Introduction

Lodgepole pine, Pinus contorta Douglas ex Louden var. latifolia Engelm. ex S. Watson (http://www.itis.gov/), is a valuable and widespread species found throughout western North America, where it is widely planted as a part of sustainable forestry practice (Wu et al., 2005; Ying and Yanchuk, 2006). However, in recent years many plantations of P. c. latifolia throughout its range have been severely impacted by generally warmer temperatures (especially in winter) and localized changes in precipitation (Folke et al., 2004). In addition to climate adversely affecting tree physiology, and in turn productivity, there has been an increase in pest activity, including widespread mortality caused by the mountain pine beetle, Dendroctonus ponderosae Hopkins (Coleoptera: Scolytidae) (Walther et al., 2002; Kurz et al., 2008; Bentz et al., 2010) and regionally important outbreaks of Dothistroma needle cast [causal agent: Mycosphaerella pini Rost. in Munk; anamorph Dothistroma septospora (Dorog.) Morelet] (Ascomycota: Mycosphaerellaceae) (Woods et al., 2005). It is hypothesized that in the future, climate change will continue and even intensify, which will further impact pine growth and result in more frequent pine pest and disease outbreaks (Coakley et al., 1999; Logan et al., 2003; Desprez-Loustau et al., 2007; Bentz et al., 2010).

To counter the potential negative effects of climate change, efforts have begun to move individuals from particular pine populations to areas whose future climates are predicted to be optimal for their growth in a process called assisted migration (O’Neill et al., 2008). Assisted migration involves moving pine populations to higher latitudinal or elevational sites because it is hypothesized that certain populations of trees are better-suited to areas which, although also warming, will remain cooler overall (Coakley et al., 1999; O’Neill et al., 2008; Marris, 2009).

It remains unclear what impacts the assisted migration of certain populations of P. c. latifolia into new areas will have on their ability to thrive in the near- and long-terms (Marris, 2009). Although attempts have been made to find suitable locations to minimize climate change effects on P. c. latifolia physiology (Cumming and Burton, 1996; Hamann and Wang, 2006; O’Neill et al., 2008), our understanding of this process is in its infancy (Wu et al., 2005; Wallis et al., 2010). Populations of P. c. latifolia that originated from different ecosystems, or that were far apart geographically, have been found to vary not only in growth and reproduction, but also in their susceptibility to diseases (Rehfeldt et al., 1999; Wu et al., 2005; Heineman et al., 2010; Wallis et al., 2010). Such distinct populations also were observed to vary in foliar secondary metabolite (SM) levels, at least when they were planted together at one seed orchard site (Wallis et al., 2010). Overall differences in the ability of populations of P. c. latifolia to synthesize SMs, such as phenolics and terpenoids, are likely to have led to observed variations in growth rates and susceptibility to pests, because such compounds have a variety of roles including photoprotection, relieving oxidative stress, reducing water stress, and protection from insect and pathogen pests (Keeling and Bohlmann, 2006; Seybold et al., 2006; Whitzell and Martin, 2008).

Populations of P. c. latifolia from lower latitudes and from warmer, wetter stands will need increased protection from biotic stresses, and therefore the possession of greater pest defense-associated SM levels (both phenolics and terpenoids) will confer fitness advantages (e.g., Franceschi et al., 2005; Wallis et al., 2008; Adams et al., 2009). In contrast, P. c. latifolia from northern, colder, and drier locales will gain a fitness advantage by putting more resources into primary metabolites, presumably at the cost of producing certain SMs, in order to deal with a shorter growing season and to outcompete neighboring plants for scarce resources (Siska et al., 2002; Andrew and Hughes, 2005; Adams et al., 2009).

However, other SM classes might play larger roles in conferring resistance to abiotic stresses than in their abilities to combat pests; for instance, P. c. latifolia from boreal forest locations may possess greater cell-wall thickening phenolics such as lignin that could confer fitness advantages in overcoming abiotic stresses such as extreme cold (Smallwood and Bowles, 2002; Wei et al., 2006). Levels of other SMs, in particularly flavonols, are positively associated with latitude presumably because they are involved in resisting abiotic stresses (Lätti et al., 2008; Stark et al., 2008; Martz et al., 2009, 2010). Day lengths, lowered temperatures, and UV exposure are all implicated in the relationship between latitude and levels of certain SMs (reviewed by Jaakola and Hohtola, 2010).

