Vegetation change is expected with global climate change, potentially altering ecosystem function and climate feedbacks. However, causes of plant mortality, which are central to vegetation change, are understudied, and physiological mechanisms remain unclear, particularly the roles of carbon metabolism and xylem function.
We report analysis of foliar nonstructural carbohydrates (NSCs) and associated physiology from a previous experiment where earlier drought-induced mortality of Pinus edulis at elevated temperatures was associated with greater cumulative respiration. Here, we predicted faster NSC decline for warmed trees than for ambient-temperature trees.
Foliar NSC in droughted trees declined by 30% through mortality and was lower than in watered controls. NSC decline resulted primarily from decreased sugar concentrations. Starch initially declined, and then increased above pre-drought concentrations before mortality. Although temperature did not affect NSC and sugar, starch concentrations ceased declining and increased earlier with higher temperatures.
Reduced foliar NSC during lethal drought indicates a carbon metabolism role in mortality mechanism. Although carbohydrates were not completely exhausted at mortality, temperature differences in starch accumulation timing suggest that carbon metabolism changes are associated with time to death. Drought mortality appears to be related to temperature-dependent carbon dynamics concurrent with increasing hydraulic stress in P. edulis and potentially other similar species.
Widespread tree mortality associated with drought, increased temperatures, and tree pest/pathogen outbreak is an emerging threat of global change now documented across all six forested continents (Allen et al., 2010). Recent die-off from tree mortality in western North America now affects over 605 000 km2 of coniferous forests from Mexico to Alaska (Allen et al., 2010). The consequences of tree mortality include impacts on community composition, structure, and ecosystem function (Allen & Breshears, 1998; Breshears et al., 2005; Koepke et al., 2010; Carnicer et al., 2011; Royer et al., 2011; Anderegg et al., 2012a). These changes in the landscape affect biological diversity (Gitlin et al., 2006; Sthultz et al., 2009), wildlife habitat (Klenner & Arsenault, 2009), carbon cycling (da Costa et al., 2010; Metcalfe et al., 2010), ecosystem goods and services (Breshears et al., 2011), hydrological function (Guardiola-Clararnonte et al., 2011, 2012; Adams et al., 2012), and susceptibility to invasion by undesirable exotic species (Kane et al., 2011). Widespread tree mortality is a potentially positive feedback that could accelerate global change through the loss of terrestrial carbon sinks and canopy cover influences on landscape energy balance (Breshears & Allen, 2002; Field et al., 2007; Bonan, 2008; Chapin et al., 2008; Kurz et al., 2008a,b; Rotenberg & Yakir, 2010). Despite these threats, no model yet exists that can mechanistically predict tree mortality in response to a changing climate (Fisher et al., 2010; McDowell et al., 2011).
There is great interest in understanding the physiological processes underlying mortality of trees. Two mechanisms have been proposed to explain drought-induced tree mortality: hydraulic failure and carbon starvation (Hacke et al., 2000; Hacke & Sperry, 2001; McDowell et al., 2008). Hydraulic failure from drought could occur if high xylem tension causes catastrophic cavitation of the vasculature by air embolism, impeding the flow of water to cells, and leading to loss of turgor and increased risk of mortality. Carbon starvation could occur when stomatal limitations caused by water loss reduce photosynthesis sufficiently to decrease carbohydrate availability needed to support respiratory demands, leading to loss of metabolic and defensive capabilities (McDowell et al., 2011). In the case of carbon starvation, both the immediate production of photosynthate and the availability of stored energy reserves may be important in regulating the time to mortality (Sala et al., 2010). Mortality by both mechanisms can be exacerbated by a number of internal and external factors, including tree pests and pathogens (McDowell et al., 2008; Kane & Kolb, 2010).
