Introduction

Oxygen isotope ratios of cellulose provide a powerful signal to decipher paleoclimate and ecohydrological processes of extant ecosystems. This tool becomes even more powerful when applied to the chronosequence of tree rings, making it possible to accurately date changes in climate or hydrology in fossil or present day samples. Recently, and based on the oxygen isotope ratios of stem cellulose from trees spanning several latitudes, it has been proposed that there is a convergence of optimal photosynthetic temperature during tree trunk biomass production [1]. Helliker and Richter [1] showed that cellulose oxygen isotope enrichment relative to ambient water increases with a decrease in mean annual temperature (or at higher latitudes). Helliker and Richter [1] ascribed this increase in cellulose oxygen isotope enrichment to a greater leaf water enrichment at geographical locations with lower mean annual temperatures (M.A.T.) compared to semitropical and temperate regions. They further argued that the leaf-air water vapor pressure gradient would have to be greater at higher latitudes to maintain the hypothetical pattern of leaf water enrichment. To sustain this high vapor gradient at higher latitudes, leaf temperature would have to be disproportionally greater relative to ambient temperature at higher latitudes. After modeling the leaf water isotopic enrichment necessary to explain the cellulose oxygen isotope ratios observations, they calculated a remarkably constant average leaf temperature of 21.4°C throughout all the latitudes. One of the assumptions critical to their conclusion was that the biochemical fractionations associated with the imprinting of the water oxygen isotope signal in cellulose is constant regardless of temperature.

The oxygen isotope enrichment of cellulose from stems and tree trunks relative to ambient water can be described by the following equation [2]:(1)in which Δ_cellulose and Δ_leaf are the respective oxygen isotope enrichment of cellulose and leaf water relative to the source water. The proportional amount of stem water relative to both leaf and stem water in the tissue where cellulose is synthesized is denoted as p_x. Because the cellulose is usually extracted from stem or tree trunks with little possibility of having leaf water, we consider p_x to be 1. As sucrose is translocated from the leaf to the stem for cellulose synthesis it exchanges approximately 40% of its oxygen with stem water before being converted to cellulose [3], [4], [5], [6]. This biochemical exchange between the oxygen in carbohydrates and water during cellulose synthesis is denoted as p_ex. This exchange occurs primarily because carbonyl oxygen exchange with the oxygen in water during carbohydrate metabolism subject to an isotopic equilibrium effect [3], [7]. This exchange can be quite rapid for triose phosphates, but slower for pentoses and hexoses [8], [9]. One factor greatly contributing to this exchange is the futile cycle which occurs in the cytoplasm where glucose-6 phosphate can cycle quickly through fructose-1,6 Bisphosphate and triose phosphates [10]. Although p_ex indicates the extent of exchange during cellulose synthesis, it does not specify the actual average isotopic fractionation for this process. The average biochemical fractionation factor for this exchange as well as other reactions leading to cellulose synthesis is specified be ε_bio. The above equation (1) can be thought of as being composed of two major biological processes: physiological processes (Δ_leaf and p_x) and biochemical processes (p_ex and ε_bio) [11]. Helliker and Richter [1] assumed that one of the physiological processes (Δ_leaf.) in equation 1 increased with lower mean annual temperature in order to account for the increase in the oxygen isotopic enrichment of cellulose relative to source water. Although the authors added some variability in the biochemical factors (p_ex and ε_bio), they were considered constant regardless of latitude (temperature). Here we test whether the biochemical components of the above equation (p_ex and ε_bio) could be altered by temperature. For example, p_ex, related to the above mentioned futile cycle, could decrease with lower temperatures. This would lower the carbonyl exchange reactions occurring at the tree trunk during cellulose synthesis, and promote a greater isotopically enriched leaf water signal in the cellulose synthesized in the tree trunk and a plausible explanation of the observed increase in Δ_cellulose at lower temperatures. The other biochemical component, ε_bio, involves an equilibrium reaction which could also be subject to temperature effects. Lower temperatures tend to increase the equilibrium fractionations non-linearly [12], which could also explain the observed pattern of oxygen isotope ratios in stem cellulose with latitude. We know of only two systematic studies of the temperature effects on the biochemical oxygen isotope fractionation during cellulose synthesis [13], [14]. The first one was a survey of aquatic marine plants ranging from Puerto Rico, the southernmost sample set, to the most northerly in Woods Hole, Massachusetts. In addition, the authors of this study grew plants in aquaria maintained at specified temperatures. Although there seemed to be a temperature effect on the wild collected plants, the aquaria experiment yielded mixed results. The other study is a dissertation [14] where 6 different plant species were grown at 20°C or 30°C daytime temperature and analyzed for oxygen isotope fractionation. The authors found no consistent pattern for each species. The 20 to 30°C temperature range is relatively small considering that mean annual temperatures in some locations where trees grow can be in the range of −5°C. We note that many of the plants included in the Helliker & Richter [1] study were growing in ambient temperatures below 20°C.

