Light inhibition of foliar respiration in response to soil water availability and seasonal changes in temperature in Mediterranean holm oak ( Quercus ilex ) forest

In the present study we investigated variations in leaf respiration in darkness (RD) and light (RL), and associated traits in response to season, and along a gradient of soil moisture, in Mediterranean woodland dominated by holm oak (Quercus ilex L.) in central and north-eastern Spain respectively. On seven occasions during the year in the central Spain site, and along the soil moisture gradient in north-eastern Spain, we measured rates of leaf RD, RL (using the Kok method), light-saturated photosynthesis (A) and related light response characteristics, leaf mass per unit area (MA) and leaf nitrogen (N) content. At the central Spain site, significant seasonal changes in soil water content and ambient temperature (T) were associated with changes inMA, foliar N, A and stomatal conductance. RD measured at the prevailing daily T and in instantaneous R–T responses, displayed signs of partial acclimation and was not significantly affected by time of year. RL was always less than, and strongly related to, RD, and RL/RD did not vary significantly or systematically with seasonal changes in T or soil water content. Averaged over the year, RL/RD was 0.66 0.05 s.e. (n = 14) at the central Spain site. At the north-eastern Spain site, the soil moisture gradient was characterised by increasing MA and RD, and reduced foliar N, A, and stomatal conductance as soil water availability decreased. Light inhibition of R occurred across all sites (mean RL/RD = 0.69 0.01 s.e. (n = 18)), resulting in ratios of RL/A being lower than for RD/A. Importantly, the degree of light inhibition was largely insensitive to changes in soil water content. Our findings provide evidence for a relatively constrained degree of light inhibition of R (RL/RD ~ 0.7, or inhibition of ~30%) across gradients of water availability, although the combined impacts of seasonal changes in both T and soil water content increase the range of values expressed. The findings thus have implications in terms of the assumptions made by predictive models that seek to account for light inhibition of R, and for our understanding of how environmental gradients impact on leaf trait relationships in Mediterranean plant communities. Additional keywords: Kok effect, leaf functional traits, leaf dark respiration, leaf light respiration, leaf mass per unit area, nitrogen, photosynthesis, plasticity, season, soil moisture, temperature. Received 27 January 2017, accepted 23 July 2017, published online 17 August 2017 CSIRO PUBLISHING Functional Plant Biology https://doi.org/10.1071/FP17032 Journal compilation CSIRO 2017 www.publish.csiro.au/journals/fpb


