Thermal acclimation of photosynthesis: a comparison of boreal and temperate tree species along a latitudinal transect



    Corresponding author
    1. Department of Forest and Wildlife Ecology, University of Wisconsin-Madison, Madison, WI 53706, USA and
    2. School of Forestry, Louisiana Tech University, Ruston, LA 71270, USA
      D. Dillaway. Fax: +318-257-5061; e-mail:
    Search for more papers by this author

    1. Department of Forest and Wildlife Ecology, University of Wisconsin-Madison, Madison, WI 53706, USA and
    Search for more papers by this author

D. Dillaway. Fax: +318-257-5061; e-mail:


Common gardens were established along a ∼900 km latitudinal transect to examine factors limiting geographical distributions of boreal and temperate tree species in eastern North America. Boreal representatives were trembling aspen (Populus tremuloides Michx.) and paper birch (Betula papyrifera Marsh.), while temperate species were eastern cottonwood (Populus deltoides Bartr ex. Marsh var. deltoides) and sweetgum (Liquidambar styraciflua L.). The species were compared with respect to adjustments of leaf photosynthetic metabolism along the transect, with emphasis on temperature sensitivities of the maximum rate of ribulose bisphosphate (RuBP) carboxylation (EV) and regeneration (EJ). During leaf development, the average air temperature (Tgrowth) differed between the coolest and warmest gardens by 12 °C. Evidence of photosynthetic thermal acclimation (metabolic shifts compensating for differences in Tgrowth) was generally lacking in all species. Namely, neither EV nor EJ was positively related to Tgrowth. Correspondingly, the optimum temperature (Topt) of ambient photosynthesis (Asat) did not vary significantly with Tgrowth. Modest variation in Topt was explained by the combination of EV plus the slope and curvature of the parabolic temperature response of mesophyll conductance (gm). All in all, species differed little in photosynthetic responses to climate. Furthermore, the adaptive importance of photosynthetic thermal acclimation was overshadowed by gm's influence on Asat's temperature response.


Numerous studies predict that anticipated climate warming (IPCC 2007) will bring about marked shifts in the geographical distributions of plant species (e.g. Prentice, Sykes & Cramer 1993; Flannigan & Woodward 1994; Sykes & Prentice 1996; Iverson & Prasad 1998; Bakkenes et al. 2002; Hickler et al. 2004; Scheller & Mladenoff 2005). To varying degrees, predictions of climate change impacts on plant demography have relied on the association between a species's current geographical distribution and corresponding climate characteristics (Woodward 1987), either explicitly or implicitly invoking hypothetical constraints imposed by extremes in temperature or moisture availability on one or more stages of a species's life cycle. Typically, however, the exact nature of such constraints is unknown, and, consequently, this overall approach has been criticized for lacking a sufficiently mechanistic basis (e.g. Loehle & LeBlanc 1996; Schenk 1996; Pearson & Dawson 2003; Ibáñez, Clark & Dietze 2008).

Accordingly, we have established common gardens along a ∼900 km latitudinal transect, extending from northern Wisconsin to southern Illinois, USA, to examine extrinsic and intrinsic factors potentially limiting the current geographical distribution of boreal and temperate tree species in eastern North America. The boreal representatives in these gardens are trembling aspen (Populus tremuloides Michx.) and paper birch (Betula papyrifera Marsh.), while the temperate species are eastern cottonwood (Populus deltoides Bartr ex. Marsh var. deltoides) and sweetgum (Liquidambar styraciflua L.). A specific goal of this ongoing study was to compare the four species with respect to mode and magnitude of metabolic acclimation to different climates. In this effort, we have adopted a narrow definition of acclimation (Atkin & Tjoelker 2003) – an adjustment of metabolic rate that compensates for a change in growth temperature, potentially resulting in metabolic homeostasis (i.e. identical metabolic rates in contrasting thermal regimes, when measured in situ). A general hypothesis underlying our research is that, relative to their temperate counterparts, boreal tree species are less capable of acclimating metabolically to increasingly warm environments, and this lack of plasticity compromises their ability to compete and persist in temperate climates.

One of our foci along the latitudinal transect has been the suite of biochemical traits underlying leaf photosynthetic performance in C3 plants, such as the maximum rate at which ribulose bisphosphate (RuBP) is carboxylated (Vcmax) and regenerated (Jmax). Photosynthetic thermal acclimation revolves in large part around the extent to which these two attributes respond to short- and long-term variation in leaf temperature (e.g. Berry & Bjorkman 1980; Medlyn et al. 2002a; Hikosaka et al. 2006; Sage, Way & Kubien 2008). The sensitivity of Vcmax and Jmax to short-term (e.g. instantaneous) fluctuation in leaf temperature is often characterized in terms of an activation energy, derived from an Arrhenius equation (Hikosaka et al. 2006) or alternative models when behaviour deviates from an exponential form (e.g. Kattge & Knorr 2007). The activation energies of both Vcmax and Jmax have been shown to vary considerably among species and across climatic gradients (e.g. Wullschleger 1993;Leuning 1997; Bunce 2000; Yamori, Noguchi & Terashima 2005; Hikosaka et al. 2006; Kattge & Knorr 2007; Bunce 2008; Way & Sage 2008).

