The photochemical reflectance index provides an optical indicator of spring photosynthetic activation in evergreen conifers

Transcription

1 Research The photochemical reflectance index provides an optical indicator of spring photosynthetic activation in evergreen conifers Christopher Y. S. Wong 1 and John A. Gamon 1,2 1 Department of Earth and Atmospheric Sciences, University of Alberta, Edmonton, AB T6G 2E3, Canada; 2 Department Biological Sciences, University of Alberta, Edmonton, AB T6G 2E9, Canada Author for correspondence: Christopher Y. S. Wong Tel: cyet.wong@mail.utoronto.ca Received: 26 August 2014 Accepted: 27 November 2014 doi: /nph Key words: acclimation, chlorophyll fluorescence, cold stress, evergreen conifers, photochemical reflectance index (PRI), photosynthesis, pigments, spring activation. Summary In evergreens, the seasonal down-regulation and reactivation of photosynthesis is largely invisible and difficult to assess with remote sensing. This invisible phenology may be changing as a result of climate change. To better understand the mechanism and timing of these hidden physiological transitions, we explored several assays and optical indicators of spring photosynthetic activation in conifers exposed to a boreal climate. The photochemical reflectance index (PRI), chlorophyll fluorescence, and leaf pigments for evergreen conifer seedlings were monitored over 1 yr of a boreal climate with the addition of gas exchange during the spring. PRI, electron transport rate, pigment levels, light-use efficiency and photosynthesis all exhibited striking seasonal changes, with varying kinetics and strengths of correlation, which were used to evaluate the mechanisms and timing of spring activation. PRI and pigment pools were closely timed with photosynthetic reactivation measured by gas exchange. The PRI provided a clear optical indicator of spring photosynthetic activation that was detectable at leaf and stand scales in conifers. We propose that PRI might provide a useful metric of effective growing season length amenable to remote sensing and could improve remote-sensing-driven models of carbon uptake in evergreen ecosystems. Introduction Boreal forests cover c Mha in the Northern Hemisphere and store c. 32% of the total carbon (C) stock of established forests (Pan et al., 2011). These boreal forests are dominated by evergreen conifers that have acted as a net C sink for the removal of CO 2 from the atmosphere (Apps et al., 1993; Ciais et al., 1995). However, the boreal region has been undergoing significant climate change, including increases in temperature (IPCC, 2007). Climate change can affect season length and shift the onset of the cold hardening and spring recovery periods (Ensminger et al., 2004), but the potential impacts on annual C balance are not understood (Richardson et al., 2010; Xu et al., 2013). An implication of a longer growing season is that net primary productivity (NPP) will increase, but field studies have not always supported this prediction, sometimes showing reduced annual productivity (Canadell et al., 2007; Le Quere et al., 2009) as a consequence of increased disturbances and environmental stresses such as drought (Ciais et al., 2005). Consequently, the future role of the boreal forest as a C sink is now highly uncertain (Kurz & Apps, 1999; Goodale et al., 2002; Kurz et al., 2008; Thurner et al., 2014). Extreme temperatures and drought lead to the down-regulation of photosynthesis, which produces a potential imbalance of light energy capture and usage. Over-excitation of photosystems leads to permanent photodamage to the photosynthetic apparatus which impairs the leaves ability to photosynthesize (Powles, 1984). To prevent photodamage in the winter, evergreen leaves undergo cold hardening signaled by reduced photoperiod and decreasing temperatures (Levitt, 1980; Vogg et al., 1998; Bigras et al., 2001). Photoprotective processes involve several changes in photoprotective pigments, which allow the leaves to dissipate excess energy via thermal dissipation to maintain an energy balance (Butler, 1978; Baker, 2008; Verhoeven, 2014). Cold hardening results in the rearrangement of pigments (Ottander et al., 1995), as well as sustained increases in carotenoid pigment pools (Bj orkman & Demmig-Adams, 1994; Demmig-Adams & Adams, 1996; Verhoeven et al., 1999). Dehardening of evergreens and the spring recovery of photosynthesis involve springtime reversal of the pigment changes, signaled by a combination of warmer temperatures and longer photoperiod (Ottander et al., 1995; Vogg et al., 1998). Changing carotenoid pigment composition clearly plays a large role in thermal dissipation. Sustained thermal dissipation is not fully understood but carotenoid pigments, including those of the xanthophyll cycle, are believed to be a key component (Horton et al., 1996; Baroli & Niyogi, 2000; Holzwarth et al., 2009; Ruban et al., 2012; Verhoeven, 2014). The xanthophyll 196

2 New Phytologist Research 197 cycle pigments enable reversible down-regulation of electron transport during conditions of short-term stress (Demmig-Adams & Adams, 1992). In winter, there is an overall increase in the pool of xanthophyll pigments, as well as sustained de-epoxidation of violaxanthin to zeaxanthin (Adams et al., 1995, 2002; Ottander et al., 1995; Demmig-Adams & Adams, 1996). Levels of lutein also increase in the winter (Garcıa-Plazaola et al., 1997; Jahns & Holzwarth, 2012; Oh et al., 2013), presumably playing a photoprotective function during sustained cold. Concomitantly, total chlorophyll a and b (Chla and b) pigment levels can decrease in concentration over the winter (Ottander et al., 1995). These seasonal pigment changes lead to shifting carotenoid : chlorophyll ratios that are detectable with spectral reflectance as the photochemical reflectance index (PRI), which may serve as an optical indicator of winter photosynthetic deactivation and subsequent spring reactivation (Wong & Gamon, 2015). Previous studies examining the spring recovery of pigments and proteins (Verhoeven et al., 1996; Ensminger et al., 2004), chlorophyll fluorescence (Porcar-Castell, 2011) and light-use efficiency (LUE) (Nichol et al., 2002; Porcar-Castell et al., 2012) have all observed a rapid spring reactivation of photosynthetic activity in evergreens. At a coarse temporal scale, these