Therefore, levels of biotic and abiotic stresses imparted on populations can positively or negatively affect distinct groups of SMs, depending on the class of the compounds and the particular roles that they play in resistance to stress. The aim of our study was to quantify the differences in foliar SM levels among stands of naturally regenerating P. c. latifolia that should be exposed to a variety of abiotic and biotic stresses. We examined variations in SM levels—including total soluble phenolics, lignin, and terpenoids—in the foliage from 12 distinct stands of P. c. latifolia. Ecological, geographical, and climatic differences among these sites were examined to observe which variables might influence the levels of these compound classes. We hypothesized that associations among SM classes and ecological, geographic, or climatic variables will be due, in part, to the particular roles they play in countering abiotic or biotic stresses.

Methods and Materials

Stand Descriptions and Sampling

A total of 12 mature stands of P. c. latifolia of natural origin were assessed across central British Columbia (Fig. 1). These stands were from a variety of habitats and climates comprising five distinct ecosystems as delineated by British Columbia’s biogeoclimatic (BEC) zone system. They included the very cold boreal white and black spruce (BWBS) zone; the very wet Coastal Western Hemlock (CWH) zone; the high elevation Englemann Spruce/Subalpine Fir (ESSF) zone; the very wet Interior Cedar/Hemlock zone (ICH); and the widespread Sub-boreal Spruce (SBS) zone (Meidinger and Pojar, 1991). Climate data for each sampling site [mean summer (June through August), mean winter (December through February) and mean annual temperatures; and mean summer (June through August) and mean annual precipitation levels] were determined by using the Climate BC program, based upon measured temperature and precipitation data from 1961 to 1990 (Wang et al., 2006) (Table 1). Sampling occurred between June 15, and July 15, 2009, and at each stand the trees were in the same phenological stage (near the completion of needle expansion from newly-formed candles). At each site, approximately 5 g of apparently insect- and disease-free, second-year needles (i.e., those that were formed in 2008) were collected from three randomly chosen branches from each of 20 naturally regenerating trees ranging from 15 to 25 year-old. The samples were immediately placed onto dry ice for transport to the lab, where they were stored at −20°C until processed for phytochemical analyses roughly 1 month later.

Fig. 1
figure 1

Locations of populations of Pinus contorta latifolia in British Columbia, Canada identified by letter corresponding to site names in Table 1; and their British Columbia biogeoclimatic (BEC) classification and sample size. Map courtesy of the BC Ministry of Forests and Range (http://www.for.gov.bc.ca/hre/becweb/)

Full size image
Table 1 Coordinates and site characteristics for the Pinus contorta latifolia stands assessed for foliar secondary metabolites
Full size table

Phytochemical Analyses

Needle samples were ground by mortar and pestle in liquid nitrogen, with 100 mg of ground tissue twice-extracted in 1 ml of methanol (Thermo-Fisher Scientific, Pittsburg, PA, USA) and another 100 mg of tissue twice-extracted in 1 ml of dichloromethane (methylene chloride) (Sigma-Aldrich, St. Louis, MO, USA) according to the protocols of Wallis et al. (2008). The methanol extracts were assessed for total phenolics by using the Folin-Ciocalteu reagent spectrophotometric method (with an instrument from Perkin-Elmer, Waltham, MA, USA) for quantifying soluble phenolics, and the pellet left behind following the methanol extraction was further processed to quantify total lignin (bound to the cell walls) according to the procedures of Wallis et al. (2008) with all reagents from Sigma-Aldrich. In brief, the pellets were washed once with 1 ml water, then 1 ml methanol, and finally 1 ml of petroleum ether and were allowed to dry overnight. To the pellet, 200 μl of 1 N NaOH were added, and this was left in a shaker at 40°C for 21 h, after which 200 μl of 1.5 M formic acid and 400 μl of methanol were added. The mixture was centrifuged at 10,000 x g for 5 min, and the supernatant removed. The pellet then was washed in 1 ml of water and re-suspended in 800 μl of 2 N HCl and 300 μl of thioglycolic (mercaptoacetic) acid heated at 86°C for 4 h. The tubes were spun down for 5 min at 10,000 x g, and the supernatant removed, with the pellet then twice-washed in 1 ml water. The pellet was then re-suspended in 1 ml of 0.5 M NaOH and left for 18 h on a shaker. Following centrifugation, the supernatant was transferred to a clean tube and the pellet re-suspended in 0.5 ml of 0.5 M NaOH. Following centrifugation, the supernatant was combined with that from the previous step to total a 1.5 ml of 0.5 M NaOH solution. To this, 300 μl of concentrated (12 N) HCl were added and allowed to stand at room at room temperature for 4 h. The solution was spun down for 5 min at 10,000 x g, and the supernatant discard. The pellet was re-suspended in 1 ml of 0.5 M NaOH, and read on a spectrophotometer (Perkin-Elmer) at 280 nm. Lignin concentrations in mg were determined by using a standard curve made from a dilution series of commercially available lignin standard from spruce (Sigma-Aldrich).