In the present study, we assess photosynthetic thermal acclimation using the approach of Hikosaka et al. (2006), who argued that the principal factors influencing responses of net photosynthesis to temperature include the activation energy of Vcmax (EV) and Jmax (EJ), as well as the balance between Jmax and Vcmax, hereafter referred to as J/V. Of the two activation energies, EV typically exerts the most influence over net photosynthesis under current atmospheric conditions, whereas EJ is an important determinant of photosynthesis at high CO2 levels and temperature extremes (e.g. Wise et al. 2004; Hikosaka et al. 2006; Sage et al. 2008). Hikosaka et al. (2006) suggested that acclimatory adjustments to photosynthetic metabolism entail an increase in EV on the order of 1 kJ mol−1 with every 1 °C rise in average air temperature. The critical outcome of this behaviour is a tracking of variation in climate by the optimum temperature for light-saturated net photosynthesis (Topt) at current partial pressures of atmospheric CO2 (Hikosaka et al. 2006).

Because, in general, Vcmax (rather than Jmax) limits photosynthesis at contemporary levels of atmospheric CO2, J/V seldom has direct implications for the temperature sensitivity of net photosynthesis under current conditions (e.g. Bunce 2000). Rather, J/V is thought to reflect the relative allocation of leaf nitrogen between light-harvesting complexes and CO2 fixation (von Caemmerer & Farquahr 1981), which has a direct bearing on the efficiency of nitrogen use in photosynthetic machinery. Hikosaka (2005) proposed that to optimize this efficiency, relatively more nitrogen should be invested in CO2 fixation [i.e. the enzyme RuBP carboxylase/oxygenase (Rubisco)] as growth temperature increases. Thus, temperature-normalized J/V should decline in warmer environments, as has been observed (e.g. Leuning 1997; Hikosaka 2005; Onada, Hikosaka & Hirose 2005; Yamori et al. 2005; Kattge & Knorr 2007).

Based, then, on the theoretical framework described above, we examined various aspects of photosynthetic metabolism in aspen, birch, cottonwood and sweetgum foliage along our latitudinal transect in order to compare the nature and extent of photosynthetic thermal acclimation between boreal and temperate species. We tested the following hypotheses: (1) overall, EV and EJ increase as a function of rising growth temperature, but the magnitude of increase is greater in temperate than in boreal species; (2) correspondingly, Topt tracks growth temperature variation more closely in temperate than in boreal species; and (3) at any given leaf temperature, J/V is lower in temperate than in boreal species, and this difference is magnified in warmer climates.


Plant material

The four tree species included in this study – trembling aspen, paper birch, eastern cottonwood and sweetgum – were chosen based on their ecological attributes and native geographical ranges. In eastern North America, aspen and birch are chiefly boreal species with ranges that extend southward into northern temperate forests, whereas cottonwood and sweetgum are temperate species (Burns & Honkala 1990). All are early successional, shade intolerant, capable of rapid juvenile growth and tolerant of a relatively broad range in soil characteristics (Burns & Honkala 1990).

Aspen and cottonwood seeds were collected from southern Wisconsin (43°N 89°W), sweetgum seeds were collected from central Kentucky (38°N 84°W) and birch seeds were collected from northern Wisconsin (45°N 89°W). We chose provenances near the southern and northern range limits for our boreal and temperate species, respectively, because, relative to sources near range centres, they are more likely to: (1) possess traits that are adaptive for warmer or cooler climates, respectively; and (2) shape the genome of propagules whose fates largely determine species demographics in nearby range margins (Davis & Shaw 2001). Seeds were germinated in flats in a greenhouse on the University of Wisconsin-Madison campus in April and May of 2007. Air temperature in the greenhouse was maintained at 23 °C (± 3 °C) both day and night. Germinants were transplanted from flats into 0.5 L pots containing, on a relative volume basis, two parts peat, one part sand and one part field soil (silt loam). All seedlings were watered daily and top-dressed with a controlled release fertilizer (Osmocote, 15-9-12, Scott's, Marysville, OH, USA). Seedlings were grown in the greenhouse to a height of 10–20 cm, at which time (late May) they were out-planted to field gardens (see below). Out-planting was completed by 15 June.

Garden descriptions

Three common gardens were established along a ∼900 km latitudinal transect from northern Wisconsin to southern Illinois (Table 1). The gardens were located on University of Wisconsin Agricultural Experiment Stations near Rhinelander, WI (45°N 89°W) and Arlington, WI (43°N 89°W), and on the University of Illinois Agricultural Center near Dixon Springs, IL (37°N 88°W). Hereafter, the common gardens are referred to as northern Wisconsin, southern Wisconsin and Illinois, respectively. All gardens were located in former agricultural fields. The northern Wisconsin garden, underlain by a loamy sand (Vilas series, Entic Haplorthod), had been maintained as a mixture of clover and winter wheat. The southern Wisconsin garden, underlain by a silty clay loam (Plano series, Typic Argiudoll), was previously maintained as a mixture of native grasses and forbs. The Illinois garden, underlain by a silt loam (Zanesville series, Typic Fragiudolf), was maintained in fescue grass prior to the study.

