Effects of mistletoe removal on growth, N and C reserves, and carbon and oxygen isotope composition in Scots pine hosts

Abstract

Most mistletoes are xylem-tapping hemiparasites, which derive their resources from the host’s xylem solution. Thus, they affect the host’s water relations and resource balance. To understand the physiological mechanisms underlying the mistletoe–host relationship, we experimentally removed Viscum album ssp. austriacum (Wiesb.) Vollmann from adult Pinus sylvestris L. host trees growing in a Swiss dry valley. We analyzed the effects of mistletoe removal over time on host tree growth and on concentrations of nonstructural carbohydrates (NSC) and nitrogen (N) in needles, fine roots and sapwood. In addition, we assessed the δ¹³C and δ¹⁸O in host tree rings. After mistletoe removal, δ¹³C did not change in newly produced tree rings compared with tree rings in control trees (still infected with mistletoe), but δ¹⁸O values increased. This pattern might be interpreted as a decrease in assimilation (A) and stomatal conductance (g_s), but in our study, it most likely points to an inadequacy of the dual isotope approach. Instead, we interpret the unchanged δ¹³C in tree rings upon mistletoe removal as a balanced increase in A and g_s that resulted in a constant intrinsic water use efficiency (defined as A/g_s). Needle area-based concentrations of N, soluble sugars and NSC, as well as needle length, single needle area, tree ring width and shoot growth, were significantly higher in trees from which mistletoe was removed than in control trees. This finding suggests that mistletoe removal results in increased N availability and carbon gain, which in turn leads to increased growth rates of the hosts. Hence, in areas where mistletoe is common and the population is large, mistletoe management (e.g., removal) may be needed to improve the host vigor, growth rate and productivity, especially for relatively small trees and crop trees in xeric growth conditions.

Introduction

About 1% of all angiosperm species are parasitic and ∼40% of them parasitize the aboveground parts of their host plants (shoot parasites) (Norton and Carpenter 1998). Mistletoes are the predominant group of angiosperm shoot parasites, and they are found widely in boreal and temperate forests, tropical rain forests and arid woodlands (Norton and Carpenter 1998). Under current predictions of global climate changes, mistletoes are expected to expand their geographical ranges (Rigling et al. 2010). As an example, Dobbertin et al. (2005a) showed that pine mistletoe (Viscum album ssp. austriacum) shifted its altitudinal distribution upward by 200 m in the Swiss Rhone Valley due to climate warming during the past century.

Most mistletoes are xylem-tapping plants that derive water and nutrients from the host’s xylem solution (Schulze and Ehleringer 1984, Ehleringer et al. 1985, Schulze et al. 1991, Popp and Richter 1998, Bannister and Strong 2001, Zuber 2004). Thus, mistletoes may exacerbate the effects of environmental stress such as drought and limited resources on host trees (Rigling et al. 2010, Sangüesa-Barreda et al. 2012), leading to increased host mortality (Dobbertin et al. 2005b, Dobbertin and Rigling 2006).

Water and nutrient relationships between mistletoes and their hosts have been studied extensively (Glatzel 1983, Orozco et al. 1990, Pate et al. 1990, Küppers et al. 1992, Flanagan et al. 1993, Hosseini et al. 2007, Zweifel et al. 2012). The parasitic plants use two pathways to receive mineral nutrients from the hosts: (i) a passive pathway caused by the parasite’s transpiration, where the transport of nutrients occurs with water flow from the host to the parasite, and (ii) an active pathway in which the haustorium actively draws nutrients from the host to the parasitic plant (Lamont 1983). Mistletoes derive nitrogen (N) from their hosts, resulting in lower N concentrations in mistletoe-infected trees than in uninfected trees within a forest community (Ehleringer et al. 1986a, Bell and Adams 2011). Nitrogen products extracted by dwarf mistletoe were found to comprise up to 20% of the total N storage in host Acacia nilotica (L.) Delile forests in Sudan (Gibson 1967). The N concentration in mistletoes was found to be three to four times higher than that in their hosts (Küppers et al. 1992, Schulze et al. 1994), suggesting that overuse of N by mistletoes may result in N-depletion in the hosts, especially under stressed growth conditions.

Compared with water and nutrients, the carbohydrate relationship between mistletoes and the host is more complicated (Bell and Adams 2011), because all mistletoes are hemiparasites, bearing evergreen leaves that carry out photosynthesis. It has been proposed that mistletoes affect the host carbon (C) balance in various ways (Bickford et al. 2005, Wang et al. 2008, Logan et al. 2013). Mistletoes are able to extract C compounds from the host’s xylem (Reblin et al. 2006) and thus directly decrease C availability in the host. In addition, mistletoes absorb water from hosts, which may induce closure of the stomata (Zweifel et al. 2012) and reduce the photosynthetic activity in the leaves of the hosts (Rigling et al. 2010). Therefore, mistletoes can directly reduce the host’s ability to gain C under drought-stressed conditions. For example, Zweifel et al. (2012) found that mistletoe infection created a leak in the water flow system of hosts and thus caused stomatal closure and lower transpiration and assimilation rates. Mistletoe-induced decreases in host leaf N may also reduce host photosynthesis, as leaf N concentrations are significantly positively correlated with photosynthetic rates (Marshall et al. 1994, Meinzer et al. 2004). These effects may result in or intensify C starvation and mortality of the hosts (Dobbertin and Rigling 2006, Rigling et al. 2010).

Previous studies have mainly concentrated on mistletoe effects on growth, nutrient and carbohydrate concentrations, and water use efficiency (WUE) at the branch level. In particular, these investigations have compared either mistletoe-infected and uninfected branches within a host tree (Hosseini et al. 2007, Lo Gullo et al. 2012, Logan et al. 2013), or infected and uninfected branch parts of a given branch (Ehleringer et al. 1986b, Tennakoon and Pate 1996, Rigling et al. 2010, Tennakoon et al. 2011). Other studies have compared the physiological properties of mistletoes with those of branches they had infected (Küppers et al. 1993, Bannister and Strong 2001, Escher et al. 2004, 2008). These branch-level experiments may underestimate the effects of resource reallocation and mobilization, as well as the feedback mechanisms within the whole host tree. Only a few studies have compared mistletoe effects on growth and physiological parameters at the whole-tree level by comparing mistletoe-infected trees with trees that were never infected (Ehleringer et al. 1986b, Marshall et al. 1994, Sala et al. 2001, Miller et al. 2003, Sangüesa-Barreda et al. 2012). However, these tree-level results may be biased by the mistletoe infection history of the infected trees and the specific biological properties (e.g., resistance) of the trees that were never infected.