Introduction
Plant respiration (R) is an integral component of the terrestrial global carbon cycle, with between 0.2 and 0.8 of the carbon assimilated by the photosynthesis during the day consumed through autotrophic respiratory processes (Gifford 1995).About half the CO 2 released in plant respiration comes from leaves (Xu et al. 2001;Atkin et al. 2007), with this proportion likely to be even greater for plants where CO 2 assimilation (A) is limited by abiotic stresses, including low nutrient availability (Grassi et al. 2002;Turnbull et al. 2005;Whitehead et al. 2005;Heskel et al. 2012Heskel et al. , 2013) ) and under drought conditions (Chaves et al. 2002).At an individual scale, mitochondrial respiration plays a key role in determining the growth and survival of plants (Hurry et al. 1995), as it is associated with the production of energy and carbon skeletons essential for cellular maintenance and biosynthesis.
Although leaf R takes place in both light (R L ) and darkness (R D ), leaf R is typically lower during the day when photosynthesis is concurrently occurring (Brooks and Farquhar 1985), even when re-fixation of respiratory CO 2 is taken into account (Pärnik et al. 2007).The metabolic basis of the light inhibition of leaf R is beginning to be better understood (Tcherkez et al. 2017), with factors such as cellular energy status, demand for tricarboxylic acid (TCA) cycle intermediates, engagement of the pentose phosphate pathway and/or rates of photorespiration (V o ) being involved in the degree of inhibition (Hurry et al. 2005;Tcherkez et al. 2005Tcherkez et al. , 2008Tcherkez et al. , 2012;;Buckley and Adams 2011).The extent to which R is reduced in the light is, potentially, highly variable among species and environments.For example, when measured at a common temperature (T), the rate of respiration in the light can be as little as 0.2 of the darkened rate (Atkin et al. 2000a;Ayub et al. 2011).In contrast, light may have little to no inhibitory effect in some cases, particularly at low measuring temperatures (Atkin et al. 2000b;Zaragoza-Castells et al. 2007).These large potential variations between measured R L and R D under different experimental conditions are important because the level to which they are actually expressed in the field will determine the extent to which variations in R impact on net CO 2 exchange in individual plants and whole ecosystems.Failure to account for light inhibition of leaf R leads to large over-and under-estimates of ecosystem respiration and net primary productivity respectively (Lloyd et al. 2002;Wohlfahrt et al. 2005;Atkin et al. 2006;Wingate et al. 2007;McLaughlin et al. 2014).
Establishing the extent to which species differ in rates of R D and R L , and the impact of natural environmental gradients on variations of both processes is crucial to successful incorporation of light inhibition of leaf R into large-scale models.Furthermore, little is known about the impact of soil water status or water stress on plant R (Atkin and Macherel 2009).The response of R D to leaf water content may lag behind that of photosynthesis and may also be biphasic i.e. it decreases in the initial stages of water stress (lower energy demands for growth) and may increase with additional stress-related demands, e.g.osmoregulation or induced senescence (Ghashghaie et al. 2001;Gulias et al. 2002;Flexas et al. 2005Flexas et al. , 2006)).The environmental response of R D will also be complicated by the fact that large shifts in daily mean T often co-vary with fluctuations in water availability, with drought often being associated with high temperatures.Several controlled-environment studies have reported large variations in leaf R D among different species (Atkin and Day 1990;Poorter et al. 1990;Loveys et al. 2003;Tjoelker et al. 2005;Atkin et al. 2008), with a smaller number of laboratory-based studies showing that genotype is also a strong determinant of R L (Villar et al. 1994;Atkin et al. 1997).R D also varies in response to changes in the abiotic environment, both in nature and under laboratory-based conditions (Larigauderie and Körner 1995;Ryan 1995;Atkin et al. 2000b;Griffin et al. 2002aGriffin et al. , 2002bGriffin et al. , 2004;;Turnbull et al. 2005;Wright et al. 2006;Xu and Griffin 2006;Tjoelker et al. 2009;Searle et al. 2011).R D often acclimates to changes in long-term growth T (Atkin et al. 2000a(Atkin et al. , 2005;;Atkin and Tjoelker 2003;Wythers et al. 2005;Ow et al. 2008a, Way andOren 2010), so that the T response of respiration to short-term and long-term changes in T is often different.Likewise, there is evidence from laboratorybased studies that light inhibition of leaf R is environmentally dependent (e.g. as a result of changes in measurement T (Atkin et al. 2000a;Zaragoza-Castells et al. 2007;Shapiro et al. 2004;Griffin and Turnbull 2013), and/or atmospheric growth CO 2 concentration (Shapiro et al. 2004;Wang et al. 2001Wang et al. , 2004;;Tissue et al. 2002)).Although we have previously shown that the balance between R D and R L changes in response to field gradients of soil fertility (Atkin et al. 2013;Heskel et al. 2012Heskel et al. , 2013) ) and imposition of water stress under controlled conditions (Ayub et al. 2011;Crous et al. 2012), to date, no study has reported how gradients in soil moisture impact on light inhibition of leaf R under natural, field conditions.We might predict that drought and/or high T induced changes in leaf metabolism, particularly in V o , would result in predictable changes in light inhibition of R (Griffin and Turnbull 2013).Moreover, we lack data on how the balance between R L and R D varies seasonally (Heskel et al. 2014).It is vital we establish the extent of inhibition of R in areas of contrasting water availability if we want to more accurately predict future rates of C exchange in water-limited environments.Overall, we lack comprehensive field-based studies that quantify variations in R D and R L among dominant species growing along natural gradients of water availability.
Mediterranean ecosystems offer excellent opportunities to test changes in respiration under water-limited conditions, such as those that now prevail (Peñuelas and Boada 2003;Peñuelas et al. 2013) and the even drier conditions that are projected for the coming decades (IPCC 2007).In Mediterranean ecosystems, the availability of water is the greatest environmental constraint on plant growth, due to the occurrence of high summer temperatures and low rainfall.Here, our aim was to determine if the dominant Mediterranean tree Quercus ilex L. (Holm oak) and co-occurring trees of Mediterranean forests responded to seasonal changes in, and natural gradients of, water availability, with changes in R (as measured by CO 2 evolution), and particularly in the inhibition of R by light.We have previously measured large seasonal shifts in T response curves of R D in Q. ilex subsp.ballota in central Spain that were consistent with thermal acclimation (Zaragoza-Castells et al. 2008).However, it is possible that seasonal variations in water availability may have contributed to such shifts in the daily T response curves (Crous et al. 2011;Rodríguez-Calcerrada et al. 2011).Our present study extends this research by quantifying seasonal changes in both R D and R L and associated leaf traits (A, M A and N content) over a 12 month period in the same forest (Villar de Cobeta, central Spain) in Q. ilex subsp.ballota trees.We also increased the geographical and species scope of our study by conducting an additional single mid-summer campaign in the Prades Mountains (Catalonia, north-eastern Spain), with Q. ilex subsp.ilex forest along a gradient of decreasing water availability from a stream course in the bottom of a valley to dry sites approaching the ridge line.In addition, at the drier end of this gradient, precipitation has been further experimentally restricted for the last 13 years, reducing soil moisture by 11% with respect to ambient conditions (Ogaya and Peñuelas 2007).Based on previous findings that conditions which tend to suppress V o (e.g.high [CO 2 ] and/or low [O 2 ]) also decrease R L /R D (Wang et al. 2001;Shapiro et al. 2004;Crous et al. 2012;Griffin and Turnbull 2013), and without considering other processes (e.g.increased use of stored carbon reserves (e.g.citrate) in the light) which might influence R L in the short term (Tcherkez et al. 2012), we hypothesised that R L /R D would be lowest where water availability was greatest, and increase at drier sites where increase photorespiration results from stomatal limitations on gas exchange.It is worth noting that in considerations of the influence of water availability on carbon assimilation, it is possible to interpret a link between photorespiration and the degree of light inhibition of R measured in the Kok effect in several ways: (i) there may be a direct effect of water availability on internal CO 2 mole fraction, which in turn affects V o and thus R L /R D ; (ii) water availability may have specific photorespiration-independent effects on R L /R D ; or (iii) water availability may affect apparent R L /R D because the Kok-effect measurement is itself influenced (at least in part) by photorespiration (Tcherkez et al. 2012).Here our focus is to determine patterns of field response rather than attempt to provide a resolution to this issue.

Study sites
Villar de Cobeta This field site was located 175 km north-east of Madrid in the Iberic System Mountain Range (950 m above sea level) at Los Cerrillos Biological Station, Villar de Cobeta (40 48 0 N, 2 12 0 W), within the Alto Tajo Natural Park (Guadalajara, Spain).The experimental work was located in two sitesa steep (20 ), south-facing upper slope which was the location for our previous work (Zaragoza-Castells et al. 2008), and an adjacent lower slope, stream-side site.Quercus ilex L. subsp.ballota dominated the vegetation at both sites.The area has a continental Mediterranean climate, with hot, dry summers and cold winters, and is characterised by large diurnal and seasonal variations in air T (diurnal variations in T in the order of 25 C are common, with leaf temperatures potentially reaching near 50 C in summer and À15 C in winter).Both sites have shallow, poorly developed soils formed mainly from Cretaceous and Jurassic limestone with a limited capacity of the soil to retain water (Valladares et al. 2008).Meteorological stations provided environmental data as described in Zaragoza-Castells et al. (2008).From December 2006 to November 2007 we monitored microclimatic conditions at each of the study sites.At the upper slope site, we installed a HOBO weather station equipped with a rain gauge (RGA-M0XX) and a soil moisture sensor (S-SMA-M003) inserted at 30 cm depth.Readings of each sensor were recorded every 30 min with a data logger (HOBO H21-001 all components from HOBO Onset Computer Corporation Bourne).At the lower slope site, we installed a second station with a soil moisture probe (ECH 2 O EC-20 Decagon Devices).Readings were recorded every 30 min with a HOBO H08-006-04 data logger.