Table 1.  Intercept (a), slope (b) and curvature (c) of the quadratic equation (y = a + bX + cX2) fit to responses of mesophyll conductance (gm, µmol m−2 s−1 Pa−1) to leaf measurement temperature (Tleaf,°C) for each species and garden
  1. Significance of the quadratic model fit for a given species and garden is indicated by Pmodel. Significance of variation among gardens in slope, curvature or intercept is indicated by Pgarden.

 Northern WI−0.80.112−0.0002<0.01
 Southern WI−9.20.703−0.01060.02
 Northern WI−5.20.364−0.0045<0.01
 Southern WI1.60.065−0.00180.16
 Northern WI−2.60.260−0.0020.01
 Southern WI23.4−1.3880.02370.51
 Northern WI−7.60.603−0.00970.31
 Southern WI−3.40.319−0.00450.12

In early spring of 2007, all gardens were treated with herbicide to eliminate extant vegetation, disced and tilled. Landscape cloth was installed to inhibit weed growth, and seedlings were planted through the cloth. Each garden was fenced with poultry netting (1 m height) and multistrand electric fencing (2 m height) to prevent mammal herbivory. All seedlings were watered twice each week throughout the study, and were top-dressed initially with a controlled release fertilizer (Osmocote, 15-9-12, Scott's) to minimize potential garden differences in soil fertility. Gardens were hand-weeded periodically. Beginning in early July, air temperature and photosynthetic photon flux (PPF) were monitored at each garden using shielded thermocouples and quantum sensors (LI-190, Li-Cor Biosciences, Lincoln, NE, USA) located 1 m above the soil surface. Instantaneous measures of temperature and PPF were recorded with a data logger (CR10X, Campbell Scientific Inc., Logan, UT, USA) every 10 min throughout the study.

Measurement of leaf gas exchange and fluorescence

Prior to leaf measurements, trees were allowed to acclimate to their respective growth environments for at least 8 weeks after out-planting to minimize possible legacies from the greenhouse. Trees grew rapidly during that period, and, at the time of gas exchange measurements, less than 10% of the tree crown was composed of foliage that had developed in the greenhouse. In early to mid-August of 2007, gas exchange was measured on leaves from 4 trees per species at each garden, using a LI-6400 portable photosynthesis system (Li-Cor Biosciences) with a fluorometer (LI-6400-40) attached to the sensor head. For each leaf, net photosynthesis (A) was measured under saturating PPF (1800 µmol m−2 s−1, provided by a red–blue LED array) at several CO2 partial pressures (pCO2) ranging from 7.5 to 120 Pa (controlled with the LI-6400 CO2 injector system). Photosynthesis was assessed first at a cuvette reference pCO2 of 40 Pa, and then again after each of three step-wise decreases in pCO2 (i.e. at 25, 15 and 7.5 Pa), and then at 60, 90 and 120 Pa CO2, respectively. The potentially confounding influences of diffusion leaks across the cuvette gasket on gas exchange calculations were taken into account after applying the manufacturer's equation to determine the gasket diffusion coefficient (Anonymous 2005).

Measurements were conducted on the youngest, fully expanded leaf of each tree. At a given pCO2, leaves were allowed to acclimate to cuvette conditions for a minimum of 2 min and a maximum of 5 min depending on when stability of photosynthesis occurred. Vapour pressure deficit in the cuvette ranged from 0.9 to 4.0 kPa. For each leaf, the photosynthetic response to intercellular pCO2 (Ci) was determined at three cuvette temperatures (25, 30 and 35 °C). The sequence of cuvette temperatures at which leaves were measured was randomly chosen at the beginning of each day. The actual range in leaf temperature during measurements was 21.3–36.8 °C, whereas, within a given ACi curve, leaf temperature varied by less than 1 °C. For every observation along the ACi curve, leaf chlorophyll fluorescence was simultaneously assessed using the leaf chamber fluorometer. Measurements began when leaves were fully illuminated each morning, and continued throughout the day, as long as stomatal conductance remained comparatively high (e.g. within 25% of the daily maximum for a given cuvette temperature). At each garden, 5 d was required to complete all measurements.

Leaves were harvested immediately after the completion of gas exchange assessments, placed on ice and transported to the lab, where they were measured for projected area using a leaf area meter (LI-3100, Li-Cor Biosciences), and then dried to a constant mass at 70 °C. Dried leaves were weighed, finely ground and analysed for nitrogen concentration using an Elementar Vario Macro CHN analyser (Elementar Analysensyteme GmbH, Hanau, Germany).

Estimation of gas exchange parameters

For individual leaves, at each of three measurement temperatures, RuBP regeneration rate (J) was estimated from chlorophyll fluorescence data using the following equation (Bernacchi et al. 2002):


where ΦPSII is the photochemical efficiency of photosynthesis, calculated as 1 − Fs/Fm′ (Fs is steady-state fluorescence yield, and Fm′ is fluorescence yield after a saturating light pulse); α1 is the leaf absorbance (set to 0.86 for all species and gardens, based on ancillary data from a greenhouse study of tree response to temperature; Serbin & Dillaway, unpublished data); β is the fraction of absorbed quanta that reaches photosystem II (typically 0.5 for C3 plants, Ogren & Evans 1993); and Q is the incident PPF.