Here we test for temperature effects in the exchange rate (p_ex) and biochemical fractionation (ε_bio) in a previously used system which mimics synthesis of cellulose in a tree trunk: cellulose synthesis in germinating wheat seeds in complete darkness [15], [16]. During germination of wheat seeds, starch, the primary carbon storage of the seed, is broken down to glucose which is principally synthesized to sucrose [17]. The sucrose is then translocated to the site of cellulose synthesis and used as a substrate. Previous measurements show a good agreement between p_ex measured with this system and those measured in wild and hydroponically grown trees [5], [6]. We germinated wheat seeds in the dark with waters showing different ¹⁸O enrichment at various temperatures. After growth we extracted cellulose and water from the culture flasks and observed the relationship between p_ex and ε_bio versus temperature. Specifically, we tested the hypothesis that p_ex will decrease and ε_bio will increase with lower temperatures. We then used these results to test whether a temperature dependent ε_bio is consistent with those observed in aquatic plants growing at different temperatures [13], [18], [19] and if it might lead to a different conclusion than that of Helliker and Richter [1] regarding convergent leaf temperature.

Materials and Methods

Data Analysis

Calculation of p_ex and ε_bio from wheat germination experiments.

We regressed the δ¹⁸O values of cellulose versus that of the media water for each temperature and compared the slopes (denoted here as p_ex) of each relationship for different temperatures using an ANCOVA test. The comparison of slopes between temperatures was Bonferroni corrected for 15 comparisons.

Since we observed only one significant difference between the slopes of the above linear regressions and no significant trend with temperature, we used a value of p_ex of 0.42, the average of all slopes observed here and in other literature (see Cernusak et al. [6] for a compilation of studies) to calculate ε_bio with the following equation:(3)Where ε_bio (i,j) is the average biochemical fractionation for the water isotopic enrichment level i and the temperature j and δ_cell (i,j) and δ_water (i,j) are the oxygen isotope ratios of cellulose and water for the seedling culture with water enrichment i and temperature j. If ¹⁸O is randomly distributed in the glucose moieties of the starch substrate, one could determine the value of the non-exchangeable oxygen (δ_NE) by analyzing the starch substrate in the wheat seeds before germination. However, previous studies indicate that ¹⁸O is probably not randomly distributed in the glucose moieties of starch [16]. We, therefore chose a value of 33.4‰ for δ_NE, which is similar to previous measurements [16] and gives the frequently observed value of ε_bio of 27‰ at 25°C. If we erred in the value of δ_NE, our absolute values of fractionations are incorrect, but the observed trends with temperature are robust. We opted for the above method of calculating ε_bio, rather than by the intercept method of the linear regression between δ_cell versus δ_water for a given temperature [3], as the intercept is highly biased towards cellulose grown in water with isotope ratios closer to 0. However, we show the results of the intercept method of calculation in Table S1. By the calculation method used here all cellulose cultures, regardless of water isotopic enrichment level, at a given temperature share equally towards the calculation of ε_bio. We tested for temperature and water enrichment effects on ε_bio with a two way ANOVA to determine if there were significant differences between ε_bio at different temperatures or/and at different culture water isotopic enrichment values. We derived an equation describing the relationship between average ε_bio and temperature by a regression analysis of the average of ε_bio for each temperature on temperature and calculating a best fit 2^nd order polynomial equation.