An initial experimental test of these mechanisms was conducted inside the environmentally controlled Biosphere 2 glasshouse with transplanted piñon pine (Pinus edulis) trees under two temperature regimes (Adams et al., 2009a). Under simulated severe drought, trees in the warmer treatment died 28% (7 wk) earlier than ambient-temperature trees (Adams et al., 2009a). Xylem tensions increased similarly for trees under both temperatures to a threshold sufficient to cause complete cavitation, as inferred from a hydraulic vulnerability curve developed for this species (Linton et al., 1998), indicating that hydraulic failure occurred before mortality (Adams et al., 2009a). However, this threshold was reached in trees of both temperature treatments at the same time, and did not explain time-to-mortality differences between temperature treatments. Trees in both treatments ceased photosynthesis rapidly in response to drought. Before mortality, all trees had cumulatively respired equivalent amounts of CO2, but elevated-temperature trees respired this CO2 in a significantly shorter period of time than ambient-temperature trees. Therefore, temperature-sensitive differences in survival during drought were linked to temperature-sensitive carbon metabolism – a result consistent with the carbon starvation hypothesis. Adams et al. (2009a) were challenged for not directly measuring stored carbohydrate resources and for failing to consider other research showing increased carbohydrate storage during drought (Leuzinger et al., 2009; Sala, 2009; Sala et al., 2010; Piper, 2011).
Subsequent assessments have led to a sophistication of initially proposed hypotheses, positing that tree drought mortality is a complex process and can occur by multiple mechanisms that are highly interrelated and hierarchical (Adams et al., 2009b,c; Sala, 2009; Sala et al., 2010; McDowell & Sevanto, 2010; McDowell, 2011; McDowell et al., 2011; Anderegg & Callaway, 2012; Plaut et al., 2012). For example, carbon starvation could occur from a reduction of tree carbohydrate reserves below a survival threshold during severe drought, or if these resources become inaccessible through inhibition of stored starch conversion to sugar (mobilization failure), or if high xylem tension prevents generation of positive hydrostatic pressure flow in the xylem (phloem failure; Münch, 1930; Minchin & Lacointe, 2005; Höltta et al., 2009; Knoblauch & Peters, 2010; Sala et al., 2010; McDowell et al., 2011). Additionally, nonstructural carbohydrates (NSCs) are also actively maintained in plant tissue for use in cryoprotection and desiccation protection (Obendorf, 1997; Ogren et al., 1997; Bansal & Germino, 2008, 2009; Bansal et al., 2011), as well as signaling between cells and tissues (Chiou & Bush, 1998; Lalonde et al., 1999; Smeekens, 2000). The functional dynamics of NSCs in these processes could complicate the relationship between pools of carbohydrates and metabolic fluxes as water stress progresses (Ryan, 2011; Zeppel et al., 2011; Sala et al., 2012).
Although there are a few cases where tree NSCs show a decline in response to nonlethal drought (Körner, 2003; Sayer & Haywood, 2006; Anderegg, 2012), observations of increased NSCs during drought are much more common in the literature (Körner, 2003; Würth et al., 2005; Sala & Hoch, 2009; Galvez et al., 2011). Increased NSC probably results when growth and growth respiration demand for carbon is reduced faster than gross photosynthetic activity, a response known as sink limitation (Körner, 2003). Sink limitation may initially seem incompatible with drought-induced tree mortality by carbon starvation, a source limitation to survival (Leuzinger et al., 2009; Sala, 2009; Gruber et al., 2012). However, mechanisms underlying drought mortality probably differ from growth responses to nonlethal water stress as a result, in part, of the time-scales of response from different physiological processes (McDowell & Sevanto, 2010; Hoffmann et al., 2011). A simple conceptual framework suggests that, during drought, trees will initially experience sink limitation and increased NSC pools, but if drought persists, then source limitation will occur and NSCs will decrease or will become inaccessible through mobilization and/or phloem failure before death (Sala et al., 2010; McDowell et al., 2011). However, the sensitivity of these processes to external drivers is poorly understood. In several recent studies of lethal drought, carbohydrates were diminished for some species, but not others: relative to survivors, NSC was lower for drought-killed Pinus sylvestris in Spain (Galiano et al., 2011) and for near-dead seedlings of Nothofagus nitida, but not for Nothofagus dombeyi, in Chile (Piper, 2011), nor mature Populus tremuloides in Colorado, USA (Anderegg et al., 2012b).
Here we present an analysis of NSCs from leaf samples collected during the piñon pine drought mortality experiment at Biosphere 2 (Adams et al., 2009a) to more fully investigate potential mechanisms of tree drought-induced mortality associated with temperature change and address previously raised concerns about aspects of Adams et al. (2009a). Notably, our study is the first repeated measurement of tree NSC response during drought through mortality for trees under two temperature regimes. We hypothesized that NSCs would be reduced during drought, and that NSC trends would reflect changes in cumulative respiration such that elevated-temperature trees would experience a faster decline in foliar carbohydrates than ambient-temperature trees.