Prades
The second study was undertaken in the Prades Mountains in southern Catalonia (NE Spain; 41 21 0 N, 1 2 0 E), also at 950 m above sea level on a south-facing slope (20 ).The soil is a Dystric Cambisol over Paleozoic schist, and its depth ranges from 35 to 90 cm.This forest has been undisturbed for the last 60 years, and the maximum height of the dominant species is ~6-10 m.Plant community composition is strongly influenced by topographical changes in soil water availability (Table 1).In this study we identified four community types: (1) riparian forestthe moistest sites along stream courses in valleys dominated by several deciduous species (Tilia platyphyllos Scop., Sorbus torminalis (L.) Crantz and Acer monspessulanum L.); (2) mid-slope In addition to the natural gradient described above, at Prades we sampled in a long-term rainfall exclusion experiment established in communities 3 and 4 (Ogaya and Peñuelas 2007) to match the conditions projected in climatic and ecophysiological models for Mediterranean forests in the coming decades (Sabaté et al. 2002;IPCC 2007).In each of these communities, four control and four treatment plots 15 Â 10 m were established at the same altitude along the mountain face.In the treatment plots, rainfall was partially excluded through the suspension of PVC strips 0.5-0.8m above the soil (covering 30% of the soil surface) and the excavation of a ditch 0.8 m deep at the upper part of the plots to intercept runoff water.Water intercepted by the strips and ditches was conducted around the plots, below their bottom edges.The other three plots received no treatment and acted as control plots.The rain exclusion treatment began in January 1999, 13 years before the sampling for this study.An automated meteorological station installed between the rainfall exclusion plots monitored T, photosynthetically active radiation (PAR), air humidity and precipitation continuously during the experiment.Soil water content was determined gravimetrically at 30 cm depth at all sites at the midpoint of the experimental campaign (July 2013).

Leaf sampling and gas-exchange measurements
At the Villar de Cobeta study site, we performed seven measurement campaigns from December 2006 to November 2007.Four individual trees were identified at both the upper and lower slope sites, with each tree representing an independent replicate.For each sampling campaign, leaf physiological measurements were made using a single south-east facing, attached, fully-expanded, mature leaf from each replicate tree.Different leaves were used on each campaign, with the measured leaves being harvested for determination of chemical-structural characteristics at the end of each sampling campaign.
Within the Villar de Cobeta seasonal study, additional measurements of the instantaneous response of R D to T were made at changing ambient air T and at ambient relative humidity (typically 35-60%) to investigate the extent of thermal acclimation of R D .Measurements of leaf R took place at regular intervals during the day and night; during the day, leaves were darkened for 30 min before measurement to avoid post-illumination transients.As Zaragoza-Castells et al. (2008) previously found that there were no differences in leaf R measured during the day and night at equivalent T; here, T response curves were fitted to data from a combination of night and day measurements.The measurements were carried out at regular intervals (approximately every 2 h) through the day (interspersed between measurements of R on the same leaves), with net CO 2 exchange (A net ) being measured at the prevailing ambient irradiance.On each sampling month, measurements were made on four replicate leaves.
The Prades study took place over a 2 week period in late July 2013 (i.e.mid-summer when the gradient of water availability was most pronounced).At each site, we sampled detached branches of sun-lit, upper canopy foliage (using a pruning pole) from six individuals of each of three of the most abundant species (Table 1) at each site along the gradient.Species composition changes along the gradient, but it was possible to sample several species on multiple, consecutive sites (Table 1).Thus, in addition to providing insights into how leaf gas exchange of the dominant species might vary among sites, our sampling strategy enabled an assessment of how individual species responded to site-to-site variations in environment (especially Q. ilex, which was sampled at five of the six sites).Sampling took place in the early evening; stems were immediately re-cut under water and the branches transported to a nearby field laboratory and allowed to equilibrate overnight for subsequent gas exchange measurements through the morning and early afternoon period.Past work has shown that leaf gas exchange rates can be comparable in cut branches and in situ leaves of forest species for many hours after removal (Mitchell et al. 1999;Turnbull et al. 2003).In addition, all gas exchange measurements were corrected for C i according to work by Kirschbaum and Farquhar (1987) (see details below), so any partial stomatal closure that may have occurred was accounted for.
For both experimental study sites (i.e. at the Villar de Cobeta and Prades field sites), light response curves of A net were measured on the most recently fully expanded leaves using LI-6400 portable photosynthesis systems (LI-COR BioSciences) equipped with a CO 2 controller and a with red-blue light source (6400-02B).The seasonal measurements at Villar de Cobeta were made at the prevailing daytime T (see Fig. 1) and the moisture gradient measurements at Prades were made with the block T set to the mid-summer prevailing T of 25 C.In all cases, light-response measurements were conducted under ambient [CO 2 ] of 400 mmol mol -1 controlled using the LI-6400 control system in a large and well ventilated field laboratory; as such, it was not necessary to correct for CO 2 diffusion through the chamber gasket (Pons and Welschen 2002).Light-saturated photosynthesis (A sat ) was measured at 1500 mmol m -2 s -1 photosynthetic photon flux density (PPFD) and a RH of 40-70%, and after leaves had been exposed to saturating irradiance in the cuvette for 10-20 min.After measurement of A sat , the irradiance response of net CO 2 exchange was measured, beginning at 1000 mmol m -2 s -1 , followed by 500, 250, 150, 100, 90, 80 and then every 5 mmol m -2 s -1 to 10 and ending at zero mmol m -2 s -1 (i.e.darkness).Additional measurements of net CO 2 exchange in darkness were conducted after a further 10 min of darknessthese were not statistical different from values of R D measured directly at the end of light response curves.Flow rate through the chamber was set to 300 mmol s -1 .Relationships between key gas exchange characteristics (A sat , R D ) and soil water content have been included for reference in Fig. S2, available as Supplementary Material to this paper.
To estimate rates of leaf respiration in the light (R L ), we used the Kok (1948) method, where R L was estimated from the y-axis intercept of a first order linear regression fitted to A-irradiance plots to measurements made over the 20-60 mmol photon m -2 s -1 irradiance range (Ayub et al. 2011); in some species, data was curvilinear at irradiances above 70 mmol m -2 s -1 PPFD (data now shown).Averaged across all sites, the r 2 values of the first order regression fits over the 10-50 mmol photons m -2 s -1 range was 0.97 AE 0.01.Using this method, intercellular CO 2 concentrations (C i ) tend to increase with decreases in irradiance, resulting in reduced rates of photorespiration and increased rates of carboxylation; this in turn results in a decrease in the slope of A net -irradiance plots in the linear region (Villar et al. 1994).Because of this, rates of R L were )  See Table 3 for results of two-way ANOVAs testing for differences with time and site.For (h), area-based rates of leaf R L plotted against corresponding rates of R D .Data shown are for individual leaves sampled during the year.The dashed line shows the 1 : 1 relationship.
adjusted (by iteration) to ensure that the intercept of plots of photosynthetic electron transport (J) against irradiance are minimised (Kirschbaum and Farquhar 1987).J was calculated according to Farquhar and von Caemmerer (1982): where G* is the CO 2 compensation point in the absence of R L (37 mmol mol -1 at 25 C (von Caemmerer and Farquhar 1981)], with G * at each measurement T calculated according to Brooks and Farquhar (1985).Rates of oxygenation and carboxylation by Rubisco (V o and V c respectively) at any given irradiance were calculated according to Farquhar and von Caemmerer (1982).

