Identifying the relevant carbohydrate storage pools available for remobilization in aspen roots

Abstract

Nonstructural carbohydrate (NSC) remobilization remains poorly understood in trees. In particular, it remains unclear (i) which tissues (e.g., living bark or xylem) and compounds (sugars or starch) in woody plants are the main sources of remobilized carbon, (ii) to what extent these NSC pools can be depleted and (iii) whether initial NSC mass or concentration is a better predictor of regrowth potential following disturbance. To address these questions, we collected root segments from a large mature trembling aspen stand; we then allowed them to resprout (sucker) in the dark and remobilize NSC until all sprouts had died. We found that initial starch mass, not concentration, was the best predictor of subsequent sprout mass. In total, more NSC mass (~4×) was remobilized from the living inner bark than the xylem of the roots. After resprouting, root starch was generally depleted to <0.6% w/w in both tissues. In contrast, a large portion of sugars appear unavailable for remobilization: sugar concentrations were only reduced to 12% w/w in the bark and 2% in the xylem. These findings suggest that in order to test whether plant processes like resprouting are limited by storage we need to (i) measure storage in the living bark, not just the xylem, (ii) consider storage pool size—not just concentration—and (iii) carefully determine which compounds are actually components of the storage pool.

Introduction

The storage and remobilization of nonstructural carbohydrates (NSCs) are important processes that allow trees to temporarily maintain a negative carbon (C) balance (i.e., utilize C at a greater rate than can be supplied by photosynthesis). Remobilizing NSCs from storage may allow trees to support growth (e.g., Cherbuy et al. 2001, Barbaroux et al. 2003, Carbone et al. 2013), tissue repair (Bucci et al. 2003), defense (e.g., Guérard et al. 2007) or flowering and seed development (Ichie et al. 2013, Han and Kabeya 2017) at critical times, such as the beginning of the growing season or during periods of high demand. NSC storage may also help trees survive long periods of minimal C gain that may occur seasonally (i.e., winter) or sporadically and unpredictably, such as during drought or other disturbances (e.g., Canham et al. 1999, Bréda et al. 2006). Despite its importance, however, we know very little about the remobilization of stored carbohydrates in trees, including how it is controlled and when it affects tree survival and recovery after disturbance (Dietze et al. 2014, Hartmann et al. 2018).

One hindrance to our understanding of remobilization and its potential for suppling C to different tree functions is that we do not know how much of the NSC pool is actually available for remobilization (Dietze et al. 2014, Hartmann et al. 2018). For instance, it is likely that the sugar pool cannot be completely depleted in living tissues, because sugars perform multiple roles in the cell and will not only function as storage molecules (Hoch 2015, Martínez-Vilalta et al. 2016). A few studies have explored the degree to which sugar and starch pools can be depleted by subjecting seedlings to light deprivation. These studies show somewhat ambiguous results, and the final minimum sugar concentrations upon death by starvation range from 1% to over 6–8% depending on the tissue and range of sugars measured (Marshall and Waring 1985, Piper and Fajardo 2016, Wiley et al. 2017, Weber et al. 2018). Some of the discrepancy may be the result of methodological differences between studies, which can be a large source of variation in NSC measurements (Quentin et al. 2015). However, some of the variation may be of biological significance, relating to differences among species, the determination of time of death (e.g., Wiley et al. 2017, Weber et al. 2018) or due to the use of different methods of quantifying different sugars (Landhäusser et al. 2018). There is more of a consensus in regard to starch remobilization; starch was largely depleted in all studies (although see Piper and Fajardo 2016). However, these studies were all conducted on seedlings, and it is unclear if these results can be extrapolated to mature trees, particularly older tree tissues (Hartmann et al. 2018). The only study that has explored lethal light deprivation in larger trees (Sevanto et al. 2014) reported minimum NSC concentrations of ~1% in young twigs and needle tissues combined for 2-m tall piñon pine at death, but individual tissue concentrations were not reported, and other organs such as roots or older tissues were not measured. The majority of organs and tissues within mature trees are much older and larger than those yet studied and may therefore be more likely to contain sequestered NSC (i.e., unable to be remobilized, sensuMillard et al. 2007) or exhibit remobilization limitations arising from increased structural complexity and distances between C sources and sinks (Hartmann et al. 2018).

Another issue that complicates the quantification of carbohydrate storage pools is that we often do not know from where in a tree NSC is remobilized to support various processes and whether organs or tissues differ in the extent to which remobilization can deplete NSC pools. This is problematic because in many studies, not all tissues and/or organs are sampled, even though there is evidence that the degree of seasonal remobilization can differ between organs (e.g., Martínez-Vilalta et al. 2016). The NSC remobilization may differ between tissues within organs as well. In trees, studies of NSC dynamics often focus on the sapwood, as it is thought to represent the largest C reserve pool in terms of total mass (Barbaroux et al. 2003, Würth et al. 2005). The inner or living bark (including secondary phloem, the cortex and the phelloderm), on the other hand, is often overlooked as a storage tissue because of its generally smaller size and the focus on its role in C transport and protection (Rosell 2016). However, the inner bark has a large proportion of parenchyma cells (Srivastava 1964). Thus, it can have a much higher NSC concentration than sapwood (Loescher et al. 1990, Landhäusser and Lieffers 2003, Zhang et al. 2014) and, in some species, a similar total NSC mass (Zhang et al. 2014) or, in seedlings, an even greater total NSC mass (Yang et al. 2016). The inner bark could therefore be a critical source of NSC during remobilization (Srivastava 1964, Rosell 2016). In addition, because the inner bark and xylem contain different proportions of living cells and perform different functions, these tissues may differ in the degree to which NSC can be remobilized to support processes like growth (Wiley et al. 2017). Therefore, knowledge of remobilization in both tissues is necessary, if only to verify that measurement of one tissue is sufficient to extrapolate to the other.

