Introduction

Trembling aspen (Populus tremuloides Michx.) has been studied extensively as it is a widespread deciduous tree species that is ecologically and economically important throughout North America (Perala 1990). Seasonal C dynamics of mature aspen stands have been reported earlier (Landhäusser and Lieffers 2003) and we continued to record the C dynamics of these same stands for 8 years, including during events of defoliation and recovery. In this article we present data on the root and branch carbohydrates of large mature trees during and after periods of C limitation as a result of insect defoliation.

There is growing interest in the causes and the underlying mechanisms of the decline of large mature trees in response to abiotic stresses such as drought and elevated temperatures and biotic stresses such as diseases or defoliation (Frey et al. 2004; McDowell et al. 2008; Adams et al. 2009; Carnicer et al. 2010). Much of the literature indicates that decline is either directly or indirectly driven by moisture limitations inducing embolism of xylem in roots and shoots, where embolism cuts off the water supply to the foliage, leading to the desiccation of tissues and ultimately to death (Tyree and Sperry 1988, 1989; Sperry et al. 2002; Bréda et al. 2006). However, an alternative hypothesis has recently been developed that suggests that carbon limitations and/or starvation as a result of C stress could also be a mechanism for tree decline (Bréda et al. 2006; McDowell et al. 2008). In particular, McDowell and coauthors (2008) argue that this mechanism could play a prominent role in isohydric species, where the xylem is protected from cavitation, because stomata close before pressure potentials reach a critical threshold for xylem embolism. In these species the C-deficit from a prolonged drought could leave the trees with insufficient fixed C to maintain respiration needs (Hogg 1999; Martinez-Vilalta et al. 2002; Frey et al. 2004). These two opposing, but non-exclusive, hypotheses have been vigorously discussed (Leuzinger et al. 2009; Sala et al. 2010; McDowell and Sevanto 2010), but there is no firm evidence for support (Adams et al. 2009; Leuzinger et al. 2009; McDowell and Sevanto 2010; Sala et al. 2010; McDowell 2011). Although there is a large body of literature on carbon dynamics in trees, there is a profound lack of information on the C dynamics in varying tissues of mature trees, particularly in response to stress that leads to reduced carbon fixation.

Aspen is considered an isohydric species (Tardieu 1993; Tardieu and Simonneau 1998; Galvez et al. 2011), showing reduced stomatal conductance with increasing atmospheric drought conditions (Dang et al. 1997; Hogg et al. 2000). Over the last decades aspen stands have declined and died in many areas of its range (Bartos and Campbell 1998; Di Orio et al. 2005; Worrall et al. 2008; Strand et al. 2009; Michaelia et al. 2010); drought and defoliation have been implicated (Hogg et al. 2008; Man et al. 2008; Worrall et al. 2010) but the full reasons for decline have not be determined (Allen et al. 2010). We believe that defoliation is a good general analog of C stress in plants as it incites a severe short-term C stress that is discrete and local; it allows for better quality-controls, than for C stress resulting from large regional and longer periods of drought. In addition, drought effects in trees are difficult to document, unless soil and plant water stress are monitored over the same time. Because of the leafless period, deciduous species are likely to have wider swings in C reserves over the season than evergreen species, which makes them very suitable for the study of C dynamics.

In this article we report the dynamics of non-structural carbohydrate reserves in the shoots and roots of mature trembling aspen trees during and after defoliation events. These results will be discussed and used to introduce a new hypothesis which proposes that carbon limitation in mature trees will affect the root system most and could be an underlying principle in the decline of trees.