Leaf structural traits and chemical composition
Leaves used for the light response curves were weighed for fresh mass, photographed (to enable subsequent calculation of leaf area using Image J software (http://rsbweb.nih.gov/ij/,accessed December 2013)) and then oven-dried at 70 C to constant mass.The mass and area data were used to determine ratios of leaf dry mass to leaf area (M A ). Subsequently, leaf samples were ground in a ball mill and analysed for tissue nitrogen and phosphorus using either a Technicon Autoanalyser II (Bran + Luebbe Pty Ltd; Villar de Cobeta samples) or a Eurovector 3011 elemental analyser (Prades samples).For the three species in sites 3-6 of the Prades forest study, we had access to previous measurements of tissue N in July over 4 different years (Ogaya and Peñuelas 2008).Based on the finding that leaf N was not significantly influenced by soil moisture content or rain exclusion treatment, we used a 4 year site average as our measure of leaf N for these species at these sites (reported only in Table 4).Leaf carbohydrate content was determined during the Villar de Cobeta study (Loveys et al. 2003), but as there was no significant explanatory power of carbohydrate status on respiratory parameters (see Fig. S1) we did not repeat these measurements in the subsequent Prades campaign.Leaf tissue relative water content (RWC) was determined according to Koide et al. (1989) for a representative leaf adjacent to the measurement leaf at the same time as gasexchange measurements.

Statistical analyses
Repeated-measures analysis of variance (ANOVA) was used to test for the effects of time of year and site in the Villar de Cobeta seasonal study and for the effects of plant species and site in the Prades gradient study using SPSS (ver.16.0; SPSS Inc.) and the R statistical platform (R Development Core Team 2008) with post hoc comparisons of sites being made using least significance difference tests (P < 0.05).Species was treated as a nominal variable, and site as an ordinal variable (i.e. to account for increasing water availability) for these analyses.Differences were considered significant if probabilities (P) were less than 0.05.

Results
Villar de Cobeta seasonal study -Q.ilex subsp.ballota Volumetric measurements of water content confirmed strong seasonal changes in soil water content from highest levels in spring (March, ~15%) to lowest levels in late summer (August-September, ~5%; Fig. 1a).The lower slope site had slightly lower volumetric water contents than the upper slope site, but there was no significant difference between the sites at the times measurements were made (Fig. 1a).This is a reflection of the fact that rainfall in the region was unusually high in winter-spring 2006-2007.Leaf mass per unit area (M A ) was significantly greater at the upper slope site and increased significantly in mid-summer, before declining during autumn (Tables 2, 3).Leaf N content did not differ significantly between site nor time of year, but phosphorus content ([P]) was significantly greater at the lower-slope site (Tables 2, 3).Leaf sugar and starch determinations were associated with relatively large  2, 3).Leaf temperatures reflected seasonal changes in ambient air T, with lowest values in late winter (February) and highest values in mid-summer (August, Fig. 1b).Maximum light-saturated photosynthetic rate (A sat ) largely mirrored changes in soil water content, with maximum rates in the spring period and declining rates in summer-autumn (Fig. 1c; Table 3).Trees in the upper slope site displayed higher rates during winter-spring (when rainfall was unusually high), but there was no difference between sites in the summer-autumn period.The pattern of response of A sat was strongly reflected in the seasonal response of g sat , with stomatal conductance declining from high values in June to low values in late summer (Fig. 1d; Table 3).This was also reflected in significant changes in internal CO 2 concentration (C i , Table 3).Foliar respiration in the dark (R D , measured at ambient seasonal temperatures at the end of light response curves) did not display a significant seasonal response (Table 3), with the exception of a significant increase at the single time point during a very hot period in August (Fig. 1e).Inhibition of R by light (i.e. the Kok effect) was consistently observed in the light-limited phase of photosynthetic light responses.As a result, we were able to estimate R L and thus calculate the extent of light inhibition of R D .Foliar respiration in the light (R L ) displayed a very similar seasonal response to R D (Fig. 1f), and so R L /R D , and displayed a range of 0.5 to 0.8 with large s.e. in some months, did not display a significant seasonal response or differ between the two sites (Fig. 1g; Table 3).Averaged over all seven months, R L /R D was 0.66 AE 0.05 s.e.(n = 14).The relationship between R L and R D was a very strong one, and did not differ between trees at the two sites (Fig. 1h; Table 3).There was no significant relationship between seasonal changes in R L /R D and soil volumetric water content (R L /R D = -0.012Â SWC + 0.76, P = 0.18) or leaf T (R L /R D = 0.007 Â T leaf + 0.51, P = 0.11).The single-point measurements of R D in Fig. 1e, when plotted against leaf T, indicate a relatively muted response in the range of leaf temperatures from 7 to 30 C (Fig. 2a).In August, when daily ambient and leaf temperatures exceeded 30 C during the period of that field campaign, R D values increased (to values in the range 2-5 mmol m -2 s -1 ).This response of R D was also reflected, in part, by short-term thermal response measurements generated over daily cycles of ambient T (Fig. 3).These show shifts in the thermal response indicative of partial acclimation to increasing temperatures from winter through spring and summer, most strongly in trees at the upper-slope site (Fig. 3a).R L /R D was not related to seasonal changes in leaf T (Fig. 2b; R L /R D = 0.005 Â T leaf + 0.56, P = 0.14) or leaf N or sugar content (Fig. S1).Significant seasonal changes in A and increases in R resulted in a significant seasonal effect on R D /A sat ratio (Table 3), and in particular a significant increase in during the high temperatures in August (data not shown).Because of the lack of systematic response of R L /R D , the ratio of leaf R L to light-saturated photosynthesis (i.e.R L /A sat ) did not vary during the year other than via direct effects on A and R (Table 3).