Finally, it is not clear whether the relative (i.e., tissue NSC concentration) or the absolute size (i.e., NSC mass or ‘pool size’) is the appropriate measure of a storage pool as it relates to plant survival (Ryan 2011, Hoch 2015). Because bigger trees (or organs) require more C to maintain tissue functions, NSC concentrations rather than total mass may better reflect a plant’s C balance and ability to survive C stress (Hoch 2015). However, Canham et al. (1999) argued that while NSC concentration will be more important for survival mechanisms like freezing tolerance, where sugars function as cryoprotectants (Steponkus 1984, Morin et al. 2007), NSC mass might be more critical when reserves are needed to replace lost or damaged tissues (e.g., resprouting). Previous studies report contrasting results, with NSC mass (Myers and Kitajima 2007), concentration (Webb 1981, Gleason and Ares 2004) or both (Canham et al. 1999, Poorter and Kitajima 2007) correlating with subsequent survival under deep shade or following defoliation. In general, it can be difficult to separate the effects of NSC concentration and total mass because they are often highly correlated, as differences in NSC mass can be caused by differences in concentration. But making this distinction is important, because if a larger NSC pool per se can improve the chance of survival, then bigger trees or root systems should have higher survival or recovery potential than smaller trees or root systems under C stress, even if they have similar (or lower) NSC concentrations.

Resprouting is a common strategy used by plants to survive disturbances that destroy or kill aboveground tissues. As this process must rely initially on previously stored reserves in the roots as the main source of C (Bowen and Pate 1993, Bond and Midgley 2001, Clarke et al. 2013), resprouting can be used to improve our understanding of NSC remobilization in trees and its relationship with survival. Trembling aspen (Populus tremuloides Michx.) is a prolific resprouter, producing sprouts (i.e., suckers) from adventitious buds that are typically preformed on roots (Farmer 1962, Peterson and Peterson 1992, Frey et al. 2003). In aspen, even detached segments of roots ranging from 0.5 cm to 3 cm in diameter will resprout readily (Maini and Horton 1966). By using root segments of known dimensions instead of intact root systems, quantification of the mass of different NSC pools before and after remobilization can be easily and more accurately assessed. In addition, by using roots of different diameter, the initial NSC mass can be manipulated independently of concentration. Finally, because we can manipulate root segments of mature trees in controlled settings, they provide an excellent yet manageable system to study the process of remobilization and its limitations in more mature tree tissues, as aspen roots of 2–3 cm in diameter are well over 20 years old (DesRochers and Lieffers 2001) and are still capable of resprouting.

While NSC storage may limit the growth of sprouts, so too may the availability of nutrients like nitrogen (N) (El Omari et al. 2003, Kabeya and Sakai 2005), which are also necessary components of all plant tissues. For instance, in aspen root segments, fertilization with NH₄NO₃ significantly increased sprout mass (Fraser et al. 2002), and sprout growth was greater on sites with higher soil N availability in the shrub Erica australis (Cruz et al. 2003). It remains largely unknown if the amount of stored—as opposed to newly assimilated N—ever limits resprouting. However, stored N is known to be remobilized and used for early spring leaf flush and fine root growth (Millard 1994, Palacio et al. 2007, Uscola et al. 2015), leaf regrowth following defoliation (Cerasoli et al. 2004) and growth during resprouting (El Omari et al. 2003). Therefore, it is feasible that the amount of stored N could limit resprout growth, particularly if soil N availability is low. In this case, root N concentrations could also affect demand for remobilized NSC for the growth and subsequent maintenance of sprouts. Therefore, examination of the initial nutrient status of roots may be important for the interpretation of the degree to which NSC pools are depleted during resprouting.

In this study, we explored relationships between initial sugar, starch and N concentrations and mass in the root bark and xylem tissues and the subsequent sprout mass produced in aspen roots to determine (i) which pools were the main sources of remobilized C and N for resprouting and (ii) whether the total mass or the concentration of storage pools was a better measure of resprouting potential. Finally, after growth stopped and sprouts died, we examined the remaining sugar and starch concentrations in the roots to better understand (iii) whether bark and xylem have different minimum NSC concentrations indicative of starvation and (iv) what proportion of the four NSC pools considered—bark starch, bark sugars, xylem starch and xylem sugars—was unavailable for remobilization and therefore not a relevant component of the C storage pool in aspen roots.