studies have sometimes implicated the xanthophyll cycle (e.g. Nichol et al., 2002). However, other more detailed physiological studies call into question the timing of, and exact role of xanthophyll cycle pigments, during photosynthetic reactivation (Porcar-Castell et al., 2012). Thus, while previous studies have indicated that the PRI can serve as a useful indicator of winter down-regulation and spring activation in evergreens (Nichol et al., 2002; Stylinski et al., 2002; Filella et al., 2009), the exact functional significance of seasonal PRI changes has remained unclear because of the multiple pigments and processes involved. From the perspective of C cycle impacts, the timing of spring activation of photosynthesis has a much greater influence on the net C uptake of northern forests than autumn deactivation, primarily because irradiance is much higher in spring relative to the autumn. Consequently, spring temperature has a dominant influence on spring activation and a significant effect on the annual C budget (Randerson et al., 1999; Ensminger et al., 2004; Monson et al., 2005; Richardson et al., 2009, 2010). In this study, we followed the seasonal processes of photosynthetic activation and deactivation for several evergreen species over the course of a year in a boreal climate. A primary purpose of this study was to monitor the spring activation of photosynthesis by following a variety of physiological parameters at the leaf level. We also examined the kinetics of the different parameters over the course of spring recovery, allowing a better understanding of the timing and mechanisms of the component processes of photosynthetic activation. An additional goal was to compare spring PRI responses at different spatial scales and in different evergreen species to see if this index could provide a reliable remote indicator of spring photosynthetic activation. pine (Pinus ponderosa Douglas ex Lawson & C. Lawson) and white spruce (Picea glauca (Moench) Voss) which were grown outdoors in a rooftop common garden. Initially, 388 P. contorta and 367 P. ponderosa 1-yr-old bare-root seedlings were planted in 2010 into 2.31-l pots (CP412CH; Stuewe & Sons, Tangent, OR, USA). In the summer of 2011, P. contorta and P. ponderosa were repotted into 2.83-l pots (TP414; Stuewe & Sons) and 25 1-yrold bare-root seedlings of P. glauca were initially planted in l pots (D40L; Stuewe & Sons). In the summer of 2013, P. contorta and P. ponderosa were repotted into large 6.23-l pots (TP616; Stuewe & Sons) and P. glauca plants were repotted into the 2.31-l pots for the remainder of the study. Healthy plants from each species were arranged immediately adjacent to each other in dense monoculture plots (c m square, with c. 15-cm spacing between plants) to simulate homogenous closedcanopy forest seedling stands. All plants were kept well watered from the spring to autumn seasons to avoid water stress. To provide additional insulation to the roots, in November 2011 extra pots of soil and a plywood frame were added to the outside edges of the plots, and before each winter, a 3-cm layer of peat moss was added. Further descriptions of plant culture are provided in Wong & Gamon (2015). Measurements described in this study were begun in summer 2012, after the plants had been well established and growing for at least 1 yr, and the primary focus was on the spring transition in A weather station located within 5 m of the common garden provided 15-min averages of light and temperature conditions (Wong & Gamon, 2015), which were later aggregated into daily total photosynthetic photon flux density (PPFD) and daily average temperature. Temperature was also expressed as daily maximum and minimum values. For P. contorta and P. ponderosa, several photosynthetic parameters were monitored from June 2012 to July 2013 to assess seasonal patterns of photosynthetic activity. Measurements were made every other week from June 2012 to February 2013 and on a weekly basis from March 2013 to July 2013 (during spring transition) during which P. glauca was incorporated. Data collection occurred on sunny days from 12:00 h to 14:00 h (within 1 2 h of solar noon), ensuring maximal sunlight and reduced diurnal variability during key sampling periods. For leaf measurements, six randomly chosen plants for each species were monitored for the duration of the study. All leaf measurements were completed on the same individual plants but not necessarily the same leaves, which were averaged together from a single day to follow the seasonal dynamics. For all leaf measurements, the youngest needles from the 2012 cohort were used throughout the study. These were selected from top-canopy needles exposed to similar light conditions, keeping growth and sampling illumination consistently high, and avoiding interference of shading effects from deeper positions within a canopy. Stand-level optical measurements were made on each plot of P. contorta and P. ponderosa, as described in the following section. Materials and Methods The study was conducted at the University of Alberta, Canada on lodgepole pine (Pinus contorta Douglas ex Loudon), ponderosa Optical sampling Leaf-scale measurements were collected using a spectrometer (UniSpec-SC; PP Systems, Amesbury, MA, USA) housing a

3 198 Research New Phytologist halogen lamp light source, equipped with a bifurcated fibre optic (UNI410; PP Systems) to measure spectral reflectance (Gamon & Surfus, 1999). A needle leaf clip (UNI501; PP Systems) enclosed a single needle attached to the tree and held the fibre tip at a fixed angle and position relative to the leaf surface, yielding a 0.6-mm-diameter sampling area per needle. Each series of five leaf measurements from an individual plant was preceded by a dark measurement and white reference scan (Spectralon; Labsphere, North Sutton, NH, USA), and reflectance was calculated from each leaf scan divided by a white reference scan after correction for dark current. PRI was calculated as: PRI ¼ R 531 R 570 R 531 þ R 570 Eqn 1 R indicates reflectance, and the subscript indicates the waveband in nm (Gamon et al., 1992, 1997). For P. contorta and P. ponderosa, five random leaves per plant on six plants were measured, providing a total of 30 samples per species. For P. glauca, two random leaves per plant on six plants were sampled, providing a sample size of 12. Needles