Prades gradient study -Q. ilex subsp. ilex and community
Volumetric measurements confirmed that there was a significant decrease in soil water content from the riparian to the shrub sites in mid-summer at Prades (Fig. 4a).Although soil moisture has been, on average, 11% lower in the rainfall exclusion plots than Table 3. Results of repeated-measures two-way ANOVA from the Villar de Cobeta seasonal measurements, with time of year (T) and site (S) as the main effects Abbreviations: M A , leaf mass per unit area; N and P, foliar nitrogen and phosphorus content, respectively; A sat , net photosynthesis measured at 1000 mmol photons m -2 s -1 PPFD; g sat , stomatal conductance measured at 1000 mmol photons m -2 s -1 PPFD; R D , leaf respiration in darkness; R L , nonphotorespiratory mitochondrial leaf respiration in the light; V o , estimated rate of photorespiration at 1000 mmol photons m -2 s -1 PPFD; C i , internal CO 2 concentration at 1000 mmol photons m -2 s -1 PPFD.See Table 2   in the control plots during the entire experimental period (TDR measurements made between 1999 and 2007 reported by Ogaya and Peñuelas 2008;and unpublished measurements since 2007), differences between these plots were not evident in the 30 cm deep gravimetric measurements made in the present mid-summer campaign.Leaf mass per unit area and N content differed significantly with species and site (Table 5).Lowest average M A values were found in the species from the riparian and slope site (T.platyphyllos and A. monspessulanum) (Table 4), with site-averaged values increasing along the gradient (Fig. 4b).M A did not vary significantly in Q. ilex over the range it occurred.
Leaf nitrogen content on a dry mass basis was significantly greater at the riparian and slope sites (species means in Table 4) and site average values decreased from the moister to the drier sites (data not shown).Leaf relative water content differed between species, but did not differ at sites along the soil moisture gradient (Tables 4, 5).Table 4 shows average rates of light-saturated photosynthesis (A sat ) expressed on a leaf area basis for each species/site combination.Significant differences were found among site averages (Fig. 4; Table 5), with Q. ilex showing a similar trend.Rates of A were highest at the riparian and slope sites and lowest at the dry and shrub sites.This trend was also displayed by light-saturated stomatal conductance (g sat ; species averages Table 4, site averages Fig. 4).Because the two parameters changed in concert, there was no significant species or site effect on internal CO 2 concentration (C i , Tables 4, 5).No significant effect of site was found on leaf R D (Table 5), although the riparian site did have a lower rate, on average, than the drier sites (Fig. 4e)as a result there was a significant species Â site interaction.Irrespective of whether site-averages were considered (Fig. 4) or whether rates of leaf R in individual leaves were compared (Table 4), R L was almost always less than R D (i.e.light inhibited leaf R), although the degree of inhibition did vary (with R L = R D (i.e.no light inhibition) in a small number of cases) as a consequence of the technical challenges in the method.R L displayed a similar response to that of R D (Fig. 4f), and thus site-averaged R L /R D did not vary significantly across the gradient (Fig. 4g; Table 5) and R L and R D were strongly correlated (Fig. 4h).The degree of light inhibition in individual species ranged from 0.62 to 0.75 (Table 4), and averaged across all species and sites, R L /R D was 0.69 AE 0.01 s.e.(n = 18).Because of the consistency of R L /R D , the ratio of leaf R L to light-saturated photosynthesis (i.e.R L /A sat ) did not vary along the gradient other than via direct effects on A. There was no significant relationship observed between R L /R D and A or the oxygenation component of A (V o ; data not shown).

Discussion
We assessed the importance of (i) seasonal changes in T and water availability and (ii) a topographically-driven gradient in water availability in determining changes in leaf respiratory CO 2 evolution (both in darkness and in the light) in a dominant tree of the Mediterranean region (Q.ilex subsp.ballota and subsp.ilex).Our results do not provide support for our hypothesis that lower soil water availability in the field would increase light inhibition of R. R L and R D were strongly correlated, with R L almost always less than R D , but R L /R D did not vary in any way that is systematically explainable by changes in T or soil moisture in the seasonal study (Villar de Cobeta site), and there was a distinct lack of site-to-site variation in light inhibition with changes in soil water content across the gradient at the Prades site.An important consequence of the lack of site-to-site variation in light inhibition was that the ratio of R L to photosynthesis, although clearly lower than that for R D , was driven primarily by assimilation rate.