Materials and methods

Experimental setup

A total of 45 live aspen roots (~30 cm long, 1–3.5 cm diameter) were hand-excavated in the field from several locations in a large mature, monospecific fire-origin aspen stand in late September 2015 near Utikuma Lake, Alberta, Canada (56°04′ 45.05′′N, 115°28′58.74′′W). Locations were 10–15 m apart, and individual roots within a location were at least 1–2 m apart. Collected roots were covered with wet paper towels and plastic wrap to prevent desiccation, transported back to the lab and stored at 4 °C for 26 days. At the start of the experiment, any attached fine roots were clipped off and the ends of each root were trimmed by at least 1 cm to remove damaged tissue. An additional 2-cm section was then clipped off at both ends of each root for initial NSC and N analysis. Root diameter was measured at both ends of the remaining roots, which ranged from 20 cm to 25 cm in length. Samples for initial NSC and N analysis were stored at −20 °C until further processing. Plastic trays (26 × 52 × 6 cm) were filled with a 2.5-cm bed of growth media comprised of a 1:2 mixture of perlite and vermiculite (PRO-MIX, Premier Tech Horticulture, Rivière-du-Loup, Canada). The 45 roots were then placed in the trays, with 3–5 roots per tray, and covered with an additional 0.5 cm of growth media. Trays were placed in a growth chamber (Conviron, Winnipeg, Canada) kept at 22 °C and 65% humidity with no light. Roots were kept well watered (approximately every second day) and were left to resprout (sucker) entirely in the dark to ensure that only stored rather than newly assimilated C was used to produce sprouts.

Before resprouting, roots were pre-assigned to one of two harvest groups: (i) roots harvested when all their sprouts stopped growing (Harvest 1; n = 20) or (ii) roots harvested when all their sprouts were dead (Harvest 2; n = 25). Roots were assigned to harvest groups ensuring a similar range of diameters. To determine when sprouts had ceased growth for Harvest 1, the height of the five tallest sprouts per root were measured every second day; the five tallest sprouts were chosen because these tended to grow most vigorously, while the shortest sprouts stopped growing early on. Growth was considered to have stopped when the combined new height growth of all five sprouts measured was less than 0.1 cm/day. For Harvest 2, sprout death was defined at the point when all sprouts on a root had at least 1 cm of necrotic tissue at their tip or base. Because the timing of sprout growth termination and sprout death varied greatly among roots, roots in each harvest group were harvested on different days. This was done to ensure that roots in each harvest group were all harvested at a similar phenological stage. For individual roots, all sprout growth ceased between 35–107 days (average = 54 days) after the experiment started; sprouts had all died by 63–157 days (average = 102 days) after the experiment started.

For both harvests, all sprouts were counted, removed from the root, oven-dried at 70 °C for 72 h and then weighed. Then, two or three root samples of 2–5 cm length (depending on root diameter) were collected for NSC analysis. At Harvest 2 (when sprouts were dead), we also calculated root water content (WC) from measurements of fresh and dry weight of another 2–4 cm length root sample to provide evidence that sprout death was driven by C not water limitation. The WC was calculated as

\begin{equation} \mathrm{WC}=\frac{\mathrm{Fresh\ weight- Dry\ weight}}{\mathrm{Fresh\ weight}}\times 100.\end{equation}

(1)

Finally, for a subset of 17 roots at Harvest 2, we used a redox indicator to determine if roots were still alive by placing thin slices of bark tissue in a solution of 0.5% tetrazolium (K & K Laboratories, Inc.), which stains respiring tissues pink/red.

Nonstructural carbohydrate and N analysis

All NSC and N root samples were separated into ‘bark’ and ‘xylem’ samples. Bark samples were mostly separated without difficulty from the xylem by hand or with a razor blade; bark samples included all inner and outer bark tissues outward from the vascular cambium. Since aspen roots only have a very thin, papery layer of dead, outer bark tissue (see Figure S1 available as Supplementary Data at Tree Physiology Online), our bark sample is more equivalent to a sample of only inner bark—which contains secondary phloem, cortex and phelloderm tissues. This is in contrast to species that have a much thicker outer bark layer, which would, if included, significantly impact NSC measurements. Xylem samples included all wood tissue inward from the vascular cambium. All samples were then dried at 100 °C for 1 h to denature enzymes, dried at 70 °C for 4 days, weighed and then ground to 40-mesh (0.4 mm) using a Wiley Mill (Thomas Scientific, Inc.). Starch and total soluble sugar concentrations were analyzed using the phenol–sulfuric acid assay described in Chow and Landhäusser (2004) and Landhäusser et al. (2018). In this assay, following extraction in 80% ethanol, sugars are oxidized by sulfuric acid and measured after phenol addition, which induces a color-producing reaction that can be measured colorimetrically at 490 nm. Starch remaining in the pellet after extraction is broken down into glucose using α-amylase and amyloglucosidase. The resulting glucose content is measured by the addition of a peroxidase-glucose oxidase-o-dianisidine solution (Chow and Landhäusser 2004), which results in a color-producing reaction that can be measured colormetrically at 525 nm. The resulting glucose concentration was then converted to starch concentration assuming a conversion factor of 0.9, based on differences in molecular weight. Sugar and starch concentrations for each root were averaged across all segments for bark and xylem, separately. Total NSC concentration is the sum of starch and sugar concentrations.