To better understand the C- and N-balance and WUE in mistletoe-infected trees, we carried out a long-term (8 years) mistletoe removal experiment in a dry valley in Switzerland. In the research area, increased summer temperatures and rain shortage (Rigling et al. 2002, Dobbertin et al. 2005b), coupled with the high transpiration rates of mistletoes (Zweifel et al. 2012), may lead to drought-induced C-depletion (Scharpf et al. 1988) and water deficit in host trees (Dobbertin and Rigling 2006). We compared mistletoe-infected Scots pine trees (MIT) with trees where mistletoes (V. album ssp. austriacum; Noetzli et al. 2003) were removed (MRT). We studied tree needle morphology, shoot growth and stem radial growth, as well as mobile C and N concentrations (starch (ST), sugars and total N) in needles, fine roots and stem sapwood. Moreover, we assessed the tree ring stable C and oxygen (O) isotope composition in tree rings produced 6 years before and 6 years after mistletoe removal. We expected that mistletoe removal would reduce the risk of C-depletion and water deficit in host trees growing in a dry environment. We tested the specific hypotheses that (I) mobile carbohydrate and N concentrations of host trees would increase upon mistletoe removal and (II) the removal of mistletoe would lead to higher assimilation rates, as well as higher stomatal conductance (g_s), in the host trees, and that these changes would be imprinted in tree ring growth and tree ring δ¹³C and δ¹⁸O signals (Scheidegger et al. 2000, Gessler et al. 2014).

Materials and methods

Study site, stand and mistletoe treatment

The study site is situated in Pfynwald (46°19′27″N, 7°34′40″E, 610 m above sea level) in the inner-Alpine valley of Valais, Switzerland. According to climate data gained from the nearest climate station to the site, mean annual precipitation was 657 mm (Zweifel et al. 2009) with large interannual variation, and mean annual temperature was ∼9.7 °C over a 50-year period (Dobbertin et al. 2010).

The Pfynwald forest (>10 km²) is dominated by Scots pine (Pinus sylvestris L.) with occasional Quercus pubescens Willd. The forest is 90–100 years old, and pine trees have a diameter at breast height (DBH) of ≥12 cm and a mean height of ∼11 m. The understory consists of shrubs and young Q. pubescens, grasses, mosses and a few dwarf shrub species. The soil type is a Rendzic Leptosol with limestone as the parent material (Rigling et al. 2010).

We selected 12 P. sylvestris trees of similar age (90–100 years old) and size (10–12 m in height and 15–20 cm in DBH), similar growth performance and similar intensity of mistletoe infection in 2004. The growth performance and mistletoe infection intensity were visibly estimated. We randomly selected 6 out of those 12 trees to completely remove all attached mistletoe plants from the branch surfaces in September 2004 (after the 2004 growing season), and we repeated removal in September 2007 and 2010. A mean fresh weight of mistletoes of 2.04 ± 0.57 kg was removed from each tree. These six trees were defined as mistletoe-removed trees (MRT). The other six trees were controls with all mistletoes intact (MIT).

Sampling and growth measurements

On 12 July 2012, one south-oriented and one north-oriented leading branch, each with a branch base diameter of >2 cm, were taken from the upper crown section of each tree, i.e., 12 branches were collected for MRT and MIT, respectively. Only unshaded, healthy and undamaged branches with intact terminal buds were selected.

Annual shoot growth was measured along the collected branches. We measured the annual shoot increment for each tree from 1999 to 2010 (i.e., spanning 6 years before to 6 years after the mistletoe treatment occurred in after the growing season 2004). Scots pine shoots can be dated accurately by counting back the internodes or age classes of the branches from the branch tip (Kozlowski 1971, Li et al. 2006). In the present study, the age of needles and shoot segments in each selected branch was determined using marks from overwintering buds on the branch and by counting annual needle cohorts (Li et al. 2001). For example, the length of shoot segments along the leading branch emerged in 2011, 2010 and 2009 was defined as the shoot increments for 2011, 2010 and 2009, respectively. Needles that emerged in 2011 (i.e., attached to the 2011 shoot segment), 2010 and 2009 were named as 1-, 2- and 3-year-old needles, respectively.

Different-aged needles (1-, 2- and 3-year-old) were sampled separately from the collected branches mentioned above for each sample tree. We only took 1- to 3-year-old needles for the present study since very few needles reached an age of 4 years. The current-year needles (i.e., needles emerged in 2012) were also not sampled because they were not mature at the sampling time (12 July 2012), and thus, their carbohydrate concentrations were considered to be highly dynamic and not stable over time (Li et al. 2002). About 25 fresh needles for each of the three needle age classes for each tree (mixed sample collected from the two branches per tree) were scanned at 600 d.p.i. resolution (Epson V500, NSK Ltd., Tokyo, Japan), and the needle length and the projected area were then calculated automatically with the software WinSeedle (Regents Instruments Inc., Quebec, Canada). The specific leaf area per dry weight (SLA) and the leaf mass per unit leaf area (LMA) were calculated.

Using a 5-mm-diameter increment corer, two tree ring cores (one from the south-stem side and one from the north-stem side) were taken from each MIT and MRT at 1.0–1.3 m stem height above the ground surface. These cores were used for measuring tree ring width and stable isotope composition. Annual tree ring width was measured from 1999 to 2010 (as done for shoot growth, spanning 6 years before to 6 years after mistletoe treatment) on a Lintab measuring table (Rinntech, Heidelberg, Germany), and analyzed using the TSAP software (Rinntech).

The tree rings in MRT formed before mistletoe removal (end of 2004) and in MIT over the whole period were very narrow. In order to obtain the minimum amount of material necessary for δ¹³C and δ¹⁸O analyses, and thereby assess differences in the isotope composition between MRT and MIT over time, we combined the tree rings formed during 3-year periods (1999–2001, 2002–04, 2005–07 and 2008–10) from two cores per tree and pooled the material as a mixed sample per tree (n = 6).

For comparing the effects of mistletoe removal (before and after treatment) in MRT, two additional tree ring cores (also one from the south-stem side and one from the north-stem side at 1.0–1.3 m stem height) were taken from each MRT. These tree rings were separated at a 1-year resolution from 1999 to 2010 (6 years before to 6 years after mistletoe removal), and wood samples with the same age were pooled for pairs of adjacent MRT (n = 3), to obtain the minimum amount of material necessary for δ¹³C and δ¹⁸O analyses.