Materials and methods

Site selection and sample collection

Ten mature aspen-dominated stands (>90% aspen), near Alder Flats, Alberta (52°58′N, 114°59′W) were initially selected in the spring of 2000 for a study investigating the seasonal carbohydrate reserve dynamics in mature aspen clones (Landhäusser and Lieffers 2003). Within a stand, one individual clone of aspen was roughly delineated from neighboring clones by the time of bud-break and bark color (Peterson and Peterson 1992). All selected clones were healthy with no signs of decline and crown mortality. The clones were 60 years old, between 0.5 and 4 km apart from each other, between 0.1 and 0.6 ha in size, and 18–20 m tall. In the late summer of 2001 one clone was lost due to the construction of a pipeline. In 2002, four clones of the nine were defoliated by large aspen tortrix (LAT) (Choristoneura conflictana (Walker)) intermixed with forest tent caterpillar (FTC) (Malacosoma disstria (Hübner)). The percentage defoliation was visually assessed. In this first defoliation, defoliated clones had more than 70% of their leaf area removed, while the un-defoliated clones retained more than 85% of their leaf area. Between 2003 and 2006, there were no significant defoliations in any of the clones. In 2007, another significant defoliation event by FTC (>90% defoliation) occurred in four aspen clones, while five clones remained un-defoliated. Only one of those defoliated clones was defoliated in both 2002 and 2007, therefore over the two defoliation events seven of the nine clones got defoliated at one time or the other. There was no significant defoliation activity in 2008 and there was some mortality of suppressed trees, but there was no mortality of dominant or co-dominant trees in the collection core area over the eight year period.

At each collection time, we took samples from three randomly chosen trees in the central portion of each clone; within a clone, samples were combined into one sample prior to analysis. From each tree, we collected whole samples from the crown (current shoots (one-year-old) and branches (between 10 and 20 mm in diameter)) and from the root system (fine roots (<1 mm diameter) and coarse roots (between 10 and 20 mm diameter). Roots were collected within 1 m of a bole at about 10 cm depth at the organic-mineral soil horizon interface. Current shoot and branch samples from the upper part of the crown were collected by shooting off branches with a shotgun. Care was taken to minimize the variation in diameter of collected branch and root samples in each collection period to remove potential dilution effects due to different amounts of tissues (bark, xylem, and phloem).

Between 2000 and 2008 we collected the non-structural carbohydrate reserves in root and shoot tissues at three important phenological stages, determined from the earlier study (Landhäusser and Lieffers 2003). We began our study with collections in 2000 and 2002 at 3 different stages: (1) late growing season—maximum leaf area (in 2000–2002, 2007), (2) dormant—root growth ceased, when soil temperature close to 0°C (based on Landhäusser and Lieffers 1998) (in 2000–2008), (3) early spring—trees still dormant but pre-flush, root carbohydrates still low. At the end of 2002 we cut back our sampling to only monitor root tissues during the dormant season (soils ~0°C)—a time when root growth stops in P. tremuloides (Landhäusser and Lieffers 1998; Wan et al. 2001). Reserves were also the most stable during the dormant season (Landhäusser and Lieffers 2003) and thus we used the dormant season collection as the most appropriate time for an assessment of the carbohydrate reserve status for the next growing season. During years with no defoliation, shoot samples were not collected as sampling was costly and took a heavy toll on human shoulder tissues, but more importantly branch carbohydrate levels were not as responsive to yearly variations in our sampling scheme. Bole tissue collections were discontinued after the first 2 years as continued sampling could have affected the health of the trees. Further, significant seasonal fluctuation would have been difficult to detect since overall seasonal fluctuations in the phloem and xylem were very minor and starch concentrations were generally very low (0.1–0.01%) compared to the other tissues collected (Landhäusser and Lieffers 2003). During and after the defoliation event in 2007, we stepped up frequency of sampling and included branch tissues again (similar to the 2000–2002 period).

To assess climatic conditions during the sampling period, precipitation and average monthly temperature data for the eight sampling years were obtained from a climate station (Breton, AB 53°10.200′N, 114°28.800′W) approximately 35 km northeast of the aspen sites (Environment Canada 2010).

