Variation in carbon availability, defense chemistry and susceptibility to fungal invasion along the stems of mature trees

Authors


Author for correspondence:

Devin W. Goodsman

Tel: +1 780 492 6827

Email: goodsman@ualberta.ca

Summary

  • If carbon (C) sinks withdraw carbohydrates as they are transported along tree stems, carbohydrate availability may depend on local sink strength and distance from sources. Defenses, including monoterpenes – a major component of resin – limit the invasibility of pines. Since carbohydrate reserves fund monoterpene synthesis, we hypothesized that monoterpene concentrations in pine stems would decrease from the crown to the lower stem, and susceptibility to fungal infection would increase.
  • Here, we measured carbohydrate and monoterpene concentrations along the stems of lodgepole pine trees (Pinus contorta var. latifolia) before inoculating with a blue-stain fungus at different heights. After 6 wk, we assessed tree responses to fungal infection based on lesion length and carbohydrate mobilization.
  • Concentrations of carbohydrates and monoterpenes in the phloem before inoculation decreased with distance from the crown, whereas lesion lengths after inoculation increased. However, trees mobilized sugars in response to fungal infection such that carbohydrate reserves near lesions were similar at all heights.
  • Despite C mobilization, the lower stem was more vulnerable than the upper stem. Consistent with predictions based on sink–source relationships, vulnerability occurred where carbohydrates were less available, and likely resulted from C withdrawal by sinks higher in the supply chain.

Introduction

Bark beetles that kill healthy mature trees when their populations are sufficiently large include the mountain pine beetle, Dendroctonus ponderosae, the southern pine beetle, Dendroctonus frontalis, and the European spruce bark beetle, Ips typographus (Franceschi et al., 2005). These beetles kill mature trees by attacking their stems, yet with the exception of one study on Pinus ponderosa, which reported resin exudation at two heights (Kolb et al., 2006), researchers have not investigated whether defenses vary along tree stems.

Carbohydrate reserves fund many processes that act as carbon (C) sinks in trees, including growth and the production of defense chemicals. Thus, carbohydrate reserves, which primarily comprise sugars and starch (Chapin et al., 1990), are converted to soluble sugars and transported to C sinks throughout the tree. Before arriving in the roots, soluble sugars produced in the crown must descend the stem, which is itself a C sink, especially during the period of wood growth (Vose & Ryan, 2002). Radial ring growth and respiration sinks extract soluble sugars from the phloem as they are transported down from the crown (Sevanto et al., 2003). This is likely the reason why the reserves in the roots of mature aspen trees that were defoliated by forest tent caterpillar (Malacasoma disstria) recovered more slowly than the reserves in the woody parts of the crown (Landhäusser & Lieffers, 2012). Moreover, sinks in tree stems likely also explain negative correlations between the length of lodgepole pine stems and the quantity of carbohydrate reserves in their roots (Goodsman et al., 2010). If a tree is attacked mid-stem, the resources used to fund the defensive response originated in the crown and were transported down the stem to the region of attack.

Although long-distance carbohydrate transport in stems is recognized as an important factor impacting defense (Christiansen et al., 1987), the mechanics of long-distance carbohydrate transport are still in dispute. There is disagreement in the literature about whether the pressure-induced flow mechanisms originally proposed by Münch (1930) can adequately describe carbohydrate transport in long stems (Thompson, 2006; Jensen et al., 2012). Thompson (2006) contends that the dual function of the sieve element–companion cell complexes, which includes both long-distance transport and lateral movement of carbohydrates to surrounding tissues, is not accounted for in the original Münch hypothesis. Apoplastic unloading is an example of lateral movement which makes phloem transport conduits appear to leak (Thompson, 2006). Leaky phloem transport in apple seedlings provides a mechanism that may explain the existence of carbohydrate reserves in the stem phloem of trees, as well as how they can be remobilized (McQueen et al., 2005). While the literature on long-distance phloem transport is unsettled, we know still less about how these concepts relate to C availability and defense in the stems of large trees.