We also explored whether the degree of depletion of glucose, fructose and sucrose (GFS) was similar to the degree of depletion observed in the total soluble sugar pool by analyzing the concentrations of GFS individually for a subset of five roots from the initial sample and from Harvest 2. For GFS analysis, tissue samples were extracted in 80% ethanol at 90 °C for 30 min. After removal of ethanol and reconstitution in water, extracts were analyzed following Landhäusser et al. (2018). Glucose was phosphorylated by adenosine triphosphate with hexokinase (Sigma G3293; Sigma-Aldrich, St. Louis, MO, USA) and subsequently oxidized to gluconate-6-phosphate by Nicotinamide adenine dinucleotide (NAD+) in the presence of glucose-6-phosphate dehydrogenase (Sigma G3293), converting NAD+ to NADH. The amount of NADH produced was then determined spectrophotometrically at 340 nm, with moles of NADH equivalent to moles of glucose. Fructose+glucose concentration was determined by the same procedure with the addition of phosphoglucose isomerase (Sigma P5381) that converted fructose-6-phosphate to glucose-6-phosphate; fructose concentration was then calculated by subtracting the glucose concentration. Finally, GFS content was determined by the addition of invertase (Sigma I9274) and a repetition of the procedure for fructose determination; sucrose concentration was then determined by subtracting the glucose and fructose concentration.

Initial N concentrations were also analyzed for bark and xylem of each root, using the same samples taken for NSC. Concentration was determined using the Dumas combustion method with the Costech Model EA 4010 Elemental Analyzer (Costech International Strumatzione) at the University of Alberta (Edmonton, AB, Canada).

Calculations and statistical analysis

Bark, xylem and whole-root NSC and N mass were extrapolated from the volume and dry weight of one of the NSC sample segments in the following way. The volumes of the NSC segment and the whole root were estimated, assuming cylindrical shape, from the length and average diameter (both ends) of the NSC segment and the root. Then, given the dry weight of the bark and xylem tissue of the NSC segment, we estimated bark or xylem dry weight as

\begin{align}& \text{Estimated root tissue wt}\nonumber\\&\text{= Estimated root volume}\kern0.5em \times\kern0.3em \frac{\text{segment tissue wt}}{\text{segment volume}}. \end{align}

(2)

Bark and xylem NSC and N mass (g) for each root were then calculated as the product of estimated root tissue weight and NSC or N concentration. Structural mass (i.e., non-NSC) was then also estimated as the difference between estimated root tissue weight and NSC mass. Whole-root NSC and N mass were calculated as the sum of bark and xylem NSC mass or N mass. We also calculated whole-root NSC concentration by taking a weighted average of bark and xylem NSC concentrations, weighted by the estimated root tissue weight.

Table 1

Relationships between total sprout mass (g) and initial N and NSC mass (g) and concentration (%).

Predictor variable	Relationship with sprout mass1	R2	P
Initial bark N mass1	+	0.69	<0.0001
Initial xyl N mass1	+	0.67	<0.0001
Initial bark NSC mass1	+	0.64	<0.0001
Initial xyl NSC mass1	+	0.69	<0.0001
Initial bark starch mass1	+	0.79	<0.0001
Initial xyl starch mass1	+	0.78	<0.0001
Initial bark sugar mass1	+	0.49	<0.0001
Initial xyl sugar mass1	+	0.55	<0.0001
PC 1 (N1 and starch1 mass)2	+	0.80	<0.0001
Initial bark N %	None	0.01	0.58
Initial xyl N %	−	0.16	0.0063
Initial bark NSC %	+	0.15	0.0092
Initial xyl NSC %	None	0.01	0.51
Initial bark starch %	+	0.07	0.079
Initial xyl starch %	None	0.004	0.67
Initial bark sugar %	None	0.01	0.50
Initial xyl sugar%	None	0.007	0.57

¹ln-transformed.

²First principal component of a PCA of ln-transformed initial root starch mass and N mass (accounted for 93% of variation).

Abbreviations: xyl (xylem). Bolded values were statistically significant (P < 0.05).

Figure 1.

Relationships between initial root starch mass and N mass and subsequent sprout mass.

To determine whether total mass or concentration was a better predictor of resprouting potential, we used linear regressions between initial NSC (starch and/or sugar) and N concentrations and mass and subsequent sprout production for Harvests 1 and 2 combined (no difference in sprout biomass between harvests: t = 0.41, P = 0.68). Sprout production was expressed as the total dry weight of all sprouts and any fine roots produced (only 6 out of 45 roots produced fine roots, which only accounted for 3–9.5% of total sprout production). In addition, we also compared initial NSC and N concentrations with relative sprout production (i.e., total sprout mass∕root volume) to determine if concentrations affected sprout production when controlling for variation in root size.

In cases where initial N and NSC (sugar or starch) measures were both strongly related to sprout production, we additionally used principal components analysis (PCA) to construct a predictor variable that combined N and NSC. Because N and NSC mass were correlated, the first principal component of root N mass and root starch mass explained 93% of the variance, with larger values indicative of both high N mass and high starch mass. Principal component values for each root were then regressed against sprout mass to determine if a combination of N and starch predicted sprout mass better than N or NSC alone. We similarly applied PCA to bark and xylem N and starch concentrations and then used the first principal component (explaining 73% of variation) to determine if a combination of N and starch concentrations predicted ‘relative’ sprout mass better than either concentration alone.

To determine the extent to which different NSC pools (bark versus xylem, sugar versus starch) were available for remobilization (or not), we used paired t-tests to compare initial concentrations and total mass and final concentrations and total mass at Harvest 2 (i.e., when all sprouts had died). To compare the difference in total NSC mass remobilized from the bark versus the xylem, we calculated the difference between initial and final NSC mass for both the bark and xylem of each root as described above. We then calculated the ratio of NSC mass remobilized from the bark to that remobilized from the xylem; we tested whether this ratio was significantly different from one (indicating greater remobilization from either tissue) by determining if the 95% confidence interval (CI) of this ratio included one. Lastly, we used linear regressions between final NSC concentrations and relative sprout mass, root WC and initial NSC and N concentrations to explore potential sources of variation in the amount of un-remobilized NSC remaining in the roots. All analyses were done in JMP 12.1.0 (SAS Institute, Inc.).

