Needle age and season influence photosynthetic temperature response and total annual carbon uptake in mature Picea mariana trees

Transcription

1 Annals of Botany 116: , 215 doi:1.193/aob/mcv115, available online at Needle age and season influence photosynthetic temperature response and total annual carbon uptake in mature Picea mariana trees Anna M. Jensen*, Jeffrey M. Warren, Paul J. Hanson, Joanne Childs and Stan D. Wullschleger Climate Change Science Institute, Environmental Sciences Division, Oak Ridge National Laboratory, Oak Ridge, TN , USA *For correspondence. Present address: Department of Forestry and Wood Technology, Linnaeus University, Växjö, Sweden. Received: 6 December 214 Returned for revision: 6 February 215 Accepted: 22 June 215 Published electronically: 28 July 215 Background and Aims The carbon (C) balance of boreal terrestrial ecosystems is sensitive to increasing temperature, but the direction and thresholds of responses are uncertain. Annual C uptake in Picea and other evergreen boreal conifers is dependent on seasonal- and cohort-specific photosynthetic and respiratory temperature response functions, so this study examined the physiological significance of maintaining multiple foliar cohorts for Picea mariana trees within an ombrotrophic bog ecosystem in Minnesota, USA. Methods Measurements were taken on multiple cohorts of needles for photosynthetic capacity, foliar respiration (R d ) and leaf biochemistry and morphology of mature trees from April to October over 4 years. The results were applied to a simple model of canopy photosynthesis in order to simulate annual C uptake by cohort age under ambient and elevated temperature scenarios. Key Results Temperature responses of key photosynthetic parameters [i.e. light-saturated rate of CO 2 assimilation (A sat ), rate of Rubisco carboxylation (V cmax ) and electron transport rate (J max )] were dependent on season and generally less responsive in the developing current-year (Y) needles compared with 1-year-old (Y1) or 2-year-old (Y2) foliage. Temperature optimums ranged from 187 to237, 313 to383 and 287 to367 CforA sat, V cmax and J max, respectively. Foliar cohorts differed in their morphology and photosynthetic capacity, which resulted in 64 % of modelled annual stand C uptake from Y1&2 cohorts (LAI 67 m 2 m 2 ) and just 36 % from Y cohorts (LAI 52 m 2 m 2 ). Under warmer climate change scenarios, the contribution of Y cohorts was even less; e.g. 31 % of annual C uptake for a modelled 9 C rise in mean summer temperatures. Results suggest that net annual C uptake by P. mariana could increase under elevated temperature, and become more dependent on older foliar cohorts. Conclusions Collectively, this study illustrates the physiological and ecological significance of different foliar cohorts, and indicates the need for seasonal- and cohort-specific model parameterization when estimating C uptake capacity of boreal forest ecosystems under ambient or future temperature scenarios. Key words: Black spruce, Picea mariana, climate change, photosynthesis, temperature adjustment, carbon assimilation, A/C i curve, leaf age, Q 1, evergreen, SPRUCE project, STELLA model, respiration. INTRODUCTION Moderate climate warming is expected to prolong the growth season in the boreal zone. For woody plants, the combination of elevated air and soil temperatures will entail shifts in foliar ontogeny earlier onset of bud break, resumption of photosynthesis and acceleration of foliar maturation (Troeng and Linder, 1982; Goulden et al., 1997; Mäkelä et al., 24; Goodine et al., 28; Fløistad and Granhus, 21; Sutinen et al., 212) all changes that would increase the net annual carbon (C) uptake. In addition, greater C assimilation rates across a warmer growth season may also initially result from photosynthesis operating closer to a localized species-specific photosynthetic temperature optimum (T opt ) (Berry and Björkman, 198; Way and Sage, 28a, b; Lin et al., 213; Way and Yamori, 214; Yamori et al., 214). Whether these positives will compensate for C losses due to increased rates of respiration with rising temperature is still unclear. Identifying and quantifying seasonal, canopy and ontogenetic variation in photosynthetic parameters is essential for modelling species and ecosystem-specific sensitivities to climate changes, including important aspects related to warming. Whereas the seasonal and canopy variations in photosynthetic capacity, such as the maximum assimilation rate (A sat ), the maximum Rubisco carboxylation rate (V cmax ) and the maximum rate of electron transport (J max ), have been studied intensively across multiple species and functional groups (e.g. Wullschleger, 1993; Niinemets et al., 24), the underlying photosynthetic temperature responses have often been assumed to be constant. This assumption persists due to lack of data, even though factors known to affect the thermal adjustment of photosynthesis, such as air temperature and nutrient or water availability, are known to vary between seasons, canopy position and foliar age (Niinemets, 22; Grassi et al., 25; Gunderson et al., 21; Salmon et al., 211). Published by Oxford University Press on behalf of the Annals of Botany Company 215. This work is written by US Government employees and is in the public domain in the US.