The samples were milled, followed by the simultaneous delignification and removal of noncellulosic polysaccharides using an acetic acid : nitric acid mixture, as described by Brendel et al. (2000). This method was used to minimize losses from the small samples. After cellulose purification, the samples were dried in the Pyrex test tubes until a constant weight was reached (Brendel et al. 2000).

Samples for chemical analysis (N and carbohydrate concentrations) included 1-year-old needles, fine roots and stem sapwood. One-year-old needles were collected from the two selected branches per tree and pooled together to get a mixed sample for each tree (n = 6 trees). To collect fine roots from MRT and MIT, we first found coarse roots originating from each tree, and then we collected fine roots (<0.5 cm in diameter) attached to those coarse roots (n = 6 trees). Two short tree ring cores (one from the south-stem side and one from the north-stem side) were taken from each tree at a stem height of ∼50 cm above the ground surface. To ensure the sapwood analyzed was formed after the mistletoe treatment, the outer six tree rings of the two cores were separated from older rings and pooled as a single sample per tree (n = 6). These samples were frozen in liquid N, freeze-dried under vacuum conditions, ground and stored at −20 °C until chemical analysis was conducted.

Tree ring δ¹³C and δ¹⁸O analyses

The dry and finely ground tree ring samples were weighed (∼0.4–0.6 mg) into tin capsules for δ¹³C analysis and were weighed (0.5–0.7 mg) into silver capsules for δ¹⁸O analysis. Prepared capsules were analyzed at the Paul Scherrer Institute in Switzerland (Ecosystem Fluxes Group, PSI, Zurich, Switzerland). Samples were combusted in a High Temperature Conversion/Elemental Analyzer (Flash EA-1112, Carlo Erba, Milan, Italy) coupled to a mass spectrometer (DELTA plus XL, Thermo Finnigan, Bremen, Germany).

Carbon isotopic values were expressed in δ notation relative to the Vienna Pee Dee Belemnite standard (Coplen 2011). For oxygen, δ notation relative to the Vienna Standard Mean Ocean Water was used. The precision for measurements as determined by repeated measurements of standards was better than 0.1‰ for both isotopes. The rates of C fixation (A) and g_s are the primary factors determining photosynthetic C isotopic discrimination and therefore C isotopic composition (δ¹³C). Thus, the intrinsic WUE can be derived from the C isotope composition according to Farquhar et al. (1982) and Seibt et al. (2008).

Total soluble sugar and starch measurements

The powdered material (0.1 g) of needles, fine roots and stem sapwood was put into a 10 ml centrifuge tube, where 5 ml of 80% ethanol was added. The mixture was incubated at 80 °C in a water bath shaker for 30 min and then centrifuged at 4000 r.p.m. for 5 min. The pellets were extracted two more times with 80% ethanol. Supernatants were retained and combined for soluble sugar (SS) determinations using the anthrone method. Ethanol was removed by evaporation and then sugars were dissolved in distilled water. An aliquot of the water–sugar solution was hydrolyzed in 5 ml of 0.4% anthrone solution (4 g anthrone in 1000 ml 95% H₂SO₄) in a boiling water bath for 15 min. After cooling, the sugar concentration was determined spectrophotometrically (ultraviolet–visible spectrophotometer 752S, Cany Precision Instruments Co., Ltd, Shanghai, China) at 620 nm (Li et al. 2008b). Glucose was used as a standard. The SS concentration in fine roots, sapwood and needles was calculated on a dry mass basis (SS_mass, % d.m.) and in needles concentrations were additionally calculated on a projected needle area basis (SS_area, g m⁻²).

The ethanol-insoluble pellet was used for ST extraction. Ethanol was removed by evaporation. Starch in the residue was released in 2 ml distilled water for 15 min in a boiling water bath. After cooling to room temperature, 2 ml of 9.2 mol l⁻¹ HClO₄ was added. Starch was hydrolyzed for 15 min. Four milliliters of distilled water were added to the samples. Samples were then centrifuged at 4000 r.p.m. for 10 min. The pellets were extracted one more times with 2 ml of 4.6 mol l⁻¹ HClO₄. Supernatants were retained, combined and filled to 20 ml with distilled water. The ST concentration was measured spectrophotometrically (ultraviolet–visible spectrophotometer 752S) at 620 nm using anthrone reagent and calculated by multiplying glucose concentrations by the conversion factor of 0.9 (Li et al. 2008b). Glucose was used as a standard. The ST concentrations in needles, fine roots and sapwood were calculated on a dry mass basis (ST_mass, % d.m.), and concentrations in needles were additionally expressed on a projected needle area basis (ST_area, g m⁻²).

Total N analysis

The total N concentration in needles, roots and sapwood was determined in finely ground oven-dried samples by the micro Kjeldahl procedure, using CuSO₄, K₂SO₄ and H₂SO₄ for digestion (Li et al. 2008a). The concentration of NH₃ was determined on an auto-analyzer using the indophenol-blue colorimetric method (Li et al. 2008a). The N concentrations in fine roots and sapwood were expressed on a dry mass basis (N_mass, % d.m.), and concentrations in needles were expressed both on a dry mass basis and on a projected needle area basis (N_area, g m⁻²).

Data analysis

The relative growth rate (shoot and tree ring) was defined as growth changes, expressed in percent, using the yearly value from 2005 to 2010 relative to the pretreatment mean value averaged over 6 years (1999–2004) for MRT and MIT, respectively. Nonstructural carbohydrate (NSC) is defined as the sum of ST and SS for each sample within each category (Chapin et al. 1990, Li et al. 2001). The SS/ST ratio is defined as the concentration ratio of SS to ST, and NSC/N ratio is the concentration ratio of NSC to N within each sample. All data (δ¹³C, δ¹⁸O, NSC, ST and total SS) were checked for normality by Kolmogorov–Smirnov tests. Repeated measurements analyses of variance (RM-ANOVAs) were used and determined that both mistletoe treatment (MRT vs MIT, all P < 0.05) and time after treatment (all P < 0.01) significantly affected tree growth (6 years after treatment for shoot and tree ring growth and 3 years after treatment for needle parameters) and values of δ¹³C and δ¹⁸O (results of RM-ANOVAs not shown due to consistently significant effects). Therefore, one-way ANOVAs were used to test differences in variables between MRT and MIT, and independent-samples t-tests (two-tailed) were performed to examine the difference in variables between pretreatment and posttreatment values within MRT or MIT. To eliminate the effects of annual environmental variations on shoot increment, tree ring width, δ¹³C and δ¹⁸O across the study period of 12 years, values of these parameters in MRT were compared with those in MIT. All statistical analyses were performed using SPSS 17 for windows (SPSS Inc., Chicago, IL, USA).