Again, for individual leaves, at each of three measurement temperatures, Vcmax and Jmax were estimated after accounting for resistances impeding CO2 diffusion into mesophyll chloroplasts. This was accomplished with an approach that combined two published methods – one for estimating mesophyll conductance (gm) (Harley et al. 1992), and another for estimating leaf biochemical properties based on the relationship between A and chloroplast pCO2 (Cc) (Ethier & Livingston 2004). Firstly, for a given leaf and measurement temperature, Γ (intercellular compensation pCO2) was determined from the lower portion of the ACi curve (when Ci < 20 Pa). Next, Π, the ratio of (non-photorespiratory) respiration (Rd) to Vcmax for an illuminated leaf, was calculated with a reformulation of component ‘c’ in the solution (Eqn 10) for the quadratic equation describing A as a function of gm and Ci in Ethier & Livingston (2004):


where O was chloroplast pO2 (assumed to equal atmospheric pO2), and Γ*, Kc and Ko were temperature-sensitive values for the chloroplastic photocompensation pCO2 and Rubisco CO2 and O2 kinetic constants, respectively, calculated from Cc-based formulae provided in Bernacchi et al. (2002). The product of Vcmax and Π was then substituted for Rd in the ‘variable J’ method described by Harley et al. (1992) for calculating gm with fluorescence-based J estimates (from Eqn 1):


Additionally, Rd was replaced by Vcmax · Π in the solution (Eqn 10) for the quadratic formula relating A, gm and Ci in Ethier & Livingston (2004). At each Ci < 20 Pa, A was then estimated with Eqn 10 in Ethier & Livingston (2004), based on a simultaneous calculation of gm with Eqn 3. This approach required the estimation of only two rather than three unknowns (Vcmax and gm, but not Rd) via a minimization of sums of squares for error (SSE) resulting from the comparison of observed and modelled A values across Cis < 20 Pa (using the ‘Solver’ function in Microsoft Office Excel, version 12, Microsoft Corp., Redmond, WA, USA).

After finding the value for Vcmax that minimized SSE, Rd was calculated as Vcmax · Π and subsequently employed to estimate gm with Eqn 3 at each Ci ≥ 20 Pa. Finally, Rd- and Ci-specific values of gm were used to derive a Cc-based estimate of Jmax (Ethier & Livingston (2004) from the upper portion of the ACi curve (when Ci > 60 Pa). Simulations of ACi curves, using various combinations of different Vcmax, Jmax, Rd and gm values across a range of leaf temperatures, verified the utility and accuracy of this approach (data not shown), which afforded Cc-based estimates of leaf biochemical traits, but, in contrast to the method of Ethier & Livingston (2004), allowed gm to vary with Ci (or Cc).

Typically, gm varied non-monotonically, by less than 1.5 µmol m−2 s−1 Pa−1, across much of the ACi curve, although it tended to decrease at the highest Cis (e.g. when Ci > 60 Pa; data not shown). When cuvette pCO2 approximated current ambient values (∼37 Pa during a summer day in 2007), gm consistently fell within the reported range for deciduous woody perennials (Ethier & Livingston 2004), and it varied among species and gardens in terms of magnitude and response to variation in leaf temperature (Fig. 1). The latter was generally parabolic in shape, and was described for each species and garden with a quadratic equation (Table 1). These equations were then used to model the temperature sensitivity of light-saturated A at ambient pCO2 (see below).

Figure 1.

Mesophyll conductance at ambient pCO2 (∼37 Pa) plotted against leaf measurement temperature for each species and garden. Individual trends were fitted with the quadratic equations reported in Table 1.

Characterization of the temperature responses of Vcmax and Jmax

The responses of Vcmax to leaf temperature were modelled using a modified Arrhenius equation, which was derived by Lloyd & Taylor (1994) and again described by Turnbull et al. (2001) and Griffin, Turnbull & Murthy (2002):


where VT0 is Vcmax at a given reference temperature T0, VT1 is Vcmax at the temperature of interest T1, EV is a value equivalent to the energy of activation for RuBP carboxylation (analogous to a Q10) and Rg is the ideal gas constant (8.314 J mol−1 K−1). Based on Vcmax values calculated at each of the three leaf measurement temperatures, VT1, VT0 and EV, were estimated for each leaf using non-linear regression (NLIN procedure, SAS Institute Inc., Cary, NC, USA). Temperature-normalized values of Jmax, as well as its energy of activation (EJ), were estimated for every leaf in a similar manner.

We evaluated the Arrhenius model, in terms of its appropriateness for our analysis, via a comparison of EV and EJ estimates generated from two different data sets – one containing all Vcmax and Jmax values, and another containing only values stemming from gas exchange measurements at the lower two cuvette temperatures (25 and 30 °C). This comparison provided a means of assessing whether full temperature responses deviated consistently from an exponential form. Estimates from the two data sets were linearly related (r2 = 0.59, P < 0.001), and there was no indication of bias, as, for both EV and EJ, the regression slope and intercept did not differ significantly from 1 and 0, respectively.

