Low root reserve accumulation during drought may lead to winter mortality in poplar seedlings


Author for correspondence:

David A. Galvez

Tel: +1 604 822 2507

Email: david.galvez@ales.ualberta.ca


  • Climate models suggest that more frequent drought events of greater severity and length, associated with climate change, can be expected in the coming decades. Although drought-induced tree mortality has been recognized as an important factor modulating forest demography at the global scale, the mechanisms underlying drought-induced tree mortality remain contentious.
  • Above- and below-ground growth, gas exchange, water relations and carbon reserve accumulation dynamics at the organ and whole-plant scale were quantified in Populus tremuloides and P. balsamifera seedlings in response to severe drought. Seedlings were maintained in drought conditions over one growing and one dormant winter season.
  • Our experiment presents a detailed description of the effect of severe drought on growth and physiological variables, leading to seedling mortality after an extended period of drought and dormancy. After re-watering following the dormant period, drought-exposed seedlings did not re-flush, showing that the root system had died off.
  • The results of this study suggest a complex series of physiological feedbacks between the measured variables in both Populus species. Further, they reveal that reduced reserve accumulation in the root system during drought decreases the conversion of starch to soluble sugars in roots, which may contribute to the root death of drought-exposed seedlings during the dormant season by compromising the frost tolerance of the root system.


Nonstructural carbohydrates (NSC) reserves play a fundamental role in plant germination, growth, reproduction, defense and survivorship under stress. Although most of these roles have been studied for more than a century (Brown & Escombe, 1898; Halsted, 1902), the interaction between NSC reserves and water transport, especially under hydraulic stress, has not received much attention until recently. Although potentially critical in understanding and projecting drought-induced tree mortality with climate change, the underlying interrelationships between carbon dynamics and water transport at the organ and whole-plant level are far from understood (McDowell et al., 2008; McDowell & Sevanto, 2010; Sala et al., 2010; Ryan, 2011). Furthermore, the notion that, under drought stress, carbon allocation and water transport are probably closely coupled processes has received growing appreciation (Galvez et al., 2011; McDowell, 2011; McDowell et al., 2011; Landhäusser & Lieffers, 2012; Sala et al., 2012). In a recent review, Sala et al. (2012) suggested that carbon reserves in trees are sustained at a minimum level, which is needed to maintain metabolic processes and hydraulic integrity, particularly under episodes of severe drought. The authors further speculated that the maintenance of hydraulic integrity is prioritized in order to avoid hydraulic failure, which could make tissue carbon reserves irretrievable (i.e. if xylem suffers catastrophic levels of embolism) after water transport ceases, and the potential remobilization of reserves from sources to sinks would not be possible.

The interaction between carbon dynamics and water transport under drought conditions is probably more complex than is currently understood. With a few exceptions (Galvez et al., 2011; Anderegg, 2012; Anderegg et al., 2012), the interaction between carbon and water has been traditionally assessed using functional proxies of whole-plant carbohydrate and hydraulic status (e.g. mass accumulation, CO2 assimilation and stomatal conductance) instead of direct measurement (i.e. content and concentration of NSC and percentage loss of hydraulic conductivity (PLC)). An understanding of the possible feedbacks between carbon dynamics and water relations (McDowell & Sevanto, 2010) from a wider, more integrative perspective is crucial to interpret current ecological phenomena, such as drought-induced sudden aspen decline across the western USA and Canada (Michaelian et al., 2010; Worrall et al., 2010; Anderegg et al., 2012).

Xylem vulnerability to hydraulic failure and stomatal behavior (e.g. the timing and cues of stomatal opening and closure) are important drivers of plant responses to drought stress, and can vary with species (Sperry & Pockman, 1993; Nardini et al., 2001). Trembling aspen (Populus tremuloides) and balsam poplar (Populus balsamifera) are both fast-growing boreal forest trees species that can coexist in sites that have similar mesic, edaphic and climatic conditions. However, both species have different tolerances to drought stress, with aspen being more tolerant than balsam poplar. During the onset of water limitation, leaf stomatal conductance (gs) in aspen decreased in parallel with soil water content, maintaining a relatively constant stem water potential (Galvez et al., 2011), whereas, in balsam poplar, gs remained initially relatively unchanged until the soil water content had decreased below a threshold, after which gs and stem water potentials changed abruptly. Both species also appear to show different vulnerabilities to hydraulic failure associated with a 50% loss of conductivity (P50) when growing under similar environmental conditions, with balsam poplar being more prone to hydraulic failure (P50 at a stem water potential of −1.3 to −1.4 MPa) than trembling aspen (P50 at a stem water potential of −2.2 to −2.3 MPa; Hacke & Sauter, 1995; Lu et al., 2010).