Conifers use carbohydrate reserves located within their stems to fund defense reactions there (Guérard et al., 2007), but it is unknown how carbohydrate reserves are allocated to attack zones at different locations along the length of the stem. Although sink–source relationships are typically used to explain the impacts of defoliation on defense within tree crowns (Honkanen et al., 1999), they are also relevant along tree stems because they explain variation in defense based on rules that govern how C sinks access C resources (Honkanen & Haukioja, 1998). The most explicit rule of this framework is that strong growth sinks lead to high sugar influx in growing tissues, resulting in strong defensive responses as a result of high carbohydrate availability. Thus, varying sink strength and sink priority lead to spatial and temporal variation in suitability for herbivores within plants (Honkanen & Haukioja, 1998). One limitation of the sink–source framework as described by Honkanen & Haukioja (1998), however, is that it is based on competing modular sinks. To our knowledge it has not been extended to include nonmodular plant organs such as tree stems. We contend that the sink–source framework is equally useful in this context, provided the concept of discrete competing modules is replaced by a vertical continuum of competing sinks along the length of the stem. As sinks in the stem likely reduce downstream carbohydrate availability, we hypothesized that long, crownless stems would create vertical hierarchies in chemical defenses dependent on distance from the crown.

The current North American mountain pine beetle outbreak motivated our study of vulnerability along the stems of mature pine trees. The mountain pine beetle is one of the most destructive pests in North American forests, owing to its ability to use aggregation pheromones to mass-attack and kill healthy pines (Safranyik & Carroll, 2006). After landing on a suitable host, beetles bore through the outer bark and inoculate the phloem and xylem with symbiotic fungi, including Grosmannia clavigera, Leptographium longiclavatum and Ophiostoma montium (Six, 2003). Although constitutive defenses in the phloem are the first obstacles encountered by invading bark beetles, pine trees also respond to attack by synthesizing defense chemicals, pitching out beetles with resin, and by forming resin-filled lesions to quarantine and kill the beetles and their fungi (Franceschi et al., 2005). Shorter lesions are believed to indicate more efficient defenses (Bonello & Blodgett, 2003; Bonello et al., 2006; Krokene et al., 2008). Monoterpene concentrations in the phloem are a relevant measure of defense in this context, as they are a major component of tree resin, and pine trees produce them when responding to invading bark beetles and their fungal symbionts (Raffa et al., 2005). The objectives of our study were to test whether gradients in defense and vulnerability to fungal attack exist in lodgepole pine stems, and to relate these to local carbohydrate availability and distance from C sources. We quantified defense by measuring monoterpene concentrations in the phloem, and vulnerability by measuring lesion lengths formed by trees after they were inoculated with G. clavigera.

Materials and Methods

Individual trees were selected in three pure fire-origin lodgepole pine (Pinus contorta Dougl. var. latifolia) stands that were 55 yr old, had similar stand density and productivity as measured by site index (height over age), and with similar understory composition. Stands were located south of Hinton, Alberta, Canada: stand 1 was located at N 53.233, W 117.360; stand 2 at N 53.224, W 117.348; and stand 3 at N 53.227 W 117.355 (WGS 84 Map datum). In each of the three stands, 12 dominant and 12 intermediate lodgepole pine trees were selected. Dominant trees were 15 m tall on average, and had crowns which were partially above those of neighboring trees. The crowns of dominant trees received full sunlight throughout most of the day. Intermediate trees had crowns with tops below the general canopy, but they still received direct light from above. The mean diameter at breast height (1.3 m) of dominant trees was 16.85 cm, whereas the mean diameter at breast height of intermediate trees was 9.19 cm. We included both dominant and intermediate trees in our study because mountain pine beetles prefer larger-diameter dominant trees when their populations are high (Safranyik & Carroll, 2006). Experimental trees were separated by a minimum of 30 m (approx. two tree lengths) to minimize nonindependence among trees. They had symmetrical crowns, and stems were free of visible signs of damage or disease. The lodgepole pine trees in these stands were well suited to this experiment, as we were able to access the full length of their crownless stems using a 13 m extension ladder stabilized with guy lines.

