficients are presented (*, 0.01 < P ≤ 0.05; ***, P ≤ 0.001).

The maximum rate of carboxylation scaled positively with leaf N content, especially when expressed on an area basis (Figure 6). In addition, Vcmax and NA scaled positively with Chl (data not shown). Leaf N content on an area basis was significantly and negatively correlated to WUEi (R2 = 0.18, P = 0.0123) but no relationship could be observed with Δ13Cbl. No correlation was observed between NA and LMA while NM was significantly and positively correlated to LMA (R2 = 0.42, P < 0.0001). Neither gm nor Vcmax was correlated to LMA. A significant and negative relationship was found between WUEi and PNUE, mainly due to the observations during MC3 (Figure 7a). Similarly, the significant and positive relationship between gm and PNUE was mainly driven by MC3 readings.

Figure 6.

Relationship between the leaf N content (NM) and the maximum rate of carboxylation (Vcmax_M) on a mass basis. The genotype-specific Spearman's correlation coefficients are presented (*, 0.01 < P ≤ 0.05; **, 0.001 < P ≤ 0.01). The insert panel shows the same relationship on an area basis, NA vs Vcmax_A. Data points indicate the first letter of each genotype (B, Bakan; S, Skado; K, Koster; O, Oudenberg; G, Grimminge; W, Wolterson) followed by the number of MCs (1–7) and represent the mean of four individuals.

Figure 7.

Relationships between (a) intrinsic water-use efficiency (WUEi) and photosynthetic nitrogen-use efficiency (PNUE); and (b) between mesophyll conductance (gm) and PNUE. Data points indicate the first letter of each genotype (B, Bakan; S, Skado; K, Koster; O, Oudenberg; G, Grimminge; W, Wolterson) followed by the number of MCs (1–7) and represent the mean of four individuals. For (a) the genotype-specific Spearman's correlation coefficients are presented (*, 0.01 < P ≤ 0.05).

Discussion

The results reported above illustrate that photosynthesis-related leaf traits, including WUEi, significantly varied during the growing season in an SRC bioenergy plantation, and that this variation was genotype dependent in poplar. Furthermore, our results indicate that the relation between Δ13Cbl and WUEi did not always hold throughout the growing season, and that water availability played a significant role in this relationship. Since we have shown that seasonal variations and genotypic differences in photosynthetic parameters are substantial in SRC poplar plantations, they need to be accounted for in future modelling studies. Seasonality of leaf gas exchange can result from leaf ontogeny and dynamic changes in environmental conditions such as light, nutrients, temperature or water availability. In our study, potential effects of leaf ontogeny were minimized by always measuring recently matured leaves emerging on the current-year axis, taking advantage of the indeterminate growth of poplars. As all measurements were performed under saturating irradiance and constant VPD, seasonal variations could be mostly attributed to variations in soil water availability. Temporal variation in photosynthetic parameters is important in determining the seasonality and magnitude of ecosystem carbon fluxes and is therefore an important factor to consider for modelling studies (Wilson et al. 2001).

Leaf photosynthetic parameters recorded in this study were consistent with data previously reported for several poplar species and hybrids (Roupsard et al. 1996, Pons and Westbeek 2004, Ripullone et al. 2004, Monclus et al. 2006, Gornall and Guy 2007, Soolanayakanahally et al. 2009, Fichot et al. 2010, 2011). Eddy covariance measurements of net ecosystem CO2 fluxes performed during the same period confirmed a net ecosystem carbon uptake until the end of September (Broeckx et al. 2013, Zona et al. 2013). This indicated a good agreement between the timing of leaf-level and canopy-level photosynthetic processes in the plantation. High photosynthetic activity until the end of September has already been reported for three different genotypes of P. alba L., P. nigra L. and P. deltoides × P. nigra (Bernacchi et al. 2003). However, this pattern contrasts with data reported for other temperate deciduous species such as ash, maple and oak, for which photosynthetic uptake and photosynthetic capacity already showed a substantial decline by early or mid-September (Wilson et al. 2000, Grassi et al. 2005, Dillen et al. 2012). Delayed senescence with sustained carbon uptake is most likely associated with the pioneering and fast-growth habit of poplar species.

