Does carbon storage limit tree growth?



Storage is a fundamental process that allows organisms to meet variable demand for resources during their development and buffer environmental fluctuations in resource supply. Storage of nonstructural carbohydrates (NSC) such as starch and sugars or other carbon (C) compounds that can be mobilized (sugar alcohols, lipids, proteins) are thought to be critical for survival under stress and disturbance, particularly for long-lived trees. Our view of tree physiology often assumes that, as the availability of CO2 limits photosynthesis, growth is in turn C limited (implying a C source limitation). However, for trees growing under environmental constraints such as nutrient limitation, cold or drought, direct restrictions on tissue formation can occur before a C shortage comes into play (C sink limitation, Körner, 2003). Recent attempts at explaining C limitation under stress suggest that priority allocation to storage could compete with growth and, thus, make assimilated C a limiting resource (e.g. McDowell, 2011; Sala et al., 2012; Wiley & Helliker, 2012). These considerations imply that under limiting availability of assimilates: (1) C storage is given priority over growth, because ultimately survival depends more on C demands for metabolism than for growth (Sala et al., 2012); (2) such prioritization is a conservative strategy that occurs frequently regardless of environmental stress (Wiley & Helliker, 2012). The latter implies that most storage by trees occurs at the expense of growth and the frequently observed build-up of NSC in trees under cold and dry conditions reflects a precautionary measure by plants, compatible with (or even indicative of) C limitation for growth (Wiley & Helliker, 2012). These concepts, particularly (2), build on theories that are, so far, not supported empirically, in contrast to the known empirical evidence of greater stress sensitivity of growth over photosynthesis.

The questions remain whether C storage by trees is under most circumstances a priority over growth, and whether growth restriction under stress is the result of a direct impact on meristems or plant-internal tradeoffs of C utilization. Here we argue that there is substantial experimental and observational evidence that growth limitation of plants under environmental stress like critically low temperature or drought is not driven by a limited supply of photo-assimilates, except perhaps for extreme situations when severe constraints on C assimilation prevail over several months. We also review the literature dealing with a C-based growth-storage tradeoff in woody plants and conclude that current evidence in support of such a tradeoff is equivocal. Consequently, although a storage-growth tradeoff for C could occur under certain circumstances when C supply becomes indeed limiting (such as after severe defoliation), we conclude that caution should be taken when generalizing such results.

Does C availability ever limit tree growth?

One way to test whether C limitation plays a role in tree growth is to expose trees to elevated CO2. After a decade of CO2 fertilization experiments, growth stimulation has only been documented under high nutrition (e.g. fertilized sites or on former agricultural land) or over short periods of time (i.e. transitorily; see reviews by Körner, 2006; Millard et al., 2007; Norby & Zak, 2011; and new data by Bader et al., 2013; Sigurdsson et al., 2013). For example, the continued response of trees at the Duke FACE (free-air CO2 enrichment) site has been explained by a combination of high nutrient availability (at least in one of the three FACE rings) and priming effects through which nutrients were released (Phillips et al., 2011). Consequently, so far, there is no empirical evidence for a long-term, sustained stimulation of tree growth by elevated CO2 in natural undisturbed settings with a natural steady state nutrient cycle.

Cell division and expansion are more sensitive to drought and cold stress than photosynthesis (Boyer, 1970; Hsiao & Acevedo, 1974; Körner, 2003; Muller et al., 2011; Dosio et al., 2011; Galvez et al., 2013). Similarly, the growth reduction found in trees when they reach their maximum height is related to hydraulic constraints directly affecting meristematic growth, rather than C availability (Sala et al., 2010; Woodruff & Meinzer, 2011). The capacity for tissue formation in cold-adapted plants becomes marginal below 5°C, as shown for xylogenesis (Rossi et al., 2008) and for root growth (Alvarez-Uria & Körner, 2007). Leaf photosynthesis, by contrast, reaches 30–50% of maximum rates under light saturation at 0°C and 50–70% at 5°C in cold adapted plants (Körner, 2012). Hence, under low temperature or drought, growth ceases long before C-gain does.