2 822 Jensen et al. Photosynthetic responses to temperature in black spruce Temperature not only affects the immediate biochemical and enzymatic activities involved in photosynthesis but also drives short- and long-term temperature acclimation (Stinziano and Way, 214). Higher growth temperatures often result in greater T opt for key photosynthetic parameters, such as V cmax and J max (Berry and Björkman, 198; Way and Sage, 28a). Several studies have looked at thermal regulation/adjustment between seasons and across a latitudinal temperature gradient (e.g. Hikosaka et al., 27; Dillaway and Kruger, 21), whereas few studies have investigated effects of ontogeny or seasonality on T opt of these key photosynthetic parameters (Medlyn et al., 22a; Gunderson et al.,21). Here we identify seasonal and ontogenetic variation of the photosynthetic temperature responses in Picea mariana (black spruce) to explore the physiological/ecological significance of maintaining multiple foliar cohorts. In P. mariana, as in many evergreen trees, the maintenance of multiple foliar cohorts is considered an adaptive trait to conserve nutrients but also to enable early-season C assimilation prior to and during shoot flushing (Troeng and Linder, 1982; Greenway et al., 1992; Öquist and Huner, 23). The latter may be especially important for boreal conifers, in which bud break occurs in May June and new foliage is only fully mature late in the growing season (Teskey et al., 1984). For such conifers, a combination of elevated air temperatures and altered precipitation regimes associated with global warming may enhance total seasonal CO 2 assimilation, thus potentially reducing the physiological significance of maintaining older foliar cohorts (Dang and Lieffers, 1989; Way and Sage, 28b). The purpose of the study was to identify and quantify seasonal, canopy and cohort C assimilation patterns of a P. mariana stand, using seasonal- and cohort-specific temperature responses. To meet these objectives we: (1) characterized seasonal temperature responses of key photosynthetic parameters (i.e. A sat, V cmax and J max ) and daytime foliar dark respiration (R d ) of two successive P. mariana needle cohorts; (2) characterized spatial differences in the key photosynthetic parameters R d, foliar morphology and nitrogen (N) status; and (3) used these data, in combination with site-specific climatic and allometric data, as model input to project annual tree canopy C uptake for the stand using dynamic cohort-specific photosynthetic temperature response functions in a simple modelling framework. The model was used to project annual net C exchange for three temperature scenarios [þ (ambient),þ45 and þ9 C] to evaluate the importance and relative influence of foliar cohort differences on total annual C gain for warmer climate scenarios. Study site MATERIALS AND METHODS The study was conducted in a mixed peatland Picea mariana (black spruce) and Larix laricina (tamarack) stand at the Marcell Experimental Forest in northern Minnesota, USA ( N, W). Elevation is 41 m above sea level. The stand was established by natural regeneration following two strip-cuttings in 1969 and 1974 (Sebestyen et al., 211). The evaluated bog (S1 Bog) had an open canopy structure, with an average tree height of 45 m and a basal area of 56cm 2 m 2 in February 213. At the site, P. mariana trees retain on average five foliar cohorts. Bud break occurs during June and the newly developed foliage (Y) becomes fully expanded towards the end of July and the beginning of August. The understorey consists of woody shrub species [e.g. Rhododendron groenlandicum (Labrador tea) and Chamaedaphne calyculata (leatherleaf)] and a ground layer of moss species (e.g. Sphagnum spp. and Polytrichum spp.) and herbaceous species [e.g. Eriophorum angustifolium (cotton grass) and Maianthemum trifolium (three-leaf false Solomon s seal)]. The climate is continental, with a 4-year average T air of 15 and 19 C during January and July, respectively, and annual precipitation averages 78 mm (Sebestyen et al., 211). Site-specific climatic conditions (T air, relative humidity, photosynthetically active radiation (PAR) and wind speed) at 2 m were measured at the south end of S1 Bog, 2 2 m from measurement trees (Table 1). Our study site is part of a longterm climate change experiment to study ecosystem response to elevated CO 2 and temperature: Spruce and Peatland Responses Under Climatic and Environmental Change (SPRUCE; mnspruce.ornl.gov/). Beginning in 215, the experimental plots will be exposed to elevated CO 2 (4 and 9 ppm) and temperatures (þ, þ45 andþ9 C). We expect the treatments to change soil water availability, photosynthesis, respiration and likely relative species composition. Here we studied spatial and seasonal pretreatment patterns of photosynthetic capacity in mature P. mariana trees. Gas exchange Seasonal patterns of the photosynthetic capacity of currentyear (Y) and 1-year-old (Y1) foliage were measured in April, May, July, September, August and October between 21 and 213. We combined data across years, as both year of collection and the interaction between season and year of collection had no significant effect (P > 5) on photosynthetic patterns. Influences of canopy position and needle age on photosynthesis were measured for three successive needle cohorts [Y, Y1 and 2-year-old (Y2)] from top, middle and bottom branches in July 211. Photosynthetic response (A/C i ) curves were produced using portable infrared gas analyser systems (LI64XT, Li- COR, USA). We generated A/C i curves at saturating light levels (PAR ¼ 1 lmol m 2 s 1 ) by sequential adjustment of the reference CO 2 concentration between 5 and 16 ppm (4, Table 1. Observed environmental data for 211, 212 and 213 at S1 Bog in northern Minnesota, USA. All data are based on half-hour data collected 2 2 m from measured trees. Climatic data collection began in late June 21; annual values are therefore not available for 21 Variable Mean annual T air ( C) Maximum annual T air ( C) Minimum annual T air ( C) Mean annual relative humidity (%) Cumulative incident PAR (mol quanta m 2 year 1 ) Annual precipitation (mm) Mean wind velocity (m s 1 ) 6 5 6