To gain a further understanding of the possible feedbacks between carbon dynamics and water relations and to further integrate these processes, we hypothesize that severe drought will not only be detrimental to gas exchange and water relations, but also to the allocation of carbon to growth and reserves in seedlings of both species; however, we expect the impact to be greater in the more drought-sensitive P. balsamifera. We further hypothesize that these effects will be lasting and that they will continue to affect the performance of seedlings in the following growing season.

Materials and Methods

Plant material

One hundred and twenty trembling aspen (Populus tremuloides Michx; hereafter Pt) and balsam poplar (Populus balsamifera L; hereafter Pb) seedlings were grown from open pollinated seed sources collected near Edmonton, AB, Canada (53.6°N, −113.3°W). Seedlings were established under well-watered conditions in a glasshouse at the University of Alberta, Canada in April of 2011. Glasshouse conditions were a 18-h photoperiod at 21 : 18°C with a humidity of c. 60%. After 4 wk, seedlings were transplanted into individual plastic pots (4 l; 15.2 cm in diameter with four equidistant perforations at the base to allow excess water to drain), with one seedling per pot, and filled with Metro Mix medium (Metro Mix 290, Terra Lite 2000; W. R. Grace of Canada, Ajax, ON, Canada).

After transplanting, plants were watered daily and fertilized with 200 ml of 10–52–10 NPK solution (1 g l−1 per pot) every 2 wk for 4 wk. After 4 wk, the plant height and stem basal diameter of all seedlings were recorded and a distribution curve for these variables was constructed for each species. Once the average of each variable per species had been calculated, 84 Pt and Pb seedlings, each with height and basal diameter closest to their respective average, were kept, and the rest of the seedlings were discarded. The 84 remaining seedlings per species were randomly selected and assigned to six groups of 12 plants each. Plants were randomly reassigned in each group until no significant differences (tested with one-way ANOVA) in initial plant height and stem basal diameter were detected among the six groups. Six plants within each group were randomly selected and assigned to a well-watered control treatment (hereafter referred to as CON seedlings), and the remaining six plants were assigned to a drought treatment (hereafter referred to as DRY seedlings). On 1 June 2011, all plants were moved outside the glasshouse into cold frames. Transparent lids were attached to the frames, allowing them to be closed in the event of rain. There were only few rain events and therefore light availability was not reduced substantially throughout the experiment. The lids did not close completely and allowed for sufficient air circulation, preventing heating within the cold frame.

Application of the drought treatment

Following the same methodology developed during past experiments (Galvez et al., 2011), the DRY seedlings were slowly desiccated in a controlled process. Starting on 26 June 2011, DRY seedlings were weighed daily using an Adam Equipment digital balance model PGW 4502e (Danbury, CT, USA). After each weight had been recorded, DRY seedlings were re-watered by adding the equivalent of half the weight that was lost from the day before. This process was repeated for 10 d. This desiccation protocol was implemented to simulate a gradual soil drying process. Leaf water potential measurements were performed at days 8, 9 and 10 on three randomly selected plants of each species using one leaf per plant (data not shown). By allowing leaf and stem water potentials to equalize, leaf water potential can be used as a proxy for stem water potential (Begg & Tuner, 1970). To accomplish this, leaves selected to be measured were kept inside an aluminum foil envelope for 2 h before the leaf water potential measurement to allow the leaf water potential to equilibrate with the stem water potential. Water potential measurements were performed using a Compact Water Status Console (i.e. a portable Scholander-type pressure chamber) model 3115P40G4 (Soilmoisture Equipment Corp., Santa Barbara, CA, USA). By the end of the desiccation period, midday stem water potential Ψmd in DRY Pt was c. −2.2 MPa and −1.2 MPa in DRY Pb. These Ψmd values are slightly less negative than the Ψmd values associated with a 50% loss of conductivity (P50) at −2.2 to −2.3 MPa in Pt and at −1.3 to −1.4 MPa in Pb (Hacke & Sauter, 1995; Lu et al., 2010). The percentage loss of hydraulic conductivity in stems (PLCstem) is a good integrator of hydraulic status at the whole-plant scale because, during the day, the stem water potential (Ψstem) is commonly less negative than Ψleaf, but more negative than Ψroot. During the desiccation protocol, daily gas exchange measurements were performed in DRY and CON seedlings of both species following the protocol described in the following subsection. After the initial 10-d period, DRY seedlings were watered with the average full amount of daily water loss determined at the 10th day of the 10-d drying period. CON seedlings were kept well watered throughout the growing season. Watering was discontinued after 15 October 2011 as soils were frozen and seedlings had shed their leaves. A layer of straw (20 cm) was placed on top and around the remaining pots to preserve soil moisture and provide some thermal insulation (see below) inside the cold frame, in addition to the natural insulation provided by the snow cover.