Carbohydrate analysis

All collected tissue samples were kept on ice in the field (<4 h), were quickly cleaned, and immediately oven dried at 68°C. Samples were ground through 40-mesh screen, using a Wiley mill. After grinding equal amounts of the fine and coarse root tissues and the current shoot and branch tissues were combined and homogenized and then analyzed for shoot and root carbohydrate reserves. For all samples the soluble sugar and starch concentrations were determined following Chow and Landhäusser (2004). Water soluble sugars were extracted with hot ethanol (85%) and sugar concentrations were determined colorimetrically using a phenol-sulfuric acid assay. Remaining starch was solubilised by sodium hydroxide from the tissue residue and hydrolysed to glucose by an enzyme mixture of α-amylase (ICN 190151, from Bacillus licheniformis) and amyloglucosidase (Sigma A3514, from Aspergillus niger). Glucose was measured colorimetrically using glucose oxidase/peroxidase-o-dianisidine solution (Sigma Glucose Diagnostic Kit 510A) and the starch equivalent determined (Chow and Landhäusser 2004). Sugar alcohol concentrations in aspen tissues were generally constant and were less than 0.5% of tissue dry mass over the sampling periods (Landhäusser and Lieffers 2003). In poplar (Sauter and VanCleve 1991, 1994) and other diffuse-porous species such as birch (Piispanen and Saranpää 2004) storage lipids appear to play a minor role in the reserve allocation and dynamics.

Data analysis

The different clones were the experimental units and the non-structural carbohydrate reserve concentration (total sugars, total starch, and their sum (total non-structural carbohydrate (TNC)) of roots and shoots was averaged for each clone prior to analysis. The clones were categorized as defoliated or undefoliated (see Fig. 1). Clones defoliated in 2002 were identified and their carbohydrate reserves were compared to the undefoliated clones for the period starting from 2000 until the dormant period of 2006. The trees defoliated in 2007 became the new class of defoliated trees and their carbohydrate status was compared to undefoliated trees from the dormant period in 2006 until the summer of 2008. The carbohydrate data were analyzed using analysis of variance procedures and planned comparisons were performed with general linear models available in SAS 6.11 (SAS Institute Inc., Cary, NC). The significance level was set at α = 0.05.

Fig. 1
figure 1

Total non-structural carbohydrate and starch concentrations (TNC) within (a) current shoots and branches combined; and within (b) fine and coarse roots combined, from collections over eight growing seasons. Time of defoliation is indicated by arrows. Bars indicate one standard error of the mean and an asterisk denotes a significant difference between defoliated (n = 4) and non-defoliated trees (n = 5). Monthly precipitation and average air temperature (c) are depicted over the same time period

Full size image

Results

The 2002 defoliation had no impact on the TNC reserves in the twigs and branches of the aspen crowns while in 2007, defoliated trees had lower TNC reserves at the first collection after the tent caterpillar defoliation (Fig. 1a). In both of defoliation events, however, the TNC reserves in twigs and branches had fully recovered by the end of the first growing season with defoliated and undefoliated trees having the same reserve concentrations in their branch tissues during the dormant season collection (Fig. 1a).

Root TNC reserves of defoliated trees were significantly lower during and after the defoliation years of 2002 and 2007 and it took up to two growing seasons for root reserves of defoliated trees to recover to levels of undefoliated trees (Fig. 1b). Root TNC reserves in the first dormant season after the defoliation of 2002 were about 25% lower than in the undefoliated clones (P = 0.012). Only by the following dormant season collection in 2003/04, the root TNC reserves were similar to levels found in the un-defoliated clones. Most of the differences in root reserves during the defoliation manifested itself in the starch concentrations as soluble sugar concentrations dropped only 15% while starch reserve concentrations were reduced by 61%. As a result after the first growing season root starch reserves in the defoliated trees were close to 0% (0.29) compared to 1.5% in undefoliated (Fig. 1b). Starch reserves in the defoliated trees, had still not recovered by the end of the following growing season. The trees defoliated in 2007 also showed the same significant decline in root TNC reserves driven by root starch reserves which were reduced by 53%, while sugar concentration remained the same (Fig. 1b). During the drier period (2000–2003 with an average of 454 mm/yr precipitation), root TNC reserves in undefoliated aspen during the dormant seasons were approximately 20% lower than during the following wetter years (2004–2007 with an average of 587 mm/yr); these differences in TNC reserves were largely driven by the starch reserves which were about 50% lower during the dry years than during the wet years compared to the soluble sugar reserves, which dropped by only 15% (Fig. 1c).