Responses to seasonal and gradient changes in environment
In the year of our seasonal measurements at Villar de Cobeta (2007), the average rainfall in the region was significantly greater than average.As a result, the lower slope site was not greatly different in soil water availability compared with the upper slope site (and we combine the sites for the purposes of further interpretation); however, significant underlying seasonal variation in T and soil water content did provide the driver for significant changes in leaf physiology which allow us to address questions on environmental influences on R D and R L .During the course of the year, leaf T ranged from ~6 to >30 C, mid-morning measured rates of A sat ranged from around 3 mmol m -2 s -1 in late summer/autumn to >10 mmol m -2 s -1 in spring-early summer and g sat ranged from maximal values in excess of 0.15 mol m -2 s -1 in spring to below 0.05 mol m -2 s -1 in late summer.When measured at prevailing mid-morning air temperatures, R D ranged around 1 mmol m -2 s -1 for much of the year, but peaked at ~3 mmol m -2 s -1 during a very hot August (Fig. 1e), when leaf temperatures exceeded 30 C. As a consequence, variation in R D /A ratios was relatively constrained ~0.08-0.18during spring, early summer and winter, and only increasing significantly (to 0.38) during a very hot August.The relatively consistent rate )  Values shown (a-g) are averages for all species at the site and values for Q. ilex only (AEs.e. of the mean).For details of sites, see Table 1.For individual species values see Table 4. See Table 5 for results of two-way ANOVAs testing for differences among species and site averages.

Arbutus unedo
128.1 ± 6.0 96.7 ± 0.6 1.05 ± 0.18 8.0 ± 0.6 0.079 ± 0.012 189 ± 26 6.38 ± 1.5 0.86 ± 0.10 0.57 ± 0.10 0.65 ± 0.06 0.11 ± 0.01 0.07 ± 0.01 of R D for six out of the seven campaigns is indicative of at least partial thermal acclimation (Atkin and Tjoelker 2003), which is supported by two separate lines of evidence.First, the relationship between single point measurements of R D from light response curves on attached leaves and leaf T is fairly muted below 30 C (Fig. 2).Second, short-term T responses curves (generated by daily changes in air temperature on intact leaves in the field; Fig. 3) indicate shifts in the thermal response consistent with thermal acclimation, at least in trees at the south-facing, drier site.
Despite the significant seasonal changes in leaf physiology above (at Villar de Cobeta), R L /R D measured mid-morning did not follow a clear seasonal pattern, and was not significantly influence by time of year.Averaged over the year, the inhibition of R was ~34% (i.e.R L /R D was 0.66 AE 0.05 s.e.).The fact that R L /R D did not change seasonally in any systematic way is somewhat surprising, since we had predicted that light inhibition would increase with increasing measuring T and lower soil water content (in summer).That it did not indicates that, in this forest type under field conditions, soil water availability has no clear impact on the extent of light inhibition of respiration (discussed further in the next section).It also shows that both R L and R D have acclimated to the seasonal shifts in daily average T. The latter is partially supported by our data, since we found that there was some seasonal shift in the R-T curves for the dry site plants (Fig. 3), but much less evidence for acclimation in the lower slope site plants.The constant R L /R D found at the lower slope site may suggest that the Q 10 is not different for the two processes, but for the upper slope site the situation is potentially a more complex mix of acclimation combined with similar Q 10 values.Another complication for interpreting our findings is that the response of R to leaf water content may also be biphasic i.e. it may decrease in the initial stages of water stress (lower energy demands for growth) and may increase with additional stress-related demands e.g.osmoregulation or induced senescence (Ghashghaie et al. 2001;Gulias et al. 2002;Flexas et al. 2005Flexas et al. , 2006)).This means that the timing of investigations during seasonal rainfall cycles is likely to be critical in terms of the state of the plant response.Further, T and water availability tend to vary in opposite directions.The syndrome of response to the combined effects of seasonal changes in water availability and T clearly requires further investigation to resolve these potentially competing drivers and responses, and is partially addressed by the companion study using the Prades gradient, where the mid-summer campaign removes the impact of T.
We found changes in leaf structure and nutrient content in species across the Prades gradient consistent with previous findings (i.e.low M A and high N content at the riparian and slope sites with greater soil water content; Tables 4, 5).Rates of photosynthesis and stomatal conductance varied substantially across the gradient (Fig. 4; Table 4), with species sampled at the lower sites exhibiting the highest rates of net CO 2 uptake.We also observed variations in R D /A ratios among species, which allows an assessment of the leaf-level balance between the respiratory cost and photosynthetic capacity in response to soil water content.R D /A was highest (0.14-0.22) in high M A species in the drier sites that exhibit low rates of mass-based photosynthesis, and lower (0.07-0.11) in deciduous species at the riparian site (Table 4).This is consistent with previous studies that have reported a higher R/A sat ratio under drought (Flexas et al. 2006;Atkin and Macherel 2009).The ratio of leaf R in the light and dark (R L /R D ) was, by contrast, very consistent despite changes in both R L and R D (Fig. 4; Table 4).R L /R D across all sites resolved to an average of 0.69 AE 0.01 s.e., meaning that R L /A ratios were consistently ~30% lower than R D /A ratios and were driven by underlying rates of A. The variation in R L /A contrasts with previous findings of greater constancy of R L /A (compared with R D /A) in controlled-environment grown plants exposed to a range of contrasting environments (Atkin et al. 2006;Ayub et al. 2011) and in a wide range of temperate rainforest species at contrasting sites differing in soil age/nutrient availability/ species composition/leaf traits (Atkin et al. 2013).