In mid-July 2010, we randomly assigned three dominant and three intermediate trees at each site to one of four treatments, such that each site had a total of 12 dominant and 12 intermediate trees assigned to treatments (Table 1). The four treatments were: inoculation near the base of the live crown; inoculation at a height of 1.3 m; inoculation halfway between the base of the live crown and 1.3 m; and no inoculation. We chose to inoculate at the base of the live crown, because the direction of carbohydrate mobilization below the crown would presumably be downward toward sinks in the stem and roots, whereas carbohydrates can be moved upward within the crown. We also inoculated at 1.3 m above the ground, because this is a standard sampling location in trees, and the location where most fungal-inoculation studies insert inoculums. Having an inoculation point between the base of the crown and 1.3 m permitted us to detect gradients along the stem. Throughout the current study we refer to the inoculation near the base of the live crown as the top inoculation; the inoculation between the base of the crown and 1.3 m as the middle inoculation; and the inoculation at 1.3 m as the bottom inoculation (Fig. 1a). Trees that were not inoculated served as a baseline for carbohydrate content at each height to be compared with the fungus-inoculated trees at the end of the study. The Grosmannia clavigera (Rob.-Jeffr. & R.W. Davidson) Zipfel, Z.W. de Beer & M.J. Wingf. that was inoculated was isolated from living mountain pine beetle (Dendroctonus ponderosae Hopkins) larvae as described in Goodsman et al. (2012) and grown on malt extract agar (1.5% agar and 1.5% malt extract by volume).

Table 1. Dominance class of lodgepole pine (Pinus contorta var. latifolia) as well as inoculation treatments (with Grosmannia clavigera inoculums) replicated at each of the three study sites
Dominance classInoculation heightTime of sampling
DominantTopMid-July and late August
MiddleMid-July and late August
BottomMid-July and late August
UntreatedLate August
IntermediateTopMid-July and late August
MiddleMid-July and late August
BottomMid-July and late August
UntreatedLate August
Figure 1.

Sampling and inoculation methods showing: (a) the location of inoculations along the crownless stem of a lodgepole pine (Pinus contorta var. latifolia) tree; (b) the location of the initial stem sample with respect to the inoculation point; and (c) the location of the late-August stem samples with respect to the lesion that formed in response to fungal inoculation. At each inoculation height (top, middle and bottom), trees were inoculated twice such that inoculations were on opposite sides of the stem.

To inoculate with G. clavigera, we bored a hole through the phloem and into the first layer of the xylem using a sterilized cork borer (0.9 cm in diameter). We then inserted a circular fungus-agar plug and held it in place with a small piece of sterilized wooden dowel. Trees were inoculated on opposite sides of the stem at the same height (two inoculations per tree). At the time of inoculation, we also collected two samples of phloem (3 cm long × 1 cm wide) from opposite sides of the stem, 4 cm to the right and 4 cm above each inoculation point using sterilized chisels (Fig. 1b). Samples were immediately frozen on dry ice in the field and remained frozen at −20°C until carbohydrate and monoterpene analysis. Inoculation treatments across all three sites were completed within 1 wk.

In late August, c. 6 wk after inoculation, we collected a 1 cm × 1 cm sample of phloem from inside each lesion above the point of inoculation and a 3 cm × 1 cm sample from 4 cm to the left of the lesion boundary (Fig. 1c). Lesions were measured from their highest to lowest extents based on discoloration of the xylem just under the vascular cambium (Fig. 2). Thus, for each inoculated tree we collected two samples from outside the lesions and two samples from inside the lesions. At this time we also sampled the untreated trees (not inoculated) at the three respective heights to be comparable to samples taken from inoculated trees. Untreated trees were not sampled when the study was initiated because we assumed that the samples we took in mid-July from trees that we subsequently inoculated would be representative of carbohydrate and monoterpene content in trees at that time. However, by inoculating, we changed the way trees allocated carbohydrates relative to unmolested trees, so we needed untreated trees to serve as a baseline in late August. All samples were placed on dry ice in the field and frozen at −20°C until carbohydrate and monoterpene analysis. Mid-July and late-August sample periods corresponded to the time when mountain pine beetles attack and colonize new hosts as well as the time of rapid radial growth in the stem.