Gas exchange and PLC measurements

On 11 July 2011, the CO2 assimilation rate (A) and leaf stomatal conductance (gs) were measured in six randomly selected DRY and CON seedlings of each Pt and Pb. All gas exchange measurements were performed using a LI-6400 portable photosynthesis system (Li-Cor, Lincoln, NE, USA). All measurements were performed between 09:00 and 11:00 h on the youngest fully expanded leaf. Chamber's reference CO2 concentration was set to 385 ppm using a 12-g Li-Cor CO2 cartridge as CO2 source. Light environment in the chamber was set to 1800 μmol m−2 s−1 after a 10-min induction period at 500 μmol m−2 s−1 using the 6400-2B red/blue LED light source of the LI-6400's chamber. The induction period was implemented to stabilize air humidity, flow and temperate prior to exposure of the measured leaf to the light-saturating photon flux density (PFD) level which the seedlings experienced in the open conditions. Measurements were taken after 3 min, when A and gs values were stable. The cuvette conditions were based on light response curves that had been determined before the measurements on three individual plants (data not shown). From these curves, the optimum induction time and the PFD to achieve maximum A were determined. These same measurements were repeated on 24 July, 8, 22 August and 5 September 2011 (i.e. 2, 4, 6 and 8 wk after the first set of measurements was taken). All measurements in the rest of this section were performed using the same number of samples and the time schedule described above.

PLC was measured using a conductivity apparatus (Sperry et al., 1988) following the standardized, and now traditional, protocol for this widely used equipment. Seedlings were cut at the stem base in the cold frames and transported to the laboratory (c. 200 m), wrapped in damp towel paper inside black plastic bags to minimize stem dehydration. Stem samples for PLC measurements were collected before 11:00 h. Stems cut from the pot were re-cut under water, separating the 15-cm stem section proximal to the original cutting site in order to remove embolisms induced by exposure of open xylem vessels to atmospheric pressure when cut from the stem base. Previous work by the authors has shown that the maximum vessel length in both species is c. 10 cm (data not shown), and hence the removal of the 15-cm section proximal to the original cutting site was considered to be adequate to remove tissue that had been embolized during sample collection. Keeping the re-cut stem under water, five consecutive 2-cm stem segments from each stem were cut using a razor blade. Segments from each stem were mounted and measured at the same time in the conductivity apparatus. The apparatus' reservoir tank was filled with filtered (0.2 μm) 100 mM KCl solution prepared in deionized water. After the initial hydraulic conductivity (i.e. the initial value of kh, expressed as ki in Eqn 1) of each stem segment had been measured, the native embolism was displaced by flushing KCl solution from the reservoir under constant pressure (120 kPa) for 2 min. After flushing, the segment's measured final hydraulic conductivity was taken as kmax. Preliminary tests were performed to ensure that changes in kmax values after repeated flushing were not different from zero (M. T. Tyree, unpublished). PLC was calculated from Eqn 1:

display math(Eqn 1)

ks was calculated from ki/Aw, where Aw is the stem cross-sectional area.

Growth and NSC measurement

At each sampling period, all tissue sampling was performed between 09:00 and 11:00 h. The heights of the selected seedlings of Pt and Pb were recorded and the plants were separated into roots, stems and leaves. Roots were carefully washed to remove all substrate. Once cleaned, roots were patted dry with paper towels and left exposed to air circulation for 5 min. After this time, the root systems were attached to a small metal clamp by the root collar and carefully submerged inside a water reservoir placed on top of a digital balance (Adam Equipment PGW-6002e, 6000 g, 0.01-g increments). The root volume was determined by water displacement, recording the weight reading before and after root immersion, and verifying that lateral roots were not pressing against the bottom or sides of the container. After the root volume had been determined, each root system was patted dry again, individually bagged, labeled, stored cooled and sent to the laboratory for immediate processing.

Leaves from each plant were detached from the stems, bagged, labeled and sent to the laboratory for processing. Stem material was processed as indicated in the previous section to perform PLC measurements. After PLC measurements had been conducted, all segments of each stem were bagged, labeled and sent to the laboratory for processing. The total leaf area per plant was measured on the same day using an LI-3000 leaf area meter (Li-Cor); after the measurements, leaf, stem and root materials were oven dried at 70°C for 72 h and weighed. During all sample processing (i.e. seedling harvest and measurements), the handling time was minimized and tissue samples were always refrigerated in between steps. All dried samples were individually ground using a Wiley Mill to pass a 40-mesh (0.4-mm) screen and the ground tissues were used to estimate water-soluble sugar and starch concentrations. Soluble sugars were extracted three times with hot 80% ethanol, followed by a reaction between the extract and phenol–sulfuric acid, and measured colorimetrically with a spectrophotometer (Genesys 10S UV-Vis; Therma Scientific, Madison, WI, USA) at an absorbance of 490 nm (Chow & Landhäusser, 2004). To measure starch concentrations, the tissue remaining after the ethanol extraction was digested with the enzymes α-amylase and amyloglucosidase, followed by a reaction with peroxidase-glucose oxidase-o-dianisidine to be measured colorimetrically at an absorbance of 525 nm (Chow & Landhäusser, 2004). Soluble sugar and starch concentrations of roots, stems and leaves are presented at the organ scale and also as an average total concentration of NSC at the whole-plant scale.