Modelling of light-saturated photosynthesis at ambient pCO2

With species- and garden-specific estimates for the photosynthetic parameters described above, we assessed the consequences of variation in photosynthetic metabolism for the sensitivity of light-saturated photosynthesis (Asat) to leaf temperature (Tleaf) at ambient pCO2 (∼37 Pa). Temperature-normalized values for Vcmax, Jmax, gm and Rd– as well as EV, EJ and temperature responses of gm and Rd– were used to calibrate a biochemical model of photosynthesis (Farquhar, von Caemmerer & Berry 1980, with refinements by Ethier & Livingston 2004). While use of the Farquhar model usually requires an estimate of Ci or Cc, we adopted an alternative approach in which photosynthesis was calculated using quadratic reformulations of the model components relying instead on ambient pCO2 and a value of stomatal conductance (gs). For each species and garden, gs was modelled based on its responses to variation in Tleaf observed during gas exchange measurements.

For a given species and garden, and across a range of Tleaf (21–37 °C), Asat was set equal to the lower of two calculated rates – one limited by RuBP regeneration and the other by Rubisco activity. The leaf temperature corresponding to maximum Asat (Topt) was then determined using the first derivative of the quadratic model fit to the relationship between modelled Asat and Tleaf (Asat = aTleaf2 + bTleaf + c), where Topt = −b/2a. The validity of this approach was examined through a comparison of observed and modelled Asat across species and gardens. The linear relationship between the two was fairly close (r2 = 0.73, P < 0.001), and the regression slope and intercept did not differ significantly from 1 and 0, respectively.

Study of potential soil effects on photosynthetic parameters

In order to assess the degree to which observed variation in photosynthetic parameters along the latitudinal transect might have been attributable to differences in garden edaphic conditions, a pot study was conducted near Madison, WI, using soil collected from each of the three gardens. Eastern cottonwood seed from southern Wisconsin was sown in flats and transplanted into 5 L pots filled with garden soils and located in full sun. Trees were top-dressed with a controlled release fertilizer similarly to those in the common gardens, and were irrigated as needed to keep the soil near field capacity. After 45 d of growth, ACi curves were generated, using the methods detailed above, with gas exchange measurements on the youngest fully expanded leaf from each of four trees at a PPF of 1800 µmol m−2 s−1 and a cuvette temperature of 25 °C. Measurements were conducted between 0900 and 1200 h under clear skies. For each leaf, Vcmax, Jmax and Asat were calculated in the same manner as described above, albeit on the basis of Ci rather than Cc, and only at a Tleaf of ∼25 °C.

Statistical analyses

For a given species, the significance of variation in leaf properties across gardens was determined with analysis of variance (anova) (procedure GLM, SAS Institute Inc.), treating individual leaves as experimental units. Variation among garden means was deemed significant when P ≤ 0.05. Linear regression was employed (Proc REG) to examine relationships between leaf properties and garden climate. The climate variable we used was an average for air temperature during the 5 d prior to gas exchange measurements at a given garden. This choice was based on: (1) our observation that, during the study period, the bulk of individual leaf development by our target tree species typically required 4–6 d; and (2) published results indicating that adjustments of leaf metabolism to climate change can occur rapidly (e.g. in a span of 1–5 d following a shift in temperature) (e.g. Teskey & Will 1999; Atkin et al. 2000; Bolstad, Reich & Lee 2003; Lee, Reich & Bolstad 2005; Hartley et al. 2006). We also note that the 5 d temperature average, hereafter referred to as growth temperature (Tgrowth), was closely correlated to each of several other climate variables, such as temperature averages for the previous day, and 2 d and 10 d (P < 0.001, data not shown). Regressions of leaf properties against climate included effects for species and species by climate interactions. Owing to the absence of true replication of a particular climate within a given species, species by garden means were used as experimental units. Relationships among leaf properties were also examined with linear regression, using individual leaves as experimental units. Regressions were deemed significant if P ≤ 0.05.


Climate variation along the latitudinal transect

Climate varied considerably along the latitudinal transect during the 2007 growing season (Table 2, Fig. 2). For example, average air temperature across the 5 d prior to leaf measurements differed between the coolest and warmest site (northern WI and Illinois, respectively) by nearly 12 °C (Table 2). The light environment also varied across sites, even though all three were in open fields. In particular, the maximum PPF decreased slightly with increasing latitude, and cloudy skies were less frequent in Illinois than in Wisconsin (Fig. 3).

Table 2.  Average air temperatures observed at the three gardens along our latitudinal transect during the study period in 2007
IntervalAverage air temperature (°C)
IllinoisSouthern WINorthern WI
  1. Values are based on instantaneous measures, using a shielded thermocouple located 1 m above the soil surface, during the 5 d prior to each leaf measurement campaign. Also included is the average of temperature measurements throughout the entire study period (from the beginning of July through the final day of photosynthesis measurements at each garden in August).

5 d Diel30.622.518.7
5 d Diurnal32.126.120.9
5 d Nocturnal27.417.916.5
Entire study28.522.219.3
Figure 2.

Temporal dynamics in air temperature at the three gardens along our latitudinal transect during the study period (July and August) of 2007. Data are instantaneous measures of air temperature, recorded every 10 min with a shielded thermocouple located 1 m above the soil surface.

Figure 3.