Discussion

Our results indicate that in defoliated trees root carbohydrate reserves remained lower for much longer after a defoliation event (up to two growing seasons) than the reserves in the crown tissue (less than 1 growing season). The reduction of root C reserves was especially evident in the starch concentrations measured during the dormant season, indicating a reduction in reserves stored for future use. To our knowledge this is the first time that carbohydrate status of different organs were assessed on large trees, over more than one defoliation event. Our findings are largely in contrast to work done on seedlings and small trees where defoliation produced only transient effects on root C reserves that were not considered detrimental to root production in trees growing in temperate climates (Dickmann et al. 1996; Kosola et al. 2001, 2002) or at the treeline (Li et al. 2002). We believe that any long-term suppression of C fixation could lead to a significant reduction in root C reserves leading to tree decline and mortality, particularly for large trees. Since this effect can be subtle (Bréda et al. 2006) it will likely only be detected, if root reserves are monitored over longer periods and during periods of carbon stress.

This study leads us to hypothesize that in tall trees C deficiency during times of stress should be first noticeable in their root systems and we argue that during times of C stress (defoliation and/or drought) roots are likely the first tissue to be limited in carbon (Fig. 2). Our hypothesis is an adaptation of a metaphor coined by Donald Fisher (cited in Thompson 2006), that the various C sinks along the phloem stream of a tree cause a local reduction in C supply analogous ‘to pigs feeding at a trough’. We further adapt this metaphor indicating that if the food is only added on one end of the trough (from the tree crown) the sinks near the crown (twigs, branches and upper bole) will have first access to the carbohydrates from the foliage (Fig. 2a), while roots are at the end of the line and will only receive adequate supplies of C during times when the other sinks cannot feed fast enough or are sated (Fig. 2b). However, despite much uncertainty in our understanding of the physiology of phloem transport in tall trees (van Bel 2003; Hölttä et al. 2006), newer refinements and models (Thompson 2006; Minchin and Lacointe 2005; Hölttä et al. 2006; De Schepper and Steppe 2010) indicate that phloem transport is mainly driven by a sink hierarchy (Wright 1989; Wardlaw 1990) rather than by a pressure-driven pipe model only. As a result the flow of carbon into a specific sink is determined by the interaction between source, alternative sinks, and the pathway to the sink itself (Minchin and Thorpe 1993).

Fig. 2
figure 2

Conceptual model of C supply to various tissues in a tall tree during (a) spring and early summer and (b) late summer when shoot growth and radial growth have largely been terminated. Boxes and ovals represent the C usage for maintenance and growth respiration and storage in the different portions of a tree. The large vertical arrow depicts the phloem stream in the stem, while the smaller horizontal arrows describe the C flow toward active sinks. The shading in the vertical bar indicates C concentration along the stem while in the horizontal arrows it indicates the flow of C toward the sinks. Lighter shading indicates lower values for both. In extremely poor years the late summer stage of C flow (b) may not ever happen

Full size image

In large trees the cambial sinks along the large branches in the crown and particularly in the long bole should remove significant amounts of C from the phloem stream, especially during growth periods when leaf area is developing and cambial activity is high. The sink hierarchy supports the idea that the root system, although considered a large sink, is at the end of the C transport chain and should have a lower priority for carbohydrates as long as there are active sinks more proximate to the source (Wardlaw 1990; Minchin 2007). This effect should be most noticeable during the first part of the growing season, even in years with good growing conditions (Fig. 2a). Later in the growing season when shoot growth is terminated and diameter growth slows, most of the phloem stream bypasses these proximate sinks and is available to replenish the more distal sinks such as the root system (Fig. 2b).