Seedling dormancy and leaf re-flush

To explore whether there was a lasting effect of the drought treatment on our seedlings, we assessed the growth performance of the seedlings after a dormant resting period. After the last growing season measurement in September 2011, all remaining seedlings were left inside their cold frames to be exposed to winter conditions until 15 January 2012. On that date, all seedlings were relocated inside the glasshouse to environmental conditions similar to those previously described during seed germination. After soils had thawed (2 d after seedlings were moved inside), a set of six DRY and six CON seedlings for each species was collected and all response variables, other than leaf measurements, were taken as described previously. All remaining seedlings (CON and DRY) were watered and maintained at or near field capacity until the end of the study. The experiment was terminated on 27 February 2012 after all CON seedlings had been flushed and had expanded their leaves. At that time, all response variables were measured. To assess the lasting effect of drought on the seedlings, tissue damage and/or seedling mortality was evaluated by the extent of stem and root necrosis and by the ability to flush or sprout from existing or newly formed buds on stems or the root system.

Experimental design and data analyses

The experimental design was analyzed as a 2 × 2 × 7 factorial design with two species, two drought treatments (DRY and CON seedlings) and seven collection times. All growth data were normally distributed and variances were equal. Three-way ANOVAs were performed for height, leaf area, leaf number, PLC, stem water potential and root dry weight response variables, using the statistical software package SigmaStat 4 (Systat Software Inc., Chicago, IL, USA). Differences between means were considered to be significant at α = 0.05. When significant differences between the means were detected, all-pairwise multiple comparisons using the Holm–Sidak procedure were performed. The CO2 assimilation rate, stomatal conductance, sugar and starch concentration datasets failed tests for normality or independence of variance. These response variables were fitted with a linear mixed-effects model using the functions lme and varIdent from the nlme R package (Pinheiro et al., 2010) to allow different variance structure for time and treatment. Once the datasets had been fitted, they were analyzed using ANOVA procedures with the statistical package R (R Development Core Team, Vienna, Austria). Differences between drought treatments, species and collection times stated in the following 'Results' and 'Discussion' sections were statistically significant at α = 0.05, unless mentioned otherwise.


Gas exchange during controlled desiccation

During the controlled desiccation period of 10 d, CO2 assimilation (A) and leaf stomatal conductance (gs) in well-irrigated seedlings were 16.7 μmol CO2 m−2 s−1 and 0.43 mol H2O m−2 s−1, respectively, in Pt and 18.4 μmol CO2 m−2 s−1 and 0.5 mol H2O m−2 s−1, respectively, in Pb, and remained fairly constant; however, A and gs were c. 16% and 10% higher in Pb than in Pt. There was an immediate negative response of stomatal conductance during the onset of drought in DRY Pt seedlings, whereas A lagged somewhat behind. Both A and gs continued to decrease until seedlings reached the desired drought target. At that point, A and gs were 4 μmol CO2 m−2 s−1 and 0.05 mol H2O m−2 s−1, respectively, in Pt and 7.35 μmol CO2 m−2 s−1 and 0.08 mol H2O m−2 s−1, respectively, in Pb (Fig. 1a,b). During the same desiccation period, seedlings of Pb showed a distinctly different short-term stomatal behavior. Assimilation and gs remained similar to the values of the well-irrigated CON seedlings for c. 8 d into the desiccation process, after which both variables declined rapidly (Fig. 1). At the end of the desiccation process and the start of the experiment, gs and A values in DRY seedlings of both species were much lower than in the CON seedlings.

Figure 1.

Mean (± SE) CO2 assimilation (a) and leaf stomatal conductance (b) of Populus tremuloides (Pt, circles) and Populus balsamifera (Pb, squares) seedlings under drought (open symbols) and well-irrigated (closed symbols) conditions, during the desiccation process (= 6). CON, well-watered control treatment; DRY, drought treatment.