Diurnal distribution of photosynthetic photon flux (PPF) at the three gardens along our latitudinal transect during the study period (July and August) of 2007. Data are instantaneous measures of PPF, recorded every 10 min with a quantum sensor located 1 m above the soil surface.

Responses of Vcmax and Jmax to leaf temperature

In general, the temperature sensitivity of Vcmax (EV) and Jmax (EJ) increased from Illinois to northern Wisconsin (Table 3), and, correspondingly, each was negatively correlated with Tgrowth across species (P ≤ 0.003, r2 ≥ 0.74, Fig. 4). The decline in EJ with increasing Tgrowth was more pronounced than that in EV. Sweetgum's behaviour contrasted somewhat with that of the others, as its respective temperature sensitivities were similar in Illinois and northern WI. Nevertheless, in neither relationship (EV or EJ versus Tgrowth) did the slope or intercept differ significantly between boreal and temperate groups (P ≥ 0.23). Overall, EV and EJ were positively correlated (where EV = 45.5 + 0.62EJ, P < 0.001, r2 = 0.76; n = 12).

Table 3.  Leaf traits measured along our latitudinal transect
Trait/SpeciesIllinoisSouthern WINorthern WIPgarden
  1. Parameters include maximum rates of ribulose bisphosphate (RuBP) carboxylation (Vcmax) and RuBP regeneration (Jmax) estimated at a leaf temperature of 25 °C (V25, J25), the temperature sensitivity of Vcmax (EV) and Jmax (EJ), leaf nitrogen content (Narea) and specific leaf area (SLA). Means and standard errors (in parentheses) are based on measures from four trees (n = 4) per garden. For a given variable and species, the corresponding Pgarden indicates the significance of variation among garden means.

EV (kJ mol−1 K−1)    
 Aspen57.1 (4.5)56.3 (1.7)70.2 (1.3)0.01
 Birch57.6 (2.4)63.9 (8.5)67.1 (1.9)0.47
 Cottonwood60.2 (1.2)62.1 (7.6)68.3 (4.8)0.54
 Sweetgum74.2 (5.3)56.9 (4.2)75.8 (2.2)0.02
EJ (kJ mol−1 K−1)    
 Aspen26.5 (2.7)17.6 (1.9)34.4 (1.3)<0.001
 Birch18.8 (2.1)21.1 (4.6)38.2 (1.5)0.003
 Cottonwood23.8 (2.1)24.8 (4.9)42.9 (2.8)0.006
 Sweetgum41.8 (7.5)26.5 (1.9)44.3 (2.9)0.03
V25 (µmol m−2 s−1)    
 Aspen199.1 (13.9)228.2 (4.2)187.2 (13.9)0.08
 Birch212.5 (13.9)200.1 (27.1)196.6 (8.8)0.82
 Cottonwood219.1 (12.5)214.8 (13.7)191.9 (19.5)0.45
 Sweetgum166.3 (10.4)175.2 (6.1)136.7 (12.6)0.09
J25 (µmol m−2 s−1)    
 Aspen205.4 (9.8)273.5 (9.7)239.5 (15.8)0.01
 Birch275.9 (12.5)267.3 (24.4)243.6 (12.9)0.44
 Cottonwood268.8 (17.8)262.3 (19.3)250.4 (19.1)0.79
 Sweetgum157.4 (8.6)202.8 (10.1)187.8 (14.8)0.29
Narea (g m−2)    
 Aspen2.07 (0.05)2.22 (0.11)1.92 (0.05)0.03
 Birch2.41 (0.09)2.33 (0.18)2.37 (0.13)0.24
 Cottonwood2.09 (0.06)2.14 (0.16)2.42 (0.11)0.14
 Sweetgum1.99 (0.06)1.86 (0.04)1.81 (0.09)0.18
SLA (m2 kg−1)    
 Aspen14.7 (0.7)12.2 (0.4)20.7 (0.9)<0.0001
 Birch13.4 (0.4)13.5 (0.5)13.7 (0.5)0.59
 Cottonwood18.0 (0.4)19.0 (1.5)19.0 (0.5)0.13
 Sweetgum12.7 (0.5)17.3 (0.9)14.0 (0.6)0.001
Figure 4.

Trends in the temperature sensitivities of Vcmax[EV (a)] or Jmax[EJ (b)] with growth temperature (Tgrowth, average air temperature during the 5 d prior to leaf measurements at each garden). Species are denoted with the following symbols: trembling aspen (○), eastern cottonwood (inline image), paper birch (inline image) and sweetgum (◆). Boreal and temperate species are depicted with open and filled symbols, respectively. For all species pooled, EV = 45 322/Tgrowth2 − 3368.2/Tgrowth + 119.9 (P = 0.003, r2 = 0.74); and EJ = 80 318.9/Tgrowth2 − 6093.6/Tgrowth + 134.7 (P < 0.01, r2 = 0.86).

Relationship between Jmax and Vcmax

The ratio between Jmax and Vcmax (J/V) was strongly and negatively correlated with variation in leaf temperature (Tleaf) during gas exchange measurements (P < 0.001, Table 4). For all species except birch, the intercept of this relationship was significantly lower in Illinois than in southern and/or northern WI. The same was true across all species, except the difference in garden intercept was slight (i.e. ∼10%). Additionally, across gardens, temperate species as a whole exhibited a higher intercept and lower slope than did the boreal group (Table 4).