Results

Host tree growth

Mistletoe removal tended to increase the needle length, mean single needle area and LMA of host trees (Table 1). Mean needle length increased by 63.1% (P < 0.05), 27.6% (P < 0.10) and 29.6% (P < 0.05) for 1-, 2- and 3-year-old needles, respectively, in MRT compared with those in MIT (Table 1). Mean single needle area increased by 76.4% (P < 0.05), 30.2% (P < 0.10) and 38.2% (P < 0.05) for 1-, 2- and 3-year-old needles, respectively, in MRT compared with those in MIT (Table 1). Similarly, the mean needle LMA in MRT was 12.7% (P < 0.05), 5.6% (P > 0.10) and 5.9% (P > 0.10) higher than those in MIT for 1-, 2- and 3-year-old needles, respectively (Table 1).

Table 1.

Mean values (±1 SD, n = 6 trees) of LMA, needle area and needle length from 1- to 3-year-old needles in MRT and MIT. One-way ANOVAs were performed to examine statistical differences in the mean values of each needle parameter between MRT and MIT within each age class. Significant differences at P < 0.05 are highlighted in bold, and changes in percent are given in parentheses and were calculated using (MRT-mean − MIT-mean) × 100/MIT-mean.

Needles age	Mean LMA (g m−2)				Mean single needle area (mm2)				Mean length (mm)
	MRT	MIT	F	P	MRT	MIT	F	P	MRT	MIT	F	P
1 year	290.49 ± 11.08	257.69 ± 9.20	2.313	0.043	26.74 ± 2.32	15.16 ± 3.79	6.781	0.026	31.10 ± 2.09	19.07 ± 3.83	7.596	0.02
2 years	314.57 ± 9.16	297.73 ± 8.13	1.375	0.199	32.04 ± 2.42	24.61 ± 3.19	3.438	0.093	35.49 ± 2.50	28.05 ± 2.97	3.676	0.084
3 years	326.78 ± 6.07	308.49 ± 11.22	1.433	0.182	35.07 ± 1.87	25.37 ± 3.31	6.52	0.029	37.42 ± 2.40	28.87 ± 2.77	5.445	0.042
Mean	310.61 (+7.9%)	287.97			31.28 (+44.1%)	21.71			34.67 (+36.9%)	25.33

Tree ring width in both MRT and MIT decreased until 2003, when it increased (Figure 1a). For the 6 years (1999–2004) before the mistletoe treatment took place, the mean annual tree ring width was 78.33 × 10⁻² mm for MIT and 70.33 × 10⁻² mm for MRT (10.2% smaller for MRT; F = 0.714, P > 0.05; Figure 1a). An opposite trend occurred after treatment, when MIT (65.94 × 10⁻² mm) had a 9.5% smaller mean tree ring width than MRT (72.86 × 10⁻² mm) for the 6 years from 2005 to 2010 (F = 1.175, P > 0.05; Figure 1a). The yearly standardized relative tree ring growth rate was always greater in MRT than in MIT after treatment (Figure 1b).

Figure 1.

Tree ring width (a) and annual relative growth rate (b) for both mistletoe-removed trees (MRT, n = 6 trees) and mistletoe-infected trees (MIT, n = 6 trees) during the period from 1999 to 2010 (mean values ± 1 SD). (a) Shown are tree ring width in MRT (open circles) and MIT (filled circles), and mean tree ring width in MRT (broken horizontal lines) and MIT (continuous horizontal lines) for the 6 years before and 6 years after mistletoe removal (conducted at the end of the growing season 2004). The relative tree ring growth rate (b) was defined as tree ring width changes, expressed in percent, using the yearly tree ring width from 2005 to 2010 relative to the pretreatment mean tree ring width averaged over 6 years (1999–2004) for MRT and MIT, respectively. Asterisks indicate a significant difference (P < 0.05) in mean values between MRT and MIT.

The mean annual shoot increment length in MRT did not significantly differ from that of MIT before treatment (MRT > MIT, P > 0.05; Figure 2a). After mistletoe removal, the mean annual shoot increment length was significantly larger (+29.1%; P < 0.05) in MRT (3.9 cm) than that in MIT (3.0 cm) (Figure 2a). The yearly standardized relative growth rate of shoot growth was always higher in MRT than in MIT after mistletoe removal (Figure 2b).

Figure 2.

The annual shoot growth increment (a) and annual relative growth rate (b) for both mistletoe-removed trees (MRT, n = 6 trees) and mistletoe-infected trees (MIT, n = 6 trees) during the period from 1999 to 2010 (mean values ± 1 SD). (a) Shown are shoot growth increment in MRT (open circles) and MIT (filled circles), and mean shoot growth increment in MRT (broken horizontal lines) and MIT (continuous horizontal lines) for the 6 years before and 6 years after mistletoe removal (conducted at the end of the growing season 2004). The relative growth rate (b) was defined as growth changes, expressed in percent, using the yearly shoot growth increment from 2005 to 2010 relative to the pretreatment mean growth increment averaged over 6 years (1999–2004) for MRT and MIT, respectively. Asterisks indicate a significant difference (P < 0.05) in mean values between MRT and MIT.

Nitrogen and mobile carbohydrate concentrations

MRT had significantly higher needle area-based concentrations of N, SS and NSC than MIT. Only ST did not display a significant difference between MRT and MIT (Figure 3). Needle N_area, SS_area and NSC_area in MRT were 28.6% (P < 0.05), 32.1% (P < 0.05) and 17.3% (P < 0.05) higher than those in MIT, respectively (Figure 3).

Figure 3.

Mean concentration (±1 SD, n = 6) of total N, SS, ST and NSC in 1-year-old needles expressed on a projected area basis (g m⁻²) in mistletoe-removed trees (MRT, white columns) and mistletoe-infected trees (MIT, black columns). Different letters indicate statistically significant (P < 0.05) differences in mean values between MRT and MIT for a given variable.