Materials and Methods
The Biosphere 2 piñon pine mortality experiment was conducted with small (mean height of 1.7 m), reproductively mature trees transplanted from northern New Mexico into 100-l containers. Estimating from the dimensions of these pots and field measurements of Pinus edulis (Engelm.) roots (Foxx & Tierney, 1987; Tierney & Foxx, 1987), transplanted trees had on average 7% of their lateral root spread and 39% of their vertical root spread. Ten trees were randomly assigned to two areas of the glasshouse set to near-ambient or elevated (by c. 4°C) growing season temperatures (for a full description of the experiment, see Adams et al., 2009a). After 4 months of irrigation to promote acclimation to these temperature regimes following transplant, we withheld water from five of the trees (randomly selected) in each temperature treatment, starting on 8 February 2008 (soil moisture is usually at maximum within a few weeks of this date; Breshears et al., 2009). One droughted tree in the elevated temperature treatment was accidentally watered during the experiment and was excluded from further analysis, resulting in a sample size of five ambient droughted trees, four elevated-temperature droughted trees, and five watered controls at each temperature. The drought treatment resulted in mortality of trees in both temperature treatments, defined as 90% foliar browning, which occurred within 1–2 wk after the first signs of foliar browning.
Physiological measurements and collection of foliar material were periodically made on the south-facing side of all trees throughout the time-course of the experiment. On the same day as foliar sampling, we measured pre-dawn stem water potential (Ψpd) by pressure chamber (PMS Instruments, Albany, OR, USA) and CO2 exchange with an LI-6400 portable photosynthesis system (Li-Cor Biosciences, Lincoln, NE, USA), before dawn to quantify respiration (R) and at midday for net photosynthesis (A). We measured stomatal conductance (gs) from c. 1 h after sunrise to sunset with a Model SC-1 leaf porometer (Decagon Devices, Pullman, WA, USA). For sampling of foliar tissue, we collected the most recent mature, green needles from a single branch on each tree, typically the same used in pre-dawn water potential measurements. At mortality, if water potential measurements were not possible because of brittle stems, needles were collected directly from these trees in the early morning. At-mortality needle material was sampled during the closest scheduled periodic collection just before, at, or just after mortality was documented. Needles were dried at 70°C and ground to a powder fine enough to pass a 40-mesh screen (0.42-mm sieve). Foliar material was analyzed for NSCs, defined as starch and soluble sugar (glucose, fructose, and sucrose), following the enzymatic digest and UV spectrophotometry methods modified from Hoch et al. (2002; Supporting Information Methods S1). Any NSC or component value result less than zero was excluded from further analysis.
We calculated NSC differentials by subtracting the NSC, starch, and soluble sugar content values for droughted trees from watered-tree means in each temperature treatment for each sampling date. Diurnal measurements of stomatal conductance were interpolated to 10 time intervals following Fritsch & Carlson (1980), which were then used to calculate daily means. Time-series data, including NSC totals, components, and differentials, stomatal conductance, photosynthesis, and respiration, were analyzed using repeated measures ANOVA (spss 19.0; IBM Corporation, Armonk, NY, USA; α = 0.05). Mean total NSC, sugar, and starch in droughted trees initially and at time of mortality for each tree were also analyzed with repeated measures ANOVA. For all repeated measures ANOVA tests, when sphericity assumptions were violated (determined by Mauchly's test for sphericity), we used the Huynh–Feldt correction to degrees of freedom of the F-statistic to determine the significance of results (Huynh & Feldt, 1976). Note that measurements ceased earlier for trees in the elevated-temperature drought treatment because these trees died earlier (Adams et al., 2009a).
We used structural equation modeling (SEM) to analyze causal relationships among water potential, gas exchange, and NSC constituents as a series of multiple linear regressions for each independent variable to calculate error terms and standardized partial correlation coefficients (spss 19.0; IBM Corporation; α = 0.05; Methods S1). We analyzed the model separately for data from trees in each of the four treatments: ambient watered control, elevated-temperature watered control, ambient-temperature drought, and elevated-temperature drought. Additionally, we analyzed the correlation of cumulative respiration (Adams et al., 2009a) with total NSCs, sugar, and starch for droughted trees in both temperature treatments separately and pooled (spss 19.0; IBM Corporation; α = 0.05).