Table 4.  Intercepts and slopes of linear relationships between J/V (ratio of Jmax to Vcmax) and leaf measurement temperature (Tleaf, °C) for each species and garden, for each garden (across species) and for boreal and temperature groups (across gardens)
  1. All models fit for a given species, garden or group were significant (P < 0.001, r2 ≥ 0.74). Probabilities Pgarden and Pgroup indicate the significance of variation among gardens or groups, respectively, in model intercept or slope.

 Northern WI2.49−0.049
 Southern WI2.69−0.058
 Northern WI2.44−0.047
 Southern WI2.85−0.060
 Northern WI2.69−0.055
 Southern WI2.70−0.058
 Northern WI2.47−0.046
 Southern WI2.18−0.040
All species  
 Northern WI2.53−0.049
 Southern WI2.62−0.055
All gardens  

Relationship between photosynthetic parameters and leaf nitrogen

Variation among gardens in leaf nitrogen content (Narea) was modest for a given species (e.g. <20%, Table 3). However, both V25 and J25 were positively, albeit weakly, related to Narea, based on an analysis pooling all individual leaf data (where V25 = 132 + 15.5(Narea), P < 0.021, r2 = 0.11; J25 = 145 + 23.7(Narea), P < 0.003, r2 = 0.17; n = 48). Neither EV nor EJ was correlated with Narea (P ≥ 0.35, data not shown).

Photosynthetic temperature optimum and its relation to other leaf traits

The optimal temperature (Topt) for light-saturated photosynthesis at current ambient pCO2 (Asat) always fell within the range of leaf temperatures observed during gas exchange measurements (Fig. 5), and it did not vary significantly across species and/or gardens (P ≥ 0.07, Fig. 6). Contrary to expectations, Topt was not correlated with EV alone. Rather, a significant amount of Topt variation was explained only by a combination of EV and the slope (b) plus curvature (c) of ambient gm's temperature response (Table 1), where Topt = 0.107EV + 39.4b + 2248.5c (P < 0.001, r2 = 0.92, n = 12). Moreover, across species and gardens, there was a close correlation between estimates of Asat at Topt and ambient gm at Topt (where Asat = 7.1 + 7.98gm, P < 0.001, r2 = 0.78, n = 12). Correspondingly, Asat was positively related to gm across all species, gardens and measurement temperatures (where Asat = 14.91gm0.604, P < 0.001, r2 = 0.74, n = 144). In light of this, we calculated the limitation on Asat imposed by gm using methods outlined by Yamori et al. (2006) and Warren (2008). The estimated limitation (l) ranged from 0 to 64% among all gas exchange measurements, and it was negatively correlated with gm (where l = −22.9ln(gm) + 41.9, P < 0.001, r2 = 0.68, n = 144).

Figure 5.

Modelled relationships between light-saturated photosynthesis at current ambient pCO2 (Asat) and leaf temperature (Tleaf), where Asat was calculated based on estimates of photosynthetic parameters generated for each species and garden. Gardens are denoted with the following symbols: northern WI (○), southern WI (inline image) and Illinois (●).

Figure 6.

The optimal temperature of photosynthesis (Topt) plotted against average growth temperature (Tgrowth) during the 5 d prior to leaf measurements. Species are denoted with the following symbols: aspen (○), eastern cottonwood (inline image), birch (inline image) and sweetgum (◆), with boreal species depicted as open symbols and temperate species as filled symbols. The relationship between Topt and Tgrowth was not significant (P = 0.07).

Soil influences on photosynthetic metabolism

No estimated photosynthetic parameter (Vcmax, Jmax or Asat) differed (P ≥ 0.39) among potted cottonwood trees grown in soils from the three gardens (Table 5).

Table 5.  Leaf photosynthetic parameters for potted cottonwood trees grown in soils from each of the three gardens along our latitudinal transect
TraitIllinoisSouthern WINorthern WIPsoil
  1. Parameters include the maximum rate of ribulose bisphosphate (RuBP) carboxylation (Vcmax), the maximum rate of RuBP regeneration (Jmax) and an estimate of light-saturated photosynthesis (Asat). Values are calculated at a leaf temperature of 25 °C. Trees were grown in a common garden near Madison, WI, for approximately 45 d prior to measurement. Means and standard errors (in parentheses) are based on n = 4 trees. For a given variable, the corresponding Psoil indicates the significance of differences among soil means.

Vcmax (µmol m−2 s−1)78.7 (7.0)75.8 (4.7)76.7 (2.8)0.92
Jmax (µmol m−2 s−1)200.4 (11.7)182.4 (5.9)192.1 (7.7)0.39
Asat (µmol m−2 s−1)25.3 (1.8)24.8 (1.2)25.9 (1.1)0.76


In general, results of this study provide little support for our hypotheses regarding differential adjustments of photosynthetic metabolism to climate in boreal versus temperate species. Namely, based on the criteria we adopted, photosynthetic thermal acclimation along our latitudinal transect was, for the most part, lacking in all species. This was exemplified by the absence of a positive relation between either EV or EJ and Tgrowth. Indeed, the only observed behaviour resembling an acclimatory response was the slight overall decline in temperature-normalized J/V with increasing Tgrowth.