Figure 2.

A lesion that formed on a lodgepole pine (Pinus contorta var. latifolia) tree responding to inoculation with Grosmannia clavigera. To determine lesion length, lesions were measured from the highest point of discoloration on the sapwood down to the lowest point, as shown. Also shown are the location of the initial stem sample collected at the time of inoculation and the location of stem samples collected 6 wk later when lesion lengths were recorded. Bar, 1 cm.

In the laboratory, we removed the outer bark and xylem from our frozen stem samples using a box cutter and split each sample in half. What remained was primarily living phloem. Half was immediately returned to the freezer and remained frozen at −20°C until monoterpene analysis. The other half was oven dried at 100°C for 1 h to stop enzymatic conversion of starch to sugar, and then dried for another 3 d at 70°C. After drying, samples were ground using a Wiley Mini Mill fitted with a 40 mesh (0.4 mm) screen (Wiley, Thomas Scientific, Swedesboro, NJ, USA). After grinding, the two samples from each tree were pooled and used to determine total starch and water-soluble sugar concentrations in the phloem according to the protocol described by Chow & Landhäusser (2004). Briefly, water-soluble sugars were extracted from 50 mg of ground tissue in 80% hot ethanol and reacted with phenol-sulfuric acid before colorimetric measurement using a spectrophotometer (Pharmacia LKB Ultrospec III, Sparta, NJ, USA) at a wavelength of 490 nm. We enzymatically digested starches remaining in the residual pellet and combined the resultant glucose hydrolyzate with peroxidase-glucose oxidase/o-dianisidine (color reagent) before measuring glucose hydrolyzate (starch) concentrations at a wavelength of 525 nm.

For the monoterpene analyses, the remaining half of each frozen phloem sample was ground in liquid nitrogen, and 100 mg of the tissue was transferred to a 1.5 ml micro-centrifuge tube. Samples were extracted with 0.5 ml dichloromethane and 0.01% tridecane as a surrogate standard. After adding the solvent, the samples were vortexed for 30 s, sonicated for 10 min, centrifuged at 11 600 g for 15 min, and placed in a freezer overnight to freeze the pellet. The following morning, we transferred the extract to GC vials and repeated the extraction on the pellet before discarding it. One microliter of extract was injected from the GC vials into an Agilent 7890A/5062C Gas Chromatograph/Mass Spectrometer (Agilent Technologies, Santa Clara, CA, USA) equipped with an HP Innowax (Agilent Technologies) column (ID, 0.25 mm; length, 30 m). The helium carrier gas flow was set at 1.0 ml min−1 and the following temperature program was applied: the temperature was set to 50°C for two min, increased to 60°C by 1°C min−1 and then elevated to 250°C by 20°C min−1. We used the following standards for quantification: borneol, pulegone, α-terpinene, γ-terpinene, α-terpineol (Sigma-Aldrich), camphor, 3-carene, α-humulene, terpinolene, α-thujone and ß-thujone, (−)-α-pinene, (−)-β-pinene, (S)-(−)-limonene, sabinene hydrate, myrcene, (−)-camphene, p-cymene (Fluka, Sigma-Aldrich), bornyl acetate, cis-ocimene, α-phellandrene (SAFC Supply Solutions, St Louis, MO, USA), β-phellandrene (Glidco Inc., Jacksonville, FL, USA). Agilent software (MSD ChemStation, Agilent Technologies, Santa Clara, CA, USA) allowed us to determine the concentrations of each chemical by integrating the area under each peak measured by the GC.