Lack of variation in the Kok effect
Past work suggests that light inhibition of leaf R may be linked to changes in cellular energy status (due to excess ATP or redox equivalents generated by the light reactions of photosynthesis, decreasing the demand for respiratory energy compared with leaves in darkness), photorespiration-dependent inactivation of the pyruvate dehydrogenase (PDH) complex (Budde and Randall 1990;Gemel and Randall 1992), or transition to a truncated TCA cycle (Igamberdiev et al. 2001;Tcherkez et al. 2005Tcherkez et al. , 2008Tcherkez et al. , 2009)).More recently, increased use of stored carbon reserves (e.g.citrate) in the light has been shown to reduce demand for flux through the TCA cycle (Tcherkez et al. 2012).Steady-state stoichiometric modelling has also indicated that light suppression of CO 2 release by the oxidative pentose phosphate pathway (OPPP) has the potential to contribute to the Kok effect (Buckley and Adams 2011).Collectively, these biochemical and modelling studies suggest Table 5. Results of two-way ANOVA of leaf traits for species growing along a soil water gradient at Prades, with species (Sp) and site (S) as the main effects Abbreviations: M A , leaf mass per unit area; A sat , net photosynthesis measured at 1500 mmol photons m -2 s -1 PPFD; g sat , stomatal conductance measured at 1500 mmol photons m -2 s -1 PPFD; R D , leaf respiration in darkness; R L , non-photorespiratory mitochondrial leaf respiration in the light; V o , estimated rate of photorespiration at 1000 mmol photons m -2 s -1 PPFD; C i , internal CO 2 concentration at 1000 mmol photons m -2 s -1 PPFD.See Table 2 Farquhar and Busch (2017), using a theoretical modelling approach, have shown it is possible to create a Kok-effect-like response in the presence of very specific conditions of changing mesophyll conductance (g m ) at low light.To our knowledge, this potential explanation for the Kok effect is yet to be supported by experimental evidence documenting the irradiance response of g m or the relative importance of irradiance-mediated changes in respiratory CO 2 release vs g m .Nevertheless, the findings of Farquhar and Busch (2017) highlight the need for cautious interpretation of Kok effect results.
The tight coupling of leaf mitochondrial metabolism in the light to that of the prevailing rate of photosynthesis (Krömer 1995;Hoefnagel et al. 1998;Noguchi and Yoshida 2008) and related processes (e.g.sucrose synthesis, phloem loading, protein turnover) might provide an explanation for variability in R L /R D ratios in the field.Despite recent work (discussed below), uncertainty remains concerning the other main potential driver that of the linkage between light inhibition of R and factors which increase the rate of photorespiration (V o ).Importantly, Tcherkez et al. (2008) found that the degree of light inhibition of R decreases when Xanthium strumarium leaves are exposed to low atmospheric [CO 2 ] for short periods (i.e.under conditions which increase the demand for TCA cycle intermediates associated with the recovery of photorespiratory cycle intermediates in the peroxisome).This relationship between R L /R D and carboxylation/oxygenation in the short term has been supported by further recent studies (Ayub et al. 2011;Crous et al. 2012;Griffin and Turnbull 2013).The literature suggests a putative link between the degree of light inhibition of R and any factors that would change rates of photorespiration, although the strength of that link under field conditions should be subject of continued investigation.
What underlying factors might explain the lack of systematic variation in R L /R D found with season at Villar de Cobeta and along the moisture gradient at Prades?At least three factors could have contributed to variations in R L /R D along the Prades gradient.First, a direct effect of soil water availability on C i and the potential rate of photosynthetic oxygenation (V o ).Griffin and Turnbull (2013) found that R L /R D decreased under conditions that suppressed light-saturated V o .Crous et al. (2012) also report a positive linear relationship between R L /R D and V o100 (the velocity of RuBP oxygenation at 100 mmol m -2 s -1 PPFD).In terms of our mechanistic hypothesis (that light inhibition of R would decrease (and R L /R D increase) at drier sites where stomatal limitations on gas exchange increase photorespiration), C i was significantly influenced by time of year at Villar de Cobeta (Table 3), but R L /R D was not significantly related to C i at either saturating PPFD (Fig. S1f) or at limiting PPFD (100 mmol m -2 s -1 ; data not shown).It is possible that the level of water limitation experienced by trees during the study was not sufficient to influence the balance between R L and R D , but the range of A sat , g sat and C i (180-330 mmol mol -1 ) does not support this.There was no effect of site on C i along the Prades gradient, which clearly reflects some level of balancing between assimilation and conductance as soil water availability becomes more limited in the dry and shrub sites.Second, gradients in soil nutrient availability might have a secondary effect on foliar N content, and thence on the degree of light inhibition.Relatively few studies have quantified the effect of nutrient supply on R L .Shapiro et al. (2004) found that deficiencies in nitrogen supply resulted in reduced rates of both R L and R D , but with the R L /R D ratios being nearly identical in high and low N grown plants (0.50 and 0.48, respectively for ambient CO 2 grown X. strumarium).We have previously found that variations in R L /R D in arctic tundra shrubs (Heskel et al. 2012) and forest species along a soil development chronosequence (Atkin et al. 2013) may be significantly related to several traits that are strong correlates of metabolic capacity, including leaf [N] and [P].In the present study, tissue N was significantly influenced by site (water availability) in the Prades study, but this had no impact on R L / R D , which supports the findings of Shapiro et al. (2004).Third, the degree of light inhibition might differ systematically among species that exhibit contrasting leaf traits, with highly productive, low M A species (which occupy the riparian and slope sites) exhibiting lower degrees of light inhibition than their high M A counterparts.Previous evidence for this is equivocalwe have previously observed weaker inhibition in high-metabolic rate species growing on young/productive sites (Atkin et al. 2013) but similar R L /R D ratios have also been observed in inherently fastand slow-growing herbaceous species (Atkin et al. 1997;Atkin et al. 2006).In the present study, we found that M A varied with season at Villar de Cobeta and site at Prades, once again with no impact on R L /R D .Thus, here we add to the existing data that does not support the hypothesis that R L /R D ratios vary systematically among species with inherently different leaf traits.