Effects of drought on seedling growth

Regardless of species and drought treatment, the height growth of all seedlings stopped c. 4 wk into the experiment, probably as a result of the natural outside growing conditions in late summer signaling the end of the active height growth period in August (Fig. 2a). However, in the first 4 wk, the height of CON Pt seedlings increased from 65.2 to 89.3 cm, whereas that of CON Pb seedlings increased from 32.2 to 47.6 cm. During and after the desiccation process, the height growth in DRY seedlings slowed substantially and the total growth in the first growing season was only 8.6 cm in Pt and 5.9 cm in Pb. After shoot expansion had ceased, no new leaves were added in the CON seedlings (Fig. 2b). During the fall, all seedlings shed their leaves. After re-flush (27 February 2012), CON Pt and Pb seedlings had leaf numbers similar to the seedlings of the previous growing season, suggesting that no substantial damage to the bud meristems had occurred over the winter period. DRY Pt seedlings steadily shed their leaves during the first 6 wk of the experiment, reducing the leaf numbers from an average of 25 to 10 leaves. Leaf numbers increased again to 16 leaves by 5 September 2011, after some dormant buds had started to flush. In the DRY Pb seedlings, leaf numbers steadily decreased over the whole growing season from 17 to nine and a re-flush of leaves did not occur. After the dormant period and re-watering of the DRY seedlings, no new leaves redeveloped in these seedlings (Fig. 2b). Concomitantly, the average leaf area of CON Pt and Pb seedlings increased during the first 4 wk of the experiment and remained stable for the rest of the growing season, whereas the leaf area in DRY seedlings decreased steadily from 923.1 to 66 cm2 in Pt seedlings and from 480.4 to 97 cm2 in Pb seedlings (Fig. 2c).

Figure 2.

Mean (± SE) plant height (a), number of leaves (b), leaf area (c) and root volume (d) of Populus tremuloides (Pt, circles) and Populus balsamifera (Pb, squares) seedlings under drought (open symbols) and well-irrigated (closed symbols) conditions (= 6). Gray area indicates dormant period (not to scale). Dotted line indicates approximate date when seedlings were moved inside the glasshouse. CON, well-watered control treatment; DRY, drought treatment.

Stem necrosis of 50% or more was evident in most of the stems of aspen seedlings, but there was little evidence of stem necrosis in Pb.

Well-watered Pt and Pb seedlings added c. 50% new root volume to the root system during the early part of the experimental period, after which the root volume remained relatively constant with a decrease in root volume after the dormant period and a rapid recovery after 6 wk into the second growing season. Visual inspection of the roots during cleaning suggested a reduction in the number of fine roots (< 1 mm in diameter) of both species during the dormant period, but no diameter-based quantification was performed (Fig. 2d). The average root volume in DRY seedlings of both species was already lower at the start of the experiment. This initial loss of roots is probably an artificial effect of the root extraction under dry soil conditions. However, throughout the experiment, there was no evidence of any new fine root growth in the DRY seedlings (Fig. 2d). The coarse root systems of all CON and DRY seedlings did not show any signs of necrosis, and phloem and cambium tissues remained light colored during the first growing season and shortly after the root system had thawed out after the dormant period. Six weeks later, the entire root system of DRY seedlings had turned necrotic, whereas the root system in CON remained light colored.

Gas exchange and water relations: response to drought

Throughout the growing season, A in CON Pt and Pb decreased steadily from 12.1 to 6.24 μmol CO2 m−2 s−1 in Pt seedlings and from 13.9 to 7.46 μmol CO2 m−2 s−1 in Pb seedlings (Fig. 3a). In 2012, after the CON seedlings had been flushed in the glasshouse, A was 7.06 μmol CO2 m−2 s−1 in Pt seedlings and 8.47 μmol CO2 m−2 s−1 in Pb seedlings. Stomatal conductance and A continued to decrease after the desiccation process and were both close to zero at the end of the first growing season (Fig. 3a). As no new leaves were produced in 2012 after the dormant period, no additional measurements of A, gs or Ψmd could be taken in DRY seedlings of both species. Initially, gs in CON Pb seedlings was much higher (0.525 mol H2O m−2 s−1) than in Pt seedlings (0.272 mol H2O m−2 s−1) and, consequently, the seasonal decline in Pb was substantially steeper than that in Pt (Fig. 3b).

Figure 3.

Mean (± SE) CO2 assimilation (a), leaf stomatal conductance (b), percentage loss of hydraulic conductivity (PLC, c) and stem water potential (d) of Populus tremuloides (Pt, circles) and Populus balsamifera (Pb, squares) seedlings under drought (open symbols) and well-irrigated (closed symbols) conditions (= 6). Gray area indicates dormant period (not to scale). Dotted line indicates approximate date when seedlings were moved inside the glasshouse. CON, well-watered control treatment; DRY, drought treatment.

Average PLC in stems of CON seedlings of both species remained below 10% for the whole first growing season; however, PLC was 35.8% in Pt and 31.2% in Pb after the dormant period (Fig. 3c). In DRY seedlings, the average PLC only remained comparable with PLC in CON seedlings for the first 2 wk and, by the beginning of September, the stem xylem was > 80% embolized in both species (Fig. 3c). The average stem water potential (Ψmd) in CON seedlings of both species remained above −1 MPa for the whole experiment. Average Ψmd decreased from −2.16 to −2.66 MPa in DRY Pt seedlings and from −1.2 to −1.5 MPa in DRY Pb seedlings (Fig. 3d).