At the start of the drought experiment, foliar total NSC concentrations were similar for droughted and watered trees (means 5.55 ± 0.33 and 4.96 ± 0.38% dry mass, respectively; P >0.05). For droughted trees in both temperature treatments we found a significant decline in NSC over the course of the experiment (Fig. 1; P < 0.05) and as a pairwise contrast between initial and at-mortality values, whether pooled between temperature treatments or analyzed separately (Fig. 2; P < 0.05). Mean NSC at mortality (pooled between temperature treatments) was 3.88 ± 0.49% dry mass, a reduction of 30% from initial measurements (Fig. 2). The decline in total NSCs in both analyses was related to a decrease in soluble sugar concentration, despite an apparent net increase in starch concentration (P <0.05). There was no effect of temperature on the declines in total NSC and sugar concentration (P >0.05; Figs 1, 2). However, post-hoc analysis of starch trends showed an initial decline in values followed by a significant increase in starch at week 12 for warmer droughted trees and at week 17 for ambient droughted trees (P <0.05; Fig. 1). A significant time × temperature interaction observed in the starch analysis indicates that this trend reversal, from a decline to an increase, occurred earlier in the warmer drought treatment than in the ambient drought treatment (P <0.05).
Foliar NSC of watered controls doubled during the course of the experiment, and was greater than drought tree means in weeks 6, 8, 12, 17, and 22 (P <0.05; Fig. 1). For sugar, we also observed a significant water × time interaction (P <0.05), although trends in sugar did not differ greatly between droughted trees and watered controls (Fig. 1). Sugar content in droughted trees was greater than that in watered controls at week 8, but was significantly lower than that in watered trees at weeks 15 and 22 (P <0.05). Similar to the trend in total NSCs in watered controls, starch concentrations increased in watered controls by between five and eight times the initial values during the course of the experiment (P <0.05). Foliar starch content was greater in watered controls than in droughted trees at weeks 6, 8, 15, 17, and 19 (P <0.05). There was no significant temperature effect on watered control tree NSC, sugar, or starch. When calculated as a differential (watered – droughted, within each temperature treatment), there were no differences between the temperature treatments in total NSC, sugar, or starch content (P >0.05). Total NSC and starch differentials increased over time (P <0.05), but sugar differentials were unchanged (P >0.05). Thus, although pooled data were very consistent between temperature treatments and within watering treatments, the timing of changes differed for some components, but not for others.
Repeated measures analysis revealed significant changes over time for Ψpd, gs, A, and R (P <0.001). There were no differences in Ψpd, gs, A, and R between temperature treatments for droughted trees (P >0.05), although when quantified cumulatively R was higher in warmer droughted trees (Adams et al., 2009a). Differences between watered controls and droughted trees (pooled between temperature treatments) were found for Ψpd, gs and A (P <0.05), but not R (P >0.05).
For watered control trees in both temperature treatments, SEM revealed positive relationships between Ψpd and gs (P <0.05; Fig. 4). A and gs were inversely related for ambient watered trees, but positively related for warmer watered trees, where A also had a positive effect on sugar concentration (P <0.05). Relationships found with SEM in droughted trees were similar between temperature treatments, where a positive effect of gs on Ψpd, positive relationships between gs and A, and a positive effect of A on sugar concentration were observed (P <0.05). Inverse effects of sugar on starch and of Ψpd on R were seen in ambient droughted trees, but not warmer droughted trees (P <0.05). Cumulative respiration was correlated positively with starch concentration for ambient droughted trees (R =0.36; P <0.05), but not with total NSCs or any other NSC component when the analysis was conducted separately on temperature treatments (Fig. S1). When droughted tree data were pooled from both temperature treatments, cumulative respiration was inversely correlated with sugar (R =−0.30; P <0.05) and positively correlated with starch (R =0.35; P <0.05), but not correlated with total NSCs (P >0.05; Figs S1, S2).