3 Jensen et al. Photosynthetic responses to temperature in black spruce 823 3, 2, 1, 5, 4, 55, 65, 8, 95, 12 and 16 ppm) to assess A across a broad range of intercellular CO 2 (C i ) concentrations. Temperature response surfaces were generated by replication of the A/C i curves at multiple temperatures. Cuvette air temperature was regulated from 2 to 5 C by Peltier adjustment of the block temperature and by use of an external temperature-controlled circulating water bath attached to the cuvette with block water jackets (LI-COR model 64-88). Temperature increments were done in steps of 3 5 C. We measured and included both full (2 5 C) and partial (2 25 Cand 2 5 C) temperature series. It took our set-up 5min to reach target temperature, except at low (2 5 C) and high (4 5 C) temperatures, where additional time was needed. We used air temperature here, as the spruce needles were not always in direct contact with the leaf thermocouple, as is the case when making measurements on broadleaved species. Leaf and air temperatures differed on average by 5 C (ranging from 2 to43 C) in the chamber. Rates were recorded 5 1 min after target temperatures were reached. Values of vapour pressure deficit (VPD) were calculated based on air temperature and relative humidity in the cuvette, and varied from 3 to 86 kpa (average 19 kpa). We minimized variation in VPD by regulating humidity manually (adjusting the scrub). At high temperature we used a compact travel humidifier to increase humidity of the inlet air. Humidity was adjusted prior to but not during the automated A/C i curve measurements. In total, 38 A/C i curves were used. Temperature responses of foliar daytime R d were measured in a similar fashion but using a darkened conifer chamber (April and September 213), with several needle cohorts within the same chamber. Rates of respiration were recorded at PAR ¼ lmol m 2 s 1 after the signal had stabilized (typically after 5 1 min). Following gas exchange measurements, foliage and branch material was shipped on ice to Tennessee, USA for analysis of leaf area, leaf mass and N content. Shoots were defoliated and projected leaf area was estimated using WinRHIZO 212b (Regent Instruments Canada Inc., Canada) and ImageJ 147 (Rasband, 212) software. Tissue was dried at 7 C then analysed for leaf N content by mass (N m ) and carbon content using an elemental analyser (Costech Analytical Technologies, Inc., USA). We calculated nitrogen per leaf area (N a )asn a ¼ N m leaf mass per area (LMA). Tree allometry To estimate canopy foliar mass distribution, eight trees 4 6 m tall were harvested during June/July 21 and 211. Each tree was divided into 1-m stem sections and dried, such that foliar mass represented all cohorts of needles within each 1-m-stem increment. Additionally, to determine the proportion of total foliar mass represented by each cohort, 15 trees were sampled in October 212. A total of 45 branches were collected at the south side of the tree at three different positions within the canopy. Foliage on each branch was separated by cohort for analysis of foliar and woody mass, LMA and N content. Data analysis We obtained values for V cmax and J max for each individual A/C i curve using the platform LeafWeb ( LeafWeb utilizes the Farquhar von Caemmerer Berry method with a novel exhaustive dual optimization approach integrating mesophyll conductance (g m )(Gu et al., 21). Values of A sat (PAR 1 lmol m 2 s 1,CO 2 4 ppm) were obtained from individual A/C i curves. By using our protocol, we generated two values of A sat per A/C i curve (see above). The two values were oftensimilar(2 6%)butwhentheydiffered(>25 %) we selected the observation that aligned best with the full A/C i curve. Only a single value of A sat was retrieved per A/C i curve. We parameterized the temperature response of A sat, V cmax and J max according to Medlyn et al. (22b) by fitting eqns (1) and (2) to gas exchange data collected from Y and Y1 needles in August and October: fðt k Þ ¼ Pð 25 CÞ exp h i! EaðT k 298Þ 298 R T k H d exp fðt k Þ ¼ P ðoptþ 1 þ exp 298 DS H d 298 R 1 þ exp HaðT k ToptÞ T k R Topt T k DS H d T k R 1 C A (1) C H d H a 1 exp H d ðt k T opt Þ A (2) T k R T opt P (25 C) is the specific parameter (A sat, V cmax and J max )at25 C (298 K), E a describes the exponential part of the function or the activation energy, R is the gas constant (83143 J K 1 mol 1 ), T is air temperature in K, DS (J K 1 mol 1 ) is an entropy term and H d (J mol 1 ) is the deactivation energy. We kept H d constant at 2 kjmol 1 to ensure model stability (Medlyn et al., 22b). The second part of eqn (1) adjusts for loss of enzyme activity at higher temperatures (Leuning, 22). Equation 2 includes the parameter T opt, which is related to DS; seedreyer et al. (21) and Medlyn et al. (22b) for a full description of the method. All values of foliar R d were measured directly. Foliar R d Q 1 values were calculated according to Linder and Troeng (1981) using a linear regression model from the natural logarithm of R d [ln(r d ) ¼ a þ b T air ]. 1 Q 1 ¼ exp ð1bþ (3) As Q 1 values may be dependent on the underlying temperature range used (Atkin et al., 25; Kruse et al., 28), we normalized R d values to 25 C(R d25 C)usingeqn (4),withaQ 1 value calculated between 15 and 3 C using our September 213 temperature response data. R d25 C ¼ R dt Q 1 ½25 TÞ=25Š Seasonal-, canopy- and cohort-specific values of A sat, V cmax and J max were normalized to 25 C (A sat25 C, V cmax25 C and J max25 C)byparameterizationofeqn (1), according to season and needle cohort (Table 3). Photosynthetic parameters for Y2 needles were normalized using the response of Y1 needles. We make this assumption based on similar photosynthetic capacities of Y1 and Y2 needles in July (see Results section, Fig. 4). Effects of needle age (Y), canopy position (C), seasonality (M) and their interactions were evaluated using an ANCOVA model, using a Tukey Kramer honest significant difference (HSD) test to separate means. Pseudoreplication was accounted (4)