The dichloromethane extracts were analyzed for terpenoid content (monoterpenoids, sesquiterpenoids, and diterpenoids) with a Varian Model 3800 gas chromatograph with a flame-ionization detector and a split line to a Saturn 2200 Ion Trap MS-MS (Agilent, Palo Alto, CA, USA) and n-pentadecane (Sigma-Aldrich) as an internal standard (Wallis et al., 2010). The column was a 30 m × 0.25 mm i.d. × 0.25 μm film thickness Agilent Ultra 2.5% phenylmethylsiloxane capillary column; the carrier gas was helium running at a constant flow of 2.1 ml/min; the split/splitless injection port was in split mode (10:1 ratio) with the temperature set at 320°C; and the detector/transfer line temperatures were set at 320°C. The temperature program started with a hold of 2 min at 40°C; ramped at 10°C/min to 200°C; and finally ramped at 50°C/min to a final temperature of 300°C (with a 5 min final hold) (total run time of 28.20 min). All peak areas of identified terpenes were summed together with others in their class (monoterpenoids, sesquiterpenoids, or diterpenoids) to give an overall value for each terpenoid group. Compound identifications were based on retention time matching to known standards and matching to mass spectral libraries with reverse search indices higher than 850 (Wallis et al., 2008). Stereoisomers of certain compounds (e.g., α-pinene, β-pinene, and limonene) went unresolved, as this study stressed changes in total terpenoid group production (i.e., the regulation upstream of individual monoterpenes, sesquiterpenes, and diterpenes). The toxicity of terpenoids to pathogens is thought to not depend on the enantiomeric composition (Wallis et al., 2008). Putative diterpenoids, labeled in this work as unknown diterpenoids, were compounds whose matches to mass spectral libraries contained a typical diterpenoid skeleton of C20 and for which the top five library matches were different diterpenoids. However, for these putative diterpenoids no single compound exceeded the reverse search index threshold set at 850.

Statistical Analyses

All statistical tests were performed with either SPSS ver. 17 (SPSS Inc., Chicago, IL, USA) or R ver. 2.10.0 (CRAN, Wirtschaftsuniversität Wien, Austria) at a threshold of α = 0.05. Normality was assessed by examining residual plots and histograms, and Levene’s tests for variance heterogeneity were performed. Pearson’s correlations were used to find associations between groups of foliar compounds. Univariate ANOVAs were used to find differences in phytochemistry between trees grown in different BEC zones. Tukey HSD tests were used to separate individual BEC zone means.

Coordinates were first converted from UTM to BC Albers, a metric that is continuous throughout British Columbia (http://geobc.gov.bc.ca) for all analyses that examined relationships between geographical variables and SM levels. Specifically, the BC Albers has an x-coordinate that represents longitude (with increasing numbers representing movement eastward) and a y-coordinate that represents latitude (with increasing numbers representing movement northward).