Foliar carbohydrate resources were diminished during drought as foliar NSCs declined by 30% through mortality (Figs 1, 2). This decline in foliar carbohydrates of droughted trees was related to a reduction in sugar concentration, despite an increase in starch. Droughted and watered trees had similar NSCs at the start of the experiment, but as the growing season progressed, foliar NSCs increased greatly for watered control trees. Increased starch concentration in watered trees was the primary driver of differences in total NSCs between watered and droughted trees. Together, the results indicate that NSC declined for droughted trees before mortality, was much lower in droughted trees than in watered controls, showed temperature sensitivity in the time-pattern that depended upon component, and was associated with changes in carbon efflux from the leaf (Adams et al., 2009a), collectively suggesting an important role of carbon metabolism in the physiological mechanism of mortality in this species. This should not be interpreted to diminish the potential role for hydraulic failure in the mortality mechanism (Plaut et al., 2012), as droughted trees in both temperature treatments also reached the threshold for catastrophic xylem dysfunction (inferred as Ψpd < −6 MPa) before mortality (Adams et al., 2009a). Ambient drought trees spent c. 7.5 wk on average below this threshold, while elevated-temperature drought trees persisted only c. 2.5 wk. However, we observed no temperature differences in hydraulic function of the xylem that predicted the differences in time to mortality.
Despite the link between cumulative respiration and reduced survival time under elevated temperatures (relative to ambient), we did not observe an earlier reduction in NSC pools in elevated-temperature trees. This may suggest that neither hydraulic failure nor foliar carbon metabolism entirely explained the observed temperature-sensitive difference in time to mortality during drought. However, the changes in the dynamics of the NSC components through time (decline in sugar, and decline followed by increase in starch through time) are indicative of shifts in carbon metabolism driven by temperature which are related to the tendency of plants to maintain positive carbon status at the primary sites of metabolic activity (Adams et al., 2009a). Our ability to partition the physiological cause of mortality is limited by the overall experimental design, which precluded detection of NSC trends in other plant organs and translocation among tissues. Improved assessment of carbohydrate status could be made through sampling NSCs in woody sink tissues (stems and roots) further from the sites of carbon fixation in the foliage (Landhäusser & Lieffers, 2012). However, sampling these tissues can cause cavitation and potentially exacerbate tree stress, and would have interfered with the primary objective of this experiment, the quantification of temperature differences in time to mortality. Creative experimental design will be required to further evaluate these metabolic and hydraulic processes.
Although we saw a 30% decline in foliar NSCs in droughted trees, there was not complete NSC depletion at mortality (Fig. 2). While the decline in total NSCs was related primarily to a decline in sugar concentrations, just < 50% of initial sugar still remained in the needles of droughted piñon pines at mortality. This result shows that, although carbohydrate resources were diminished during drought, they were not entirely exhausted at mortality. However, there is much uncertainty surrounding the definition of reserve exhaustion, as reduction to zero NSC under any physiological circumstances seems highly unlikely (McDowell, 2011). Although starch initially declined in droughted trees, starch was higher at mortality than pre-drought levels, a result that is not initially consistent with carbon starvation by reserve depletion. However, leaf carbon balance is a function of many coordinated changes in pools and fluxes, and the increase in starch late in the drought may indicate an inability to translocate sugar from the chloroplast or the entire cell (Sala et al., 2010). The mechanism for such feedback inhibition of photosynthesis leading to starch accumulation is fairly well known; foliar starch accumulation in the chloroplast occurs when sugar sink strength is low in the cytoplasm and RuBP regeneration is low (Kozlowski & Pallardy, 1997; Myers et al., 1999). Development of low sink strength in the droughted trees could have resulted from reduced phloem function, preventing the export of sugars to other tree tissues. Thus, RuBP regeneration was limited by a lack of phosphorus translocator exchange with sugar, leaving starch synthesis as the only path available to any assimilated or re-assimilated carbon. Our observation of increased foliar starch concentration concurrent with the lack of complete sugar depletion could indicate that reduced carbon sink strength in other tree tissues caused by multiple factors affecting growth led to a decrease in phloem flow and sugar export from the foliage as mortality approached. However, this explanation assumes that, during drought, the sink strength of other tissues exceeded that of the foliage. It is also possible that typical tree tissue sink-strength relationships shifted or reversed during this severe drought experiment and woody tissues became an NSC source to sink foliage. Our dynamic NSC results highlight the role of mobilization, translocation and phloem function during the mortality process as high xylem tension can disrupt whole-plant processes and affects carbon sink strength in nonfoliar tissues (Sala et al., 2010; McDowell et al., 2011). Determining carbohydrate source and sink tissues during severe drought and distinguishing among the causes of reduced carbon sink strength are future challenges in resolving the mechanism of tree mortality.