Drought-induced variations in leaf photosynthesis can be mediated by stomatal closure, by changes in mesophyll conductance to CO2 and by alterations of photosynthetic capacities. Reduced gm was observed concomitantly with reduced gs-sat during the period of low water availability, as already documented for different species (Roupsard et al. 1996, Warren et al. 2004, Galmes et al. 2007, Flexas et al. 2008). However, in the present study not all poplar genotypes responded in the same way to drier soil conditions, indicating substantial genotypic variation in the degree of this response. Local measurements of soil water potential around the mast could not exclude the possibility of genotypic differences in soil water potential related to genotypic differences in total LA. Bigger trees encounter a more rapid and more severe water shortage due to their high transpiratory water loss. In contrast, the values of Vcmax and Jmax did not show any clear pattern during the dry period, suggesting that the decrease in photosynthesis was mostly caused by diffusional limitations (stomatal and non-stomatal) rather than by biochemical limitations. This is consistent with the idea that biochemical limitations become quantitatively important only during severe droughts (Grassi and Magnani 2005, Galmes et al. 2007). The values of Vcmax were tightly correlated with leaf N contents expressed either on a unit mass or on an area basis, suggesting that leaf N contents were a reliable estimator of photosynthetic capacities across all poplar genotypes along the growing season (Zhu et al. 2011). The constant NA observed during the period of low water availability is in line with the absence of a reduction in Vcmax, and suggests that there was no marked seasonal change in N allocation to the photosynthetic apparatus (i.e., Rubisco and chlorophyll) (Demarez et al. 1999, Montpied et al. 2009).

Stomatal conductance remained higher than gm during the drier period, suggesting that stomatal conductance was actually the most limiting process to photosynthesis at this time. This is in line with other studies (Wilson et al. 2000, Grassi and Magnani 2005, Niinemets et al. 2005, Limousin et al. 2010, Flexas et al. 2012). The relative contribution of stomatal vs mesophyll conductance to total limitations may however vary with species, with drought intensity and also with canopy position (Diaz-Espejo et al. 2007, Galmes et al. 2007, Cano et al. 2013, Flexas et al. 2013). This was confirmed in our study by the fact that gs reached higher values than gm by the end of the growing season. The curvilinear relationship between gm and gs-sat substantiates the modelled relationship predicted by Tholen et al. (2012). Significant relationships between gs and gm have also been reported in other studies (Douthe et al. 2011, Egea et al. 2011, Buckley and Warren 2014).

Variations in WUEi originate from variations in either A, gs or both (Farquhar and Richards 1984). Previous studies on poplars have suggested that variations among genotypes are generally driven by variations in gs (Monclus et al. 2006, Fichot et al. 2011, Rasheed et al. 2011, Cao et al. 2012) although one opposite result has been reported (Rasheed et al. 2013). Our results suggest that variations in WUEi across dates and genotypes were primarily driven by variations in gs-sat. This was supported by the fact that while WUEi and gs-sat were negatively related—as expected—WUEi and Asat were also negatively related which was at first counter-intuitive. This negative relationship can be explained by the fact that variations in Asat were actually overridden by larger parallel variations in gs-sat. In addition, WUEi and gs-sat were negatively and positively related to δ18Obl, respectively. Our results suggest that variations in δ18Obl reflected a significant part of variations in gs-sat. The oxygen in organic matter is derived from water and the δ18O of organic matter is primarily affected by source δ18O and by evaporative processes (Scheidegger et al. 2000, Roden and Farquhar 2012). As we did not measure the source δ18O in the present study, we have no evidence for differences in the source δ18O. We know, however, that the different genotypes experienced different water table depths, which significantly and spatially varied throughout the plantation (L.S. Broeckx, unpublished data). So we hypothesize that the different genotypes acquired water from different soil horizons, considering the observed genotypic differences in plant size (Duursma et al. 2011) and assuming genotypic differences in rooting depth in response to the varying water table depths. The response of rooting depth to water table depth is a trait adaptive to the native riparian habitat of poplars (Rood et al. 2003).