We were able to detect this priority for proximal sinks by examining the seasonal carbohydrate status of branches and roots in trees under normal growing conditions. Sinks near the foliage fill earlier in the season and roots are last to be filled; in aspen, roots only gain significant reserves after shoot growth had stopped in late summer (Landhäusser and Lieffers 2003). The fact that following defoliation, the roots systems took much longer to recover their reserves than the branches in the crown (Fig. 1a, b) is consistent with the hierarchy that proximate sinks near the source have higher priority for C than roots.

Another argument supporting our hypothesis is that aspen root tissues appear to have a larger capacity to store C reserves than branches during the dormant season (Martens et al. 2007; Landhäusser and Lieffers 2002). Roots therefore, appear to be adapted to store large quantities of reserves during good growing years in order to sustain C demands during poor years. Given that roots are at the end of the phloem stream, in some years roots may not have an opportunity to completely refill their C reserves and are therefore more in need of C storage to meet their needs. Over the 8 years of study, we did see much wider seasonal and inter annual fluctuations in dormant root tissues than in tissues of the crown (Fig. 1); however, we did not always measure the branch TNC.

There is some indirect evidence from the literature that a decline of C reserves from the crown to the roots is a reasonable conclusion for tall trees. Our current research indicates that there is a decline in root C reserves with an increase in sapwood area below the crown (Goodsman et al. 2010). Similarly, tall Douglas-fir trees that had a live crown ratio of 0.39, the C concentrations in the stem (xylem and phloem combined) decreased by 33% from the base of the crown down to the base of branchless bole over a distance of 27 m (Pruyn et al. 2005). Galiano et al. (2011) found a marked decline in stem reserves of Pinus sylvestris growing in dense mature stands; however, this decline was not detectable in open-grown Ponderosa pine stems where crowns ratios were much higher (Pruyn et al. 2005; Sala and Hoch 2009).

We believe that the scarcity of studies examining carbon reserve dynamics in large trees at different phenological stages has clouded our understanding of the mechanisms of C allocation to the different tissues in trees during times of stress. Results based on seedlings, small trees, or even large trees with long live crowns will yield different understanding of C dynamics than results from trees with long branchless boles, topped by small crowns. A tree with a low live crown ratio (crown length/total height) is the most common form of trees found in dense and mature forests. The transferability of results from research on seedlings has also been questioned for studies investigating the impact of increased CO2 concentrations on large trees and continuous forests (Körner et al. 2005).

It has been argued that branches in tree crowns are carbon autonomous (Sprugel et al. 1991), especially in mature trees, where the flushing of the current buds is more or less independent of reserves from sources further away (Newell et al. 2002; Landhäusser and Lieffers 2003; Hoch 2005; Schädel et al. 2009). As a result, the reflushing of leaves after a defoliation event might not be as costly as one would expect. In abscised aspen shoots, stored reserves were lowered only by 14% during budflush and with shoot reserves being rebuilt after only 7 days after the flush of new leaves (Landhäusser 2011). It appears that in seedlings and small trees, the bole is small and will behave more like a branch on a mature tree where sinks are smaller and C reserves can accumulate even during periods of stress (Ludovici et al. 2002; Sanz-Pérez et al. 2009). As a result, smaller woody plants under drought stress and near zero assimilation actually accumulated reserves, likely due to close sinks and the down regulation of growth and respiration; therefore these plants are more likely to die of desiccation through hydraulic failure than to die of carbon starvation (Galvez et al. 2011). Results from evergreen species with their more stable C economy likely also are very different from those found in deciduous species, as these trees can photosynthesise over a longer period of the year than deciduous trees. The timing of sampling different tissues for C reserves is also a crucial element in detecting C stress in deciduous trees and plays a significant role in the interpretation of results. Synchronizing sampling with defined phenological stages of trees will likely provide more conclusive evidence for assessing C dynamics.