Although uncommon, the EV trends observed along our transect are not unprecedented in the literature. While positive relationships between EV and growth temperature have been found in various studies (e.g. Hikosaka, Murakami & Hirose 1999; Yamori et al. 2005; Hikosaka et al. 2006; Bauerle, Bowden & Wang 2007; Ishikawa et al. 2007; Kositsup et al. 2009), EV has responded minimally to climate variation in several others (e.g. Medlyn et al. 2002a; Bunce 2008; Way & Sage 2008), and negative trends similar to ours have also been reported (e.g. Medlyn, Loustau & Delzon 2002b; Kattge & Knorr 2007; Warren 2008). This same body of literature also contains a lack of empirical consistency with respect to the temperature responses of EJ and J/V. For example, several studies have reported marked decreases in temperature-normalized J/V with increasing growth temperature (Hikosaka et al. 1999, 2006; Yamori et al. 2005; Kattge & Knorr 2007), whereas a number of others corroborate our finding that the ratio varies modestly, if at all, across climate gradients (Ferrar, Slayter & Vranjic 1989; Bunce 2000; Medlyn et al. 2002b; Ishikawa et al. 2007; Way & Sage 2008).

Particularly with respect to J/V, the amount and allocation of nitrogen in a leaf are often examined as a possible determinant of climate-mediated adjustments in photosynthetic biochemistry (e.g. Medlyn et al. 2002b; Hikosaka et al. 2006; Ow et al. 2008; Way & Sage 2008; Kositsup et al. 2009). In our study, variation in leaf nitrogen content appeared to play only a minor role in photosynthetic adjustments. In any case, the mechanisms underlying such adjustments remain unclear, although there is evidence that those concerning Vcmax may revolve primarily around modulation of Rubisco's activation state at high temperatures (Cen & Sage 2005; Sage et al. 2008; Way & Sage 2008).

Regardless of its underlying causes, variation in photosynthetic metabolism along our transect contributed to differences in key facets of photosynthetic performance under current ambient pCO2. This was manifested by the positive influences of EV and V25 on Topt and Asat at Topt, respectively. Both of these relationships are anticipated from a theoretical standpoint (Hikosaka et al. 2006; Sage et al. 2008), and have emerged to varying degrees in other empirical analyses, depending on the exact shape of Vcmax's temperature response, along with the manner in which other factors, such as Rd, covary with that trend (e.g. Medlyn et al. 2002b; Hikosaka et al. 2006; Ishikawa et al. 2007; Kattge & Knorr 2007; Way & Sage 2008). While most attention in these analyses is usually focused on acclimatory shifts in Topt, we concur with others (e.g. Berry & Bjorkman 1980; Medlyn et al. 2002b; Kositsup et al. 2009) that variation in photosynthetic rate at that optimum temperature is equally critical for homeostasis of plant carbon balance across climate gradients.

Importantly, the influences of EV and V25 on photosynthetic temperature responses along our transect were muted considerably owing to the often substantial limitation on Asat imposed by gm. The close coupling of Asat and gm is apparent in a comparison of trends in Figs 1 and 5. In aspen, for instance, Asat in northern Wisconsin exceeded that in Illinois – especially at higher temperatures – and the same was true for gm. On the other hand, there was no clear correspondence between these trends and garden-to-garden variation in EV or V25 (Table 3). In fact, results of simulations across species and gardens confirmed that the anticipated response of Topt to a change in EV[∼0.5 °C (J mol−1 K−1)−1] was diminished 60% or more by the observed behaviour of gm (data not shown). In light of this, we agree with Warren (2008) that the benefits of photosynthetic thermal acclimation may often be minimized or obviated by gm and its temperature response.

Collectively, our findings, in conjunction with those of others (e.g. Medlyn et al. 2002b; Warren 2008; Way & Sage 2008), point to the possibility that photosynthetic thermal acclimation may not occur as commonly as was once thought, at least among tree species. Moreover, to date, there is no indication that mesophyll conductance acclimates to growth temperature (Yamori et al. 2006; Warren 2008). Consequently, many species may not adjust photosynthetic metabolism or mesophyll conductance in a manner that maximizes leaf carbon gain in increasingly warm environments. This would certainly have important implications for forest response to anticipated climate change (IPCC 2007). We acknowledge, however, that many facets of plant physiology, morphology and resource allocation are typically responsive to climate variation (e.g. Tjoelker, Olkesyn & Reich 1998; Loveys et al. 2002; Atkin, Scheurwater & Pons 2006). Hence, the ramifications of variation in leaf photosynthetic metabolism for tree carbon balance along our transect are fully revealed only after a comprehensive analysis of growth and its physiological, morphological and allocational determinants, which is the topic of forthcoming publications.


We would like to thank Bronwyn Aly of the University of Illinois for help with the southern Illinois garden, and Bryan Bowen (University of Wisconsin-Madison) for help with the northern Wisconsin garden. We appreciate the help of three anonymous reviewers in improving the quality of this paper. This study was supported by the National Science Foundation (award #0802729) and McIntire–Stennis formula funding (project #WIS01227).