from mature P. contorta were also monitored to incorporate naturally grown plants to examine potential experimental artifacts from pot culture. For mature P. contorta trees, five random leaves exposed to the sun were measured per plant from five separate individuals, providing a total sample size of 25. These mature trees were located in the grounds of the University of Alberta campus, within 90 m of the common garden plants, so were exposed to similar irradiance and temperatures. Stand-scale measurements were collected using a dual detector field spectrometer (UniSpec-DC; PP Systems) equipped with two fibre optics. An upward facing fibre (UNI686; PP Systems) was attached to a cosine receptor (UNI435; PP Systems) for incoming irradiance. A downward-looking fibre (UNI684; PP Systems) was fitted with a field-of-view restrictor (UNI688; PP Systems) to limit the field of view to c.15. Measurements on the potted P. contorta and P. ponderosa plants were performed c. 0.5 m above the tallest canopy position. PRI was calculated using Eqn 1. For each species, 12 measurements were taken with the fibre pointed at a different section of the stand, combined to provide an average measure of several individual canopies within the stand. The intention of these stand measurements was to provide a sample of a 1-m pixel, a common scale in airborne sampling. Pigment assay Leaves were collected near solar noon on sunny days to maximize light conditions during sampling. Leaves were removed from the top of the plants within 30 min of the leaf-scale optical sampling, cut into three 1-cm-long segments, measured for leaf area with calipers, immediately stored in liquid nitrogen, and later transferred to an 80 C freezer for long-term storage. These steps were conducted under ambient light as quickly as possible, within 10 s, so leaves would be photosynthetically activated while minimizing potential changes in xanthophyll cycle epoxidation state. For high-performance liquid chromatography (HPLC; 1260 Infinity; Agilent Technologies, Santa Clara, CA, USA), batches of six leaf samples from different plants from a given date and time were pooled together for pigment analysis. Pigment extraction and HPLC analysis following the Thayer & Bj orkman (1990) method were used to determine the pigment concentrations of various carotenoid and chlorophyll pigments. The HPLC system was calibrated using commercial pigment standards (DHI LAB Products, Hørsholm, Denmark). Xanthophyll cycle pigment pools (VAZ) were determined as the sum of violaxanthin (V), antheraxanthin (A) and zeaxanthin (Z) concentrations. Total carotenoids (Car) were determined as the sum of VAZ, neoxanthin (N), lutein (Lut), and b-carotene concentrations. Total chlorophyll (Chl) was determined as the sum of Chla and b concentrations. Carotenoid pigments were expressed individually on a leaf area basis (lmol m 2 ) and also normalized to total chlorophyll levels (mmol mol 1 ). The epoxidation state (EPS) of the xanthophyll cycle was expressed as: EPS ¼ V þ 0:5A V þ A þ Z Eqn 2 Chlorophyll fluorescence Chlorophyll fluorescence was monitored using a portable fluorometer (Mini-PAM; Walz, Effeltrich, Germany). The fluorometer was fitted with a fibre optic and leaf clip holder (2030-B; Walz), which kept the fibre tip at a fixed angle and position for repeatability. Leaves were carefully bundled together and clamped within the leaf clip holder as close to their original orientation as possible to maintain ambient illumination during sampling. A short 0.8-s pulse of saturating light of c lmol photons m 2 s 1 was provided and the maximal fluorescence, Fm 0, was obtained. The fluorometer calculated the fluorescence yield parameter (DF/Fm 0 ), an indicator of the photosynthetic efficiency of photosystem II (Baker, 2008). Yield ¼ u ¼ DF F 0 m Eqn 3 This method allowed for the calculation of the relative electron transport rate (ETR) (lmol m 2 s 1 ). This parameter provided a more direct estimation related to photosynthetic activity (Maxwell & Johnson, 2000; Baker, 2008). ETR ¼ u PPFD 0:5 f PPFDa Eqn 4 To calculate ETR, the PPFD was obtained from the built-in quantum sensor from the leaf clip. The fraction of absorbed light (f PPFDa ) was obtained from the leaf-scale optical reflectance measurements calculated as 1 reflectance, assuming no transmittance in these optically thick needles, and was expressed as the average light absorptance from 400 to 700 nm. f PPFDa differs between species and changes seasonally with pigment changes, so by incorporating a flexible value, the fluorescence-based estimation was

4 New Phytologist Research 199 corrected for changing light absorption (Maxwell & Johnson, 2000). Incident sunlight on leaves of random orientation was used to derive a nonlinear light response curve of ETR, and this curve was used to estimate the light-saturated ETR at 1500 lmol m 2 s 1. Each species had a sample size of six with measurements on three leaf bundles per plant from six individuals. Gas exchange During the spring transition, photosynthesis was measured using a portable photosynthesis system (Li-6400; Li-Cor, Lincoln, NE, USA). Six needles from an individual plant were placed inside the gas exchange chamber and a photosynthetic light response curve was then measured. The light curve was set to monitor photosynthesis at light intensity values of 1500, 1000, 800, 600, 400, 100, 50 and 0 lmol photons m 2 s 1. For each stage, there was a 1-min minimum and 3-min maximum acclimation time before measurement and changing to the next light level. The chamber environmental settings were set to match current ambient temperatures and the reference CO 2 was set to 395 lmol mol 1, which was approximately the same as atmospheric concentrations. Air flow was set to 400 lmol s 1. The same five plants were measured for each pine species over the course of spring recovery from March 2013 to July 2013 on a weekly basis. After averaging the photosynthetic light response curves together, the light-saturated photosynthetic rate (P s ; lmol CO 2 m 2 s 1 ) was determined at 1500 lmol photons m 2 s 1 (near the saturating region of the curve) and the LUE on an incident light basis was calculated from the initial slope at low light intensities. Kinetics All the methods incorporated in this project provided indicators of seasonal transition, including autumn