Results

Effects of initial NSC and N concentration and mass on sprout production

Overall, initial NSC mass, N mass, bark sugar concentrations and sprout mass increased with increasing root diameter, while initial N and starch concentrations decreased with root diameter (see Table S1 available as Supplementary Data at Tree Physiology Online). Across both harvests, total sprout mass was correlated most strongly with initial starch mass and N mass (Table 1), particularly when bark and xylem were combined (i.e., root starch and N; Figure 1). The first principal component of a PCA of both N mass and starch mass was also highly correlated with sprout mass (Table 1), although this measure did not explain more variation than starch mass alone (Figure 1). Unlike the N and NSC mass, initial N and NSC concentrations were generally unrelated to sprout production, with the exceptions of xylem N % (negatively related) and bark NSC % (positively related); however, both were only weakly correlated with sprout mass (Table 1).

Though initial root N and starch concentrations were weak predictors of total sprout mass, they were both highly correlated with ‘relative’ sprout mass (i.e., sprout mass per root volume; Table 2), especially when bark and xylem were combined (i.e., root concentrations: a weighted average of bark and xylem concentrations; Figure 2). A PCA of bark and xylem N and starch concentrations yielded a principal component that was even more highly correlated with relative sprout mass than either starch concentration or N concentration alone (Table 2), indicating that both resources could be limiting sprout mass production. Initial xylem sugar concentration was unrelated to relative sprout mass; however, in the bark, initial sugar concentration and relative sprout mass were negatively correlated (Table 2).

Table 2

Relationships between relative sprout mass (g per cm³ root) and initial N and NSC concentrations.

Predictor variable	Relationship	R2	P
Initial bark N%	+	0.51	<0.0001
Initial xyl N%	+	0.48	<0.0001
Initial bark NSC%	+	0.13	0.0171
Initial xyl NSC%	+	0.58	<0.0001
Initial bark starch%	+	0.59	<0.0001
Initial xyl starch%	+	0.61	<0.0001
Initial bark sugar%	−	0.35	<0.0001
Initial xyl sugar%	None	0.02	0.40
1PC 1 (root N% and starch%)	+	0.77	<0.0001

¹First principal component of a PCA of initial root starch and N concentrations (accounted for 80% of variation).

Abbreviations: xyl (xylem). Bolded values were statistically significant (P < 0.05).

Figure 2.

Relationships between initial root N and starch concentrations and subsequent relative sprout mass at both harvests combined. Relative sprout mass is the total sprout mass divided by the total root volume. Whole-root concentrations are dry mass-weighted averages of bark and xylem concentrations.

Nonstructural carbohydrate pool remobilization and minimum concentrations at sprout death

By Harvest 2 when the sprouts had died, the amount and degree of NSC remobilization differed substantially between the bark and xylem. On average, over four times more total NSC mass had been remobilized from the bark than from the xylem (Figure 3a and b). Despite this greater remobilization in absolute terms, NSC concentrations were reduced proportionally more in the xylem than in the bark (68% versus 45%; paired t = 9.91, P < 0.001; Figure 4). Sugar concentrations were reduced by 51% in the xylem, significantly more than in the bark where sugar concentrations were only reduced by 19% (paired t-test: t = 9.1, N = 25, P < 0.001). The starch pool, on the other hand, was largely depleted in both tissues (92% and 88% reduction in bark and xylem, respectively). From the estimates of NSC mass loss (i.e., sugar and starch pools combined), we also estimated the loss of structural mass (i.e., non-NSC mass) during resprouting. Structural mass declined by a relatively similar amount: 11.9% ± 2.5% standard error (SE) in the bark and 9.8% ± 2.6% in the xylem.

Figure 3.

NSC mass remobilized from the bark and xylem. (a) Remobilized NSC mass as a function of root volume. Remobilized NSC mass was calculated as the difference between initial NSC mass and the final NSC mass at Harvest 2, when sprouts had died. (b) Average ratio of NSC mass remobilized from the bark to NSC mass remobilized from the xylem. Error bars represent 95% CI, n = 25. The CI lies far above 1, indicating a significantly greater remobilization from the bark than the xylem.

Figure 4.

Nonstructural carbohydrate concentrations in the bark and xylem. The NSC concentrations were measured at the start of the experiment (initial; n = 45), at Harvest 1 (when sprout height growth ceased; n = 20) and at Harvest 2 (when all sprouts had died; n = 25). The hashed portion of each bar represents starch concentration, with the remainder (open portion) representing sugar concentration. Error bars = SE for sugar and starch concentrations.

During resprouting, starch concentrations declined more than sugar concentrations did in both tissues, indicating that more mass (and hence more C) was remobilized from the starch pool than the sugar pool. In the bark, starch concentrations declined by 4.2 ± 1.0% (Δ%) more than sugar concentrations (paired t-test: t = 4.10, N = 25, P < 0.001). In the xylem, there was a similar pattern with starch concentration averaging a decline of 1.27 ± 0.25% more than sugar concentration (paired t-test: t = 5.07, N = 25, P < 0.001).