Calculation of Δ_cell for Aquatic plants.

We selected three studies where oxygen isotope ratios of cellulose from aquatic plants were determined along with the lake/ocean water they grew in at several geographical locations [13], [18], [19]. Two of these studies [13], [18] also grew aquatic plants under different temperatures in aquaria, but aquarium data was only used from the Sauer et al. study [18], since only in this study it was certain that biomass was produced during the temperature treatment and no reserves from its previous growth were used to produce biomass. For the Sternberg study [19], lake water and the cellulose oxygen isotope ratios as well as the specific location and altitude of sampling were given. We derived temperature from the closest weather stations. In the case of Andean samples in the Sternberg [19] study, we used data from the closest station and corrected for altitude differences using the temperature lapse rate of intermediate to dry and saturated air. For the Sauer study [19] we assumed that plants were operating at a minimum temperature of 5°C. This is the optimal photosynthetic temperature observed in Antarctic lichen [22]. The DeNiro and Epstein survey [13] provided the growing season temperature for each site where marine plants were collected. Further information on these samples can be found in the supplementary material (Table S2). Since there is no transpiration in aquatic plants Δ_cell = ε_bio. We subtracted the δ¹⁸O values of lake/sea water/aquarium water from those of cellulose to derive the Δ_cell, plotted Δ_cell versus water temperature, and used regression analysis to calculate the best fit 2^nd order polynomial equation.

Re-evaluation of Tree Trunk Cellulose Oxygen Isotope data.

We used data reported in the supplemental of Helliker and Richter [1] to calculate leaf water oxygen isotope enrichment relative to source water (Δ_leaf) under two assumptions: 1) the ε_bio is constant and equal to 27‰, as it was assumed in their study, and 2) ε_bio varies with temperature according to our polynomial fit. We used equation 1 to calculate Δ_leaf by rearranging the terms of equation 1:(4)For the calculations in which ε_bio varied with temperature, we assumed both leaf and trunk had the same average temperature as the mean annual temperature. If the M.A.T. was below 5°C, we assumed values of ε_bio to be that at 5°C. In the above calculation we exclude three points out of 70 in a site which in the Helliker and Richter [1] study were shown to be outliers. We compared the above values of Δ_leaf with those actually measured in a global samples set from the Moisture Isotopes the Biosphere and Atmosphere (MIBA) network sponsored by the International Atomic Energy Agency [23]. The description and sampling months for each of the sites are shown in Table 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Leaf and Stem water sampling stations of the MIBA network used to calculate isotopic enrichment of leaf water in Figure 3.

https://doi.org/10.1371/journal.pone.0028040.t001

Results

The δ¹⁸O values of cellulose from the seed germination experiment were highly correlated with those of the water having a slope (p_ex) averaging 0.42±.01 and an average intercept of 31.03±0.21 (Table 2; see also Table S1). The slopes of the cultures at 5 and 10°C were slightly higher than those of higher temperature, but the differences were not significant (Table 2). After the Bonferroni correction, only slopes for cultures growing at 15 and 10°C were significantly different from each other. There were no significant differences between slopes from cultures growing at the extremes of temperature (5 and 30°C). A correlation analysis between slope and temperature did not show a significant trend in slope as a function of temperature (r = 0.73, P>0.05).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Linear regression parameters between oxygen isotope ratios of seedling cellulose and culture water.

https://doi.org/10.1371/journal.pone.0028040.t002

The average biochemical oxygen isotope fractionation for the seedling growth experiment varied from ∼25‰ at the highest temperature to ∼31‰ for the lowest temperature. The average ε_bio varied less at temperatures between 20 and 30°C and increased more steeply at lower temperatures (Figure 1). There was a significant effect of temperature in the ε_bio (F = 5.85, P = <0.01) but no effect of the culture water isotopic enrichment level (F = 0.7, P>0.05). A best fit polynomial regression of average ε_bio on temperature gave a significant fit (r = 0.92, P<0.05).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. ε_bio versus temperature.