Broad-scale mortality of trees and forests has been observed worldwide and is thought to be driven by drought stress, exacerbated by insect attacks and warming climate (Allen et al. 2010). Although this study was not intended to test hypotheses related to the mortality of trees (we did not observe any die-back of dominant or co-dominant trees during the study period) and provide hard evidence for the discussion around carbon starvation and hydraulic failure (see Sala et al. 2010; McDowell and Sevanto 2010; McDowell 2011), it addresses issues raised by the above authors on how carbohydrate reserves could vary and are mobilized at the whole plant level particularly for mature trees. The results from this current study suggest that hydraulic failure and carbon starvation are likely interrelated, making it difficult to separate both mechanisms from each other. Our hypothesis provides a simple but versatile explanation for many of the symptoms observed and described for declining and/or dying mature trees which is preceded by root reserve reduction and driven by a loss of roots once root reserves fall below a critical threshold. In fact the defoliation event in 2002 coincided with a drier period (2000–2002) and it appears that the drier conditions suppressed root carbohydrate reserves in both the undefoliated and defoliated aspen (Fig. 1) as the dormant season root reserves during the wetter period were about 20% higher compared to the drier period earlier. However, regardless of the potential impact of the drier period on overall aspen reserves, it is clear from our data that the defoliation events played a much larger role in C dynamics during this period than the climate, as in both instances defoliation reduced root starch reserves to near zero. Nonetheless, this also suggests that long distance transport of reserves was not impaired during drought (Sala et al. 2010).

Although mature trees have the ability to store large quantities of C reserves that can be mobilized (Chapin et al. 1990), it appears that under carbon limitation, root reserves in mature aspen are not necessarily mobilized to tissues that are low in storage reserves. We therefore believe that under conditions of low C assimilation in mature aspen trees (caused by drought and/or defoliation) root reserves are the tissues affected most and are much slower to recover. If a limitation in assimilation persists, the fine roots of these root systems are likely not renewed (Meier and Leuschner 2008) and/or entire portions of the root system are abscised (Landhäusser and Lieffers 2002). Soluble sugar concentrations in cells need to be maintained at a minimum level to maintain basic cell functions and it is likely that below a critical threshold, sugars are no longer available for growth or respiration (citations in McDowell and Sevanto 2010; Sala et al. 2010). In our study root soluble sugars never dropped below about 8% (Fig. 1) which coincides with sugar levels we found in short root sections containing living tissues along aspen roots that were dead on either end (unpublished data from Landhäusser and Lieffers (2002) and Wachowski, Landhäusser and Lieffers). In these same root tissues, starch concentrations were below the detection limit.

A decline in the root system size will result in a cascade of negative effects, limiting water and nutrient uptake, which in turn will limit C assimilation and exacerbate the risk of mortality through xylem cavitation, continuing C-starvation, and/or reduced defense. Repeated defoliation of mature aspen over several years has been shown to reduce growth and to increase their susceptibility to disease and mortality (Batzer et al. 1954; Hildahl and Reeks 1960; Hogg et al. 2002; Man et al. 2008). This could explain why trees are slow to recover from defoliation or drought and why tree decline and death sometimes become noticeable 3–5 years after an inciting event (Battaglia et al. 1998; Le Dantec et al. 2000; Bigler et al. 2007; Allen et al. 2010), including in aspen (Worrall et al. 2010; Michaelia et al. 2010). We hope that this work will stimulate further research in C dynamics in large trees, the long distance transport mechanisms of C, the distribution of C to sinks under different conditions, and the storage pools for non-structural C reserves in crowns, boles and roots of trees.