Situations leading to long-lasting negative C balance could, however, lead to C limitation (Quirk et al., 2013). For instance, trees fully and repeatedly defoliated over 11 yr, and left to recover for 6 yr, had lower growth, but similar NSC concentrations and higher starch concentrations in their stems relative to undefoliated trees (Palacio et al., 2012). However, while these results are consistent with a C limitation of growth induced by preferential allocation to storage, the conditions in this study were quite extreme. Further, they could also be explained by a sink limitation to growth due to bud (Palacio et al., 2008) or nutrient limitation (Millard et al., 2001) after sustained, severe defoliation, or by the lack of a demand-signal from the canopy for cambial activity. Similar to root growth, the cambium may not produce xylem if there is no demand, that is, no terminal branch growth and transpiring foliage (Zimmermann & Brown, 1974). Detailed investigations on the growth and C allocation processes of long-term defoliated trees and their response to increased C-availability are needed to ascertain which explanation is more plausible. If defoliation systematically leads to C-limitation of growth, then elevated CO2 should ameliorate growth of defoliated trees. However, results from previous studies are inconsistent with this prediction (Lovelock et al., 1999; Handa et al., 2005; Huttunen et al., 2007) indicating that growth in defoliated trees is not necessarily C-limited.

A closer look at the literature: does increased storage ever reduce tree growth?

Does allocation of C to storage systematically incur a cost in terms of tree growth in the long term, as argued by Wiley & Helliker (2012)? A closer look at the studies cited in support for a competition between C allocation to storage and growth in trees (Chapin et al., 1990; Kobe, 1997; Canham et al., 1999; Lacointe et al., 2004; Myers & Kitajima, 2007; Poorter & Kitajima, 2007; Silpi et al., 2007; Genet et al., 2010) reveals inconclusive evidence. For example, Chapin et al. (1990) cited studies on wild herbs and sugar beet but none on trees and concluded that C allocation to storage over growth ‘has been critically demonstrated in only a few studies, making it difficult to detect any broad-scale ecological pattern’. Lacointe et al. (2004) compared growth and carbohydrate allocation in the branches of young walnut trees in shade or full sunlight. Shaded branches grew less but had similar NSC concentrations in sun or shade. They concluded that reserve levels were maintained in shaded branches ‘possibly at the expense of growth’. However, this assumes that maximizing growth in the shade is an efficient alternative, which is not necessarily the case. Even so, their results also showed that shaded branches allocated slightly more labeled C to growth than to storage, thus casting some doubt on their assertion that allocation to storage occurred at the expense of growth. Further, growth in the shade could be regulated to prevent excessive self-shading without invoking competition with storage.

Adaptation to disturbance such as fire can lead to differences in C allocation and growth. Resprouter species or ecotypes generally show greater allocation to NSC (i.e. larger NSC concentrations and pools) than seeders (Bell, 2001; Verdaguer & Ojeda, 2002), and generally exhibit overall slower growth (Bell, 2001). However, results are not consistent across families and species (Knox & Clarke, 2005; Clarke & Knox, 2009). For instance, Chew & Bonser (2009) analyzed differences in growth rate and allocation to NSC of seedlings of eight pairs of species, each including a resprouter and a seeder. They found no differences in growth rate or starch content between regeneration strategies of comparable lifespans, and concluded that differences in life history, rather than in NSC allocation, were controlling their respective growth rates.

Finally, the majority of studies that evaluated the existence of a tradeoff between C allocation to NSC (measured as change in NSC pools or concentrations) and growth in trees used correlation analyses. We note however, that while some studies document a negative correlation between growth and NSC allocation (Myers & Kitajima, 2007; Silpi et al., 2007; Chantuma et al., 2009; Genet et al., 2010), others do not (Canham et al., 1999; Poorter & Kitajima, 2007; Imaji & Seiwa, 2010; Piper, 2011; Palacio et al., 2012). Overall, results are species-specific (Genet et al., 2010) or vary through ontogeny (Myers & Kitajima, 2007). Furthermore, correlation does not imply causation, and negative correlations between NSC concentrations or pools and growth do not provide unequivocal evidence of a tradeoff between the two. Alternatively, direct constraints on growth (sink limitation) could lead to increased NSC availability for storage (Genet et al., 2010; Hoch & Körner, 2012; Palacio et al., 2012). For example Silpi et al. (2007) and Chantuma et al. (2009) showed that rubber gum trees tapped or treated to increase latex production had higher NSC concentrations, but grew less than control trees. Because latex is rich in C, this was interpreted as evidence of active C allocation to storage in response to tapping, thereby limiting growth. However, other factors, such as turgor maintenance or nutrient supply, could limit growth and enhance C availability for storage and latex production (Junjittakarn et al., 2012). One would expect that elevated CO2 would enhance latex production, but when tested, the opposite was found, namely a reduction in latex production while NSC concentrations rose in response to elevated CO2 (Häring & Körner, 2004). To evaluate the potential consequences of increased C allocation to storage on tree growth we need to move beyond correlative studies and experimentally determine if tree growth is limited by C-availability and the mechanisms leading to the observed increases in C-storage pools.