The incomplete depletion of NSCs upon mortality is not altogether unexpected, but highlights the importance of understanding the many carbon metabolic processes that affect NSC dynamics. One potential explanation for the lack of complete depletion of foliar NSCs could be related to maintenance of sugars for nonmetabolic functions that can lead to the decoupling of carbohydrate pools from the external carbon fluxes (Hill et al., 2011; Ryan, 2011; Zeppel et al., 2011; Sala et al., 2012; Wiley & Helliker, 2012). In studies of subalpine conifer seedlings at treeline, photosynthesis and respiration only explained c. 40–50% of the variation in NSCs (Bansal & Germino, 2008, 2009, 2010a), and NSC dynamics during the growing season were not attributed to growth (Bansal & Germino, 2010b). Instead, researchers suspected that environmental influences on active maintenance of sugars for cryoprotection were linked to NSC dynamics (Strimbeck et al., 2008). Sugars also aid in desiccation tolerance during drought through osmotic adjustment and stabilization of membranes and proteins, and their active maintenance for such protection in leaf tissue is well recognized (Obendorf, 1997; Ogren et al., 1997; Nelson & Bartels, 1998; Murakeozy et al., 2002; Oliver et al., 2011; Sergeant et al., 2011). In addition, the use of sugar for signaling of gene expression between tissues also contributes to its maintenance for nonmetabolic functions (Koch, 1996; Chiou & Bush, 1998; Lalonde et al., 1999; Smeekens, 2000; Hill et al., 2011). An actively maintained threshold in sugar may have contributed to the lack of complete carbohydrate exhaustion in piñon pines (Fig. 1), despite respiration demands and a lack of supply from photosynthesis (Fig. 3). Fully resolving the interrelated contribution of hydraulic failure and carbon starvation to drought mortality will require improved understanding of the active control of NSCs. Without estimates of NSC thresholds for maintenance of osmoprotection and signaling, the contribution of carbon starvation to mortality may not be easily determined.
Previously published results from this experiment showed that trees in the warmer drought treatment died c. 30% faster, and had higher cumulative respiration than the ambient-temperature droughted trees (Adams et al., 2009a). Therefore, we hypothesized that NSCs would decline earlier in warmer droughted trees than in the ambient droughted trees. However, there was no difference in total NSC and sugar decline between droughted temperature treatments, indicating that the respiration difference was not reflected in changes to foliar NSCs, but probably attributable to a component that we did not have the ability to measure. Moreover, there was no temperature difference in NSCs between watered control tree treatments. The only significant difference in NSC trends between temperature treatments was that starch increased earlier for warmer than ambient droughted trees (Fig. 1), which could indicate earlier onset of mobilization constraints and phloem functional changes, or temperature-sensitive differences in photosynthetic feedback inhibition in foliar cells, as discussed above. Correlation of cumulative respiration with total NSCs and components was weak, and only significant for starch in ambient droughted trees or when data were pooled across temperatures for sugar and starch (Figs S1, S2). Correlations with cumulative respiration were negative for sugar and positive for starch, consistent with droughted tree trends in NSC components over time during the experiment (Fig. 1).
There were also no temperature-driven differences in transpiration or photosynthesis as these fluxes declined to near-zero before mortality in droughted trees (Fig. 3). Therefore, maintenance of equivalent foliar total NSC and sugar pools despite differences in respiration, but not photosynthesis, suggests changes to internal tree NSC dynamics. Potential explanatory processes include translocation of NSCs between other tree tissues and leaves, and conversion of other carbon macromolecules into NSCs (Hoch, 2007). Additionally, changes in the concentrations of other compounds within the leaf can cause NSC concentrations to shift even though the absolute abundance of NSCs remains constant (Bansal & Germino, 2008). For these processes to explain similar NSCs in all droughted trees during drought, they would have to occur differently between the ambient and warmer droughted trees. If the increase in foliar starch in droughted trees was the result of any form of sink limitation or translocation constraint, then its earlier occurrence in warmer droughted trees could explain the similar decline in sugar and total NSCs despite respiration differences between droughted tree temperature treatments.