Pearson’s correlations were used to examine potential relationships between geographic or climatic variables and foliar SM levels. One analysis compared the mean levels of the SMs for each site with these variables, whereas another used the SM values for individual trees. The first analysis was performed to emphasize the geographic or climatic effects on a typical individual in the tree’s population, whereas the latter emphasized effects on multiple individuals in the same stand of trees. This was also performed in part due to concerns over the large sample size (greater than 250 replicates overall) of SM levels potentially skewing these associations, considering that 12 estimates were made for each of the geographical and climatic variables. An additional analysis compared the levels of SMs from individual trees vs. geographic or climatic variables with data from the SBS zone only, as this zone was represented far more (7 times) than the other sites and is arguably the primary ecosystem type in which P. c. latifolia is found in British Columbia.

Linear regression analyses were performed to observe which combination of variables best explained the variance in foliar secondary metabolites. Using the “Rcmdr” package within R, a linear regression model for each SM group as explained by all the geographic and climatic variables was loaded. These models then underwent the “subset model selection…” application under Rcmdr’s “Model” menu, which output a list of the variables included in the models that had the lowest Schwarz’s Bayesian Information Criteria (BIC) values (with the lowest values representing the best models) (Schwarz, 1978). The strongest model matched that selected via Stepwise Regression Analysis based on BIC; however, the next two top models also were selected for comparison as they could not be dismissed due to their possessing similar BIC values (within ~2) to the top model (Burnham and Anderson, 2002). For additional evaluation, the Akaike Information Criteria (AIC) values also were assessed (Akaike, 1974). These regressions were performed on the mean SM levels at each site; on SM levels for all of the individual trees; and on the SM levels for all individual trees for the SBS zone alone.

Results

Secondary Metabolite Identities and Associations

In this study, we assessed a total of 14 monoterpenoids (sabinene, α-pinene, camphene, β-pinene, β-myrcene, α-phellandrene, δ-3-carene, p-cymene, β-phellandrene, δ-2-carene, linalool, camphor, bornyl acetate, and verbanone), seven sesquiterpenoids (α-cubenene, α-copanene, aristolene, β-cadinene, τ-cadinol, germacrenol, and an unidentified sesquiterpenoid), and five diterpenoids (all of which were unable to be identified to specific compounds). We also assessed total phenolic (with Folin-Ciocalteu reagent) and total lignin content.

Foliar phenolic levels were significantly and positively associated with the foliar levels of all three groups of terpenoids; and sesquiterpenoid levels were positively associated with levels of both monoterpenoids and diterpenoids (Table 2). No other significant correlations were observed.

Table 2 Correlations among the different classes of secondary metabolites analyzed from the foliage of Pinus contorta latifolia
Full size table

BEC Zone and Subspecies Effects on Foliar Secondary Metabolites

Populations of P. c. latifolia from the Coastal Western Hemlock (CWH) and Interior Cedar Hemlock (ICH) zones had greater levels of soluble phenolics (41% to 45% more) than the Boreal White/Black Spruce (BWBS) zone (F = 8.348, P < 0.001, N = 237) (Fig. 2a). Lignin levels were greater in the BWBS zone (11% to 22% more) than the SBS or Engelmann Spruce/Subalpine Fir (ESSF) zones, and trees from the ICH zone had greater lignin levels (14% more) than those in the ESSF zone (F = 4.591, P = 0.001, N = 238). Total monoterpenoid levels were greater in P. c. latifolia from the ICH zone (72% to 242% more) than from any other zone (F = 10.105, P < 0.001, N = 226) (Fig. 2b). Total sesquiterpenoid levels were greater in the CWH zone (60% to 147% more) than any other zone, and levels were greater in the ICH (38% more) and BWBS (37% more) zones than the SBS zone (F = 38.467, P < 0.001, N = 233). Total diterpenoid levels were greater in the CWH population (39% to 857% more) than any other population, and diterpenoid levels were also greater in populations in the BWBS and ICH zones (93% to 588% more) than in the ESSF or SBS zones (F = 38.763, P < 0.001, N = 224).

