Jasmonic acid induces rapid changes in carbon transport and partitioning in Populus

Authors


Author for correspondence: Benjamin A. Babst Tel: +1 617 627 2758 Fax: +1 617 627 3805 Email: ben.babst@tufts.edu

Summary

  • • Here, we tested whether rapid changes in carbohydrate transport and partitioning to storage organs would be induced by jasmonic acid (JA), a plant-produced signal of herbivore attack known to induce resistance.
  • • Carbon-11, introduced as 11CO2, was used to track real-time carbohydrate transport and partitioning nondestructively in Populus species before and 12 h after application of JA to a single leaf.
  • • Jasmonic acid resulted in more rapid [11C]-photosynthate export from both local and systemic leaves, as well as greater partitioning of [11C]-photosynthate to the stem and roots. In Populus tremuloides, following JA treatment, leaf starch decreased, but there was no change in photosynthetic rates or leaf soluble sugar concentration, indicating that recent photosynthate was diverted from starch accumulation in the leaf to other plant organs.
  • • Increasing the supply of photosynthate to roots and stems may shield resources from folivorous predators, and may also facilitate both storage and nutrient uptake, and ultimately lead to greater tolerance, either by enhancing regrowth capacity or by replacing nutrients consumed by herbivores.

Introduction

Many plants have defensive responses to herbivory, rapidly increasing secondary defensive compounds that are toxic and/or repellent (Karban & Baldwin, 1997), as well as mechanical mechanisms such as toughness or trichomes (Traw & Dawson, 2002; Dalin & Björkman, 2003). Much evidence suggests that induced resistance can be attributed to specific signaling molecules or pathways (Creelman & Mullet, 1997), initiated by herbivore chewing and saliva (Halitschke et al., 2001). Most studies have examined induced resistance over a relatively short time, although there is evidence that the responses can be longer lasting (Thaler et al., 1996), and can influence the community of herbivores that can damage the plant (Kessler et al., 2004).

Recently, gene expression studies suggest that plant responses to herbivory may be much more complex than previously thought, perhaps leading to an entire metabolic reorganization (Hui et al., 2003). Indeed, in longer term studies plants may respond to herbivory by increasing photosynthesis or by reallocating stored resources to growing shoots to compensate for lost or damaged photosynthetic tissues (Strauss & Agrawal, 1999). In most of these long-term studies, plants were manually defoliated, rather than exposed to real herbivory or to defense-signaling elicitors. Hence, it is unclear whether these are direct responses to herbivore cues mediated by signaling pathways, or indirect responses to the imbalance in root–shoot allocation following leaf tissue loss.

Although photosynthetic rates may increase in remaining leaves immediately following herbivory or defoliation (Bassman & Dickmann, 1982; Houle & Simard, 1996; Mabry & Wayne, 1997), there has been relatively little exploration of whole-plant physiological responses immediately following herbivory. Partitioning of resources within the plant is of particular interest because changes in partitioning among organs may provide insight into changing patterns of resource investment in different plant functions. For example, in grasses there is evidence that moderate levels of herbivory can increase transport of resources from leaves to roots (Holland, 1995), increasing root respiration and nutrient uptake (Holland, 1996). In poplar, manual defoliation resulted in changes in carbon partitioning both over the long term and short term (Bassman & Dickmann, 1985). Whether resistance-inducing signal molecules can elicit changes in whole-plant partitioning has not, however, been tested previously.

We hypothesize that changes in resource partitioning and allocation may be specifically induced by herbivore cues, such as those that induce resistance (Creelman & Mullet, 1997). The discovery that jasmonates can mimic induction of plant resistance by herbivores (Baldwin, 1996; Thaler et al., 1996) provides us with the ability to separate the effects of tissue removal from the effects of induction. Because Populus species are known to increase secondary defensive chemicals in response to jasmonates (Constabel et al., 2000; Arnold & Schultz, 2002), we examined the effects of exogenous jasmonic acid (JA) treatment on carbon transport and partitioning in Populus tremuloides and Populus nigra (clone NC5271) using carbon-11 tracer. The high-energy emissions and rapid decay of carbon-11 allowed us to examine the partitioning and transport dynamics of recently fixed carbon noninvasively and repeatedly, and to determine whether plants may change their carbohydrate partitioning in response to JA, even in the absence of tissue removal.