Average biochemical oxygen isotopic fractionation (ε_bio±S.E.M., n = 5) in ‰ units for cellulose synthesized heterotrophically in wheat seedlings versus temperature (°C).

https://doi.org/10.1371/journal.pone.0028040.g001

The biochemical oxygen isotope fractionation of aquatic and marine plants and organisms varied from 26 to 31‰ from lakes at lower latitudes having temperatures close to 30°C to lakes at cooler regions having temperature of 5°C respectively (Table S2). The fractionation was highly correlated with temperature yielding a polynomial regression equation similar to the one derived for the seedling growth experiment (Figure 2; also see Figure S1 for specific locations).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Aquatic Δ_cell versus temperature.

Cellulose isotopic enrichment relative to source water (‰ units) for aquatic plants grown submerged from various studies [13], [18], [19]. Information for each data point is provided in the supplement (Table 2). Bold line is the best fit polynomial regression while the stippled line is for the best fit from wheat seedling experiment (Figure 1).

https://doi.org/10.1371/journal.pone.0028040.g002

As expected and as proposed by Helliker & Richter [1] the Δ_leaf increases with lower M.A.T. when one assumes a constant ε_bio of 27‰ (Figure 3, black circles). On the other hand, Δ_leaf calculated with a temperature sensitive ε_bio indicated a constant Δ_leaf averaging (17.1±0.3‰, empty circles) regardless of M.A.T. Observed Δ_leaf from the MIBA samples indicated no specific trend as well (Figure 3, red circles; also see Table S3 and Figure S2 for specific location), with a greater degree of scatter and lower enrichment values averaging (12.2±1.3‰, red circles), but with many values matching the predictions of Δ_leaf calculated with the temperature sensitive ε_bio.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 3. Δ_leaf versus mean annual temperature.

Oxygen isotopic enrichment of leaf water relative to source water based on three types of calculations as described in the text: (•) based on a constant ε_bio of 27‰; (○) based on a temperature sensitive ε_bio adjusted to M.A.T.; (•) based on actual leaf water and stem water measurements. The relationship between Δ_leaf and M.A.T. was only significant with Δ_leaf calculated with a constant ε_bio.

https://doi.org/10.1371/journal.pone.0028040.g003

Discussion

Our first hypothesis regarding differences in p_ex with temperatures was shown false (Table 2). Significant temperature effects in p_ex were only observed between 10 and 15°C, nor was there a significant trend in p_ex as a function of temperature. One would expect that since p_ex is a function of the futile cycle: glucose – 6 P <$>\raster(70%)="rg1"<$>fructose 1–6 P <$>\raster(70%)="rg1"<$>Gylceraldehyde-3 P +Dihydroxyacetone P; that it would occur to a greater extent at higher temperatures than lower temperatures. However, what actually determines the extent of exchange during this futile cycle would be the lifetime of each of these components, particularly the triose phosphates; i.e. the greater the lifetime, the greater the exchange. Although high temperatures could increase the reaction rates which lead to the above intermediates, it will also drive respiratory rates and possibly cellulose synthesis up, which consumes these intermediates and decreases the lifetime of the exchange intermediates and the oxygen isotope exchange. It is likely that any increase in the futile cycle rate is offset by a decrease in the lifetime of the futile cycle intermediates, thus causing p_ex to remain constant with temperature.

In contrast to p_ex, there was a significant increase in ε_bio with a decrease in temperature (Figure 1). This would be the expected pattern for an equilibrium reaction. One of the principal reactions responsible for the exchange of oxygen between carbohydrates and water during cellulose synthesis is the carbonyl hydration reaction [3], [7]. Previous measurements for the isotopic fractionation during the carbonyl hydration reaction in a model molecule (acetone) indicates values similar to the fractionation between cellulose and water as well as a significant temperature effect at higher temperatures [7]. Equilibrium fractionation factors are often a direct function of the ratio of the partition coefficients for each isotopes which become closer to each other with an increase in temperature [12], hence the observed decrease in fractionation with higher temperature for equilibrium fractionation factors (see Clark & Fritz [24] for a compilation of equilibrium factors). Our results explain why a previous study using 6 different species of plants failed to observe a temperature effect [14]. Zhou [14] observed no temperature effect for ε_bio for plants grown at 30 and 20°C. This range of temperature is precisely the range of temperature with the least effect on the biochemical fractionation (Figure 1).