Fig. 2
figure 2

Concentrations (mean ± s.e.) of a phenolic and b terpene secondary metabolites in the foliage of Pinus contorta latifolia collected from various British Columbia biogeoclimatic (BEC) zones. ANOVA statistics are given, and letters indicate significant differences by Tukey HSD tests. BEC zones are as follows: Boreal White/Black Spruce (BWBS); Coastal Western Hemlock (CWH); Engelmann Spruce/Subalpine Fir (ESSF); Interior Cedar Hemlock (ICH); and Sub-Boreal Spruce (SBS)

Full size image

The foliar levels of almost every individual compound, with the exceptions of sabinene, β-pinene, and linalool, were significantly related to BEC zone (Table 3). Pines from the ICH zone possessed greater levels of camphene, δ-3-carene, and β-phellandrene than pines from the ESSF zone (F > 4.088, P < 0.010, N = 233). Pines in the CWH zone had greater levels of α-cubenene, aristolene, β-cadinene, and germacrenol than those growing in either the ESSF, ICH, or SBS zones (F ≥ 5.247, P < 0.001, N = 233). Pines growing in the BWBS zone possessed greater levels of the unidentified sesquiterpenoid than those growing in the ESSF or SBS zones (F = 5.355, P < 0.001, N = 233). Pines growing in the CWH zone possessed greater levels of all five diterpenes than pines growing in the ESSF or SBS zones (F ≥ 4.766, P < 0.001, N = 233).

Table 3 Mean compound levels from foliage of Pinus contorta latifolia (μg/g FW) and ANOVA statistics showing British Columbia’s biogeoclimatic zone differences, with letters indicating separation by Tukey HSD tests
Full size table

Geographic and Climatic Variables and Secondary Metabolite Levels

By using the mean compound levels from each site for the correlation analyses, foliar levels of phenolics were negatively associated with longitude (i.e., phenolic levels were lower in more easterly sites); levels of lignin were positively associated with latitude (i.e., lignin levels were higher in more northerly stands); levels of sesquiterpenoids were positively associated with winter temperature and annual precipitation; and levels of diterpenoids were negatively associated with longitude but positively associated with latitude and winter temperature (Table 4). No other significant correlations were observed.

Table 4 Pearson’s correlations between geographic or climatic variables and mean compound levels per site or individual tree compound levels from an analysis of foliage metabolites from Pinus contorta latifolia
Full size table

Considering the foliar levels of SMs in each tree and not using site means for the correlation analyses, levels of phenolics were negatively associated with elevation, longitude, summer temperature, and summer precipitation but positively associated with latitude and winter temperature; levels of lignin were negatively associated with longitude, summer temperature, annual temperature, and annual precipitation but positively associated with latitude; levels of all terpenoid classes were negatively associated with elevation and longitude but positively associated with latitude, summer temperature (with the exception of the diterpenoids), winter temperature, annual temperature, and annual precipitation (with the exception of the monoterpenoids); and levels of monoterpenoids were negatively associated with summer precipitation, whereas levels of sesquiterpenoids and diterpenoids were positively associated with summer precipitation (Table 4). No other significant correlations were observed.

For trees in the SBS zone only, total phenolics were negatively associated with elevation, longitude, summer temperatures, annual temperature, summer precipitation, and annual precipitation but positively associated with latitude; lignin levels were negatively associated with elevation, longitude, summer temperatures, annual temperatures, and summer precipitation but positively associated with latitude; monoterpenoid and sesquiterpenoid levels were positively associated with latitude; and diterpenoid levels were negatively associated with elevation, longitude, summer precipitation, and annual precipitation but positively associated with latitude (Table 4). No other significant correlations were observed.

Linear Regression Modeling to Explain Variance in Foliar Secondary Metabolite Levels

Whereas the correlation analyses related individual climatic and geographic variables with the observed levels of the SMs, linear regressions were made to combine multiple factors together to explain the observed levels. Mean foliar phenolic levels at each site were best explained [by having the lowest Bayesian Information Criterion (BIC)] by longitude; lignin levels were best explained by elevation, mean annual temperature, and mean winter temperature; monoterpenoid levels were best explained by mean summer precipitation, mean summer temperature, and mean winter temperature; sesquiterpenoid levels were best explained elevation, longitude, mean annual precipitation, mean annual temperature, and mean summer precipitation; and diterpene levels were best explained by elevation, latitude, longitude, mean annual precipitation, and mean summer precipitation (Table 5).