Materials and Methods

Plant material

We used P. tremuloides and P. nigra clone NC5271 for these studies. Populus tremuloides seeds were sown in flats of fafards #2 potting soil (Fafard, Agawam, MA, USA) in early spring in a glasshouse the same year as the experiments. When seedlings were 3–5 cm tall, they were transferred to 2.83-l ‘tall-one’ treepots (Stuewe & Sons, Corvalis, OR, USA), and were used in experiments when about 0.35 m high (c. 60 d after transplanting). For root imaging, a subset of P. tremuloides seedlings were transferred to hydroponics (Hoaglands nutrient solution) several weeks before the experiments. The P. nigra clone was chosen for its inducible resistance against gypsy moth caterpillars (Havill & Raffa, 1999; for lineage description see Robison & Raffa, 1994; Havill & Raffa, 1999). Cuttings of the P. nigra clone were dipped in TakeRoot 0.1% indole-3-butyric acid (Schultz, St Louis, MO, USA) and rooted in fafards #2 potting soil in 2.83-l ‘tall-one’ treepots. P. nigra clones were used for experiments approximately 45–60 d after sprouting, when stems were approximately 0.4 m in height.

Jasmonic acid induction protocol

Jasmonates are often applied to the whole plant in herbivory-related studies (Thaler et al., 1996; Agrawal et al., 1999). To more closely mimic herbivory, JA (1 mm) in 1% aq. acetone and 0.125% (v : v) Triton-X 100 (as in Arnold & Schultz, 2002) was sprayed onto a single leaf at night when many herbivores feed (e.g. gypsy moth; Leonard, 1970) about 12 h before post-treatment measurements (one plant was tested after 6 h). The same solution without JA, 1% aqueous acetone with 0.125% (v : v) Triton-X 100, was sprayed on a single leaf for control plants. We saw no evidence of decreased photosynthesis (see the Results section) that would indicate a stress-type response, or signs of early leaf senescence. Further, preliminary evidence suggests that low-level herbivory elicits a similar response as our JA treatment (B. A. Babst et al. unpublished), indicating that our JA treatment was in the appropriate range to mimic JA-signaling following herbivory. The treated leaf was the focus of measurements in experiments designed to look at local effects of JA. To test for systemic responses, an orthostichous leaf younger than the treated leaf was targeted (Fig. 1a,b).

Figure 1.

Diagrams of experimental set-up for 11C experiments and sample data set used to calculate transport parameters. (a) Local induction was tested by applying 1 mm jasmonic acid (JA) to the same leaf that would be exposed to 11C. (b) Systemic induction was tested using 11C exposure to a leaf younger than the JA-treated leaf. In both cases, two detectors (d1 and d2) were placed on each stem below the 11C-target leaf. (c) As [11C]-photosynthate was transported through the stem, the radiation was detected. Radioactivity counts were decay-corrected and used to calculate export speed, transport speed, and loading time. The time at which radioactivity reached half of its maximum was used to calculate transport times and speeds, because the half-max gave a more precise and objective time than when the curve began to rise or when it reached its maximum.

Transport and partitioning of 11C-labeled photosynthate

Carbon dioxide labeled with 11C (t1/2 = 20.4 min) was produced by irradiating a nitrogen gas (N2) target with 17.4 MeV protons on the BNL 41-inch cyclotron (Japan Steel Works, Inc., Tokyo, Japan) to induce the 14N(p, α)11C nuclear transformation (Ferrieri & Wolf, 1983). The 11CO2 was pulse-fed to the target leaf for 30 s inside of a photosynthesis cuvette fitted with red/blue LED lighting (PAR = 600 µmol m−2 s−1), and chased with ordinary air (Ferrieri et al., 2005). The emission of radioactivity by the 11C-labeled plants requires special containment, limiting analysis to a single plant at a time. This reduces typical sample sizes to two to four plants per treatment (Dyer et al., 1991; Minchin et al., 2002).

One or two cadmium-telluride gamma radiation detectors (Radiation Monitoring Devices, Inc., Watertown, MA, USA) were positioned on the stems of plants below the focal leaf (Fig. 1a), which allowed us to calculate carbon-11 export speed, loading time, and transport speed through the phloem. Export speed was determined by dividing the distance between the leaf and the first detector (d1) by the time between 11C pulse and 11C arrival at d1 (see Fig. 1c for explanation of d1 and d2). Arrival at the detector was defined as the time at which the radiation was at half of its maximum because the half-max gave a more precise and objective time than the first discernible rise in radiation (Fig. 1c; also see Ferrieri et al., 2005). If the radiation had not reached a plateau by the end of the 4000-s measurement window, the radiation at 4000 s was averaged over the final 10 min and that average used as the maximum. Export speed was broken into two components, transport speed and loading time.