deactivation and spring photosynthetic recovery. To examine the kinetics of spring recovery, data from 8 March 2013 to 17 July 2013 were selected, spanning the transition from winter- to summer-adapted states. To facilitate comparison, the values of all the parameters were normalized from zero to one. A sigmoidal fit incorporating stable winter and summer days and transitional spring recovery was estimated using commercial plotting software (IGOR PRO; WaveMetrics, Portland, OR, USA). The half-point of the recovery (estimated with a 95% confidence interval) was used to compare the timing of activation of different photosynthetic components. Results Daily mean temperature and daily total PPFD exhibited large seasonal changes characteristic of a boreal climate (Fig. 1a; see Supporting Information Table S1 for ambient conditions during measurements). For descriptive purposes, we defined four distinct seasons based on temperature, and indicated by contrasting shading. The daily mean temperatures in the summer, from June to August, were typically above 18 C and consisted of high PPFD as a result of a long photoperiod and high sun angle. During the autumn season, from September to November, both daily mean temperature and total PPFD declined gradually. The winter season, from December to March, had daily mean temperatures below 0 C and low PPFD as a result of a short photoperiod and low sun angle. PPFD rose in February and gradually increased until May. Daily mean temperature increased rapidly in April and May. PRI exhibited seasonal variability for both pine species roughly in parallel with temperature (Fig. 1b). Maximum values of c were observed in the summer from June to August. In the autumn from September to November, PRI gradually decreased. During the early winter, from December to January, deep cold temperatures below 4 C resulted in an abrupt and temporary increase in PRI as a result of decreased leaf albedo (discussed in Wong & Gamon, 2015). This increase reversed later in the winter (February April), when average winter temperatures rose above 4 C. PRI during the remainder of the winter season had minimum values of 0.27 and 0.23 for P. contorta and P. ponderosa, respectively. The lowest PRI values were observed in late April, when temperatures remained low, but when daily total PPFD values exceeded 30 mol m 2 d 1. In the spring, PRI rapidly increased from the minimum winter values to the maximum summer values within 4 wk (late April to early May). Relative to the autumn transition, when PRI seemed to anticipate the decline in temperature, early spring PRI values lagged temperature during spring, recovering only when temperatures were consistently above 0 C. This led to an asymmetry in the annual PRI response, clearly visible in Fig. 1. ETR values had similar patterns to seasonal temperature, and were more symmetrical than PRI trends (Fig. 1c). The highest values of ETR (c. 300 lmol m 2 s 1 ) were observed in the summer, and these values declined from October to November. During early winter, when average temperatures were c. 10 C in December and parts of January, the ETR dropped to zero, indicating complete inactivity. When winter temperatures slightly increased near 0 C, we observed a slight increase in activity. Spring recovery in ETR occurred from mid-april to early-may over c. 4 wk. Photosynthetic pigments exhibited seasonal changes in composition and pool size that resembled the seasonal PRI patterns. The EPS of the xanthophyll cycle gradually decreased from September to November (Fig. 1d). The lowest EPS values were exhibited in the winter season, indicative of high relative zeaxanthin content (the photoprotective form of the xanthophyll cycle pigments). The absolute lowest EPS occurred in early April, just before the start of the spring recovery of temperature. Spring recovery of the EPS occurred rapidly within about a week (in mid-april) when it increased by over fivefold from c to > 0.50 (Fig. 1d). The total concentration of Chla and b varied, being highest in the growing season and lowest in the winter (Fig. 1e). Several carotenoid pigment pools, including lutein (Fig. 1f), b-carotene (not shown), xanthophyll cycle pigment pools (VAZ; Fig. 1g) and total carotenoids (Fig. 1h), exhibited nearly identical seasonal patterns that were rough mirror images of the PRI trends (see Tables S2 S5 for individual pigment concentrations). The lowest concentrations of these carotenoid pigments were observed in the summer and the highest

5 200 Research New Phytologist (c) (d) (e) (f) (g) (h) Fig. 1 Seasonal variation of key variables from summer 2012 to summer Daily mean temperature (solid red line), daily min/max temperature (dashed red line) and daily total photosynthetic photon flux density (PPFD) (black line). Comparison of Pinus contorta (red) and Pinus ponderosa (black) for the remaining panels. (b h) Time series of photochemical reflectance index (PRI), (c) electron transport rate (ETR), (d) epoxidation state of the xanthophyll cycle (EPS), (e) total Chla+bconcentration, (f) lutein : Chl, (g) xanthophyll pool size on a chlorophyll basis (VAZ : Chl), and (h) carotenoid pool size on a chlorophyll basis (Car : Chl). Error bars represents standard error of the mean. Grey regions denote summer and winter, and white regions denote transitional autumn and spring seasons. Panels, and (h) are modified from Wong & Gamon (2015) with additional data added (panels c g). concentrations occurred between late autumn (October) and late spring (May). Interestingly, lutein, VAZ and carotenoid pool size (Car : Chl) exhibited brief peaks in early winter and before spring recovery (Fig. 1f h), and these coincided with PRI minima (Fig. 1b). Note that the pigment pools remained high well into May, about a month after the EPS began to increase (compare Fig. 1d,g,h). Thus, reduction of these high carotenoid concentrations in spring lagged 1 month behind the EPS recovery (Fig. 1d), occurred at the end of the spring recovery period, and was more nearly coincident with PRI recovery (Fig. 1b). Correlations for the entire study period between the various photosynthetic parameters (PRI, pigments and ETR) are shown in Table 1 The annual linear correlation coefficients (r 2 ) of key parameters for Pinus contorta and Pinus ponderosa with deep cold days (< 4 C) removed r 2 ETR PRI EPS N : Chl V : Chl A : Chl Z : Chl Lut : Chl b : Chl VAZ : Chl Car : Chl ETR 1 PRI 0.762**** 1 EPS 0.567**** 0.427**** 1 N : Chl V : Chl **** A : Chl * 0.558*** 1 Z : Chl 0.761**** 0.660**** 0.895**** ** Lut : Chl 0.646**** 0.904**** 0.311* * 0.564*** 1 b : Chl 0.730**** 0.760**** 0.311* ** 0.665**** 1 VAZ : Chl 0.374** 0.751**** * *** 0.301* 0.897**** 0.503*** 1 Car : Chl 0.618**** 0.894**** 0.252* ** 0.494*** 0.989**** 0.711**** 0.925**** 1 ETR 1 PRI 0.728**** 1 EPS 0.802**** 0.562*** 1 N : Chl 0.368** * 1 V : Chl 0.508*** **** 0.336** 1 A : Chl * 0.493*** 1 Z : Chl 0.797**** 0.789**** 0.875**** ** Lut : Chl 0.566*** 0.774**** 0.475** **** 1 b : Chl 0.522*** 0.517*** 0.438** *** 0.743**** 1 VAZ : Chl *** *** 0.393** 0.794**** 0.441** 1 Car : Chl 0.471** 0.707**** 0.377** * 0.668**** 0.980**** 0.786**** 0.842**** 1 ETR, electron transport rate; PRI, photochemical reflectance index; EPS, epoxidation state of xanthophyll cycle; N : Chl, neoxanthin : chlorophyll; V : Chl, violaxanthin : chlorophyll; A : Chl, antheraxanthin : chlorophyll; Z : Chl, zeaxanthin : chlorophyll; Lut : Chl, lutein : chlorophyll; b : Chl, b- carotene : chlorophyll; VAZ : Chl, xanthophyll pool size on a chlorophyll basis; Car : Chl, carotenoid pool size on a chlorophyll basis. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P <

6 New Phytologist Research 201 Table 1 (P. contorta) and Table 1 (P. ponderosa). Deep cold days (< 4 C) were removed from the analysis because of decoupling in the PRI response caused by decreased albedo at these temperatures (see Wong & Gamon (2015); full data correlations provided in Table S6). For the remaining dates (temperatures > 4 C), PRI had significant correlations with ETR (r 2 = for P. contorta, andr 2 = for P. ponderosa) and total carotenoid pools (r 2 = for P. contorta, andr 2 = for P. ponderosa). Of all the carotenoid groups examined, the highest correlations with PRI were observed in Lut : Chl for P. contorta (r 2 = 0.904) and in Z : Chl for P. ponderosa (r 2 = 0.798). EPS showed a lower correlation with PRI, in part reflecting its different timing during spring transition (Fig. 1d,f). In P. contorta, EPS had weak correlations with most parameters, except for V : Chl, Z : Chl and ETR (Table 1). Spring transitions A main study focus was the spring transition from inactive to active photosynthesis. Gas exchange provided a direct measurement of photosynthesis during the course of spring recovery (the gas exchange system was not available before spring). Both lightsaturated photosynthetic rate (Fig. 2a) and LUE (Fig. 2b) displayed similar patterns following the increasing spring temperature. During the late winter (March), both parameters had values near zero, illustrating a near-complete photosynthetic deactivation. Spring recovery was initiated in mid-april and fully completed by mid-may for LUE and late May for light-saturated photosynthetic rate. Relatively stable values of c (LUE) and c lmol CO 2 m 2 s 1 (midday photosynthetic rate) were maintained for the rest of the summer (Fig. 2). For both species, the relative response of PRI and pigment parameters exhibited similar differences in timing during spring recovery (Fig. 2c,d). In both species, EPS recovered the earliest, followed by PRI and later Chl : Car (inverted from Fig. 1 to display a positive slope during recovery). To better evaluate the timing and kinetics of spring recovery, sigmoidal fits were applied to all parameters, with half-maximum values during transition providing a metric of the date of spring reactivation for each parameter (Fig. 3). The order and rate of spring recovery for each parameter can be more clearly seen in Fig. 4, which overlays and normalizes the sigmoidal fits for a better visual comparison of the responses shown in Fig. 3. The spring recovery in 2013 of all photosynthetic parameters (Figs 3, 4) occurred rapidly during April and May as temperature and PPFD increased (Fig. 1a). Each photosynthetic parameter exhibited different kinetics, measured as speed of recovery and timing of the half recovery values for both species (Figs 3, 4; Table 2). EPS recovered abruptly in mid-april, showing only one transitional point (Figs 3e, 4). This recovery was coincident with a brief rise in VAZ : Chl values (Fig. 3f), suggesting de novo synthesis of xanthophyll cycle pigments. ETR and LUE responded a bit later, at the end of April (Fig. 4). Light-saturated photosynthetic rate recovered the most gradually, with a half value at the beginning of May for P. contorta and mid-may for P. ponderosa (Fig. 4). The date of PRI recovery matched that of photosynthesis very closely, recovering in early to mid-may (PRI was slightly behind photosynthesis for P. contorta, and ahead for P. ponderosa). Carotenoid pigment pools, expressed relative to chlorophyll, followed PRI (the pigment ratios were inverted to display a positive slope during recovery). Note that Lut : Chl and VAZ : Chl are excluded from Fig. 4 because their variation was closely coincident with that of Car : Chl (half recovery times are included in Table 2). The timing of spring recovery (e.g. half recovery times; Table 2) influenced the correlations between the various parameters (Table 3). PRI exhibited significant correlations with all photosynthetic parameters for P. contorta (Table 3a) and P. ponderosa (Table 3b). The weakest correlation was with EPS, which had the most different timing from that of PRI and other parameters except V : Chl and Z : Chl (Table 2). Photosynthesis had strong correlations with PRI (Table 3), reflecting their similar timing during reactivation (Fig. 4; Table 2). Scaling studies To examine the remote detectability of the seasonal PRI transitions, we compared the leaf-level PRI responses to stand-level (c) (d) Fig. 2 Photosynthetic parameters during the 2013 spring recovery of photosynthesis in Pinus contorta and Pinus ponderosa. The lightsaturated photosynthetic rate and light-use efficiency measured using a gas exchange system (n = 6). (c, d) Relative response of photochemical reflectance index (PRI) and key pigment parameters for (c) P. contorta and (d) P. ponderosa. The 5-d running mean of temperature shows spring recovery temperature trends (dotted line, from Fig. 1). Error bars denote SE of the mean. Grey regions denote winter and summer, and the white region denotes the transitional spring season.