Why results from Arabidopsis cannot be directly extrapolated to trees

To date, the only evidence of plant growth being limited by preferential C allocation to NSC storage comes from light-limited, C-starved Arabidopsis plants (Smith & Stitt, 2007; Gibon et al., 2009). However, up to which point can results from Arabidopsis be extrapolated to trees? Trees are long-lived plants with a complex structure, composed of multiple, semi-autonomous modules (Sprugel et al., 1991). Carbon and nutrients are allocated to storage in specialized organs and tissues over many years. By contrast, the monocarpic senescence exhibited by Arabidopsis plants relies on NSC stored largely in foliage (assimilatory starch in chloroplasts) and short-lived axial tissue, for seed production. In essence, most of the C-storing tissue of Arabidopsis photosynthesizes, while the majority of C-storage organs in woody species do not. When C becomes limiting in herbaceous plants it is hence possible that all stores within the plant are mobilized and consumed (Smith & Stitt, 2007). Also, because of the monocarpic senescence, C invested in starch in leaves might contribute directly to reproductive fitness. In trees, C-starvation may be found at the leaf or even the branch level without depletion of NSC pools at the whole tree level (Würth et al., 2005). In contrast to Arabidopsis, it is probable that a large proportion of the C found in NSC pools in large trees is actually sequestered (and hence not functional), rather than stored (Millard & Grelet, 2010). Also, if starch serves as a source of compatible solutes for hydraulic functions (Sala et al., 2012), plants may prevent falling below certain minimum levels of stored NSC. If so, not all measured NSC may be available for growth and respiration and starch pools do not need to be completely exhausted before trees have no NSC storage left available for growth and respiration. Consequently, measuring NSC pools may inform about the overall C-supply status of trees (particularly when used in comparative terms, e.g. Hoch & Körner, 2005), but cannot estimate the size of the readily available C stock. The different dynamism of NSC pools in trees as compared to annual plants leaves some leeway for the occurrence of a tradeoff between C allocation to storage vs growth in trees, even when starch pools are present. It also illustrates the difficulties in translating results from Arabidopsis to trees.

Conclusions and open questions for future research

There is experimental evidence showing that: (1) closed tree stands on undisturbed forest soil do not show a sustained stimulation of growth by CO2 fertilization; and (2) cell growth processes are commonly more sensitive to stress than processes related to C gain. This suggests that growth of mature trees is seldom limited by C availability. By contrast, evidence in support for a tradeoff between C storage and growth in trees is, to date, inconclusive. Under cold or drought stress, there are physiological limitations to tissue formation, such as a critical cell turgor, that prevent structural growth irrespective of C supply. However, during certain developmental or phenological stages and under environmental conditions leading to true C limitation (e.g. sustained severe defoliation, deep shade) a tradeoff between storage and growth could be possible. The issue is now identifying these situations in different organs and their relevance for the C balance and survival of trees (Galvez et al., 2013; Hartmann et al., 2013).

The fact that genes which control basal functions in all plants are sensitive to sugar and starch (Gibon et al., 2009), suggests that such ‘sensing’ is likely universal in plants. The consequences of this for trees and their C balance are, however, completely unknown. Although we currently know a lot about the molecular regulation of C-storage at the cellular level, there is still a large gap of knowledge at the level of the whole plant, especially for large plants like trees. System-level approaches integrating data from transcriptomic, enzymatic, metabolomic and growth analysis similar to those already implemented in Arabidopsis (Smith & Stitt, 2007) or agricultural crops (Muller et al., 2011) but applied to woody plants, could offer promising tools to unravel the timing of events leading to the coordination of C allocation and growth in trees (Stitt & Zeeman, 2012).