The extent of remobilization during resprouting also varied between different components of the sugar pool. The combined GFS concentration declined significantly between the beginning of the experiment and Harvest 2 in both the xylem and the bark (paired t-tests: t = 3.78, N = 5, P = 0.019 and t = 4.1, N = 5, P = 0.001; Figure 5). In contrast, the non-GFS sugar concentration did not decline significantly, although there was a trend for a reduced concentration in the xylem (t = 1.91, N = 5, P = 0.13; Figure 5). In both tissues, the decline in GFS concentration was due to the reduction in sucrose; glucose and fructose concentrations were low and did not change (Figure 5).

Figure 5.

Concentrations of sugar pool components: GFS and the remainder of sugars (non-GFS). Hashed areas represent the sucrose fraction of GFS. Asterisks indicate significant differences between initial and final concentrations after sprouts had died (Harvest 2) according to paired t-tests. The significant declines in GFS were driven by changes in sucrose concentration in both bark and xylem (P < 0.05). There were no changes in glucose or fructose concentration in either tissue (P > 0.44). Error bars = SE for GFS and non-GFS sugars, n = 5.

The final concentration of NSC in the roots after sprouts died was quite high, with 12.1% remaining in the bark and 2.7% in the xylem (Figure 4). The remaining NSCs were primarily sugars, with final starch concentrations averaging only 0.58% (range: 0–3.9%; median = 0.38%) and 0.61% (range: 0–2.8%; median = 0.28%) for bark and xylem, respectively (Figure 4). Final GFS concentrations of a subset of five roots averaged 1.8% and 1.3% dry weight for the bark and xylem, respectively; final sucrose concentrations averaged 0.9% in both tissues (Figure 5). Therefore, in the bark, the vast majority of the remaining sugar pool was composed of non-GFS sugars (85.5%).

Though final root NSC concentrations were highly variable when sprouts died, the cause of this variation was unclear (see Table S2 available as Supplementary Data at Tree Physiology Online). Final NSC concentrations in the bark or xylem were unrelated to relative sprout mass (see Figure S2 available as Supplementary Data at Tree Physiology Online), indicating that greater Sprout production did not lead to greater depletion of NSC pools. Final bark and xylem starch concentrations were negatively related to final WC (R² = 0.35 and 0.32, respectively; P < 0.05), but final sugar concentrations were not (P > 0.05). Final bark and xylem sugar concentrations increased significantly with root diameter, but the relationships were not strong (R² = 0.22 and 0.21, respectively; P < 0.05).

Additionally, there was no difference in final bark NSC concentrations between roots that were fully stained by tetrazolium (i.e., respiring/living tissues), partially stained or unstained (univariate analysis of variance: P = 0.77, N = 7, 4, 6). However, final xylem NSC concentration tended to increase from 2.25% to 3.4% with increasing degree of staining (i.e., increasing root viability; F = 3.24, P = 0.07). Starch concentration increased significantly with degree of staining from 0.15% to 0.43% to 1.41% in the xylem (ln-transformed, F = 7.30, P = 0.007) and from 0.26% to 0.45% to 1.37% in the bark (ln-transformed, F = 5.06, P = 0.022). Final root WC also differed with degree of staining; roots that were fully stained and living had a significantly lower WC than partially stained and unstained roots combined (linear contrast: F_1,14 = 5.25, P = 0.038).

Discussion

The majority of remobilized NSC was from the bark and from starch pools

We found that resprouting aspen roots remobilized four times more NSC mass from the inner bark tissues than from the xylem, demonstrating the critical role of bark tissue in carbohydrate storage. Recent studies have emphasized the multifunctional nature of bark tissue in woody plants (Paine et al. 2010, Rosell 2016), and though its potential role for carbohydrate storage and remobilization has been acknowledged, bark remains underrepresented in carbohydrate studies. However, previous studies have demonstrated that in, some species, bark can contain nearly as much NSC mass as the wood in mature trees (Zhang et al. 2014) and can contain even more NSC mass than wood in seedlings (Yang et al. 2016). And it appears that for aspen roots that typically resprout after disturbance (<4 cm diameter; Schier and Campbell 1978, DesRochers and Lieffers 2001), the living bark tissues provide the majority of remobilized C. Of course, bark NSC storage may not be so prominent in some species—or in larger diameter roots of aspen; however, this study demonstrates that the dominance of xylem storage cannot be taken for granted in mature trees.

Despite the fact that sugars made up the largest portion of the NSC pool in aspen roots, starch concentrations declined much more than sugar concentrations, indicating that starch was the main source of remobilized C during resprouting, as has been found in other species (e.g., Luostarinen and Kauppi 2005, Smith et al. 2018). The greater remobilization of starch explains why the initial starch measures were better predictors of total and relative sprout production than either sugar or total NSC measures. Ultimately, we do not know if remobilized starch was utilized for sprout growth directly. It is possible that the initial sugar pool directly fueled sucker growth and was partially replenished by the remobilization of starch. However, if some of the remobilized sugar must be replaced by the breakdown of starch, the size of the starch pool still determines the degree to which sugars can be remobilized for sprout growth and maintenance.