Transport speed through the phloem was calculated by dividing the distance between the two detectors by the time of 11C transport between the two detectors (see ‘d1-d2 time’ in Fig. 1c). Loading time, which includes all processes between exposure to 11CO2 through phloem loading, was calculated by multiplying d1–d2 transport speed by the distance between the leaf and the first detector to obtain the leaf-to-detector transport time, and then subtracting that out from the time between the 11C pulse and 11C arrival at the first detector.

The distribution of radiolabelled photosynthate was detected 2 h after 11C tracer administration using phosphor plate imaging of positron emissions, and quantified using Science Laboratory 99 Image Gauge software (Fuji Photo Film Co., Ltd, Tokyo, Japan). Since each JA- and control-treated plant was used as its own control, no correction was necessary for differences in tissue absorption of positrons. A partitioning ratio (basipetal activity divided by acropetal activity) was compared between baseline and post-treatment for each plant (JA, n = 7; control spray, n = 6). To maximize our replication for partitioning in potted plants, several additional control plants (n = 3) were treated with JA on the second night and measured again for a third day. For hydroponic plants, because the entire plant was imaged, partitioning was calculated as the per cent of whole-plant radioactivity.

Leaf properties

A LiCor infrared gas analyser (Model 6262; Li-Cor Inc., Lincoln, NE, USA) was connected in-line with the 11CO2 pulsing system to measure photosynthetic rate of the focal leaf immediately before 11CO2 pulse-exposure.

The two major carbon pools in the leaf, sugars and starch, were measured in a second experiment for JA-treated and control seedlings of P. tremuloides. To test for changes in sugar and starch content a single leaf of each plant was sprayed with either JA or a control spray in the morning, and the treated leaves were harvested 8 h later, dried, and ground with a ball mill to a fine powder. Leaf powder was extracted in 80% aqueous methanol and supernatant was analysed by the phenol–sulfuric acid method to measure soluble sugars (Dubois et al., 1956). Absorbance was measured at 490 nm using a microtiter plate reader (Bio-Rad, Hercules, CA, USA), and d-glucose was used as the standard. To measure starch, the pellets were digested overnight at 55°C with amyloglucosidase (Haissig & Dickson, 1979), and the resulting free sugars were quantified as described above using the phenol-sulfuric acid method.

Statistical analysis

Statistical comparisons were made between baseline and post-treatment measurements using a paired t-test, where nondestructive measures made this possible. Since the partitioning data with potted plants had a skewed distribution, partitioning ratios were log-transformed for paired t-test comparisons between pre- and post-treatment partitioning ratios, testing the response to JA and the response to the control spray in separate tests. For transport and partitioning, there were not enough P. nigra replicates to perform a separate statistical analysis on local and systemic treatments. Local and systemic plants were combined to allow statistical testing for a P. nigra response to JA treatment. Since partitioning data for hydroponic plants were expressed as per cent of whole-plant activity, data were arcsine transformed for statistical analysis. Comparisons of leaf traits between treatments were made using the t-test. All statistics were done with systat 10 (SPSS Inc., Chicago, IL, USA).

Results

JA Effects on 11C transport

Following JA application, radiation was detected in the stem sooner after 11CO2 labeling and at higher levels than were observed before JA application for both P. tremuloides and P. nigra (see Fig. 2a for a representative example in P. tremuloides). When 11CO2 was applied to JA-treated leaf, export speed of 11C increased in response to JA but not in response to a control spray without JA for P. tremuloides and the P. nigra clone (Fig. 2b). For a separate set of P. nigra plants, 11CO2–labeled leaves distant from the JA-treated leaf showed a response similar to the treated leaves, indicating such responses can be systemic (Fig. 2b). Separate statistical analyses could not be performed on local and systemically labeled P. nigra plants separately because the sample size was two plants each. After combining systemic and local plants into one JA treatment, the response to JA by the P. nigra clone is significant (T = −9.5, df = 3, P = 0.002; Fig. 2b).

Figure 2.