4 824 Jensen et al. Photosynthetic responses to temperature in black spruce for when repeated measurements occurred (e.g. needle cohorts from the same branch or tree). We used an ordinary least squares model to test relationships between temperatures, foliar N, needle age and R d, V cmax25 C, J max25 C, foliar mass, LMA and C:N ratios. Model assumptions were tested using a Shapiro Wilk normality test and a Bartlett test of homogeneity of variances. Test outputs are given as R 2 and F d.f.,n,withanindication of significance. Data analyses were carried out in R version 3 ( All data (environmental, allometric and physiological) are available at gov (Jensen et al.,215). Estimating annual carbon uptake by foliar cohorts We estimated the net annual total foliar C uptake contribution of each individual foliar cohort of P. mariana trees by using the derived seasonal- and cohort-specific temperature response functions applied to a simple model of canopy photosynthesis. Annual net C uptake was interpolated and integrated over a calendar year using an hourly time step model for P. mariana cohorts coded for each cohort class using STELLA 13 modelling software (ISEE Systems Inc., NH, USA). This photosynthetic module has been successfully used as a component within an ecosystem level C and water cycle model (INTRASTAND; for more information see Hanson et al., 24, 25) and was used here only to estimate net foliar C uptake based on site-specific photosynthetic and R d measurements and environmental drivers (Tables 1 and 2). The photosynthetic module estimated annual C uptake from the coupled Farquhar/ Ball Berry photosynthetic and stomatal conductance model as described by Harley et al. (1992), incorporating the temperature response functions (eqn 1). Farquhar model and R d variables were obtained from the gas exchange measurements described above (Table 2). Using values of daytime R d likely underestimates extrapolated night-time respiratory C losses, as mitochondrial respiration can be suppressed by light (Atkin et al., 25). Measured half-hourly site meteorological conditions at 2 m (T air, relative humidity and PAR) were applied as the effective conditions at the leaf surface (Table 1). Daytime wind velocities at the site averaged 6ms 1, which reduced the boundary layer conductance sufficiently to support this assumption. We also assumed that T responses were constant for different light levels. This protocol ensured integration of both diurnal and seasonal conditions, including short-term changes due to clouds or rain events. Given the open nature of the S1 Bog forest, all foliage was assumed to be fully sunexposed for the purposes of annual C gain calculations. There certainly was some shading of the older cohorts, especially those >3 years old, which were assumed not to be a significant component of C uptake due to their low residual biomass. Canopy foliar biomass per unit ground area was derived from mass allometric relationships developed specifically for the SPRUCE site for trees with diameter >13m at breast height. Site-specific tree diameter data for 21 defined plot areas (113 m 2 plot 1 ) combined with the allometric relationship data provided a ground-area-based estimate of total standing above-ground biomass for February 213. Live foliar mass data multiplied by corresponding leaf mass per unit area yielded the leaf area index (LAI) for the defined annual Table 2. Needle cohort and physiological and mass characteristics used in the annual C estimation. Values of A sat, V cmax and J max are normalized to 25 C according to season and needle age Variable Y1&2 Y V cmax25 C DOY DOY DOY >18 71 DOY DOY >24 36 J max25 C DOY DOY DOY > DOY DOY > g o g l R d25 C DOY59 14 DOY DOY > Q 1DOY DOY DOY > LAI (m 2 m 2 ) Needle mass (g m 2 ground) LMA (g m 2 needle) DOY, day of year. cohort(s) (Table 2). At the SPRUCE site, foliage older than three cohorts makes up only a small fraction of P. mariana foliar mass (7 %). Because such older foliage has reduced photosynthetic capacity (see also Niinemets, 22), these cohorts were not considered an important contribution to P. mariana annual C uptake. Net C uptake was modelled for three different temperature assumptions [þ (ambient), þ45 and þ9 C] using site-specific environmental data from 211, 212 and 213. These temperature scenarios were chosen based on the temperature treatments within the SPRUCE experiment, enabling future physiological and model comparisons. While 45 C is within the expected range of future temperature for the boreal north, 9 C is high, but used here to determine potential threshold responses. Total annual C uptake was calculated for the sum of individual cohort groups (multiple foliar cohort simulation), but was also estimated for a simplified single cohort (Y1) to judge the significance of multi-cohort data (single foliar cohort simulation). RESULTS Climate and shoot phenology Climatic conditions between 211 and 213 were similar to the 4-year monthly means reported by Sebestyen et al. (211). Measured mean annual T air ranged from 22 to52 C, with maximum and minimum temperatures around 34 and 36 C (Table 1). Annual cumulative PAR averages were 7929, 8162 and 775 mol quanta m 2 year 1 in 211, 212 and 213, respectively, whereas annual precipitation averaged 645, 66 and 586 mm in 211, 212 and 213, respectively (Table 1).

5 Jensen et al. Photosynthetic responses to temperature in black spruce 825 Table 3. Parameters of the temperature response of A sat, V cmax, J max and TPU by cohort and season 1. Rates were generated using eqns (1) and (2) and given on projected leaf area. All estimated parameters, with the exception of A sat E a of Y1 needles, were significant (P51) P (25 C) (lmol m 2 s 1 ) P (opt) (lmol m 2 s 1 ) H d (kj mol 1 ) E a (kj mol 1 ) DS (J mol 1 ) T opt ( C) Residual s.e. (P (25 C)/P (opt) ) D.f. (P (25 C)/P (opt) ) A sat (lmol m 2 s 1 ) August Y1 92(3) 93(3) 2 3(25) (31) /18 62/64 Y 89(4) 87(2) 2 67(31) 6399 (23) /15 7/72 October Y 64(4) 59(2) 2 129(49) 6439 (24) /11 6/62 V cmax (lmol m 2 s 1 ) August Y1 77(23) 121(28) 2 37(32) 628 (23) /144 6/62 Y 334(13) 386(12) 2 183(34) 631 (27) 35 8/79 7/72 October Y 355(2) 48(14) 2 498(78) 6478 (25) /77 6/62 J max (lmol m 2 s 1 ) August Y1 1128(22) 1351(27) 2 176(21) 6259 (23) /164 55/57 Y 71(2) 85(18) 2 144(23) 6256 (28) /12 62/64 October Y 938(48) 929(257) 2 325(57) 649 (25) /118 31/33 TPU (lmol m 2 s 1 ) August Y1 85(2) 99(2) 2 159(19) 6238 (25) /12 55/57 Y 52(2) 62(2) 2 161(26) 624 (28) /11 7/72 October Y 7(2) 62(2) 2 332(38) 6472 (15) 296 9/9 6/62 TPU, triose phosphate utilization. 1 Parameters for April were not estimated due to low responsiveness. 