The lack of substantial remobilization from the sugar pool—especially in the living bark tissues—supports the idea that a relatively high sugar concentration may be necessary to maintain root and tissue functions and keep cells alive and that many sugars are not functioning as storage (Sala et al. 2012, Sevanto et al. 2014, Martínez-Vilalta 2016). This appears to be particularly true for non-GFS sugars, at least in aspen roots. When we broke the sugar pool down into GFS and non-GFS components, sucrose was the only portion to significantly decline. Therefore, while sucrose—most of which may be stored in vacuoles—was largely available for remobilization, other sugars perhaps should not be considered, or measured, as storage molecules. In fact, in the living bark, initial sugar concentrations were even ‘negatively’ correlated with relative sprout growth. This relationship likely arose because bark sugar concentration was negatively correlated with bark starch (R² = 0.26, P < 0.001), which strongly predicted sprouting. It is unclear why bark sugar and starch were negatively related; possibly, the negative correlation reflects tradeoffs in allocation in the bark between storage functions (e.g., starch synthesis) and other functions to which sugars are allocated. In any case, the lack of remobilization of the non-GFS sugar pool supports the suggestion by Martínez-Vilalta et al. (2016) that by including sugars in estimates of carbohydrate storage, we may be overestimating the amount of C available for remobilization. However, the substantial decline in sucrose content also warns against completely excluding all sugars from storage measurements. Furthermore, as this study was conducted on a single species, it is unclear whether these storage classifications will be consistent across all species or highly variable. Clearly more work is needed to determine which sugars can be remobilized and which cannot.

While the inclusion of the whole nonstructural sugar pool may overestimate the size of storage pools in aspen, this may be in part counterbalanced by the exclusion of other potential sources of remobilized C not in NSC form. In our experiment, after accounting for changes in NSC mass, we estimated that structural root mass also declined by 11%. While some of this mass likely represents the remobilization of other nutrients, like N in the form of amino acids or proteins (which contain C), it is possible that other sources of non-NSC C, which we did not measure, were remobilized. For example, under light deprivation, Scots pine seedlings switched their respiratory substrates away from carbohydrates to lipids and/or proteins (Fischer et al. 2015). In addition, some C from hemicellulose is remobilized seasonally in certain species (Schädel et al. 2009), and the extent to which these alternative storage compounds can be remobilized is not known (Fischer et al. 2015, Piper and Fajardo 2016).

Minimum NSC tissue concentrations indicate a lack of sugar depletion

To better recognize mortality from C starvation in the field, recent studies have used light deprivation in seedlings to attempt to identify minimum tissue NSC concentrations that would be indicative of starvation (Wiley et al. 2017, Weber et al. 2018). This study represents one of the first attempts to identify such threshold concentrations in older tissues of mature trees. Our results generally confirm observations made in starved aspen seedlings: due to a lack of sugar depletion, minimum NSC concentrations can be quite high and vary between tissues/organs (Wiley et al. 2017), with ~ 4× greater minimum concentrations in the bark than in the xylem. Wiley et al. (2017) found the minimum NSC concentration in roots were ~ 5–6% in seedlings; in roots ranging from 1–3.5 cm in diameter, we find a minimum concentration of 8.7 ± 0.3% (range: 6–11%; N = 25; weighted average of xylem and bark), suggesting that minimum concentrations can be, if anything, slightly higher in tissues of mature, large trees. That minimum NSC concentrations may increase with size is somewhat supported by the correlation between final bark NSC and root diameter. Contrastingly, Weber et al. (2018) report much lower NSC concentrations at starvation, with sugar concentrations depleted to below 1% in many tissues for four different species. The discrepancies between studies may simply be the result of huge variation in NSC measurements that can be found among labs (Quentin et al. 2015), or, in our case, it may relate to the different NSC quantification methods used (Landhäusser et al. 2018). The phenol–sulfuric acid assay used here measures a much wider range of sugars—including oligosaccharides—than the enzymatic method used by Weber et al. (2018), which only measures GFS. And in fact, our measurements of GFS sugars only yielded final concentrations of only 1–2%, which are more consistent with Weber et al. (2018).

In contrast to sugars, we found no evidence that mature tree tissues contain a substantial amount of sequestered starch that is unavailable for remobilization (Millard et al. 2007). By the time sprouts died, most root tissue contained less than 0.4% starch, confirming findings in seedlings that starvation completely depletes starch pools (Marshall and Waring 1985, Wiley et al. 2017, Weber et al. 2018). Our data therefore provide evidence that results from starvation experiments in seedlings—at least those regarding starch remobilization—can be extrapolated to mature tree tissues. Of course, depletion was not universal in our experiment, and 6 out of 25 roots maintained >1% starch in either the bark, xylem or both when suckers died. However, all six of these roots were still alive according to tetrazolium assay, and the significantly lower starch concentrations of dead versus live roots suggest that further depletion of the highest starch concentrations recorded would likely have occurred before those roots died.

Root NSC concentrations at sprout death were quite variable, and this variation needs to be understood if we want to use NSC concentrations as indicators of starvation. Some of this variation may relate to genetic differences, though as we collected roots from a somewhat limited area within a large aspen stand, the genetic diversity among roots was likely reduced. Importantly, we must ensure that such high NSC concentrations do not result from a restricted ability to utilize C (sink limitation) and that they ‘are’ therefore representative of concentrations under C-limiting conditions. In our experiment, roots were unlikely water limited, as all roots at the final harvest maintained a WC above 50% (gravimetric WC: 100–190%). We did find that final starch concentration was negatively associated with final root WC when sprouts died, which might suggest that some roots had died of desiccation before they could remobilize all starch, leading to higher starch levels. However, this was apparently not the case, as those roots that were still living (i.e., stained by tetrazolium) were the ones with significantly ‘lower’ WC and higher starch concentrations.