Published observations of oxygen isotope enrichment of cellulose relative to ambient water of aquatic plants at different temperatures are consistent with our experimental results; there is an increase in enrichment with a decrease in temperature (Figure 2). In the case of aquatic plants, because they are submerged and there is no transpiration, the oxygen isotope enrichment of cellulose relative to the source water is the same as the biochemical fractionation (i.e. Δ_leaf in equation 1 is equal to 0). The best fit relationship is also a second order polynomial fit having r = 0.91 and p<0.01 compared to the linear fit (r = 0.88and p<.01). The polynomial fit for the ε_bio from the seed germination experiments fits remarkably well with the relationship observed for aquatic plants (Figure 2). Because the Calvin cycle end product (Glyceraldehyde-3 P) is one of the intermediates of the futile cycle mentioned above, it is likely that biochemical fractionation for heterotrophic and autotrophic cellulose synthesis are very similar, as it is often assumed in the literature [5]. Measurements of translocated sucrose which is autotrophically synthesized, indeed, show that the oxygen isotope ratios of autotrophically synthesized sucrose is ∼27‰ more enriched than that of the leaf lamina water and similar to heterotrophic cellulose synthesis [25].

Estimation of leaf water ¹⁸O enrichment based on a constant ε_bio of 27‰ with the previously published data set predict that Δ_leaf would increase with lower M.A.T. (r = 0.82, p<0.01). Helliker & Richter [1], therefore, invoked a greater vapor pressure gradient in colder regions, which can only be achieved by a greater increase in leaf temperature compared to ambient temperature during biomass production. However, when we calculate Δ_leaf using the temperature sensitive ε_bio, no trends of Δ_leaf in relation to M.A.T. (r = 0.24, p>0.05) were observed. The Δ_leaf extrapolated using the temperature sensitive ε_bio is similar to those calculated with the constant ε_bio of 27‰ at warmer temperatures but lower at cooler temperatures. The calculations of actual Δ_leaf based on a global sample set initiated by the International Atomic Energy Agency (MIBA) gives an even greater discrepancy between observed Δ_leaf as a function of M.A.T. and those based on a constant ε_bio, although there were several values similar to those extrapolated based on a temperature sensitive ε_bio. We have no experimental data to explain some of the lower than expected values of leaf water isotope enrichment for the MIBA samples, but speculate here that lower values were caused by the presences of non lamina leaf water in the MIBA samples. Previous studies showed a good correlation between the oxygen isotope ratios of phloem carbohydrates with lamina water. The authors of this study [25] went through considerable trouble to remove the main veins from leaves and only extract lamina water. Although the leaf sampling protocol for MIBA advises the removal of petiole and the main vein, it is likely that second order veins were not removed during leaf sampling leading to a more depleted leaf water values than the leaf lamina. In addition, because the MIBA samples involved spot sampling it would likely lead to greater variation in the isotopic composition than that extrapolated from tree trunk cellulose, which integrates leaf water over a considerable period of time and photosynthesis occurring throughout the whole plant.

Our data on the temperature sensitivity of the biochemical fractionation during cellulose synthesis indicates that there is an increase in ε_bio with lower temperature. Hence, it is not necessary to invoke greater leaf water enrichment at colder habitat to explain the greater cellulose isotopic enrichment at high latitudes. An analysis of leaf and stem water samples from several latitudes, indeed, show no trending increase in leaf isotopic enrichment with a decrease in M.A.T. It is well recognized that photosynthesis in plants is limited to a temperature window. For example, plants in boreal climates do not have significant photosynthetic biomass production during the winter. Likewise, it is also well known that in hot climates plant photosynthetic responses include the so called “mid-day depression”, with stomatal closure and lower photosynthesis, which avoids the hotter and dryer times of the day [26]. However, there is also a large body of literature showing that plants native to a colder habitat will have a lower temperature photosynthetic optima [22]. Further, many plants can acclimate to different temperatures so as to have optimal photosynthesis over a wide range of temperatures [22]. Acclimation is an important and likely costly evolutionary process which would be wasted if plants produce the bulk of their biomass at a constant temperature of 21.4°C across a wide range of biomes.

Supporting Information