2 P-values > 5. Growing season length [frost-free period (Monson et al., 25; Viereck and Johnston, 199)] averaged 175 d between 21 and 213 at S1 Bog. Seasonal temperature response of photosynthesis Season and needle cohort affected photosynthetic (A sat, V cmax and J max ) responsiveness to temperature and CO 2,typically displaying flat response surfaces early in the season and in younger foliage (Fig. 1A I). Table 3 gives the estimated parameters of the temperature response of A sat, V cmax and J max according to season and needle age. In April, when air temperatures were low, photosynthetic capacity (A sat, V cmax and J max ) of overwintering needles (Y1) showed a low responsiveness to CO 2 and temperature, compared with later in the season (Fig. 1, Table 3); as a result, temperature responses were not considered in April. Projected leaf area values of A sat and its response to temperature were similar for Y and Y1 needles but differed across seasons, averaging , and lmol m 2 s 1 at 25 C for Y1 August, Y August and Y October, respectively (Fig. 1B andcandtable 3). In contrast, both rates and responses of the biochemical parameters (V cmax and J max ) depended on foliar age and season. In August, values of rates for V cmax and J max were much more responsive to temperature in Y1 needles compared with the new (Y) foliage (Fig. 1E, H and Table 3). For Y foliage, values of P (25 C) of V cmax and J max increased with time (Fig. 1E, F, H, I and Table 3). Activation energy (E a ) ranged between 183 and 498 kjmol 1 for V cmax and between 144 and325kjmol 1 for J max, generally increasing as needles aged. For example, E a values for V cmax and J max were 2- to 3-fold greater in October compared with August in Y foliar cohorts. Temperature optima (T opt ) differed between cohorts and across seasons (Table 3). In August, T opt values for V cmax and J max were 34 and 12 C lower in Y foliage compared with Y1 foliage. In contrast, T opt values for V cmax and J max decreased in Y needles by 37and68 C, respectively between August and October. Temperature response of foliar R d, on a leaf area basis, differed seasonally (M T air : F 1,8 ¼ 246; P51), with Q 1 values of 189 and 225 (between 15 and 3 C) during April and September 213, respectively (Fig. 2). While LMA was greater in September ( g m 2 ) compared with April ( gm 2 ), temperature response on a mass basis still differed (M T air, F 1,8 ¼ 131; P51). No sharp R d reductions were observed at higher temperatures, and Q 1 values (calculated in steps of 1 C) were similar over the range of temperatures used (data not shown). Seasonal patterns in needle photosynthetic capacity and morphology Photosynthetic capacity at 25 C (A sat25 C, V cmax25 C, J max25 C), N a and LMA were affected by both foliar age and season (Fig. 3). Mean (6 s.e.) values of A sat25 C ranged from to and from to lmol m 2 s 1 for Y1 and Y needles, respectively. Whereas rates of A sat25 C differed with foliar age (part of the season, Y, F 1,21 ¼ 132, P51; M, F 5,21 ¼ 21,

6 826 Jensen et al. Photosynthetic responses to temperature in black spruce April August October A sat 1 A Y1 Y B C D E F V cmax 1 5 G H I J max T air ( C) T air ( C) T air ( C) FIG. 1. Seasonal temperature response of (A C) A sat,(d F)V cmax and (G I) J max in 1-year-old (Y1) and current-year (Y) P. mariana needles at S1 Bog in northern Minnesota, USA. The scatter shows individual values of A sat, V cmax and J max and lines (solid, Y1; broken, Y) denote the curve fittings (eqn 1) including all observations; see Table 3. ln R d 2 y = 2 + 2x R 2 = T air ( C) y = x R 2 = 658 P < 1 April, Y Y2 September, Y Y2 FIG. 2. Temperature response of foliar daytime respiration rates (R d ) in Y Y2 needle cohorts during April and September P51; Y M, F 1,21 ¼ 12, P51), similar rates were observed for July and August (Fig. 3A). For overwintering needles (Y1), mean values of V cmax25 C increased from to lmol m 2 s 1 between April and May, and were stable around lmol m 2 s 1 until August. Current year (Y) needles had 67 % and 58 % lower V cmax25 C values compared with Y1 foliage in July and August, respectively. While V cmax25 C of Y needles increased from to lmol m 2 s 1 between July and September, they remained well below V cmax25 C of Y1 needles during the growing season (Fig. 3B). For Y1 foliage, J max25 C increased significantly between April and May from to lmol m 2 s 1 (Fig. 3C). Seasonal maximum values of J max25 C were and lmol m 2 s 1 in Y1 and Y foliage, occurring in August and October, respectively (Fig. 3C). Values of N a differed between cohort and season (Y, F 1,62 ¼ 445, P51; M, F 5,62 ¼ 54, P51) and increased over time from to and to mmolm 2 in Y1 and developing

7 Jensen et al. Photosynthetic responses to temperature in black spruce 827 A sat25 C V cmax25 C J max25 C N a (mmol m 2 ) LMA (g m 2 ) A B C D E April June August Months Y1 Y: F (1) = 13 2 Y M: F (5) = 2 1 Y M: F (1) = 12 n = 21 Y: F (1) = 21 3 M: F (5) = 79 1 Y M: F (1) = 5 ns n = 287 Y: F (1) = M: F (5) = 75 6 Y M: F (1) = 8 ns n = 29 Y: F (1) = 44 5 M: F (5) = 5 4 Y M: F (1) = 2 9 ns n = 62 Y: F (1) = 12 3 M: F (5) = 41 2 Y M: F (1) = 1 9 ns n = 63 October V cmax25 C J max25 C R d25 C 2 A Top Middle Bottom 1 B 2 1 C Needle age (years) 2 Y: F (2) = 5 9 C: F (2) = 6 1 Y C: F (3) = 8 ns n = 22 Y: F (2) = 4 1 C: F (2) = 4 4 Y C: F (3) = 6 ns n = 2 Y: F (2) = 12 9 C: F (2) = 9 ns Y C: F (3) = 8 ns n = 28 FIG. 3. Seasonal patterns in (A C) photosynthetic capacity, (D) N a and (E) LMA in 1-year-old (Y1) and current-year (Y) P. mariana needles. Values of A sat, V cmax and J max were normalized to 25 C based on season and needle age (Table 3). Data were collected and combined over a 4-year period. Year of collection was not a significant factor for any of the variables. Values are means (6 1 s.e., variation often smaller than the symbols) and ANCOVA output [F d.f. values and n (total number of A/C i curves across cohorts and months)], with significance levels for cohort (Y), month (M) and their interaction (Y M) estimates. Y needles, respectively (Fig. 3D). Similarly, LMA was affected by cohort and season (Y, F 1,63 ¼ 123, P51; M, F 5,63 ¼ 412, P51), with average values ranging from to and to gm 2 in Y1 and Y needles, respectively (Fig. 3E). Photosynthetic capacity and R d by needle age and canopy position In July of 211, V cmax25 C and J max25 C were affected by needle age (Y, F 2,22 ¼ 59, P55 and F 2,2 ¼ 41, P55, respectively) and canopy position (C, F 2,22 ¼ 61, P55 and F 2,2 ¼ 44, P55), but