As expected from the theory (Farquhar et al. 1982), the values of WUEi and Δ13Cbl were significantly and negatively related, which confirms previously reported observations for various crop species (Farquhar and Richards 1984, Meinzer et al. 1990) and for woody species (Guehl et al. 1995, Ponton et al. 2002, Ripullone et al. 2004). The significant scatter in and disturbance of the observed relationship may be explained by several things. Firstly, WUEi values correspond to virtually instantaneous measurements, while Δ13Cbl reflects a temporal integration of WUEi over the course of leaf formation and recent photosynthetic activity. Secondly, Asat and gs-sat were measured under saturating conditions after the sampled leaf had acclimated to the chamber conditions, such that the values of WUEi reflected ‘maximal’ functioning. This optimal functioning is obviously not maintained during the entire leaf lifespan. Thirdly, finite but variable gm can affect WUEi and influence the relationship between WUEi and Δ13Cbl (Warren and Adams 2006, Flexas et al. 2008, Seibt et al. 2008, Soolanayakanahally et al. 2009). The hyperbolic relationship observed between WUEi and gm/gs-sat supports this line of reasoning and is consistent with both theory (Flexas et al. 2013) and data reported for different species (Galmes et al. 2010, Flexas et al. 2013). In addition, the observed negative Δ13Cbl–WUEi-sat relationship varied significantly among genotypes and with timing throughout the growing season. This observation confirms the effects of both species and water availability on the relationship between Δ13Cbl and WUEi that were previously reported for poplar (DesRochers et al. 2007, Xu et al. 2008, Larchevêque et al. 2011). The absence of a correlation in genotype Wolterson is most likely explained by lower water availability experienced as a consequence of low(er) total LA (Broeckx et al. 2012, 2012), hence the lower transpiration and reduced soil water depletion. The lack of a significant correlation observed between Δ13Cbl and δ18Obl, although δ18Obl was significantly related to gs-sat and WUEi (see the discussion above), also reinforces the idea that the WUEi–Δ13Cbl relationship was partly influenced by gm.

The economics of N and water use during photosynthesis is primarily interlinked through their mutual dependence on stomatal conductance. Especially during drought stomatal closure contributes to increasing WUEi on the one hand, while decreasing PNUE on the other hand resulting in a trade-off between both traits (Field et al. 1983, Warren and Adams 2006). Our results were consistent with this concept. The reduced assimilation rate caused by a decrease in stomatal conductance with constant N allocation increased the N cost per unit of carbon gain, suggesting maximization of resource-use efficiency depending on the most limiting resource (Reich et al. 1989, Xu and Baldocchi 2003, Han 2011). However, when the early July data (i.e., when low soil water availability had the largest effect on leaf gas exchange) were discarded from the analysis, WUEi and PNUE were not significantly related. This suggests that WUEi and PNUE are uncoupled under optimal conditions, as already observed for other poplar species (Soolanayakanahally et al. 2009). As suggested by Soolanayakanahally et al. (2009), this might be expected if gs-sat, gm and other factors influencing net assimilation rate vary independently, as was apparently the case in our study under non-limiting conditions (Figure 3).

In conclusion, our results showed significant seasonal evolution in photosynthesis, in WUEi—as quantified by δ13C and δ18O—and in PNUE of poplars grown under a high-density SRC regime. The seasonal evolution was mostly explained by variations in soil water availability and by stomatal control, but was strongly genotype dependent. This study suggests taking genotypic differences in seasonal evolution into account in future modelling studies.

Supplementary data

Supplementary data are available at .

Conflict of interest

None declared.

Funding

This research has received funding from the European Research Council under the European Commission's Seventh Framework Programme (FP7/2007–2013) as ERC grant agreement no. 233366 (POPFULL), as well as from the Flemish Hercules Foundation as Infrastructure contract ZW09-06. Further funding was provided by the Flemish Methusalem Programme and by the Research Council of the University of Antwerp. Funding to pay the Open Access publication charges for this article was provided by the European Commission's Seventh Framework Programme (FP7/2007–2013) as ERC grant agreement n° 233366 (POPFULL).