MRT needles had significantly lower ST_mass (−26.5%, P < 0.05; Figure 4c) and NSC_mass (−11.4%, P < 0.05; Figure 4d) but higher SS_mass (+5.5%, P < 0.05; Figure 4b) than MIT needles. Nitrogen (Figure 4a) and ST (Figure 4c) concentrations in sapwood did not differ between MRT and MIT (both P > 0.05), but MRT sapwood had significantly higher SS (+31.3%, P < 0.05; Figure 4b) and NSC (+23.3%, P < 0.05; Figure 4d) concentrations than MIT sapwood. MRT roots had significantly lower N (−19.4%, P < 0.05; Figure 4a) and SS (−4.7%, P < 0.05; Figure 4b) concentrations but significantly higher ST (+27.6%, P < 0.05; Figure 4c) and NSC (+8.7%, P < 0.05; Figure 4d) concentrations than MIT roots.

Figure 4.

Mean concentration (±1 SD, n = 6) of total N (a), SS (b), ST (c) and NSC (d) in 1-year-old needles, stem sapwood and fine roots in mistletoe-removed trees (MRT, white columns) and mistletoe-infected trees (MIT, black columns). Different letters indicate significant differences in mean values between MRT and MIT for a given variable (P < 0.05).

MRT had a significantly higher SS/ST ratio than MIT in needles and wood, but a significantly lower ratio in roots (see Figure S1a available as Supplementary Data at Tree Physiology Online). The needle NSC/N ratio did not differ between MRT and MIT, but MRT had a significantly higher NSC/N ratio in sapwood and roots than MIT (see Figure S1b available as Supplementary Data at Tree Physiology Online).

δ¹³C and δ¹⁸O values

In MRT tree rings, the mean δ¹³C value for wood formed during the 6 years (2005–10) following treatment was significantly lower than the mean value for wood formed in the 6-year period (1999–2004) before treatment (P < 0.05; see Figure S2a available as Supplementary Data at Tree Physiology Online). In contrast, the mean δ¹⁸O value was significantly higher after mistletoe removal (P < 0.05; see Figure S2b available as Supplementary Data at Tree Physiology Online). However, the mean δ¹³C value in the combined 3-year tree ring material (1999–2001, 2002–04, 2005–07 and 2008–10) decreased for both MRT and MIT from the pretreatment period to the posttreatment period (Figure 5a). No significant difference in δ¹³C between MIT and MRT was observed for either the pre- or posttreatment period (Figure 5a). Therefore, the trend of decreasing δ¹³C values in both MRT and MIT suggests that this pattern (Figure 5a, Figure S2a available as Supplementary Data at Tree Physiology Online) was caused by environmental variations over time and was not a result of the mistletoe treatment, since MIT was not subjected to mistletoe removal.

Figure 5.

Mean values (±1 SD, n = 6) of δ¹³C (a) and δ¹⁸O (b) in tree rings pooled for 3 years in both mistletoe-removed trees (MRT) and mistletoe-infected trees (MIT) during the period of 1999–2010. Shown are isotopic values in MRT (open circles) and in MIT (filled circles), and pretreatment and posttreatment mean values of δ¹³C and δ¹⁸O in MRT (broken horizontal lines) and in MIT (continuous horizontal lines). Asterisks indicate a significant difference (P < 0.05) in mean values between MRT and MIT.

There was no statistically significant difference between MRT and MIT in the mean δ¹⁸O values in the combined 3-year tree ring material over the pretreatment phase (MRT 25.06‰ vs MIT 25.20‰; P > 0.05) (Figure 5b). The mean δ¹⁸O values for the posttreatment phase (2005–10) decreased in MIT but increased in MRT (Figure 5b), indicating that mistletoe removal resulted in an increase in δ¹⁸O values in tree rings. The mean δ¹⁸O values of the entire posttreatment phase were significantly (P < 0.05) higher in MRT (25.27‰) than in MIT (24.79‰) (Figure 5b). This result indicates that the increase in δ¹⁸O observed in MRT after mistletoe removal (see Figure S2b available as Supplementary Data at Tree Physiology Online) was a result of the treatment.

Discussion

Effects of mistletoe removal on host growth

Mistletoe removal resulted in greater needle size, tree ring width and shoot growth in MRT (Table 1, Figures 1 and 2). Consistent negative effects of mistletoes on host growth and survival have been reported in many previous studies (Reid et al. 1994, Stanton 2006, Shaw et al. 2008, Rigling et al. 2010, Logan et al. 2013, Marias et al. 2014). At the same site as our study, Rigling et al. (2010) found that the mean needle length decreased by 20% in mistletoe-infected branches compared with noninfected branches, due to decreased water availability. Moreover, mistletoe infection increased needle loss, leading to a decline in the crown area and the photosynthetically active leaf area (Dobbertin et al. 2010, Zweifel et al. 2012). Mistletoe infection caused a leaf mass loss of >70% in Scots pine at the same site (Rigling et al. 2010) and in other species such as Eucalyptus trees in Australia (Reid et al. 1994). Mistletoe infection-induced leaf loss (Rigling et al. 2010) and increased crown transparency (Dobbertin and Rigling 2006) are thus considered to be important factors causing the decline in shoot and diameter growth rates of host trees, as observed in our study (Table 1).

Previous studies described that a reduction in leaf N concentration resulted in smaller leaf size (Logan et al. 1999, Robinson and Gessner 2000). Smaller leaf size and lower leaf N concentration cause a decline in photosynthetic production (Meinzer et al. 2004), which could in turn lead to lower shoot growth rates in mistletoe-infected trees (Ehleringer et al. 1986b, Cechin and Press 1993). Increased N availability tends to result in biomass accumulation in branches and wood (Xia and Wan 2008). A recent study showed that the length of current-year shoots decreased by ∼60% in both young and mature Qinghai spruce (Picea crassifolia Kom.) trees infected by dwarf mistletoe (Arceuthobium sichuanense) compared with those of the noninfected trees (Xia et al. 2012). Stanton (2006) showed that severe dwarf mistletoe infection reduced the radial growth rate in hosts by 28–43% compared with moderately or lightly infected trees (ponderosa pine). Therefore, mistletoe removal can enhance the growth rate of host trees. In our study, both the relative tree ring growth rate (see also Shaw et al. 2008) and the relative annual shoot growth rate were higher in MRT during the posttreatment phase than in MIT during the same time period (Figures 1 and 2).

Effects of mistletoe removal on N and C balance in host trees

We expected to find an overall increase in total N concentrations in MRT after treatment (Hypothesis I). However, mistletoe removal resulted in an increase in N in source tissues (i.e., needles; Figures 3 and 4a) but a decrease in N in sink tissues (i.e., sapwood and roots; Figure 4a) compared with MIT. This increase in MRT needle N may be mainly a result of decreased N absorption by mistletoes from the host canopy because mistletoes were removed. The decrease in N in MRT sapwood and roots might be related to N redistribution, accumulation and utilization in the larger photosynthetically active needle mass (or area) in MRT relative to MIT (Table 1). Rigling et al. (2010) measured needle biomass in the same forest stand and found that the needle dry mass of a mistletoe-infected branch was only one-quarter (6.2 g) the amount of a noninfected branch (26.6 g) (see also Tennakoon and Pate 1996, Meinzer et al. 2004).

Previous studies have revealed that host N is consumed by mistletoes and that mistletoe leaves accumulated N significantly, which resulted in lower leaf N concentrations in mistletoe-infected hosts (Schulze and Ehleringer 1984, Ehleringer et al. 1986b). For example, the mean leaf N ratio of parasite to host was ∼1.6 for Phoradendron leucarpum (Raf.) Reveal & M.C. Johnst. on a variety of its hosts (Panvini and Eickmeier 1993), and for Loranthus europaeus Jacq. on Quercus sp. (Lamont 1983). Ehleringer et al. (1986b) found that leaf N was 10.4% higher in uninfected Juniperus osteosperma (Torr.) than that in Juniperus infected by Phoradendron juniperinum Engelm. ex A. Gray. Tennakoon and Pate (1996) noted that mistletoe tissues generally have much higher N concentrations than any parts of the host. In our study, mistletoe removal and the related disappearance of the parasite N-sink increased leaf N concentrations in host trees (Figures 3 and 4a). This effect is beneficial for the hosts because it probably increases C assimilation and thus facilitates growth (Figures 1 and 2), especially under the dry conditions in the present study.

The processes of litterfall and nutrient cycling at the stand level may also be influenced by mistletoe infection. Mistletoe infection-induced host leaf mass loss (Reid et al. 1994, Rigling et al. 2010) led to an increase in accumulation of litter surrounding mistletoe-infected trees. Moreover, mistletoe litterfall contributes to the overall increase in litterfall. March and Watson (2007) found that litter from mistletoes significantly increased overall litterfall by up to 189% in adult Eucalyptus forests infected by Amyema miquelii (Lehm. Ex Miq.) Tiegh. in Australia. Increased litterfall may imply an increase in nutrient availability in the rhizosphere around mistletoe-infected trees, since nutrients released from litter decomposition are critical for the resupply of soil nutrients (Sullivan et al. 2007). This effect may also have contributed to the increased N concentrations in sapwood and roots of MIT (Figure 4a), since increased litterfall due to mistletoe infection has also been reported at our study site (Rigling et al. 2010). On the other hand, mistletoe-infected trees may allocate more N from the canopy to the stem and belowground part to promote root growth (McCarthy and Enquist 2007) in order to reach and take up more water for both the host tree and the attached mistletoes. Such an effect may be especially important under dry conditions because mistletoes transpire up to nine times as much water as their hosts (Küppers et al. 1992, Schulze et al. 1994, Zweifel et al. 2012).

Nitrogen is a major component of photosynthetic enzymes that regulate photosynthetic activity (Ripullone et al. 2003), and insufficient N directly decreases photoassimilation (Gruber and Galloway 2008). Loss of needle mass (Xia et al. 2012) in combination with decreases in leaf N concentrations (Figures 3 and 4a) may thus lead to lower concentrations of mobile carbohydrates in MIT relative to MRT, as stated in our first hypothesis. In line with our hypothesis, MRT showed significantly higher mass-based NSC concentrations in sapwood and roots (Figure 4d) and a higher area-based NSC concentration in needles (Figure 3) compared with MIT. However, the mass-based NSC concentration in needles was lower in MRT than in MIT (Figure 4d). This decline in needle mass-based NSC concentration (Figure 4d) may be mainly caused by a dilution effect of increased LMA in MRT compared with MIT (Table 1). In contrast, this greater LMA in MRT led to a smaller specific leaf area (SLA = 1/LMA), and thus, MRT had higher needle area-based NSC concentration than MIT (Figure 3).

Higher NSC concentrations in MRT may be caused by higher photosynthesis, lower NSC utilization or both (Li et al. 2002). Lower NSC utilization in MRT was obviously not the case in our study because these trees needed to use more NSC to support greater growth rates compared with MIT (Figures 1 and 2, Table 1). Increases in photosynthesis in MRT may have mainly resulted from a larger number of photosynthetically active needles, as shown in Rigling et al. (2010), larger needle area per needle (Table 1) and increased leaf N concentrations (Figures 3 and 4a). Meinzer et al. (2004) found that hemlock (Tsuga heterophylla (Raf.) Sarg.) branches infected by dwarf mistletoe (Arceuthobium tsugense (Rosendahi) G.N. Jones ssp. tsugense) had only half as much leaf area as noninfected branches, and needles of the infected branches had only half the N concentration and maximum assimilation rate (A_max) as needles of the noninfected branches.

Zweifel et al. (2012) reported leaf stomatal closure of Scots pine infected by V. album as a response that reduced water loss in hot and dry growth conditions. This stomatal closure is an important mechanism to prevent cavitation, and is thus essential for host survival, but it also reduces assimilation and C gain (Polle et al. 2001). Meinzer et al. (2004) estimated that the C accumulation rate of T. heterophylla heavily infected with A. tsugense was up to 60% lower than in uninfected trees. Hence, the C fixation decreases with increasing mistletoe infection level.

A higher ST concentration in MIT needles (Figure 4c) may imply that glucose made by photosynthesis was not invested into growth processes (Figures 1 and 2) but was converted into ST as a storage molecule (Li et al. 2013). An increase in ST concentration may, in turn, downregulate photosynthesis and glucose production (Brugnoli and Farquhar 2000), leading to lower NSC concentrations in MIT compared with MRT (Figure 4d). Watling and Press (2001) proposed that parasites provide a sink for carbohydrates produced by hosts and that some host plants can compensate by increasing photosynthesis via increases in leaf area, delayed leaf senescence and increased Rubisco concentration. In the host trees examined here, however, lower N concentrations observed in MIT suggest that such a compensation did not occur under the dry conditions in the study region.

Not only decreased photosynthesis (Polle et al. 2001, Meinzer et al. 2004, Zweifel et al. 2012) but also extraction of mobile carbohydrates by mistletoes can affect the NSC levels in infected trees. Xylem-tapping mistletoes extract a large amount of carbohydrates from the host (Heizmann et al. 2001, Escher et al. 2008). It has been reported that ∼35–78% of the C in mistletoes were directly extracted from the host (Johnson and Choinski 1993, Marshall et al. 1994, Popp and Richter 1998, Escher et al. 2004, Wang et al. 2008). Moreover, previous studies have suggested that branches infected by dwarf mistletoes have higher dark respiration rates than uninfected branches of Picea mariana (Mill.) BSP (Clark and Bonga 1970), which might also contribute to higher C consumption in infected trees. Extraction of carbohydrates by mistletoes combined with higher dark respiration would accelerate the depletion of new assimilates and C storage in the host, particularly in drought-stressed trees with decreased photosynthesis (Mcdowell et al. 2008). This mechanism may explain our observations that stem sapwood and roots had lower ST and NSC levels in MIT than in MRT (Figure 4c and d).

Effects of mistletoe removal on δ¹³C and δ¹⁸O in host trees

In principle, the dual isotope approach combining δ¹³C and δ¹⁸O measurements allows investigators to distinguish the effects of changed g_s and altered maximum assimilation rates (A_max) on δ¹³C (Scheidegger et al. 2000, Roden and Siegwolf 2012). We hypothesized that the removal of mistletoe would lead to higher A_max, which is supported by the observed higher needle N concentrations (Figures 3 and 4a) and the higher needle g_s (Zweifel et al. 2012). The latter point is based on the assumption that the termination of uncontrolled water consumption by the mistletoe allows host trees to control their stomata less strictly and thus also increase C gain. Tree ring δ¹³C did not differ significantly between MIT and MRT before and after treatment, indicating that the mistletoes did not significantly affect intrinsic WUE (defined as A/g_s and directly related to δ¹³C; Farquhar et al. 1982). However, this result could either be due to no change in both A and g_s or to concerted proportional changes in both parameters causing the quotient A/g_s to remain constant. The increase in δ¹⁸O in MRT compared with MIT in the posttreatment phase (Figure 5b) (accompanied by no change in δ¹³C) is interpreted by the conceptual Scheidegger model (Scheidegger et al. 2000) as a decrease in both A_max and g_s due to the removal treatment. This finding fully contradicts our initial hypothesis that the removal of mistletoe will lead to higher assimilation rates and higher g_s in the host trees, and it also contradicts field measurements of g_s made in Scots pine needles at an adjacent site a few hundred meters away (Zweifel et al. 2012). The pine trees in the study of Zweifel et al. (2012) were the same age as and had similar dimensions to the trees in our study. Moreover, stand structure, soil type and general climatic conditions were fully comparable. Furthermore, the measurements of Zweifel et al. (2012) were performed during the 2004 growing season, which was covered by our tree ring assessments. As a consequence, it is highly reasonable to assume that the gas exchange patterns observed in this previous study are representative for the trees in our present study. In addition, it is highly unlikely that A_max should decrease given the observed increases in leaf N since leaf N is known to be positively related to assimilation rates.

When applying the Scheidegger model, several points need to be taken into account for the interpretation of the results. (i) The general assumption of the Scheidegger model is that the source water isotopic signature is comparable among the different plant individuals tested (e.g., Roden and Siegwolf 2012). Since we chose comparable trees for the control and removal treatment in the same stand, we consider it as a valid assumption that all trees had access to the same water resources. (ii) The Scheidegger model assumes that leaf water evaporative enrichment and consequently δ¹⁸O of organic matter are mainly controlled by the water vapor pressure deficit (VPD) of the air, which in turn controls g_s. Gessler et al. (2009a, 2009b) extended the model to enable comparisons between individual trees growing under conditions with comparable VPD in the same stand, as in our study. According to this model extension, the increase in δ¹⁸O of MRT in the posttreatment phase was due to a reduced leaf level transpiration rate as a result of lower g_s (at the same VPD for MIT and MRT). The reduction in transpiration causes an increase in leaf water δ¹⁸O (and consequently in δ¹⁸O of the assimilates produced within that water). This is due to a relatively larger contribution of evaporatively enriched water from the sites of evaporation in the leaf and a relatively reduced contribution of water from the xylem, which is not evaporatively enriched, to lamina leaf water. This mixing phenomenon, based on the advection of nonenriched xylem water as supposed by back-diffusion of enriched water from the sites of evaporation in the leaf, is referred to as the Péclet effect (cf. Farquhar and Cernusak 2005).

The mixing and thus the Péclet effect, however, are not only driven by transpiration but also by the scaled effective path length (L) for water movement within the leaf (Farquhar and Lloyd 1993, Werner et al. 2012). It is known that L is variable and is determined by leaf morphological (Wang et al. 1998) and physiological (Ferrio et al. 2009) traits. Only recently, Marias et al. (2014) determined the difference in L between mistletoe-infected and noninfected individuals of T. heterophylla, and observed three times higher values in infected plants. Moreover, the authors were able to derive from a sensitivity analysis that the higher δ¹⁸O values in tree rings observed in the uninfected trees in their study (and thus comparable to our results) were most likely not due to differences in g_s but due to the variation in L. Even though we have no L values available for our trees, the fact that experimentally determined g_s increased after mistletoe removal (Zweifel et al. 2012) might point to the same mechanism observed by Marias et al. (2014) and Ferrio et al. (2009). We assessed how changes in L would affect δ¹⁸O of leaf water and thus leaf assimilates based on a detailed dataset on Scots pine leaf water evaporative enrichment in previous year needles published by Barnard et al. (2007). According to these data, a reduction of L by a factor of less than two after mistletoe removal would explain the observed differences in δ¹⁸O between MRT and MIT in the period 2005–10 (Figure 5b), thus being in line with the observation of Marias et al. (2014). Changes in needle anatomy and in the effectiveness of the needle endodermis, which surrounds the vascular strand in conifer needles and might provide hydraulic separation between the xylem and the leaf mesophyll (Zwieniecki et al. 2007), could affect the tortuosity of water flow in the leaf and thus alter L (cf. Roden et al. 2015). In fact, mistletoe infection has been shown to alter the vascular architecture and the distance between the vascular bundle and the endodermis in pine (Pinus contorta) needles (Chhikara and Ross Friedman 2008), which supports this mechanism.

A recent study suggested that, in a leaf with limited ¹⁸O exchange, rather discrete pools of water control the oxygen isotopic composition of leaf water and that the Péclet effect might not be important (Song et al. 2015). However, these authors noted that other studies presented evidence in support of the use of the Péclet correction for modeling δ¹⁸O in organic matter pools and they concluded that more work is needed to reconcile the discrepancy in results.

A potential alternative explanation for the increase in δ¹⁸O in MRT (after the treatment) compared with MIT, which does not require the Péclet effect, is related to the timing of organic material deposited in the tree ring. It is possible that the mistletoe-removed pine trees were able to continue to fix C during periods of water stress, which are often characterized by low relative air humidity, while the MIT closed stomata and did not or only negligibly fixed C (cf. Roden and Siegwolf 2012). Such a situation might have caused the production of cellulose with increased δ¹⁸O in mistletoe-removed trees. This scenario would, however, require that tree ring δ¹³C values were not affected by mistletoe removal, since we observed a difference in δ¹⁸O but not in δ¹³C.

An additional point that might need to be considered is the exchange of organic oxygen atoms with unenriched xylem water during cellulose synthesis (Barbour and Farquhar 2000). Until recently, it was assumed that the proportion of exchangeable oxygen during cellulose synthesis (p_ex) is constant and amounts to ∼42% (Cernusak et al. 2005). Only recently, it has been shown that p_ex might vary during the growing season (Gessler et al. 2009a, 2009b) and can be affected by the turnover time of the NSC pool (Song et al. 2014). The latter authors assumed that a decrease in the C pool turnover time, because of a smaller pool size in relation to the fluxes into (e.g., supply via the phloem) and out of that pool (cellulose production), would give rise to lower probability for hexose molecules to undergo triose cycling (cf. Hill et al. 1995)—which allows carbonyl oxygen to be exchanged with the surrounding water—during cellulose synthesis. In our study, the removal of mistletoes increased growth and depleted the sugar pools in stems, together pointing to faster C turnover. With a simple calculation, we can estimate the change in p_ex required to cause the difference between MIT and MRT in the posttreatment phase (Figure 5). We assume a δ¹⁸O of xylem water of −10‰ as observed for larch trees in the Valais (Treydte et al. 2014) and a p_ex of 0.42 for the MIT. The equilibrium fractionation factor (ϵ_wc) results in carbonyl oxygen being 27‰ more enriched than water (Sternberg and DeNiro 1983) and thus an average leaf water enrichment of 13.4‰ is needed to obtain the mean posttreatment δ¹⁸O in tree rings of MIT. Keeping everything else constant, a decrease of p_ex to 0.38 would explain the observed increase in δ¹⁸O in the MRT. We might thus speculate that such a small decrease in p_ex (0.42 to 0.38) (cf. Gessler et al. 2009a, 2009b) has already resulted in water enrichment in tree rings of trees without mistletoes.

In conclusion, rather than allowing us to accept or reject our initial Hypothesis II, our results clearly point to the limitation of the dual isotope approach proposed by Scheidegger et al. (2000). The dual isotope approach is not applicable if changes in the pathway of leaf water movement superimpose the effects of g_s/transpiration on δ¹⁸O, as recently pointed out by Gessler et al. (2014) and Barbour and Song (2014). Also, changes in p_ex that might be induced by faster C turnover could lead to misinterpretations. Moreover, this method is not suitable when tree rings of two treatments do not record comparable periods during the growing season.

If we assume that g_s increased after mistletoe removal (Zweifel et al. 2012), only an increase in A would explain the unchanged δ¹³C values in the tree rings of MRT compared with MIT. Thus, exactly the opposite result as the one indicated by the dual isotope approach is the most probable.

Sala et al. (2001) investigated needle δ¹³C values in both Douglas fir (Pseudotsuga menziesii Mirb.) and western larch (Larix occidentalis Nutt.) trees with various intensities of mistletoe infection compared with trees that were never infected and found that the uninfected trees had the highest leaf δ¹³C values within each species, and leaf δ¹³C values decreased (became more negative) significantly with increased levels of mistletoe infection for both species. In tree rings of T. heterophylla, δ¹³C was also significantly lower in infected compared with uninfected trees (Marias et al. 2014). Logan et al. (2013) found that δ¹³C values of needles from branches of white spruce (Picea glauca (Moench) Voss) infected by Arceuthobium pusillum Peck were significantly lower (more negative) than values of needles from uninfected branches.

From our results, we may speculate that our mistletoe removal experiment at the whole-tree level at a dry site increased g_s and assimilation rate to a similar extent. As a result, A/g_s remained constant, whereas in most other cases reported in literature, one of the two responses dominated.

Conclusion

Tree growth is a consequence of many complex physiological processes, among which C fixation is considered the most important factor (Samuelson et al. 1998). Under the dry conditions prevailing at our experimental site, mistletoe removal clearly increased shoot and stem growth of host trees. This growth increase was related to an increase in needle N concentration found in the present study and to an increase in needle g_s found in the same tree species at the same site (Zweifel et al. 2012), which points to increased photosynthetic capacity after mistletoe removal. The NSC concentrations in tree tissues partially support this assumption but preclude straightforward interpretations because of increased biomass production and the related dilution effects in MRT, as well as differences in the water status between MRT and MIT. The dual isotope approach, from which we aimed to derive retrospective information on assimilation rate and g_s, did not produce results that could be interpreted in this way. Our results strongly point to the need for further research on the effects of biotic and abiotic conditions on the scaled effective path length for water movement in the leaf, which can affect δ¹⁸O in tree rings but is not accounted for in the Scheidegger model (Scheidegger et al. 2000). Moreover, a deeper understanding of the variability of p_ex and thus the relative contribution of leaf water enrichment to the tree ring δ¹⁸O signal is necessary. However, combining the tree ring δ¹³C data of MRT and MIT with the observations of Zweifel et al. (2012) that needles from uninfected branches of Scots pine showed higher g_s than infected ones indicate that an increase in assimilation rate as a consequence of the removal treatment is most likely.

Supplementary data

Supplementary data for this article are available at Tree Physiology Online.

Conflict of interest

None declared.

Funding

C.-F.Y.’s research visit in Switzerland was supported by the Sino-Swiss Science and Technology Cooperation (SSSTC) program (Project No. EG44-092011). The study was partly supported by the Chinese State Key Laboratory of Forest and Soil Ecology (Grant No. LFSE2013-01) and the National Natural Science Foundation of China (Grant No. 41371076).