Total nonstructural carbohydrate concentration at the whole-plant level

The average combined NSC concentration (sugar + starch) across all plant organs (TotConcplant) increased throughout the growing season and into the dormant season in the CON seedlings of both species (Pt, 19.6%; Pb, 23.5%). There was a marked decrease in TotConcplant in both species during the leaf flush period (Fig. 4). In DRY Pt seedlings, TotConcplant increased during the first 4 wk, reaching concentrations similar to those of the CON Pt seedlings (19%), but then decreased continually to c. 7% with no further drop during the re-flush period. In DRY Pb seedlings, TotConcplant showed a 2-wk lag before TotConcplant increased; however, TotConcplant was less (15%) than in the CON Pb seedlings (19%). TotConcplant for the remainder of the season stayed constant, only to decrease during the dormant period and re-flush period to 12% (Fig. 4).

Figure 4.

Mean (± SE) whole-plant total nonstructural carbohydrate (NSC) concentration (sugars + starch) of Populus tremuloides (Pt, circles) and Populus balsamifera (Pb, squares) seedlings under drought (open symbols) and well-irrigated (closed symbols) conditions (= 6). Gray area indicates dormant period (not to scale). Dotted line indicates approximate date when seedlings were moved inside the glasshouse. CON, well-watered control treatment; DRY, drought treatment.

Concentration of soluble sugars and starch at the organ level


The average soluble sugar concentration in leaves (SugConcleaf) in CON seedlings of both species increased steadily throughout the growing season from 13% to 21%. Leaf starch concentrations (StaConcleaf) were generally low and, only later in the growing season, did StaConcleaf increase to a maximum of 2% (Fig. 5a,b). In DRY seedlings, SugConcleaf increased during the first 4 wk of the experiment, but then leveled out at approximately 17%. Average StaConcleaf in DRY seedlings remained below 1% for the entire experiment (i.e. at the detection limit; Fig. 5a,b).

Figure 5.

Mean (± SE) sugar (left column) and starch (right column) concentration in leaves (a, b), stems (c, d) and roots (e, f) of Populus tremuloides (Pt, circles) and Populus balsamifera (Pb, squares) seedlings under drought (open symbols) and well-irrigated (closed symbols) conditions (= 6). Gray area indicates dormant period (not to scale). Dotted line indicates approximate date when seedlings were moved inside the glasshouse. CON, well-watered control treatment; DRY, drought treatment.


Stem soluble sugar concentrations (SugConcstem) in CON seedlings of both species increased slightly over the growing season, with a prominent peak during the dormant season, followed by a substantial drop during the leaf flush period. SugConcstem in CON Pb seedlings tended to be 2% higher than in CON Pt (Fig. 5c). During the first growing season, the stem starch concentration (StaConcstem) in CON Pt and Pb seedlings increased from < 1% to 4.5%. In both species, this increase was followed by a drop in StaConcstem to 1% in the dormant period and was close to the detection limit during leaf flush (Fig. 5d). During the first 6 wk of the experiment, SugConcstem in DRY seedlings also increased to levels similar to those in CON seedlings; however, DRY seedlings of both species did not exhibit a peak SugConcstem during the dormant season, but rather a decline (Fig. 5c). StaConcstem of DRY Pt and Pb seedlings increased slightly early in the growing season, but stayed well below StaConcstem of the CON seedlings and was at the detection limit during the dormant season (Fig. 5d).


Root sugar concentrations (SugConcroot) of CON and DRY seedlings were rather stable (7%) during the first growing period and were no different between the treatments (P = 0.081; Fig. 5e). However, the same cannot be said for the starch concentrations in the root system (StaConcroot). Although CON seedlings of both species accumulated substantial starch resources in their root systems, increasing from 5% to 19% over the growing season, roots of both the DRY Pt and Pb increased their starch reserves from < 2% to only 7% at the end of the growing season (Fig. 5f). During the dormant period, SugConcroot in CON seedlings increased by more than twofold to 21% in Pt and to 25% in Pb seedlings, whereas StaConcroot in these seedlings dropped to 5% (Fig. 5e,f). This abrupt change (switch) in sugar and starch concentrations during the dormant period was not detectable in DRY seedlings of both species, where SugConcroot and StaConcroot dropped continuously towards the re-flushing period. (Fig. 5e,f).