Conclusions
The balance between photosynthetic carbon fixation and respiratory oxidation of photosynthates is of great important to tree C balance (Amthor 1989;Loomis and Amthor 1999;Alt et al. 2000).Since mitochondrial respiration plays a key role in determining the growth and survival of plants, and it is associated with the production of energy and carbon skeletons essential for cellular maintenance and biosynthesis, respiratory responses need to be considered relative to the effects on carbon gain to elucidate the overall effect on plant performance.Our estimates based on the Kok method demonstrate that the degree of light inhibition of R is relatively constrained seasonally and is fairly consistent across sites differing in soil water availability and community composition in this Mediterranean forest type (with an average R L /R D of ~0.7).This level of inhibition is consistent with recent findings (Ayub et al. 2014;McLaughlin et al. 2014).This points to R L /A ratios being consistent, but around 30% lower than, R D /A ratios, with both being primarily driven by changes in A. Notwithstanding the possibility that other methods of determining the degree of light inhibition of R might provide slightly different estimates, or that more extreme gradients of soil water availability may elicit changes not observed here, these findings have important implications for predictive models that seek to predict rates of leaf R L using more commonly measured rates of R D and associated leaf traits such as M A and foliar [N] (Mercado et al. 2007).Failure to account for light-induced reduction in leaf R will clearly lead to errors in predicted rates of ecosystem CO 2 exchange (Wohlfahrt et al. 2005;Wingate et al. 2007;McLaughlin et al. 2014).Although further work is needed to establish the precise metabolic and environmental drivers of variations in R L /R D , there is now some evidence for stand and ecosystem models to assume a relatively constant relationship between R D and R L along gradients of soil water availability.

Fig. 1 .
Fig.1.Seasonal changes in (a) soil volumetric water content, (b) leaf temperature, (c) A sat ; i.e.net CO 2 uptake measured 1000 mmol photons m -2 s -1 , (d) g sat ; maximum stomatal conductance at A sat , (e) foliar respiration rate in darkness (R D ) and (f) in the light (R L ), (g) the ratio of leaf respiration measured in the light to that in darkness (R L /R D ) in measured in Quercus ilex seven times during the course of a year at an upper and lower slope site at Villar de Cobeta (central Spain).(a-g) values shown are averages for each time point (AEs.e. of the mean).For (b), measurements were made at the prevailing daytime T during the active period on any given day (1100 to 1400 hours).See Table3for results of two-way ANOVAs testing for differences with time and site.For (h), area-based rates of leaf R L plotted against corresponding rates of R D .Data shown are for individual leaves sampled during the year.The dashed line shows the 1 : 1 relationship.

Fig. 2 .
Fig. 2. Relationships between (a) foliar respiration rate in darkness (R D ) and (b) the ratio of leaf respiration measured in the light to that in darkness (R L /R D ) and leaf temperature in Quercus ilex seven times during the course of a year at an upper (closed symbols) and lower slope (open symbols) site at Villar de Cobeta (central Spain).

Fig. 3 .
Fig. 3. Seasonal changes in the instantaneous thermal response of foliar R D in Quercus ilex at an (a) upper and (b) lower slope site at Villar de Cobeta (central Spain).Responses were determined on intact shoots using ambient changes in air temperature during the course of the day and night in April, June, July, September and November 2007.Each point represents a replicate tree at each time point.

Fig. 4 .
Fig. 4. Changes in (a) soil water content, (b) leaf mass per unit area, (c) A sat ; i.e.net CO 2 uptake measured 1500 mmol photons m -2 s -1 , (d) g sat ; maximum stomatal conductance at A sat , (e) foliar respiration rate in darkness (R D ) and (f) in the light (R L ), (g) the ratio of leaf respiration measured in the light to that in darkness (R L /R D ) for the six study sites along the soil moisture gradient at Prades (north-eastern Spain).Values shown (a-g) are averages for all species at the site and values for Q. ilex only (AEs.e. of the mean).For details of sites, see Table1.For individual species values see Table4.See Table5for results of two-way ANOVAs testing for differences among species and site averages.(h) Area-based rates of leaf R L plotted against corresponding rates of R D .Data shown are for individual leaves sampled along soil water availability gradient.The dashed line shows the 1 : 1 relationship.
Fig. 4. Changes in (a) soil water content, (b) leaf mass per unit area, (c) A sat ; i.e.net CO 2 uptake measured 1500 mmol photons m -2 s -1 , (d) g sat ; maximum stomatal conductance at A sat , (e) foliar respiration rate in darkness (R D ) and (f) in the light (R L ), (g) the ratio of leaf respiration measured in the light to that in darkness (R L /R D ) for the six study sites along the soil moisture gradient at Prades (north-eastern Spain).Values shown (a-g) are averages for all species at the site and values for Q. ilex only (AEs.e. of the mean).For details of sites, see Table1.For individual species values see Table4.See Table5for results of two-way ANOVAs testing for differences among species and site averages.(h) Area-based rates of leaf R L plotted against corresponding rates of R D .Data shown are for individual leaves sampled along soil water availability gradient.The dashed line shows the 1 : 1 relationship.

Table 1 . Site characteristics and species selected for measurements at Prades (Catalunya, north-eastern Spain)
The elevation is the altitude above sea level.Height refers to the maximum height of the dominant trees, giving an indication of the heights from which the shoots were sampled *Acer monspessulanum * *

Table 2 . Seasonal changes in average (þs.e., n = 4) values of leaf structural and chemical traits for Quercus ilex subsp. balotta at two sites at Villar de Cobeta (central Spain)
error, and thus displayed only modest significant responses to site and time of year (Tables

Table 4 . Average (þs.e., n = 6) values of leaf traits for each species growing at each site along a soil water gradient at Prades
A , dry mass per unit area; RWC, relative water content, N, nitrogen content; A sat , rate of net photosynthesis at 1500 o , estimated rate of photorespiration at 1500 light inhibits respiratory CO 2 releaseconsistent with the Kok effectand that reduced rates of CO 2 release by the TCA cycle and OPPP could both contribute to cases where R L < R D .What is less clear, however, is how environmental factors in the field influence the level of inhibition of R.More recently, that