We used the R program to analyze and graph our data (R Development Core Team, 2011) as well as the nlme package (Pinheiro et al., 2011) for mixed models in R. Our statistical models were either ANOVA or regression models. ANOVA models had two main effects: inoculation height along the stem (three levels) and dominance class (two levels). There were never statistical interactions between the two main effects and, based on Akaike Information Criteria (AIC) values, statistical models with separate random intercepts (mixed-effects models) for each site were never better than models in which all sites were pooled together. For our untreated trees, which were sampled at all three heights, having a random intercept for each tree improved the models. All of our data were normally distributed or were transformed to resemble normally distributed data. When present, heterogeneity of variance at different inoculation heights was accommodated in the models using weighting functions (Zuur et al., 2009).

For F-tests, F-values followed by numerator and then denominator degrees of freedom were reported along with associated P-values. Sample sizes used to calculate estimates were not perfectly balanced, because of missing data. Our sample size for inoculated trees included seven intermediate trees inoculated at the top position; nine intermediate trees inoculated at the middle position; seven intermediate trees inoculated at the bottom position; nine dominant trees inoculated at the top position; eight dominant trees inoculated at the middle position; and nine dominant trees inoculated at the bottom position (= 49). Our sample size for untreated trees was balanced. We therefore had nine dominant and nine intermediate trees that were sampled at three heights along their stems corresponding to the top, middle and bottom inoculation positions (= 54). Rather than perform a myriad post-hoc tests, we used confidence intervals (CIs) to visually compare means. Details of this approach and its relationship to fixed alpha testing are given in Cumming & Finch (2005). We used the boot package (Canty & Ripley, 2012) for bootstrapping confidence intervals in R. To bootstrap confidence intervals we resampled our data with replacement such that we had a bootstrapped sample size equivalent to our true sample size. We resampled the data in this way 2000 times and fitted our ANOVA models to each of the 2000 resampled datasets. From the refitted models we acquired 2000 estimates for each coefficient (mean) corresponding to the treatments. We calculated the 95% CI using these estimated coefficients. A similar procedure was used to estimate the 95% CI for untreated trees.

Results

Constitutive carbohydrate reserves and monoterpenes

While there was a decreasing trend in soluble sugar concentrations with distance from the live crown in mid-July (Fig. 3a), starch concentrations in the phloem did not differ at different heights along the stem (Fig. 3b). Dominant trees had higher overall starch concentrations in their phloem than intermediate trees (Fig. 3b, F1,43 = 35.6, < 0.001) but there was only weak evidence that they had higher overall soluble sugar concentrations in mid-July (Fig. 3a, F1,43 = 3.24, = 0.079).

Figure 3.

Mid-July concentrations of soluble sugar (a) and starch (b) in lodgepole pine (Pinus contorta var. latifolia) phloem sampled at the base of the live crown (top), halfway between the top and bottom (middle), and 1.3 m above the ground (bottom) in dominant (closed circles) and intermediate (open circles) trees. Points represent means, and whiskers indicate width of bootstrapped 95% CI.

In mid-July, constitutive monoterpene concentrations in the stem phloem were higher near the top inoculation than near the lower inoculations in both dominant and intermediate trees (Fig. 4a, F2,43 = 10.9, < 0.001). Moreover, dominant trees, which had higher overall carbohydrate concentrations, also had higher overall concentrations of constitutive monoterpenes than intermediate trees (Fig. 4a, F1,43 = 12.4, = 0.001). This pattern of higher monoterpene concentrations in dominant than in intermediate trees was consistent for the three most abundant monoterpenes produced by our study trees (β-phellandrene, β-pinene, and 3-carene). Furthermore, at the time of inoculation, concentrations of each of these monoterpenes were consistently higher near the top inoculation than near the lower inoculations in both dominant and intermediate trees (Fig. 4b,d).

Figure 4.

Mid-July concentrations of constitutive (preformed) monoterpenes in lodgepole pine (Pinus contorta var. latifolia) phloem at the time of inoculation in mid-July: (a) total monoterpenes, (b) β-phellandrene, (c) β-pinene, and (d) 3-carene. Samples were taken below the base of the live crown (top), halfway between the top and bottom (middle), and 1.3 m above the ground (bottom) in dominant (closed circles) and intermediate (open circles) trees. Points represent means, and whiskers indicate width of bootstrapped 95% CI.