Effect of jasmonic acid (JA) on 11C transport in Populus tremuloides and Populus nigra. (a) Radioactivity was measured at a fixed point on the lower stem, as shown in this representative set of traces from a P. tremuloides seedling before treatment (open circles) and after JA spray (closed circles). (b) Export speed was increased in plants following JA application (shaded bars) above baseline levels (open bars) measured before treatment (n = 3–4 plants). (c) Transport speed was unaffected by JA, but loading time, which is the time of 11C transit through the leaf to the phloem, was reduced following JA application (closed bars) compared with baseline measurements (open bars) before treatment (n = 2 plants). Error bars are standard errors. A paired t-test comparing baseline with post-treatment was made where sample size was at least 3. Data for P. nigra local and systemic plants, which responded similarly, were pooled (n = 4 plants) to allow statistical test of a JA effect; *P < 0.05, **P < 0.01.

Export speed incorporates many components, including all of the processes from photosynthesis, to photosynthate transport through the leaf mesophyll, to phloem loading, through phloem transport to the first detector. For a subset of the P. nigra plants, two detectors were set up on each plant to give (1) the transport speed through the phloem, and (2) the loading time, which is the transit time of 11C through the leaf from exposure through phloem loading. The JA treatment dramatically decreased loading time, but had little effect on transport speed through the phloem (Fig. 2c).

Change in leaf properties in response to JA: photosynthesis and carbohydrates

To further understand the mechanism by which JA increased export speed, we examined the responses of photosynthesis, leaf sugars, and leaf starch to JA. In our radioisotope experiments, JA had no significant effect on photosynthesis in either P. tremuloides or in the P. nigra clone (Table 1). In the second experiment with P. tremuloides only, JA resulted in lower foliar starch concentrations, but had no effect on foliar sugar concentrations (Table 2).

Table 1.  Changes in photosynthesis (µmol m−2 s−1) from day 1 (baseline) to day 2 (post-treatment) for control sprayed and jasmonic acid (JA)-treated plants in Populus tremuloides and Populus nigra
TreatmentP. tremuloidesP. nigra clone
Baselinepost-treatmentBaselinePost-treatment
  1. Data are means ± SE; n = 4 for controls and n = 5 for JA treatment. There were no significant changes in photosynthesis following treatment.

Control4.36 ± 1.104.14 ± 1.183.78 ± 0.324.23 ± 0.38
JA4.15 ± 0.893.72 ± 0.453.18 ± 0.364.11 ± 0.05
Table 2.  Effects of jasmonic acid (JA) on sugar and starch concentrations in Populus tremuloides leaves after 8 h
TreatmentSugar (mg g−1)Starch (mg g−1)
  1. Data are means ± SE; n = 5. Results in bold type are significantly different at P < 0.05 for comparison between control and JA treatment using a t-test.

Control129.5 ± 6.392.6 ± 4.3
JA129.3 ± 6.776.7 ± 4.6

Partitioning response to JA

The ratio of basipetal-to-acropetal radioactivity substantially increased after JA treatment, locally for P. tremuloides (T =−4.3, df = 4, P = 0.01), but not for plants treated with a control spray (T = −0.39, df = 3, P = 0.72; Fig. 3a). The partitioning ratio also increased for the P. nigra clone both locally (Fig. 3b) and systemically (Fig. 3c). When local and systemic P. nigra plants were combined to achieve a sufficient sample size for statistical analysis, the change in the ratio of basipetal-to-acropetal radioactivity in response to JA was statistically significant (T = −4.5, df = 4,P = 0.01), but was not significant in response to control spray (T = −8.9, df = 4, P = 0.42). The values of the basipetal-to-acropetal radioactivity ratios tended to be lower for the systemic leaves of the P. nigra clone. This was expected since these leaves were younger, and photosynthate is transported more acropetally from younger source leaves (Larson & Gordon, 1969). With hydroponically maintained P. tremuloides we could observe 11C partitioning within the whole plant (i.e. imaging showed tracer in the roots). Again, increased basipetal distribution of radioactivity was observed following JA treatment, with more 11C being transported to the lower stem and roots and less to the apex (Fig. 4).

Figure 3.