Table 5 Top three models as selected by lowest Bayesian Information Criterion (with Akaike’s Information Criterion also given) explaining secondary metabolite response variables from an analysis of the foliage of Pinus contorta latifolia with the means from each site
Full size table

By using the chemical levels found in foliage from individual trees and not using their site means for analyses, foliar soluble phenolic levels were best explained by longitude, elevation, and mean annual temperature; lignin levels were best explained by elevation, mean annual temperature, and mean winter temperature; monoterpenoid levels were best explained by mean summer precipitation, mean summer temperature, and mean winter temperature; sesquiterpenoid levels were best explained by longitude, elevation, mean annual precipitation, mean annual temperature, and mean summer temperature; and diterpenoids were best explained by longitude, latitude, elevation, mean annual precipitation, and mean summer temperature (Table 6).

Table 6 Top three models as selected by lowest Bayesian Information Criterion (with Akaike’s Information Criterion also given) explaining secondary metabolite response variables from an analysis of the foliage of Pinus contorta latifolia with data from each individual tree
Full size table

By using the chemical levels found in foliage from individual trees from the SBS zone only, foliar soluble phenolic levels were best explained by latitude, elevation, and mean summer temperature; lignin levels were best explained by mean annual precipitation, mean annual temperature, and mean winter temperature; monoterpenoid levels were best explained by longitude and mean summer temperature; sesquiterpenoid levels were best explained by elevation and mean summer precipitation; and diterpenoids were best explained by longitude and elevation (Table 7).

Table 7 Top three models as selected by lowest Bayesian Information Criterion (with Akaike’s Information Criterion also given) explaining secondary metabolite response variables from an analysis of the foliage of Pinus contorta latifolia with data from each individual trees from the SBS zone only
Full size table

It should be noted that in all cases the top three models for each variable had close BIC and AIC values, meaning that one model cannot be considered necessarily better than the others. Additional studies and data collection would be necessary to assess which of the top models would be best for explaining the levels of foliar SMs.

Discussion

Ecosystem type (represented by the BEC zone classifications), geographical location, and climate all significantly impacted at least some of the foliar secondary metabolite levels in the P. c. latifolia that were analyzed in this study. Thus, P. c. latifolia foliar SMs, which likely contribute to overall fitness, varied with ecosystem, locale, and climate.

BEC zones were associated with differential levels of almost every SM that we analyzed. Trees from the CWH and ICH zones often possessed greater foliar SM levels than trees from other zones, albeit significant differences varied depending on the compound. The CWH and ICH are both defined, in part, as rainforests (Meidinger and Pojar, 1991). These ecosystems likely allow P. c. latifolia to have a longer growing season and a climate where certain abiotic stresses (e.g., drought) are reduced. Both longer growing season and reduced stress could have resulted in additional resources for the pines to invest in their secondary metabolism (Herms and Mattson, 1992). Furthermore, these ecosystems could have been more favorable to pest development (Allen et al., 1996), which would, due to more frequent pest encounters, impose selective pressure on the lodgepole pines to produce more defense-associated compounds in their foliage (Wallis et al., 2010). This latter hypothesis could be overtly tested by attempting to find associations between foliar pest outbreaks and SMs, potentially with dendrochronological techniques to estimate historical pest outbreak frequency (Welsh et al., 2009). Regardless, it would appear that the need to counter biotic stresses in these ecosystems resulted in selection pressure leading to greater levels of defense-associated SMs, namely monoterpenoids for the Interior Cedar-Hemlock ecosystem and both sesquiterpenoids and diterpenoids for the Cedar-Western Hemlock ecosystem.

Precipitation likely had some influence on the measured levels of SMs because these compounds were extracted from 0.1 g of fresh and not dried material. However, the differences in terpenoid levels from the wettest sites would presumably be even greater than those from drier sites if samples were dried first, as wetter sites (i.e., those from the CWH and ICH) would likely have greater water content in their tissues.

Regardless of ecosystem type, stands in more northerly and westerly locations, and at lower elevations, were associated with higher foliar SM levels. These results do not support the hypothesis that more southerly stands would contain greater levels of SMs (Siska et al., 2002; Andrew and Hughes, 2005; Adams et al., 2009). However, in this study, northern, western, and lower elevation sites included some of the wettest and warmest sites, including the CWH and ICH zones. Indeed, prior studies observed that greater pest activity at lower elevations was associated with plants having increased levels of SMs (e.g., Hobbs and Partridge, 1979; Hengxiao et al., 1999; Heineman et al., 2010), and the CWH and ICH sites were at the lowest elevations of all the sites sampled.