Lesion length and chemistry

Lesion length depended on height of inoculation along the stem (F2,45 = 4.97, = 0.011). In both dominant and intermediate trees, lesions were longer near the bottom inoculation and shorter near the top inoculation (Fig. 5). In addition, near the top and middle inoculations there was no relationship between lesion length and local soluble sugar concentration (Fig. 6a,b). However, near the bottom inoculation, local lesion length showed a strong negative relationship with the quantity of sugars nearby before inoculation (Fig. 6c).

Figure 5.

Lesion lengths formed on lodgepole pine (Pinus contorta var. latifolia) stems just under the vascular cambium at inoculation points near the base of the live crown (top), halfway between the top and bottom (middle), and at 1.3 m (bottom) in dominant (closed circles) and intermediate (open circles) trees. Points represent means, and whiskers indicate width of bootstrapped 95% CI.

Figure 6.

Lesion lengths formed on lodgepole pine (Pinus contorta var. latifolia) stems in relation to the concentration of soluble sugars in a phloem sample taken from the inoculation point at the time of inoculation in mid-July. Points represent measurements from single dominant (closed circles) or intermediate (open circles) trees: (a) at the base of the live crown (top); (b) halfway between the top and bottom inoculations (middle); and (c) at 1.3 m above the ground (bottom).

Concentrations of induced monoterpenes inside lesions were 10 times higher than constitutive monoterpene concentrations and they did not vary based on inoculation height (Fig. 7). Thus, within a tree dominance class, large and small lesions had similar internal monoterpene concentrations. However, dominant trees did have higher overall monoterpene concentrations inside their lesions than intermediate trees (Fig. 7, F1,43 = 14.7 < 0.001).

Figure 7.

The total concentration of monoterpenes accumulated inside lesions in the lodgepole pine (Pinus contorta var. latifolia) phloem. Samples were collected at inoculation points near the base of the live crown (top), halfway between the top and bottom (middle), and 1.3 m above the ground (bottom) in dominant (closed circles) and intermediate (open circles) trees. Points represent means, and whiskers indicate width of bootstrapped 95% CI.

Carbon sinks created by fungal infection

In infected trees, soluble sugar concentrations in the phloem outside the lesions did not differ depending on inoculation height, whereas they were still lower near the middle and bottom than near the top in uninfected trees (Fig. 8a, gray points, F2,34 = 11.1, < 0.001). Note that in uninfected trees, concentrations of soluble sugars and their downward trend were similar to those in our inoculated trees before inoculation (Fig. 3a), but the 95% CIs were much narrower as a result of within-tree replication. Like soluble sugar concentrations, starch concentrations in the phloem of untreated trees decreased with distance from the crown (Fig. 8b). Moreover, soluble sugar and starch concentrations in defense zones at all inoculation heights were consistently higher than in samples taken from corresponding heights in uninfected trees (Fig. 8). Near the bottom inoculation, dominant and intermediate trees allocated more soluble sugars (concentration after – concentration before) to defense zones when lesions were longer (Fig. 9).

Figure 8.

Late-August concentrations of soluble sugar (a), and starch (b) in the phloem of lodgepole pine (Pinus contorta var. latifolia) trees that were inoculated (inoc.) with Grosmannia clavigera and in trees that were uninfected (no inoc.). Phloem was sampled at the base of the live crown (top), halfway between the top and bottom (middle), and 1.3 m above the ground (bottom) in dominant (dom.) and intermediate (int.) trees. Points represent means, and whiskers indicate width of bootstrapped 95% CI.

Figure 9.

Changes in soluble sugar concentrations as a function of lesion length near inoculation points at the bottom of lodgepole pine (Pinus contorta var. latifolia) stems (1.3 m above the ground) 6 wk after inoculation with Grosmannia clavigera in dominant (closed circles) and intermediate (open circles) trees. The fitted line is a logistic curve fitted only to the data for dominant trees.