Analysis (SEM) of tree physiology during the experiment revealed that respiration was decoupled from photosynthesis for both watered controls and droughted trees at both temperatures. The positive interrelationship between Ψpd and gs in watered control trees suggests that the physiology of these trees was sensitive to moisture, although A was not limited by gs (Fig. 4). Sugar content responded positively to increased A in warmer watered trees, but not in ambient watered trees. For droughted trees, the relationship between gs and A was stronger than for watered controls. However, the lack of an effect of Ψpd on gs in droughted trees is probably a result of the linear nature of the model: low Ψpd had a strong threshold-like influence on gs (Fig. 3). Photosynthesis was positively related to sugar concentration in droughted trees at both temperatures. However, the inverse relationship between sugar and starch content observed in the repeated measures analysis of trends (Fig. 1) was only significant for ambient droughted trees in the SEM analysis (Fig. 4). For watered control trees, SEM did not explain the increase of starch that these trees exhibited over the course of the experiment (Fig. 4). It is possible that the timing of our experiment, which decoupled temperature from daylength, extending the growing season, and transplant shock affected growth phenology and caused starch accumulation in watered trees (Kim et al., 1999).
Expanding physiological insight into the mechanism of drought-induced tree mortality from carbon exchange dynamics, we show that foliar carbohydrate resources were diminished, but not exhausted, for piñon pine during drought through mortality. Our results and those of Adams et al. (2009a) illustrate that the mechanism of tree drought mortality can be complex and highly temperature sensitive, incorporating multiple interrelated physiological processes. As xylem tensions in droughted trees in both of our temperature treatments simultaneously reached a threshold inferred to cause complete cavitation, hydraulic failure, while probably a cause of mortality, did not appear to explain the earlier mortality of elevated-temperature trees (Adams et al., 2009a). Rather, temperature sensitivity of mortality appears to be related to differences in carbon metabolism. Although we did not detect temperature differences in the decline, foliar NSC was reduced by 30% in droughted trees, while NSC increased in watered control trees. In addition, we observed increased cumulative respiration associated with temperature in these trees, along with shifts in the timing of change in some NSC pools. All of these suggest a role for carbon metabolism in the temperature sensitivity of drought mortality of these trees. Additionally, the functional dynamics of starch depletion and accumulation by treatment highlight temporal differences in carbon metabolism as potentially affected by whole-plant dynamics impacted by water status and temperature. We acknowledge that our results were limited by constraints in our experimental design which led to a lack of NSC assessment in stems and roots, where carbon starvation by reserve exhaustion is more likely (Landhäusser & Lieffers, 2012). Further physiological resolution of drought-induced mortality mechanisms will probably require increased experimental sophistication to determine active NSC maintenance, account for NSC translocation among tissues, link temperature sensitivity to physiological mechanism, and assess trends in whole-plant NSCs during drought through mortality. Consequently, temperature sensitivities need to be further evaluated for more tree species while simultaneously pursuing drought-induced mortality mechanisms. Nonetheless, our results support and refine the conclusions of Adams et al. (2009a) of a role of carbon metabolism in the temperature sensitivity of drought mortality, at least for P. edulis and perhaps in other physiologically similar species.
The authors would like to thank Biosphere 2 staff, research technicians, and interns for assistance with the experiment, and particularly Genna Gallas and Bhawika Sharma Lamichhane for help with sample preparation and NSC analysis. We thank Mohammad Torabi with Statistical Consulting for statistical assistance. We also thank Rebecca Minor, Andrew Moyes, David Tissue, Melanie Zeppel and four anonymous reviewers for helpful reviews of the manuscript, in addition to the community at large currently tackling this issue. This research was funded by the Philecology Foundation, US Department of Agriculture Cooperative State Research Education and Extension Service Grant 2005-38420-15809, US Department of Energy National Institute for Climate Change Research Grant DE-FC02-06ER64159, and US National Science Foundation Grants DEB-043526, EAR-0724985, and EPScoR 0814387. This publication was developed under STAR Fellowship Assistance Agreement no. FP-91717801-0 awarded by the US Environmental Protection Agency (EPA). It has not been formally reviewed by EPA. The views expressed in this publication are solely those of the authors, and EPA does not endorse any products or commercial services mentioned in this publication. This paper has been peer reviewed and approved for publication consistent with USGS Fundamental Science Practices (http://pubs.usgs.gov/circ/1367). Any use of trade, product, or firm names is for descriptive purposes only and does not imply endorsement by the US Government.