After the onset of drought stress in Pt and Pb seedlings, our results suggest that a cascade of responses occurred in the drought-exposed seedlings: (1) soil desiccation limited A and gs, which (2) limited stem growth and, with that, the production of new leaves and possibly fine roots; (3) as seedlings stopped growing, photoassimilates were initially re-directed to accumulate in stem and root tissues; and (4) although hydraulic conductivity continued to be compromised and the season progressed towards dormancy, reserves did not continue to accumulate substantially and were only half that of the non-drought-exposed CON seedlings during the dormant season. However, a decrease in photosynthesis and an increase in reserves in woody tissues also appear to be natural seasonal processes in seedlings growing under outside conditions. Under well-watered conditions, assimilation also declined in these seedlings throughout the growing season, whereas NSC in stem and root tissues accumulated once height growth had terminated (Fig. 5).

After shoot growth had ceased, root growth increased in the CON seedlings. The cues for these seasonal changes in Populus are probably the shortened day length and cooler night temperatures, which are known to induce tissue hardening and dormancy (Ibáñez et al., 2010). Similar late-seasonal root growth has been observed in mature boreal aspen stands (Landhäusser & Lieffers, 2003). At the end of the growing season and before the dormant season, NSC reserves in the drought-exposed seedlings did not reach or surpass the NSC levels of the CON seedlings. This pattern was in contrast with an earlier study that described a substantial increase in NSC reserves in seedlings growing under glasshouse conditions (Galvez et al., 2011). In that study, the well-watered Pt seedlings did not have environmental cues to terminate height growth, and therefore grew continuously, and the newly acquired carbon was probably used to maintain height growth, whereas the drought-exposed seedlings stopped growing as a result of the drought stress and started to accumulate NSC reserves in their tissues. This substantial difference between these two studies clearly highlights the importance of phenological stage as a controlling factor for seasonal variation in NSC acquisition, accumulation and allocation. In ecophysiological field and glasshouse studies, this impact is commonly overlooked, in particular when investigating complex perennial plants such as trees (Landhäusser & Lieffers, 2012).

Drought-exposed seedlings in the current study showed a clear decline in starch concentration of the root system; however, its potential impacts were only revealed after the dormant period. After re-watering following the dormant period, it became clear that the roots in the drought-exposed seedlings were dead. This could not have been detected prior to or during the dormant season measurement, as roots did not show any visible signs of necrosis of their tissues, even after they had thawed out after the dormant season. However, shortly thereafter, the root systems turned black and no live root segments, new root tips or root suckers were detected on these root systems. Interestingly, even at this stage, root NSC reserves were not depleted completely, supporting the hypothesis that minimum reserve requirements exist in plant tissues (McDowell, 2011; Sala et al., 2012).

The low concentration of NSC and the heavily embolized stems and roots probably compromised the re-initiation of leaves on the shoot and/or the development of root suckers, and new root growth after the dormant period. Although no PLC measurements were performed on roots, it is possible that roots suffered severe embolization prior to or during the dormant period (Pittermann & Sperry, 2006), compromising root function. However, the low NSC reserves in the roots could also have resulted in poor frost protection of the root tissues. No profound conversion of starch reserves to soluble sugars during the dormant period was detected in the DRY seedlings, whereas CON seedlings demonstrated this conversion. This conversion has been described for many different perennial plant species, and is considered to be a mechanism for the frost protection of tissues (Levitt, 1980).

Combining the effect of hydraulic failure and disruption of carbon reserve accumulation

By the end of the first growing season, the drought-exposed seedlings had suffered catastrophic hydraulic failure; PLC was above 90% in Pt and above 80% in Pb (Fig. 3c). At that time, we observed substantial stem necrosis in both species, which affected a large portion of the stem, but remained in all seedlings well above the root collar. Stem necrosis and phloem tissue damage are well-known responses of woody stems to severe drought stress, but these alone were not sufficient to kill seedlings in the short term (Lu et al., 2010). In the light of the heavily embolized stems, the likelihood of flushing from these necrotic shoots in both species was very small; however, as the non-necrotic portions of the lower stem had lateral buds and the coarse root system did not show signs of necrosis (e.g. maintained the potential to the root sucker), seedlings should have been able to re-sprout from these living tissues (Lu et al., 2010). Lu et al. (2010), however, only desiccated 1-yr-old Pt seedlings in a short-term drought until all leaves had been shed and the stems were necrotic (PLC averaged 90%); when re-watered, these plants were able to re-sprout from axial buds or from their roots. The shoot symptoms matched the conditions of our drought-exposed seedlings in September 2011. However, in their study, seedling carbon reserves and the performance after a dormant season were not measured. Both studied Populus species have the ability to regenerate from adventitious root sprouts; therefore, a still functioning root system with sufficient NSC reserves should have been able to produce new shoots (Landhäusser et al., 2006; Snedden et al., 2010). Clearly, this was not the case. To our knowledge, this is the first study to present experimental data detailing the complex set of physiological processes interacting with seasonal abiotic signals. Further, our results indicate that hydraulic damage during drought impairs carbon accumulation in the roots, which might, in turn, have hampered root survival of winter conditions if the stress survived that long. Nevertheless, uncertainty remains with regard to the exact timing of seedling death, which may have occurred before the dormant period. Under this scenario, changes in NSC could be interpreted as a consequence of death, rather than a mechanism leading to mortality. Interestingly, the change in total NSC (i.e. soluble sugars + starch) concentration at the whole-plant scale (TNSCplant) as a result of drought was predominantly driven by the NSC concentration in the root tissues (Figs 4, 5); hence, any stress affecting NSC accumulation in the roots of the seedlings probably has a direct effect on the whole-plant performance (Landhäusser & Lieffers, 2012).