Discussion

Our results demonstrate the utility of the sink–source framework for explaining patterns in defense in nonmodular tree stems which are outside its traditional domain. Consistent with the predictions of the sink–source framework as outlined by Honkanen & Haukioja (1998), we observed variation in susceptibility to fungal infection along the stems of lodgepole pines that depended on C import from sources to sinks. Before inoculation with G. clavigera, carbohydrate concentrations in the phloem and constitutive monoterpene defenses decreased with distance from the crown. At the time of inoculation, higher sugar influx rates near the crown likely led to superior defensive responses relative to lower influx rates in the stem, as lesions were smaller near the crown. After inoculation, sink priority in the stem changed and trees imported more carbohydrates to the phloem near large lesions than near small lesions, even though larger lesions were generally further from the crown. We discuss these findings in detail in the following sections.

Constitutive carbohydrates and monoterpenes

Carbohydrate concentrations in the phloem of dominant and intermediate trees decreased with distance from the crown in both the July and August samples. As peak cambial growth occurs in the summer, gradients in carbohydrate concentrations in stems were likely the result of active growth sinks along their length. Sinks closer to C sources diminish the quantity of carbohydrates available to more distal sinks, provided their sink strengths are similar (Minchin & Lacointe, 2005). Therefore, in accordance with a conceptual model proposed by Landhäusser & Lieffers (2012), we observed diminishing carbohydrate availability in the crownless stem with distance from the crown.

We did not measure reserves in the stem xylem, even though it is widely regarded as the principal storage organ in mature trees (Hoch et al., 2003; Sala et al., 2012) as a result of its size. However, we contend that concentrations of starch and soluble sugars in the phloem are better indicators of carbohydrate availability. This is partly because of the phloem's role as an avenue for carbohydrate transport, but also because the carbohydrate content of the xylem may be disconnected from C availability if carbohydrates have become sequestered there (Millard et al., 2007). In aspen (Populus tremuloides), phloem carbohydrate concentrations, and in particular starch concentrations, are consistently higher throughout the year than in the xylem (Landhäusser & Lieffers, 2003). Carbohydrate concentrations in the phloem of ponderosa pines (P. ponderosa) are five to six times higher than those in the xylem (Pruyn et al., 2005). Because carbohydrates are transported along concentration gradients, sinks in the xylem are likely supplied by resources in the phloem. However, because C import into local sinks in the xylem is likely constrained by carbohydrate availability in the nearby phloem, we believe that the concentrations of starch and soluble sugars in the outer rings of the xylem would exhibit similar patterns to those we found in the phloem.

Constitutive monoterpene concentrations were highest where carbohydrate availability was highest – near the tops of the crownless stems. We anticipated that monoterpene production would depend on local carbohydrate availability because the mevalonate-independent pathway for monoterpene synthesis occurs in plastids where carbohydrate reserves are stored (Chappell, 2002). Honkanen et al. (1999) found that growth rates and concentrations of monoterpenes in Scots pine needles were positively related, as they had predicted based on sink–source relationships. The sink–source framework also provides a viable explanation for high monoterpene concentrations in the stem phloem near active growth regions: high carbohydrate availability coupled with strong growth sinks near the tops of stems likely resulted in high rates of soluble sugar import and, consequently, high availability of carbohydrate reserves for monoterpene synthesis.

Lesion lengths and sugar concentrations

Lesions on the stem were shorter near the crown, where local carbohydrate concentrations were high, and longer near its base, where local carbohydrate concentrations were lower. One explanation for this finding is that the rate of soluble sugar import into sinks created by fungal infection likely dictates the efficacy of defensive responses (Lieutier et al., 1993). Sugar concentrations in tissues near inoculation points before fungal infection likely corresponded to rates of soluble sugar import in those regions and, consequently, they also impacted the speed of response to fungal invasion. Thus, we propose that longer lesions on the lower stem resulted from delayed sugar import into defense zones as a result of lower initial concentrations of soluble sugars in nearby reserves.