In contrast to the other SMs, levels of lignin were highest in the boreal BWBS ecosystem type, whereas the other metabolites were often at the lowest levels in trees from that zone. Lignin is involved with cell wall thickening and strengthening, which contributes to cold temperature tolerance (Smallwood and Bowles, 2002; Wei et al., 2006). However, lignin was not associated with lower winter temperatures, so other selection pressures might be at play.

When only the SBS P. c. latifolia populations were considered (because most pine plantations are found in that ecosystem type), increases in total soluble phenolics and lignin levels were associated with lower temperatures and rainfall averages, suggesting greater abiotic stress. By contrast, monoterpenoids and sesquiterpenoids were not associated with climatic variables at all, perhaps implying low variance in biological stresses that influence their production in the SBS zone. Increasing latitude was positively associated with the production of all SMs for trees in the SBS zone. Studies are needed to determine the specific factors that might explain this trend, such as increased summer day length, decreased growing season duration, or increases in cold temperature extremes that occur with higher latitude that imparts a selection pressure for greater SM production in pine populations.

Although we observed that geographic and climatic variables did affect foliar SM levels in P. c. latifolia populations, the physiological and genetic mechanisms behind these observations remain unclear. Our attempts to form testable regression models proved inconclusive, as the top models did not vary much according to their information criteria (both BIC and AIC values), likely because of the inter-relatedness between all of these variables. This also suggests that these geographic and extrapolated climatic variables are perhaps not the best explanatory variables to use for the levels of foliar SMs.

Future studies should attempt to correlate historic and current abiotic and biotic stress levels of stands of P. c. latifolia with their secondary metabolites. Associations with current stress levels would indicate that populations might have responded to local abiotic and biotic stresses by shifting SM production or responding via systemic induced responses (Herms and Mattson, 1992; Koricheva et al., 1998). Alternatively, associations with historical stress levels might reveal that different pine populations may have evolved to produce greater SM levels in response to selection pressures caused by adverse abiotic and biotic stressors (Wu et al., 1996; Hodkinson, 2005; Wallis et al., 2010).

Viewed in light of current efforts to assist the migration of P. c. latifolia populations from warmer, southerly areas, to colder, northerly areas (O’Neill et al., 2008), these results are not in and of themselves sufficient to suggest that such efforts are or are not worthwhile. However, these results reveal that variation exists between different populations of P. c. latifolia, especially those in different broadly-defined ecosystem types. Previous studies have shown that when grown in the same location, pines originating from different stands and ecosystems still retain their variance in both SM levels and pest resistance (Wu et al., 1996, 2005; Wallis et al., 2010). Therefore, the variance in SM concentrations observed in pine populations in this study could be predicted to remain for at least one generation if trees undergo assisted migration.

Climate change is predicted to increase both biotic and abiotic stresses on pine populations. Bark beetle populations are the major cause of area wide mortality for P. c. latifolia, and these populations are predicted to increase regionally with climate change (Bentz et al., 2010). Coupled with changes in temperature and rainfall patterns, bark beetles could force pine ecosystems to higher latitudes and elevations (Bentz et al., 2010). In this study, terpenoid levels, and to some extent the soluble phenolics and lignin, were greater in ecosystems that were wetter and warmer (the CWH and ICH versus the SBS and ESSF), i.e., those likely facing greater biological stresses (Allen et al., 1996). Therefore, migrant pines from these populations would be expected to have a fitness advantage over native pines (i.e., trees whose ancestors originate from the same area) at higher latitude or elevation sites when predicted bark beetle, and possibly pathogen, outbreaks expand into those areas (Bentz et al., 2010). Hypothetically, migrant pines would possess a greater capacity to produce defense-associated compounds. These pines might also be more tolerant of certain abiotic stresses, such as increased temperature extremes and more frequent droughts. However, only longer-term monitoring of the variation of secondary metabolites in multiple stands can confirm if this is the case, or if the native pines might fare little better than migrant trees.