not their interaction, whereas R d25 C was only affected by needle age (Y, F 2,28 ¼ 129, P51) (Fig. 4A C). Low rates of V cmax25 C and J max25 C were generally observed in Y needles and in lower parts of the canopy (Fig. 4A, B), although there was substantial variation in values due to environmental conditions, such as periodic shading from adjacent trees. The Y foliage exhibited variation in V cmax25 C and J max25 C that was at least partially the result of differences in N a content, although linear regressions only explained 25 % of the variation (V cmax25 C, R 2 ¼ 24, P51; J max25 C, R 2 ¼ 25, P51). FIG. 4. Photosynthetic capacity [(A) V cmax25 C and (B) J max25 C]and(C)foliar day-respiration rates (R d25 C) by needle age and canopy position measured over a 4-day period in July 211. Values are means (6 1 s.e.) and ANCOVA output (F d.f. values and n), with significance levels for cohort (Y), canopy position (C) and their interaction (Y C) estimates. Foliar respiration (R d25 C) was 73 84, and % greater in Y than Y1 and Y2 needles for branches at the top, middle and bottom of the canopy, respectively (Fig. 4C). There was a negative association between N a and R d25 C (R 2 ¼ 41, P51) but only in the developing Y foliage. As LMA was % greater in Y1 and Y2 needles, there was a positive correlation between R d25 C and N on a leaf mass basis (data not shown). Canopy allometric and foliar morphology Needle mass distribution varied according to canopy position and cohort, as represented by a 6-m tall P. mariana tree whose foliage was sequentially harvested in 212 (Fig. 5). For the eight P. mariana trees harvested in June and July of 21 and 211, 2 3 %, % and % of the total needle mass ( kg) was distributed in the bottom, middle and top canopy layers, respectively. Towards the end of the 212 growing season 8 % of the total foliar mass belonged to the three youngest foliar cohorts (Y, Y1 and Y2) (Fig. 6A), displayed towards the more exposed tips of the branches. Branches in the top of the canopy had a significantly greater proportion of foliar mass in the Y cohort, while the proportion of foliar mass in other cohorts was not affected by height, and declined with needle age (Fig. 6A). Older needle cohorts were observed to have higher LMA (Y, F 5,238 ¼ 24, P55) and C:N ratios (Y, F 5,194 ¼ 29, P55) compared with Y needles, whereas canopy position had no significant effect (Fig. 6B, C).

8 828 Jensen et al. Photosynthetic responses to temperature in black spruce Tree height (m) >Y3 Y2 Y1 1 Y Needle mass (g) FIG. 5. Needle mass distribution within the canopy (m) of a 6 -m tall representative P. mariana tree, according to needle age. Whole trees were harvested in June/July 21 and 211; cohort proportions were estimated in October 212. Proportion of total foliar mass LMA (g m 2 ) C:N ratio A B C Top Middle Bottom Needle age (years) Annual carbon uptake components by foliar cohort Y: F (5) = C: F (2) = 2 3 ns Y C: F (1) = 6 3 n = 239 y = 5182e 646x R 2 = 963 Y: F (5) = 2 4 C: F (2) = 2 ns Y C: F (1) = 7 ns n = 238 y = x R 2 = 473 Y: F (5) = 2 9 C: F (2) = 4 ns Y C: F (1) = 1 5 ns n = 194 FIG. 6. (A) Proportion of total needle mass distribution, (B) leaf mass per area (LMA) and (C) C:N ratios by cohort and canopy position, October 213. Values are means (6 1 s.e.), regression lines and the ANCOVA output (F d.f. values and n), with significance levels for cohort (Y), canopy position (C) and their interaction (Y C) estimates. Temperature-sensitive Farquhar/Ball Berry model projections of annual net C uptake across cohorts (see Methods) indicated that foliar cohort contributions to total annual C uptake were different, and thus should be estimated independently by cohort groups having similar characteristics. Table 4 illustrates the model-estimated annual net C uptake for the S1 Bog P. mariana stand, integrating cohort-specific photosynthetic temperature responses (multiple foliar cohort simulation) or assuming similar photosynthetic temperature responsiveness across all cohorts (single foliar cohort simulation) for three different temperature conditions. When integrating cohort-specific characteristics, we estimated the total (Y þ Y1&2) annual C uptake for the P. mariana stand at ambient T air to be 551, 57 and 534 gc m 2 year 1 for 211, 212 and 213, respectively (Table 4). Although Y LAI, at 52 m 2 m 2, was relatively large compared with 67 m 2 m 2 for Y1 and Y2, current year foliage contributed only 36 % of the total annual C uptake (Tables 2 and 4). To evaluate the importance and relative influence of foliar cohort differences on annual C gain, we estimated annual foliar C uptake assuming all needles belonged to a single cohort (Y1 in this example). By comparison, the single foliar cohort simulation estimated annual C uptake for the P. mariana stand at ambient T air to be 11 % greater (614, 563 and 589 gc m 2 year 1 for 211, 212 and 213, respectively) than the multiple foliar cohort simulation. In addition, and as a result of the relatively low photosynthetic responsiveness to CO 2 and temperature in Y needles (Fig. 1 and Table 3), the relative influence of foliar cohort differences on total annual C gain became more pronounced when warmer temperature assumptions were applied (þ45andþ9 C) (Table 4). DISCUSSION In P. mariana, where seasonal shoot development is slow, maintaining multiple foliar cohorts is of key importance for total annual C uptake. Results presented here demonstrate that (1) photosynthetic capacity and temperature responsiveness are significantly lower during midsummer in the developing Y cohort compared with older cohorts; and (2) estimates of canopy C uptake without integrating seasonal- and cohort-specific photosynthetic temperature response functions could result in overestimation of annual C uptake, relative to each cohort s LAI contribution. Furthermore, and as result of the relatively low photosynthetic responsiveness to temperatures observed in Y needles, the relative C contribution by Y needles decreased when warmer climate assumptions were applied (Table 4). The results illustrate the importance of accounting for underlying seasonal- and cohort-specific differences when estimating current and future ecosystem C uptake capacities in boreal evergreen forest ecosystems. Seasonal dynamics of photosynthetic capacity The cohort-specific seasonal patterns of A sat25 C values were similar to rates reported earlier in P. mariana and other boreal conifers, such as Picea abies and Pinus sylvestris (Fig. 3A) (Troeng and Linder, 1982; Bergh and Linder, 1999). While our A sat25 C estimates, on average, were greater than earlier published values for P. mariana (Goulden et al., 1997; Bronson and Gower, 21; Hébert et al., 21), it may reflect sitespecific differences in growth temperature and resource availabilities. For example, values of A sat25 C presented here are 2- to 25-fold greater than values reported by Bronson and Gower (21) for P. mariana saplings grown in Manitoba, Canada (13 km north of S1 Bog). It is generally assumed that A sat declines with needle age, at least on a leaf area basis

9 Jensen et al. Photosynthetic responses to temperature in black spruce 829 Table 4. Extrapolated annual net C exchange (gc m 2 ground area year 1 ) for Picea mariana needles for cohorts for three temperature assumptions [þ (ambient), þ45 and þ9 C] and three years of environmental driver data (211, 212 and 213). Annual positive values of C exchange are net uptake into the foliage from the atmosphere, with and without considering cohort-specific differences in the model. See Materials and methods section for model description Treatment ( C) Year (A) Simulation of multiple foliar cohorts (B) Simulation of single foliar cohort (B) (A) difference (%) Y1&2 Y All foliar cohorts All Y1 foliar þ þ63 (114) þ þ56 (11) þ þ55 (13) þ þ88 (142) þ þ82 (143) þ þ8 (134) þ þ128 (19) þ þ124 (21) þ þ117 (184) (e.g. Hébert et al., 21); however, our results contrast with this, in general having lower values of A sat in younger needles. This reduced photosynthetic capacity is mainly a result of the cohort-specific temperature response of the photosynthetic biochemical component (V cmax and J max ), whereas foliar morphological development may partly counter this effect in P. mariana (Figs 1 and 3). This is illustrated by the relatively similar overall values and temperature response of A sat between Y and Y1 needles, in spite of significantly lower rates of V cmax and J max in Y needles (Figs 1B, E, H and 3A, B, C), a discrepancy that can be explained by a shift in mesophyll conductance (g m )withleafmaturation(gu et al., 21). Developing needles may therefore, at least to some extent, be able to compensate for this reduced capacity by adjusting mesophyll conductance (Busch et al., 213). Springtime resumption of photosynthesis in overwintering needle cohorts and bud break is a function of plant-available water and soil and air temperature in many conifer species (Bergh and Linder, 1999; Tanja et al., 23). The release of photosynthetic dormancy in overwintering needles entails changes in chloroplast ultrastructure and elevated levels of chlorophyll, photosystem II, Rubisco and downstream co-/ enzymes of the Calvin cycle (Öquist and Huner, 23; Monson et al., 25). Similarly, we observed a substantial increase in V cmax25 C and J max25 C in P. mariana Y1 needles during spring. However, rates of recovery of V cmax25 C and J max25 C (e.g. times to seasonal optima) differed, suggesting that growth temperatures and/or plant-available water (indirectly soil/peat temperature) stimulate production and activation of Rubisco and the electron transport chain differently (Fig. 3B, C) (Monson et al., 25; Sevanto et al., 26). Whereas an increase in these biochemical parameters is often positively correlated with an increase in foliar nitrogen (e.g. Greenway et al., 1992), N concentrations in P. mariana Y1 needles were stable during the period of photosynthetic resumption (April to June; Fig. 3B D). Using N concentration as a proxy of the Rubisco content, our findings suggest activation of Rubisco rather than an increase in foliar content. The photosynthetic capacity (V cmax25 C and J max25 C)of the developing Y cohort of needles increased during the growing season, but did not reach the capacity of the Y1 needles measured in August despite equal or greater LMA and N a (Fig. 3A C). Although we did not quantify the photosynthetic temperature response functions of Y1 needles in October, our results indicate that the photosynthetic apparatus of Y needles may require more than one growth seasontomature,andthereforemayhelpexplainthedifference in the springtime rates of recovery of V cmax25 C and J max25 C. Seasonality and needle age affected both T opt and temperature response curve shapes (Fig. 1 and Table 3). Temperature response curves for V cmax and J max were noticeably more bellshaped and had greater thermal optimums and greater activation energies (E a ) in 1-year-old (Y1) needles than in new (Y) needles. However, with time and towards the end of the growth season, Y needles response curves became more similar to response curves observed for Y1 needles (Fig. 1 and Table 3). Observed values of T opt in P. mariana are of the same magnitude as reported for Quercus robur and Betula pendula (Dreyer et al., 21) but 1 C greater than observed for Pinus pinaster (Medlyn et al.,22a). Multiple factors are correlated with the shifts in T opt of V cmax and J max, including growth temperature, ontogeny and leaf nutrient status (Berry and Björkman, 198; Martindale and Leegood, 1997; Medlyn et al., 22a; Khaembah et al., 213). While changes in growth temperatures and light availability may partially explain seasonal differences in T opt of V cmax and J max, foliar nutrient status and ontogeny may also account for the observed cohort-age-dependent differences. We note that, as field conditions made it difficult to maintain a consistent VPD across temperatures, A/C i curves generated at higher temperatures are less reliable (Medlyn et al., 22a). Thermal adjustment of the photosynthetic apparatus occurred in both old and developing needles, sometimes in parallel with shifts in N allocations (Fig. 3A D; Öquist and Huner, 23). Although changes in air temperature likely induce similar physiological responses for all cohorts, their capacity to respond and thus adjust would be dependent on the biochemical status of the tissue. To our knowledge, ontogenic thermal adjustment capacity has never been described for the temperature response of V cmax or J max.however,itis outside the scope of the study to quantify thermal adjustment capacity at different developmental stages.