A pressing need is to identify the fraction of NSC that is available (potentially mobilized) in trees (Millard & Grelet, 2010; Rocha, 2013). This implies estimating the dynamics of NSC pools and their lifetime plus the fraction of NSC that is sequestered (i.e. never recovered). Time-integrated tracing techniques, such as using bomb carbon-14 (14C)-tracers, offer promising tools to unravel the turnover of NSC pools and their potential sequestration. Richardson et al. (2013) used bomb 14C to model NSC dynamics in trees and found that the inclusion of two NSC pools, one with a slow turnover (up to 31 yr) and one with a more dynamic nature gave the best prediction of radial growth and the amount and mean residence time of storage C. Also Carbone et al. (2013) used the radiocarbon (14C) ‘bomb spike’ to estimate the age of C used by red maple trees for stem respiration, tree ring growth, and stump production after harvest. They concluded that younger NSC is preferentially used for growth and current metabolic demands, while more recently stored NSC (c. 1–2 yr-old), contribute to annual ring growth and metabolism during the dormant season. Bomb 14C can also be used to determine the age of NSC remobilized after disturbance before tree survival is impaired. For instance, Vargas et al. (2009) found that new root production of tropical trees after hurricane damage relied on C assimilated up to 11 yr earlier, while that of unaffected trees used newly assimilated C. Similarly, Carbone et al. (2013) estimated that stump sprouts of harvested red maple trees relied on C up to 17-yr-old. From stable carbon isotope labeling it is known that NSC pools are turned over rapidly (Würth et al., 1998), and new C is mixed with old C before becoming invested in structural growth (Keel et al., 2007).

We also need to determine if trees have a critical threshold for NSC, below which survival is at risk. There is evidence that species and genotypes of drought-sensitive and resistant trees exhibit differential mobilization of NSC in response to water shortage (Regier et al., 2009; Piper, 2011). In these studies, drought caused a decrease of NSC concentrations in drought-sensitive plants, but an increase in resistant plants. This could indicate a differential ability to regulate baseline NSC as well as growth thresholds in genotypes differently adapted to stress. We need experimental studies that quantify minimum C pools in plants, and evaluate whether these thresholds vary depending on the organ, the species, and the environment. For example, Sevanto et al. (2013) estimated the theoretical threshold of soluble sugar concentrations in bark and phloem tissues required to produce osmotic pressure equal to the pre-dawn leaf water potential. They subsequently compared such theoretical thresholds with the solute contents of drought-stressed pine seedlings and concluded that hydraulic failure could be associated with loss of adequate tissue carbohydrate content required for osmoregulation.

We should move beyond correlative approaches and determine if observed decreases in the growth of trees under long-term C-imbalances are due to a C allocation to storage. Experiments with CO2 fertilization as an additional treatment should be carried out (e.g. Handa et al., 2005; Duan et al., 2013). Such experiments should be performed at the whole plant level and recording temporal changes in C allocation to growth and NSC pools after the imposed treatment (e.g. defoliation, drought) at different levels of CO2 availability. For example, Duan et al. (2013) recently analyzed the effect of drought and increased temperature on the growth, NSC concentrations and C-balance of Eucalyptus globulus seedlings grown at different CO2 concentrations. Their results show that under moderate drought increased CO2 availability promotes both C allocation to growth and NSC storage. However, such beneficial effects of CO2 fertilization disappear when drought becomes severe.

Finally, we need to evaluate if a potential tradeoff between NSC storage and growth actually matters for the C balance and survival of a tree in the long term, and whether and to what extent this effect depends on tree age, habitat and growth conditions. Without filling these important knowledge gaps it seems difficult to critically evaluate if and to what extent C storage can compete with growth and particularly, whether it matters under stress. To date we are left with the empirical evidence that drought and temperatures below 5°C prevent growth, irrespective of C supply.


S.P. was funded by a Juan de la Cierva contract (MEC, Spain) and projects CGL2011-26654 (MCI-Spain) and ARBALMONT (786-2012; Organismo Autónomo de Parques Nacionales, Spain). G.H. received funding from European Research Council (ERC) grant no. 233399 (project ‘TREELIM’ to C.K.). The authors have no conflict of interest to declare.