Effect of jasmonic acid (JA) on basipetal-to-acropetal partitioning of 11C. Individual plants are represented each by a line, with control plants on the left and JA-treated plants on the right column. (a) Local leaves of Populus tremuloides; (b) local leaves of Populus nigra clone; (c) systemic leaves of P. nigra clone. Distribution of activity within the plant was quantified using autoradiography 2 h after 11C administration. The partitioning ratio was calculated as 11C transported basipetally divided by 11C transported acropetally (11Cstem/11Capex). In every plant, the partitioning ratio was higher following application of JA. Paired t-test showed statistically significant responses to JA in P. tremuloides (t = −4.3, df = 4, P = 0.01) and in pooled local and systemic P. nigra plants (t = −4.5, df = 4, P = 0.01), but no significant effect of control spray for P. tremuloides (t = −0.39, df = 3, P = 0.72) or the P. nigra clone (t = −8.9, df = 4, P = 0.42).

Figure 4.

Jasmonic acid (JA) effects on whole plant partitioning of 11C for Populus tremuloides grown in hydroponic nutrient solution. Plants were removed from nutrient solution for phosphor plate imaging 2 h after a 30-s 11CO2 pulse. Representative images from one plant before and after JA treatment are shown here. The blue coloration indicates low radioactivity, with increasing radioactivity for yellow and then red. Partitioning was calculated as radiation in the apex, source leaf, lower stem or roots as a per cent of whole-plant radiation, and are shown as mean ± SE (n = 3). Percentages were arcsine transformed for statistical analysis with a paired t-test. Significant differences between baseline and post-treatment measurements are indicated by an asterisk (α = 0.05).

Discussion

We have demonstrated for the first time in Populus a previously unreported response to JA, a plant signal molecule involved in induction of resistance. As hypothesized, following treatment with JA, newly fixed carbon (i.e. 11C) was exported from mature leaves at a faster rate. Further, JA resulted in decreased leaf starch content and a greater partitioning of 11C to the lower stem and roots. This suggests that JA, a hormonal signal associated with herbivore attack, triggers a shunting of carbon below-ground, which might be stored and used later for regrowth or may be used to increase nutrient uptake. Further study is required to test for consequences of this change in partitioning.

Our data suggest some possible points of action for JA in effecting changes in carbon transport. It appears that the crucial changes occurred in leaves rather than in the plant's vascular system. The observed increase in export speed includes all processes from the time of 11CO2 pulsing until the 11C-photosynthate passes the radiation detector on the stem. Data from two detectors on the stem indicate that transport speed through the phloem was not altered by JA, but processes within the leaf decreased phloem loading time and caused the observed increase in export speed. Following photosynthesis, there are many steps carbon would pass through before reaching the phloem. An increase in the apoplastic supply of sugars could cause an increased rate of phloem loading directly (Minchin et al., 2002). However, we found neither an increase in photosynthesis in P. tremuloides or the P. nigra clone, nor an increase in sugars in P. tremuloides leaves, indicating that increased sugar concentrations are not necessary to drive the JA-induced decrease in loading time. Since sucrose is the major transport sugar in Populus, increased expression or activity of sucrose transporters could have speeded 11C delivery to the phloem. Within the chloroplasts of Populus, a substantial fraction of fixed carbon (up to 30%) is stored as starch during the day and then catabolized to sugars at night for export (Dickson, 1987). Late in the day, when our samples were taken, starch concentrations in leaves were maximal because night-time starch catabolism and export had not begun (Dickson, 1987). The low starch concentrations in the JA-treated leaves suggests a strong decrease in starch synthesis, with carbon having been diverted away from starch production and channeled directly through export pathways.

Past reports indicate that high levels of JA can induce senescence (Parthier, 1990), but lower concentrations of jasmonates act as signals without senescence or toxicity (Staswick et al., 1992). Unlike previous senescence experiments, which often use a continuous whole-plant exposure to high levels of jasmonates, we applied a one-time pulse of JA to a single leaf to more closely approximate local herbivory. The lack of a decline in photosynthesis in our experiment, paired with our observation that there was no early onset of senescence in leaves treated with JA even after 6 wk, suggests that our treatment stimulated a narrower response to the JA signal, and was not a general stressor or senescence-causing treatment. Moreover, preliminary data indicate that herbivory by gypsy moth larvae (< 10% defoliation) generates a response similar to that reported here (B. A. Babst et al. unpublished).