7 202 Research New Phytologist (c) (d) (e) (f) (g) (h) Fig. 3 Sigmoidal fits for all parameters during the 2013 spring recovery (March July) for Pinus contorta (red) and Pinus ponderosa (black). The metrics include electron transport rate (ETR), photosynthesis, (c) light-use efficiency (LUE), (d) photochemical reflectance index (PRI), (e) epoxidation state of the xanthophyll cycle (EPS), (f) xanthophyll pool size on a chlorophyll basis (VAZ : Chl), (g) lutein : Chl and (h) carotenoid pool size on a chlorophyll basis (Car : Chl). P < Vertical dashed lines denote half recovery times (Table 2). responses for the two pine species (P. contorta and P. ponderosa). PRI at the leaf and stand scales exhibited parallel patterns during spring recovery, resulting in strong correlations between leaf- and stand-level PRI (Fig. 5). This provides clear evidence that seasonal PRI shifts associated with leaf-level changes in pigments and photosynthetic activity can be detected at stand scales. To evaluate the generality of these seasonal responses across species and growth stage, we also sampled a third species (P. glauca), and measured mature P. contorta trees in the ground (Fig. 6). In these comparisons, all cases yielded similar spring increases in PRI regardless of culture (pot versus ground), age (seedling versus mature tree), or species (Pinus spp. versus Picea sp.). Discussion The spring recovery of photosynthesis in evergreen conifers is a complex process involving several components that differ in timing and kinetics. Variations between the timing of parameters leading to low correlations can be viewed as a decoupling of component processes during the spring recovery (Porcar-Castell et al., 2012). In our study, EPS recovered early in spring, indicating that the reactivation of the diurnal xanthophyll cycle occurred first and was decoupled from subsequent pigment pool size changes. ETR recovered next, indicating that the photosystem II activation was an early step. The slowest parameters

8 New Phytologist Research 203 Fig. 4 Overlaid sigmoidal fits (from Fig. 3), showing the relative response of photosynthetic parameters during the 2013 spring recovery (March July) for Pinus contorta and Pinus ponderosa. The dates range from the late winter in March to the summer in July. EPS, epoxidation state of the xanthophyll cycle; ETR, electron transport rate; PRI, photochemical reflectance index; Chl : Car, carotenoid pool size on a chlorophyll basis. Table 2 The half recovery times in Julian days for the sigmoid fit of photosynthetic parameters during the 2013 spring recovery (March July) for Pinus contorta and Pinus ponderosa (note that these are arranged in order of recovery for P. contorta) P. contorta P. ponderosa EPS ETR LUE Photosynthesis PRI VAZ : Chl Chlorophyll : carotenoid Lutein : chlorophyll EPS, epoxidation state of xanthophyll cycle; ETR, electron transport rate; LUE, light-use efficiency; PRI, photochemical reflectance index; VAZ: Chl, xanthophyll pool size on a chlorophyll basis; Car: Chl, carotenoid pool size on a chlorophyll basis. to recover were PRI and several pigment pool sizes (expressed as ratios), probably because of the energy and time constraints associated with synthesis of chlorophyll and carotenoid pigments. Photosynthetic gas exchange recovered the most gradually, and only fully recovered once all other parameters had done so. These findings show that these several parameters associated with photosynthesis recover with different kinetics, and that kinetic comparison can be used to infer mechanism. The similar timing of PRI recovery to that of pigments pools, and the different timing from EPS, indicates that PRI during spring transition is largely driven by pigment pools and is less affected by actual conversion of the xanthophyll cycle pigments. This contrasts with short-term (e.g. diurnal) studies where changes in EPS largely drive the rapid changes in PRI (Gamon et al., 1992, 1997; Pe~nuelas et al., 1995; Garbulsky et al., 2011). These two responses have been termed the constitutive and facultative responses, respectively, as they reflect different mechanisms operating on different time-scales, and with different spectral responses (Gamon & Berry, 2012; Wong & Gamon, 2015). Leaf pigments exhibited at least two seasonal responses that could be associated with photoprotection. Changes in the xanthophyll cycle composition, as measured by the reduction in EPS, increase the potential for photoprotection (Demmig- Adams & Adams, 1996; Horton et al., 1996; Verhoeven et al., 1996, 1999; H uner et al., 1998; Oquist & H uner, 2003). The more slowly changing carotenoid : chlorophyll pigment ratios could provide further sustained photoprotection with increased concentrations of carotenoids and loss of chlorophyll, which could decrease light absorption and avoid photooxidation of chlorophyll and enhance energy dissipation (Ottander et al., 1995; Demmig-Adams & Adams, 1996; Vogg et al., 1998; Gilmore & Ball, 2000; Adams et al., 2002, 2004). Of the different carotenoids studied, several exhibited particularly close timing (and hence strong correlations) with PRI. In particular, lutein : chlorophyll, VAZ : chlorophyll and total carotenoid : chlorophyll values more closely matched the PRI transitions than EPS, indicating that it is these pigment pools (relative to chlorophyll) rather than the xanthophyll cycle per se that largely drive the seasonal PRI changes. The fact that PRI slightly precedes pigment pool changes can probably be explained by the early EPS transition, which surely influences PRI. We propose that the PRI transitions are a weighted average of two different processes, a reactivation of the xanthophyll cycle (indicated by an increase in EPS, with a small influence on PRI), and the re-establishment of summer-active pigment concentrations (indicated by pigment pool size changes, with the predominant influence on PRI). The net effect of this is that the spring PRI increase is nearly coincident with the spring reactivation of photosynthesis, offering an optical indicator (within a few days) of the effective start of the growing season. Consistent with previous reports in winter evergreens (Demmig-Adams et al., 2012; Verhoeven, 2014), our results provide evidence that the xanthophyll cycle is more active during warmer seasons (spring and summer), and enters a period of sustained inactivity in winter. Our results also suggest that several carotenoids in addition to VAZ may be involved in wintertime down-regulation and photoprotection. Our findings are consistent with other studies suggesting a role for lutein and other carotenoids in winter down-regulation and photoprotection (Ottander et al., 1995; Oh et al., 2013). Lutein is the most abundant carotenoid and was significantly correlated with PRI. Lutein undergoes seasonal cycles with increased concentrations in the winter and lowered concentrations in the summer (Esteban et al.,