The average leaf soluble sugar concentration (SugConcleaf) in DRY seedlings increased during the first 4 wk of the experiment, even though A in DRY seedlings remained at least 50% lower than that in CON seedlings. This initial increase in SugConcleaf may suggest the onset of osmotic adjustment, a well-documented response to drought in Populus species (Gebre et al., 1994, 1998) and other tree species (Tschaplinski et al., 1998). We hypothesized that the increase in SugConcstem of DRY seedlings, measured during the growth period, could also have been an osmotic response that could have been used to up-regulate the xylem pressure potential, which has been suggested as a potential mechanism for the repair of xylem embolism (Secchi et al., 2011). The maintenance of NSC in the stems is also important because shoots need access to reserves for the new leaf flush after the dormant season (Landhäusser, 2011). The fact that, by the end of the experiment, only CON seedlings had produced new leaves and increased their root volume highlights the relevance of starch accumulation during the growing season and starch-to-sugar conversion during the cold hardening period (Levitt, 1980; Sauter, 1988; Fig. 4). Starch is a compound with no other known biological function in plants apart from storage, and it is needed to buffer periods of stress (Kozlowski & Pallardy, 2002). Starch-to-sugar conversion is a temperature-dependent adaptive mechanism well studied in Populus (Sauter, 1988; Sauter & van Cleve, 1991). This conversion plays an important role in the maintenance of cell membranes at low temperatures and increases freezing tolerance (Levitt, 1980). Recently, Hennon et al. (2012) have suggested that frost damage to fine roots as a result of reduced snow depth and soil drainage could be a main driver of the extensive yellow cedar die-off events in Alaska. Snow cover is probably an important factor protecting root systems from freezing, even in boreal forests (Hennon et al., 2012). In our study, after CON seedlings had thawed, SugConcroot in CON seedlings declined as apical and lateral meristems became active and new leaves expanded, a process previously observed in other Populus and Salix species (Sauter, 1988; Von Fircks & Sennerby-Forsse, 1998).

Although Pt and Pb can overlap on sites with similar mesic, edaphic and climatic conditions (Peterson & Peterson, 1992; Landhäusser et al., 2002; Landhäusser & Lieffers, 2003), they thrive in distinctive habitats (Burns & Honkala, 1990). Pt forms extensive stands in mesic to dry mesic upland sites, whereas Pb occupies the moister (flood plains or seepage areas) and cooler extremes (Rood et al., 2003, 2007). This difference in habitat is also reflected by the different hydraulic (Tyree et al., 1994) and stomatal adaptations to drought. Stomatal behavior in response to drought was distinctively different between Pt and Pb seedlings during the desiccation process (Fig. 1). During the drying period, gs in DRY Pt seedlings declined earlier and more rapidly than in DRY Pb seedlings. The decline in gs in DRY Pt seedlings suggests a more isohydric behavior (i.e. leaf stomatal conductance decreased as soil desiccation progressed; Tardieu & Simonneau, 1998), in comparison with a much slower response in DRY Pb seedlings, which maintained similar gs to well-irrigated CON seedlings for a period of 8 d before gs started to decrease (Larchevêque et al., 2011).

To our knowledge, our work is the first study to present detailed experimental data illustrating a complex feedback between stomatal behavior, gas exchange, water relations and carbon reserve accumulation dynamics. These responses and feedbacks are clearly influenced by seasonality, making the role of drought stress as a driver of plant mortality dependent on the interaction between phenology and physiology, and very likely the ontogeny of plants.


This research was made possible by research grants from the Canadian Forest Service, Natural Sciences and Engineering Research Council (S.M.L. and M.T.T.), a Discovery Grant and an Alberta Research Institute and Alberta Ingenuity Equipment Grant. M.T.T. wishes to thank the US Forest Service for salary support whilst working at the University of Alberta. We thank Pak Chow in the Department of Renewable Resources at the University of Alberta for assistance during the measurements of NSC. The authors also thank the personnel of the glasshouse facility at the Faculty of Agricultural, Life and Environmental Sciences for help whilst growing the seedlings, Eckehart Marenholtz for the design and construction of wood frames, and three anonymous reviewers for helpful comments and suggestions on an earlier version of this work.