For both Populus species, partitioning of newly fixed carbon was shifted toward lower stem and roots after JA treatment. Partitioning between apex and roots could change over a longer chase period (Kleiner et al., 1999), but our results are consistent with a previous report of increased accumulation of radioactive carbon-14 in Populus roots 72 h after a severe defoliation (Bassman & Dickmann, 1985). The defoliation-induced effect reported by Bassman and Dickmann (1985) may have been a passive response to damage, driven by reduced above-ground sink strength, since young leaves were targeted for defoliation. Since that time, evidence has accumulated that implicates a central role for jasmonates in plant responses to herbivory (Baldwin, 1996; Thaler et al., 1996), allowing us to separate the effects of tissue removal from the effects of induction. Our study demonstrates that JA signaling, in the absence of leaf tissue loss, causes greater carbon partitioning to roots, which suggests that the change in partitioning occurs as a direct response to perceived herbivory via the JA-signalling pathway.

Mechanistically, partitioning could be influenced by export rate from the leaf, or by changes in sink demands in the apex, stem, and roots (Minchin et al., 1993). There is evidence that young leaves of hybrid poplar (Populus deltoides × P. nigra) directly treated with JA become stronger sinks for carbon-based resources, and that a substantial portion of the imported carbon is used in production of resistance chemicals (Arnold & Schultz, 2002). Jasmonates themselves may be translocated (Zhang & Baldwin, 1997) and may play an indirect (Ryan, 2000), or perhaps a direct role (Stratmann, 2003) in systemic defense responses. Our study suggests that sink demands in roots might also increase in response to JA. Sink leaves and roots appear to compete for limited carbohydrates (Arnold et al., 2004). Differences between our partitioning responses to JA and those reported by Arnold and Schultz (2002) might be due to a difference in location of ‘attack’, since we applied JA to a mature leaf, whereas Arnold and Schultz (2002) applied JA to young expanding leaves. Since signal translocation is dictated by patterns of photosynthate translocation (Davis et al., 1991; Jones et al., 1993; Orians et al., 2000), the outcome of competition between apical and root sinks may be a function of the pre-existing flow of photosynthate from the treated leaf, which would be more basipetal in older leaves (Larson & Gordon, 1969). Additionally, the duration of treatment may be important, since our study focused on the response within 12 h, whereas Arnold and Schultz (2002) made observations after 1–2 wk of treatment. Future studies should address whether the response to JA shifts qualitatively over time, initially increasing root demand for carbon imports and later increasing apical demand for carbon imports.

Implications

When a plant is attacked by leaf chewing herbivores there is a spike in JA levels, which causes induction of resistance (Karban & Baldwin, 1997). The results presented here suggest that, in addition to the well known JA-induced changes in resistance, there might be JA-induced changes in resource allocation to enhance plant functions related to tolerance. Further research is being conducted to test the hypothesis that Populus plants respond to JA by shifting carbon and nitrogen-based metabolites to stem and root storage. Consistent with this hypothesis, sustained exposure to jasmonates can induce increased storage protein accumulation in stems of Populus (Beardmore et al. 2000). This temporary storage of resources away from foliage-consuming herbivores may allow induced defensive compounds to reach high enough concentrations relative to nutritive compounds that they end the current attack before regrowth is initiated. Further, resources stored in the stem and roots would be better protected from herbivores, especially folivores, and possibly from other stresses as well. Those stored resources could provide a competitive advantage to plants, by causing enhanced growth once the threat from herbivores has passed. Alternatively, changes in carbohydrate translocation can serve as long-distance signals from leaves to roots, affecting various functions, including nutrient uptake and assimilation (Koch, 1996; Rolland et al., 2002). Future research should determine whether nutrient uptake and assimilation are increased by JA application to mature leaves in Populus. Perhaps partitioning responses in annual plants, with lower vegetative storage capacity but similar requirements for soil nutrients, could be instructive in understanding the nature of the rapid change in 11C partitioning reported here.

Acknowledgements

This research was supported by a laboratory-directed research and development grant awarded by BNL (to R. A. F.), and in part by the US Department of Energy, Office of Biological and Environmental Research under contract DE-ACO2-98CH10886, a grant from the Andrew Mellon Foundation to C. M. O., and from the Hrdy Fund of Harvard University to M. T. L. We thank Dr Kenneth Raffa for valuable comments on our manuscript and for supplying cuttings of the P. nigra clone, and Tara Bledsoe, Amy Zanne, Nathan Alden, Monisha Sharma and Sylvie Krinsky for plant care.

Ancillary