9 204 Research New Phytologist Table 3 The linear correlation coefficients (r 2 ) of key parameters for Pinus contorta and Pinus ponderosa during spring recovery (March July) r 2 ETR Ps LUE PRI EPS N : Chl V : Chl A : Chl Z : Chl Lut : Chl b : Chl VAZ : Chl Car : Chl ETR 1 P s 0.878**** 1 LUE 0.902**** 0.734**** 1 PRI 0.832**** 0.838**** 0.824**** 1 EPS 0.566** 0.521** 0.481** 0.313* 1 N : Chl V : Chl *** A : Chl *** 1 Z : Chl 0.815**** 0.756**** 0.711*** 0.594** 0.885**** * Lut : Chl 0.678*** 0.714**** 0.682*** 0.921**** ** 1 b : Chl 0.779**** 0.734**** 0.794**** 0.919**** ** 0.863**** 1 VAZ : Chl 0.438** 0.460** 0.519** 0.777**** ** **** 0.749**** 1 Car : Chl 0.630*** 0.654*** 0.662*** 0.908**** * 0.379* 0.988**** 0.880**** 0.937**** 1 ETR 1 Ps 0.830**** 1 LUE 0.904**** 0.783**** 1 PRI 0.861**** 0.833**** 0.867**** 1 EPS 0.829**** 0.717*** 0.757**** 0.587** 1 N : Chl * V : Chl 0.453** * **** A : Chl ** 1 Z : Chl 0.862**** 0.880**** 0.864**** 0.830**** 0.874**** ** Lut : Chl 0.568** 0.785**** 0.651*** 0.813**** 0.405* *** 1 b : Chl 0.671*** 0.714*** 0.751**** 0.840**** 0.425* *** 0.858**** 1 VAZ : Chl * 0.306* 0.547** ** 0.353* 0.836**** 0.632*** 1 Car : Chl 0.492** 0.701*** 0.578** 0.774**** 0.306* * 0.623*** 0.986**** 0.857**** 0.902**** 1 ETR, electron transport rate; Ps, light-saturated photosynthesis; LUE, light-use efficiency; PRI, photochemical reflectance index; EPS, epoxidation state of the xanthophyll cycle; N : Chl, neoxanthin : chlorophyll; V : Chl, violaxanthin : chlorophyll; A : Chl, antheraxanthin : chlorophyll; Z : Chl, zeaxanthin : chlorophyll; Lut : Chl, lutein : chlorophyll; b : Chl, b-carotene : chlorophyll; VAZ : Chl, xanthophyll pool size on a chlorophyll basis; Car : Chl, carotenoid pool size on a chlorophyll basis. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P <

10 New Phytologist Research 205 Fig. 5 The 2013 spring recovery (March July) of photochemical reflectance index (PRI) at the leaf (circle) and stand (triangle) scales for Pinus contorta and Pinus ponderosa. Inserts show correlations between leaf- and stand-level PRI. Error bars denote SE of the mean. Fig. 6 Leaf-scale photochemical reflectance index (PRI) responses in three evergreen seedlings and mature Pinus contorta. Error bars denote SE of the mean. 2008; Oh et al., 2013). Its role in photoprotection has been suggested to be to quench triplet chlorophyll (Jahns & Holzwarth, 2012). b-carotene was also strongly associated with PRI and has been noted to quench singlet oxygen (Telfer, 2005). Together these major carotenoids undergo seasonal changes in concentration in evergreen leaves, presumably to protect various photosynthetic components from photodamage. Further work is needed to clarify the individual photoprotective roles under winter conditions. However, PRI appears to detect the effect of all carotenoid and chlorophyll pools combined (as expressed in the carotenoid : chlorophyll ratios). Most individual pigments increased their correlations with PRI when examined on a chlorophyll basis rather than expressed as concentrations on a leaf area basis (compare Tables 1 and 3 with Tables S7 and S8). Interestingly, differences were observed in the total chlorophyll concentrations between the two summers, probably as a result of differences in leaf age and nutrient status, but PRI, EPS and pigment ratios remained similar between summers (Fig. 1). This suggests that PRI will be a robust measure of spring transition as it gives consistent annual results regardless of the developmental stage or nutrient status of the leaves. Clearly, PRI has the potential to detect both the rapid xanthophyll cycle (a facultative process) and seasonal change in the relative concentrations of carotenoid and chlorophyll pigments (a constitutive process) (Gamon & Berry, 2012; Wong & Gamon, 2015). Consequently, a previous explanation that PRI detects EPS (and thus LUE) during spring transitions for boreal forests (Nichol et al., 2002) may not be entirely correct from a mechanistic standpoint. The poor link between PRI and EPS observed here is consistent with the poor relationships between PRI and nonphotochemical quenching recently reported for boreal forests in Finland (Porcar-Castell et al., 2012), suggesting that the spring activation affecting PRI is not primarily or exclusively mediated by the xanthophyll cycle. Instead, our results demonstrated that seasonally changing PRI was more strongly correlated with gradually changing pigment ratios than with EPS, whether evaluated annually or during spring recovery, which is consistent with previous reports for other evergreen vegetation indicating a strong role for pigment pool sizes in seasonal PRI patterns (Stylinski et al., 2002; Filella et al., 2009; Garrity et al., 2011; Porcar-Castell et al., 2012). We note that large albedo shifts during periods of deep winter cold can diminish the annual correlations between PRI and pigment pools (Tables 1, S6), and should be considered as a special case of PRI transitions that can be distinguished spectrally from EPS or pigment pool size effects (Wong & Gamon, 2015). Remote sensing provides several possible metrics of ecosystem photosynthetic activation, including PRI (Nichol et al., 2000) and solar-induced fluorescence (Meroni et al., 2009; Frankenberg et al., 2011). In theory, these parameters should respond in parallel, if all the photosynthetic components are activated simultaneously. However, in our study, the reversal of wintertime fluorescence quenching (and corresponding ETR reactivation) preceeded PRI, as did EPS and xanthophyll cycle activity. Thus, it appears that photosystem II and electron transport activity are revived before PRI or gas exchange recovery during spring photosynthetic activation. Solar-induced fluorescence (SIF) may respond differently from pulse-amplitude modulated fluorescence (Porcar-Castell et al., 2014), so future studies should consider potential differences in relation to the timing of photosynthetic recovery. Further development of PRI and SIF could lead to direct remote sensing approaches for estimating seasonal photosynthetic activity by detecting physiological responses to the onset of winter and subsequent spring recovery. However, as shown here, the evergreen photosynthetic response to changing seasons is complex, showing multiple components. Future studies involving optical detection of seasonal activation and deactivation