10 83 Jensen et al. Photosynthetic responses to temperature in black spruce Seasonal temperature adjustment of foliar R d The adjustment of foliar respiration to temperature in mature tissues is common and is thought to lead to plant C balance homeostasis (Reich et al., 1998). In contrast to Way and Sage (28a), we found seasonal adjustment of Q 1 R d in P. mariana needles between April and September, when assessed on both a leaf area and a mass basis. This adjustment reflects differences in metabolic activity (mitochondrial activity and substrate/ co-enzyme availability) (e.g. Atkin et al., 25; Kruse et al., 28). Here we measured the instantaneous foliar R d temperature response of multiple needle cohorts, but as metabolic activity is often greater during rapid growth and expansion (Machado and Reich, 26), Q 1 values are likely greater for the Y cohort. This is partly supported by the % greater R d values observed during July (211) for Y needles compared with Y1 and Y2 needles (Fig. 4C) (Teskey et al., 1984). In agreement with earlier findings (Way and Sage, 28a; Tjoelker et al., 1998), lower levels of foliar N were associated with lower rates of respiration per leaf mass in P. mariana. Annual carbon uptake contribution by foliar cohorts and model implications Our simulations differ from earlier studies (Girardin et al., 28; Hall et al., 29) in that we modelled total annual C uptake separately for each foliar cohort based on seasonal- and cohort-specific photosynthetic temperature responses (Smith and Dukes, 213). For S1 Bog, this reduced the total estimated annual C uptake, as the Y needle cohort contributed only % of the total annual C uptake (Table 4). Multi-cohort modelling illustrates the physiological/ecological significance of retaining several needle cohorts, not only by allowing for enhanced CO 2 assimilation earlier in the season (resumption of photosynthesis) (Öquist and Huner, 23; Tanja et al., 23; Hall et al., 29), but also by maintaining significantly greater photosynthetic capacity throughout the growth season. At tree and stand level, the physiological significance of retaining several cohorts should be put in the context of a cohort s total lifespan C uptake: from sink (net A sat is negative) to source (net A sat is positive) and finally sink again (A sat declines). Although Y needles contribute significantly to the total C uptake towards the end of the growth season, our results clearly demonstrate that maximum photosynthetic capacity is only reached in the second year. This finding underlines the need to better understand interactions between ontogeny and environmental drivers. However, in a future warmer climate the relative importances of different cohorts may shift. Over the last century mean annual air temperature in the North American boreal zone has increased by C, and global atmospheric model simulations suggest this trajectory will continue (Christensen et al., 27). For P. mariana, a combination of elevated air temperatures with increased soil water availability during spring will likely prolong the growth season length by (1) earlier initiation of bud break in developing shoots (Y); (2) accelerated shoot maturation; (3) earlier full photosynthetic recovery in overwintering needles; and (4) a shift in T opt of key photosynthetic parameters. Together, these changes have the potential to significantly increase net annual C uptake (Goodine et al., 28; Fløistad and Granhus, 21; Sutinen et al., 212). Indeed, our simulations indicate an overall increase in net foliar C uptake under warmer climate scenarios, a positive effect that remained after integrating enhanced C losses as a result of increased respiration rates with rising temperature. It is worth noting that we most likely underestimated respiratory C losses, as values of R d were measured during daytime, as Q 1 values are often lower in light than in darkness (see Table 2 in Atkin et al., 25). For model simplification, we assumed homogeneous light availability (full sun) between needle cohorts and throughout the canopy, likely resulting in an overestimation of the total canopy C uptake. Even so, we observed little evidence of selfshading between Y and Y1 (Fig. 3A, B) or between Y1 and Y2 at branch level. Vertical light stratification influenced within-canopy photosynthetic capacity (Fig. 4A, B), suggesting that, for this relatively open S1 Bog P. mariana secondary forest, within-canopy self-shading induces greater C estimation errors compared with branch-level self-shading. However, the assumption of homogeneous light availability across multiple needle cohorts may induce greater C estimation errors in other systems, such as productive upland forests with denser canopies (Dewar et al., 212). Conclusions The findings presented here have two major implications for modelling C assimilation at the canopy and stand levels, especially for boreal conifers with prolonged foliar ontogeny (weeks to months). Firstly, it is evident that individual foliar cohort contributions to annual C uptake must be estimated for cohort groups having similar characteristics, requiring high-resolution species- and site-specific-allometric and physiological information at the leaf, canopy and stand levels. Secondly, the underlying photosynthetic temperature response depends on foliar age (ontogeny) and seasonality. We suggest that, in spite of being very labour-intensive, species-, seasonal- and age-specific temperature response dynamics should be integrated and linked to growth temperatures. Taken together, our findings show the physiological significance of maintaining multiple foliar cohorts, not only during bud break but also throughout the growth season. Our results thus illustrate the need for seasonal- and cohort-specific model parameterization when estimating the C uptake capacity of boreal forest ecosystems under ambient or future temperature scenarios. ACKNOWLEDGEMENTS The authors appreciate fieldwork, development of allometric relationships and data analysis from Carla Gunderson, Kelsey Carter, Lianhong Gu, Donald E. Todd, Deanne Brice, Jana Phillips, W. Robert Nettles and Les Hook. This material is based upon work supported by the US Department of Energy, Office of Science, Office of Biological and Environmental Research, under contract DE-AC5-OR LITERATURE CITED Atkin OK, Bruhn D, Hurry VM, Tjoelker MG. 25. The hot and the cold: unraveling the variable response of plant respiration to temperature. Functional Plant Biology 32: