Experimental design and growth conditions

The study was conducted on European beech trees. In 2006, beech seeds were collected in several forests in the Lorraine region of France and sown in 2007 in biodegradable horticultural pots made of wood fiber and filled with a peat and sand mixture. The seedlings were grown for one year in a nursery (INRA Grand-Est Nancy, France). In 2008, about 1000 of the seedlings were transplanted and grown for 7 more years in open ground at the INRA Grand-Est nursery (Champenoux, France, 48°75’N, 6°34’E, 229m asl). In 2014, a rain exclusion system was built above the 8-year-old trees: a semi-rigid structure supporting a transparent roof built with polycarbonate sheets and nets installed around the roof to intercept lateral rain. Light intensity, air temperature and vapor pressure deficit were monitored below the roof. The trees under the roof were subjected to four different treatments for two years (2014, 2015): (1) control (C) in which the trees were regularly irrigated; (2) defoliation (D) in which the trees were submitted to a yearly defoliation and regularly irrigated: manual defoliation of the trees in treatment D was done each year in June; 75% of the total foliage was removed and the removal was homogeneously distributed throughout the tree crown; (3) moderate drought (MD) and (4) severe drought (SD), where the trees were submitted to two levels of soil water deficit. The two drought stress levels were not designed to realistically simulate a climate change scenario, but rather to create drought conditions that were so unfavorable that they would likely cause beech mortality. In fact, lateral rain entering under the roof created some variability in soil water status in the drought treatment at the time of labeling and this allowed us to select trees with contrasting levels of water stress. The soil in the drought treatments was isolated by a rigid waterproof plastic sheet 1.80 meters depth buried vertically around the area. The water status of the seasonal sub-sets of trees in each treatment (24 trees per treatment) was checked by measuring pre-dawn water potential in twigs (ψpd) at day of year (DOY) 150, 176 and 198 in We sampled the twigs (one per tree) before sunrise and performed the ψpd measurement with a pressure chamber (PMS Instruments, Albany, OR, USA). 

Soil characteristics and soil water measurements

The studied site was characterized by 60cm-deep homogenous soil with an average texture (Silt: 61 ± 1.28%; Clay: 27 ± 0.98%; Sand: 12 ± 0.66%), a pH comprised between 7.5 and 8, an organic matter content between 12.1 and 14.9 g.kg-1(E Silva 2010) and a total N comprised between 0.54 to 0.87 g.kg-1. Below 60cm, the grey marl of the Jurassic inferior (Lotharingian) era was characterized by a swelling heavy clay soil with a relatively high bulk density.We used neutron probes (TROXLER TX 4301, Research Triangle Park, NC, USA) to measure the volumetric water content of the soil. Three neutron probe access tubes (aluminum, closed at their base) were installed in each of the four treatment areas in order to quantify water content at different depths: two ranged from 0-1m in depth and one ranged from 0-1.6m. During the growing season, measurements were carried out every two weeks. Counts were logged every 10 cm for the upper 100 cm, and every 20cm below that.For each depth i (thickness ti), Total Available Water soil Content (TAWC in mm) was calculated by estimating the characteristic points from pedotransfer classes for gravimetric soil moisture at field capacity (θfc) and gravimetric soil moisture at wilting point (θwp). The characteristic points were checked and adjusted with probe measurements, during winter for volumetric soil moisture at field capacity and during summer for volumetric soil moisture at wilting point. Soil bulk density was assessed with the cylinder method. Relative Extractable soil Water (REW in %) was calculated according to Bréda et al., (1995) as follows: REW=100* TAWC-R TAWC where R is the actual volumetric soil water content in mm, and total soil extractable water content down to 1.60m is estimated to 310 mm. The soil in the C and D treatments was irrigated regularly throughout the experiment with an automatic drip watering system which delivered between two and four liters per tree two to three times a week. We adjusted the amount of the water according to the REW measurements in order to avoid any water shortage (REW >0.4), with 40% of the REW corresponding to the critical threshold where trees start to avoid water loss by closing their stomata (Granier et al., 1999).

Choice of the branch

One branch per tree was selected at chest height for the 15N labeling experiment. Branch diameter and the distance from the base of the branch to the apex of the tree were measured as were the number of leaves on the branch and individual and total leaf area. A significant global treatment effect was observed on the number of leaves and individual leaf area. Total leaf area was influenced by season (p<0.01) with a significant treatment x season interaction (p<0.01). Defoliation did not decrease branch growth in 2014, but in 2015, branch diameter was significantly reduced in the defoliated trees. In the spring 2015, after the 2014 defoliation, the number of leaves on defoliated branches was similar to the number on the water-stressed and control branches. In the summer 2015, the selected branches in the D trees were thicker (diameter significantly larger) than the ones in the C trees and also contained more leaves than controls before 75% of defoliation.

Foliar 15N labeling procedure

Labeling experiments were performed at two dates (Figure VI.1): 1) in spring, at the end of May 2015 (DOY: 148) when the leaves of all the trees were fully expanded, and 2) in summer, in July 2015 (DOY: 187) one month after defoliation had been performed on the D trees and when radial growth was maximal. At each date, spring and summer, a set of 12 trees per treatment (C, D, MD and SD) were randomly chosen for labeling (giving 48 trees for each date and 24 trees per treatment for the whole labeling experiment).On each tree, a branch bag made of polyethylene was placed over the labeled branch to isolate it from its local environment. In the late afternoon, an aqueous solution of 15N-urea was sprayed inside the bag onto the leaves with a hand sprayer (Zeller et al., 1998). Urea is considered to be the most suitable form of N for foliar application due to its non-polarity, rapid absorption, low phyto-toxicity and high solubility (Knoche et al., 1994, Zeller et al., 1998). Leaves absorb N in urea much faster than N in mineral form (Zeller et al., 1998) and urea N is rapidly converted into amino acids used for leaf protein synthesis or for N export (Dilley et al.,1961, Dong et al., 2002). The 15N-urea solution (10.4 atom%, 5.0 g.L-1) was sprayed in a fine mist, which limited the formation of drops and ensured a homogeneous labeling of the leaves. After the labeling, the branch bag was kept on all night, then very carefully removed the next morning to avoid any contamination among trees. The timing of the foliar labeling is summarized in Figure VI.1.

Sampling protocol

Leaf and twig samples were collected from 12 of the selected trees per treatment before labeling to determine their natural 15N baseline abundance. After 15N labeling, leaf material from each labeled branch was sampled over a chase period of 14 days. At three sampling dates (day 0.5, 1, 2), five small leaf disks (38 mm²/disk, 1 disk per leaf and per date) were removed with a hole-puncher from five randomly selected leaves along the labeled branch. For the three dates, this sampling procedure was repeated on the same set of five leaves. At day 4, the remaining part of the five punched out leaves was taken. One additional leaf was also collected from six trees per treatment to assess total protein quantification and 15N assimilation into proteins. We chose day 4 as a sampling date because the turnover of foliar proteins, especially Rubisco, is rapid enough to allow an incorporation of 15N into foliar proteins within that time, as shown fornnn 15N labeled rice leaves by Suzuki et al., (2001). At day 7, a second set of five leaves per branch was hole-punched as described above. At day 14, the whole branch (both wood and remaining leaves) plus several leaves from the terminal apical twig of the selected trees were harvested (Figure VI.1).Each sampled branch was split by annual growth unit [current year (Y), 1-year-old (Y-1) and older (<Y-1)]. All samples were gently and carefully washed in distilled water, frozen in liquid nitrogen and stored at -80°C. The samples were then freeze-dried [Dura-Top (r), Dura-Dry (r), FTS Systems (r), Stone Ridge, NY, USA], weighed and ground into a fine powder in a ball mill (CEPI SODEMI CB2200, Cergy, France).