Near the crown, where concentrations of sugars in the phloem, and presumably soluble sugar influx rates from the crown were high, there was no relationship between local sugar concentration before infection and lesion size. Near the base of the stem, however, where influx from the crown was likely diminished, lesion length was negatively related to sugar concentrations in the phloem before inoculation. The finding that trees with more local reserves displayed more effective responses near their bases suggests that defensive responses depended on the mobilization of nearby reserves to the region of attack. This assertion is corroborated in a tracer study on Scots pine saplings inoculated with a blue-stain fungus, which showed that saplings used reserves located in the stem to fund defense reactions (Guérard et al., 2007). When carbohydrate reserves in the phloem were lower, moving sugars over a greater distance likely delayed defensive responses, allowing longer lesions to develop.

Carbon sinks created by infection

Because the allocation of soluble sugars to defense zones increased with lesion size, the decreasing trend in soluble sugar concentrations with distance from the crown was no longer evident after inoculation. Moreover, sugar accumulation near lesions likely reflected the cost of battle rather than preparedness for war. The lesions of dominant trees contained higher concentrations of monoterpenes than those of intermediate trees, and dominant trees also had higher concentrations of soluble sugars and starch near defense zones. Moreover, carbohydrate concentrations in the phloem near longer lesions increased more than in the phloem near shorter lesions, even though longer lesions were further from the crown. Thus, when locally stressed, the size of sinks created by lesions superseded the influence of sink proximity to the crown as a determinant of carbohydrate import. High quantities of carbohydrate reserves have also been recorded near latex tapping locations at the base of rubber trees (Silpi et al., 2007; Chantuma et al., 2009). Based on the sink–source framework, we hypothesize that after normal rates of sugar influx along the stem have been perturbed by attack, local sugar accumulation corresponds to the sink created by tree responses to fungal infection.

Ecological implications

Terpenes and phenolics are C-based defense chemicals (Keeling & Bohlmann, 2006), and thus the ability to mobilize carbohydrates to defense zones is likely critical for their synthesis in response to attack. For this reason carbohydrate mobilization along tree stems is likely a highly adaptive response, as long branchless stems are vulnerable to attack. Our study showed that the sinks created by tree responses to G. clavigera inoculations eventually lead to carbohydrate mobilization to the lesion front. However, trees initially had low carbohydrate content and constitutive defenses in their lower stem relative to their upper stem. The delay associated with importing the necessary carbohydrates likely impeded defensive responses in the lower stem, and resulted in heavier damage to the xylem and phloem. Thus, we propose that the lower stem is in a poor position to quickly mobilize the large quantities of carbohydrates necessary for the production of the defensive compounds used to contain attacks. This may lead to higher mountain pine beetle colonization of the lower stem, as observed by Cole & Amman (1983). Although mountain pine beetles preferentially attack healthy dominant trees when their populations are large (Safranyik & Carroll, 2006), we have confirmed that dominant trees within a stand are better defended than intermediates. When bark beetles mass-attack, however, the mobilization of large quantities of C to reaction zones in dominant trees likely results in rapid depletion of carbohydrate reserves. Thus, during mass attacks by bark beetles, the defensive responses elicited by bark beetle fungi may exhaust carbohydrate reserves and hasten tree death (Lieutier et al., 2009).

Acknowledgements

This research was funded by a PGS-D2 scholarship granted to D.W.G. from the Natural Sciences and Engineering Research Council of Canada (NSERC), and separate NSERC Discovery grants awarded to N.E., S.L., and V.L. We are also grateful to Alberta Sustainable Resource Development, West Fraser, Weyerhaeuser, and the University of Alberta for logistical support, and to Eckehart Marenholtz, Iain Grant-Weaver, Lucas Veldhoen, and Pak Chow for help in the field. We thank Jen Klutsch for proofreading the manuscript, as well as two anonymous reviewers for constructive comments on our work.

Ancillary