The high, final NSC concentrations (and the degree of variation) also do not appear to be the result of a C sink limitation to sprout growth. While higher initial xylem N concentrations were associated with lower minimum NSC (and sugar) concentrations—suggesting a potential sink limitation to NSC remobilization caused by N limitation—the lack of relationship between final NSC concentrations and relative sprout mass indicates that growing more did not lead to lower minimum NSC levels. Instead, the correlation between initial N concentration and final NSC may come about through their shared correlations with initial bark sugar concentration. Therefore, given that a large portion of sugars appear to be unavailable, roots with higher minimum bark sugar concentrations at death are those that also had initially higher sugar concentrations.

Storage pool mass not concentration is a better predictor of resprout potential

In this study, we found initial mass not concentration for both starch and N to be the best measure of storage relevant for predicting sprout growth of aspen roots. This finding supports the argument of Canham et al. (1999) that NSC mass will be a better determinant of regrowth for survival than NSC concentration, and it is consistent with results from seedlings showing initial NSC mass and not concentration determine leaf recovery from defoliation and survival under low light (Myers and Kitajima 2007). Simply, the greater the mass of reserves—irrespective of concentration—the greater is the mass of C available to build new tissues. The fact that initial NSC concentration can correlate with sprout mass (Schier and Zasada 1973, Landhäusser and Lieffers 1997) may not contradict this; if plant (or root system) size is relatively similar, differences in NSC concentration will lead to differences in NSC mass as well. Consequently, in our study, we found that starch concentration did correlate well with ‘relative’ sprout mass, which controlled for differences in root size. However, resprouting is not always correlated with NSC mass. Wachowski et al. (2014) found that starch concentration not mass predicted sprout growth of aspen root segments ranging from 1–4 cm diameter, similar to our range of 1–3.5 cm. It is possible that concentrations may become more predictive of sprout growth than mass at low concentrations, if, as it appears from our study, large portions of the NSC pool are not available for remobilization. In Wachowski et al. (2014), initial root NSC concentrations averaged between 10.7% and 15%, whereas ours averaged around 17.5%. It may only be the mass of NSC above a certain threshold of NSC concentration that determines the amount of sprout growth possible.

Positive correlations between relative and total sprout production and starch and N mass and concentrations suggest that resprouting was limited by both the amount of starch and N storage. That both were limiting is supported by the fact that the combination of starch and N concentration was more strongly related to relative sprout mass than either variable separately. However, it remains an unanswered question whether stored resource availability generally limits resprouting potential in woody plants (Clarke et al. 2013), as initial starch or NSC concentrations often do not correlate with subsequent sprout growth (Sparks and Oechel 1993, Cruz et al. 2003). In part, this may relate to the fact that, for resprouting in the field, at some point, current photosynthates replace storage as the main source of C, and this may override any initial effect of storage. This can happen very quickly in the case of budflush or resprouting (Landhäusser and Lieffers 2002, Landhäusser 2011). The cases where NSC concentrations do not correlate with subsequent sprout biomass may in part reflect this difference from resprouting in darkness, particularly on longer time scales. However, as we show, it was the mass of NSC, specifically starch, not the concentration that determined the total amount of regrowth; therefore, a lack of correlation between initial NSC concentration and resprouting might not be evidence that storage is not limiting. The NSC mass (pool size) must also be considered when assessing C limitation; however, this is much less common and more difficult to obtain, particularly in larger plants like trees. Further, even when NSC mass is considered, if the entire NSC pool cannot be remobilized, the relationship between the remobilizable fraction of NSC and resprouting potential may be obscured. A similar argument could be made for examining N limitation in resprouting.

Conclusion

We found that root sprout growth of aspen in complete darkness was determined by the root’s initial mass of starch and N, while the initial concentration of these resources was either unrelated or only weakly related to growth. Therefore, when assessing whether storage limits survival or recovery potential in any species, it may be necessary to consider not just the concentration of storage compounds but their mass. During resprouting in aspen roots, the main source of remobilized C was starch not sugar, with four times more mass coming from the living root bark rather than the xylem. After all sprouts had died, root starch pools in both bark and xylem were largely depleted, while sugar pools were not. The large remaining pool of sugars, mostly non-GFS, appears to be unavailable for remobilization, suggesting that this sugar fraction should not be considered a part of the storage pool in aspen roots. Whether all these patterns hold true for other woody species remains unclear. However, our results highlight that in order to accurately quantify C storage—both at the individual tree level and at the stand level—and its influence on tree or stand function, we need a better understanding of which sugars and non-NSC compounds are functioning as storage molecules and which are unavailable for remobilization, and how this may vary among species.

Acknowledgments

We are grateful to Pak Chow, Fran Leishman, Ashley Hart, Morgane Merlin and Kyle Le for sample processing and assistance during the experiment. We are also very grateful for the helpful comments from two anonymous reviewers. This work was funded by the Natural Sciences and Engineering Research Council of Canada.

Conflict of interest

None declared.

Author notes

Present address: Department of Biology, University of Central Arkansas, AR, USA