11 206 Research New Phytologist should consider these component mechanisms in a wider array of species as a means of understanding the physiological changes driving changing optical signals. Synchrony in the leaf and closed-canopy stand responses shows that PRI probably detects pigment pool size changes associated with photosynthetic reactivation at the two scales (Fig. 5). This observation agrees with other studies that showed coincident leafand stand-level PRI responses for closed-canopy stands (Gamon & Qiu, 1999; Stylinski et al., 2002). Consequently, we conclude that the large change between the winter and summer values provides a clear optical signal that can be detected even at the stand scale, showing promise for remote detection of spring PRI transitions. This conclusion is consistent with previous reports that airborne PRI measurements detect spring transitions in photosynthetic LUE in boreal forest stands (Nichol et al., 2000, 2002). The large seasonal PRI response was not an artifact of pot culture, as mature P. contorta in a more natural setting also exhibited similar responses (Fig. 6). The response also appeared to be common to all three species studied, suggesting that it may be a general response of all conifers exposed to extreme winter cold, although further studies on more species are needed to confirm this. However, we caution that additional variations in the PRI signal can arise when monitoring at the ecosystem level because of varying background material, canopy structure and illumination (Barton & North, 2001), so further ecosystem-level studies coupling airborne and satellite remote sensing with ground-measured physiology and fluxes during seasonal transitions are warranted. This study helps clarify the mechanism of PRI transitions associated with spring photosynthetic activity in evergreen conifers. This constitutive PRI response driven primarily by seasonally changing pigment ratios could help improve photosynthesis models driven from remote sensing. The MODIS satellite sensor has a band #11 that can serve as the PRI 531-nm band, and a band #12 near 550 nm that can provide a suitable reference (Rahman et al., 2004; Middleton et al., 2009). While these ocean colour bands are not normally processed for terrestrial regions, further work should explore the utility of these MODIS time bands for monitoring spring activation of photosynthesis in evergreen-dominated ecosystems. At the same time, a variety of automated sensors capable of detecting seasonal PRI transitions are now available and could assist ground validation efforts. We recommend that such work be conducted at flux tower sites to facilitate comparison between optical signals and ecosystem C fluxes. If routine observations of PRI seasonal transitions can be extended to spaceborne satellite measurements, this would provide a potent global monitoring tool for improving our detection of changing evergreen phenology, a topic that has been difficult to assess using remote sensing. Detecting the largely invisible seasonal down-regulation and reactivation of photosynthesis can provide a measure of growing season length, which is changing for northern latitudes (Myneni et al., 1997; Randerson et al., 1999; Slayback et al., 2003). Systematic PRI measurements in conjunction with ecosystem C flux measurements could help reduce the significant remaining uncertainties in the biospheric global C budget associated with changing growing season length (Goulden et al., 1996; Le Quere et al., 2009). Acknowledgements This work was supported by funding to J.A.G. from the Natural Sciences and Engineering Research Council of Canada (NSERC), Alberta Innovates (icore and AITF) and the Canada Foundation for Innovation (CFI). We thank Steve Williams for his advice on and assistance with plant care. Thanks to Morgan Randall and Saulo Castro for the initial set up of the plants and Olga Kovalchuk for assistance in summer 2013 data collection. We also would like to thank Keith Tierney and his research group for generous usage of laboratory equipment and sample storage and David Hik, Rolf Vinebrooke and the three anonymous reviewers for helpful comments on the manuscript. Thanks to Ingo Ensminger and Laura Junker for advice on HPLC calibration. Lee Vierling and Steve Garrity provided stimulating discussions on invisible evergreen phenology. References Adams WW, Demmig-Adams B, Rosenstiel TN, Brightwell AK, Ebbert V Photosynthesis and photoprotection in overwintering plants. Plant Biology 4: Adams WW, Demmig-Adams B, Verhoeven AS, Barker DH Photoinhibion during winter stress: involvement of sustained xanthophyll cycle-dependent energy dissipation. Australian Journal of Plant Physiology 22: Adams WW, Zarter CR, Ebbert V, Demmig-Adams B Photoprotective strategies of overwintering evergreens. BioScience 54: Apps MJ, Kurz WA, Luxmoore RJ, Nilsson LO, Sedjo RA, Schmidt R, Simpson LG, Vinson TS Boreal forests and tundra. Water Air and Soil Pollution 70: Baker NR Chlorophyll fluorescence: a probe of photosynthesis in vivo. Annual Review of Plant Biology 59: Baroli I, Niyogi KK Molecular genetics of xanthophyll-dependent photoprotection in green algae and plants. Philosophical Transactions of the Royal Society of London Series B Biological Sciences 355: Barton CVM, North PRJ Remote sensing of canopy light use efficiency using the photochemical reflectance index model and sensitivity analysis. Remote Sensing of Environment 78: Bigras FJ, Ryyppo A, Lindstrom A, Stattin E Cold acclimaion and deacclimation of shoots and roots of conifer seedlings. In: Bigras FJ, Colombo SJ, eds. Conifer cold hardiness. Dordrecht, the Netherlands: Kluwer Academic, Bj orkman O, Demmig-Adams B Regulation of photosynthetic light energy capture, conversion, and dissipation in leaves of higher plants. In: Schulze E-D, Caldwell M, eds. Ecophysiology of photosynthesis. Berlin, Heidelberg, Germany: Springer, Butler WL Energy distribution in photochemical apparatus of photosynthesis. Annual Review of Plant Physiology and Plant Molecular Biology 29: Canadell JG, Le Quere C, Raupach MR, Field CB, Buitenhuis ET, Ciais P, Conway TJ, Gillett NP, Houghton RA, Marland G Contributions to accelerating atmospheric CO 2 growth from economic activity, carbon intensity, and efficiency of natural sinks. Proceedings of the National Academy of Sciences, USA 104: Ciais P, Reichstein M, Viovy N, Granier A, Ogee J, Allard V, Aubinet M, Buchmann N, Bernhofer C, Carrara A et al Europe-wide reduction in primary productivity caused by the heat and drought in Nature 437: