Probing the effects of light and temperature on diurnal rhythms of phytosiderophore release in wheat

Authors


Author for correspondence: Suzie Reichman Tel: +64 3325 2811, ext. 8235 Fax: +64 3325 3607 Email: s.reichman@lincoln.ac.nz

Summary

  • • The existing literature is ambiguous as to whether the diurnal pulse in phytosiderophore (PS) release in the Poaceae is mediated by light or temperature, or both.
  • • Here, wheat (Triticum aestivum cv. Yecora Rojo) seedlings were grown in Fe-sufficient (pFe = 16.5) and Fe-deficient (pFe = 17.8) chelator-buffered nutrient solutions. Six different light/temperature regimes were tested over 8 d in paired growth chambers.
  • • Phytosiderophore release patterns under a square-wave light regime were similar, irrespective of whether temperature was varied diurnally or held constant, but PS release was negligible when the light was removed. Release patterns of PS for Fe-deficient and Fe-sufficient plants grown under the square-wave vs ramped light and temperature regimes were similar in the corresponding Fe treatments.
  • • Our results strongly support the notion that the diurnal pulse in PS release in the Poaceae is mainly mediated by changes in light rather than temperature. Our comparison of square-wave with more natural ramped light/temperature regimes suggests that the diurnal response patterns of PS release in wheat can be confidently studied using traditional square-wave regimes, and this is likely to be the case with other Poaceae as well.

Introduction

The diurnal pattern of phytosiderophore (PS) release in the Poaceae was first documented in 1984 in Fe-deficient barley (Hordeum vulgare) (Takagi et al., 1984). Subsequently, this daily pulse in PS release has been validated in other Fe-deficient Poaceae, for example, wheat (Triticum aestivum) (Zhang et al., 1991), red fescue (Festuca rubra) (Ma et al., 2003), and 28 noncultivated grass species (Gries & Runge, 1992). In addition, diurnal patterns in PS release have been documented under Zn deficiency in wheat (Zhang et al., 1989) and under Cu deficiency in Hordelymus europaeus (Gries et al., 1998). The only grass species found to not have diurnal patterns of PS release under Fe deficiency are maize (Zea mays), which had low but constant release (Yehuda et al., 1996), and two noncultivated grass species for which no phytosiderophore release was observed irrespective of the time of day (Gries & Runge, 1992). It has been suggested that the lack of a diurnal cycle in maize might be the reason for this species’ low Fe efficiency (Yehuda et al., 1996). Further evidence for the diurnal pattern of PS release in Fe-deficient Poaceae includes coordinated diurnal patterns of PS increase in apical root cells of barley (Walter et al., 1995). In comparison, the apical roots of Fe-sufficient barley and the shoots of both Fe-deficient and Fe-sufficient barley contained relatively constant concentrations of PSs throughout the day (Walter et al., 1995). Moreover, a number of genes induced by Fe deficiency and showing diurnal patterns in expression have been identified in barley and rice, including some genes involved in PS synthesis (Negishi et al., 2002; Nozoye et al., 2004).

A review of the literature, combined with modeling, has suggested the diurnal pulse in PS release is likely to result in rhizosphere PS concentrations up to 100 µm with Fe-deficient wheat or barley (Reichman & Parker, 2005). However, to meet plant nutrient demands, with 1 : 1 Fe : PS binding, a rhizosphere concentration of only c. 1 µm PS might be needed (Crowley & Gries, 1994). So, why do grasses excrete a large diurnal pulse of PS into the rhizosphere? A number of factors potentially affect the efficacy of PS release, for example, competition by other metal ions for binding sites on the PS molecule (Murakami et al., 1989) and biodegradation of the PS by microorganisms (Takagi et al., 1988). Peak production in the morning might also allow for coordination with peak periods of transpirational water flux, which would occur several hours later, and could convectively transport PS-solubilized Fe from the rhizosphere soil to the root surface for subsequent uptake and utilization (Parker et al., 2005). Thus, the diurnal pulse of PS release results in large, temporary concentrations of PS in the rhizosphere, which could maximize Fe uptake with minimal energy expenditure by the plant.

When Takagi et al. (1984) first documented the pattern of diurnal PS release, they only sampled root exudates four times in each 24 h period. Hence, while their results clearly indicate a diurnal rhythm, they are more qualitative then quantitative with respect to the timing of the cycle. Takagi et al.'s (1984) results did suggest that the daily pulse in barley had commenced by c. 2 h after the start of the light period, that it peaked at c. 4 h, and that it was down to negligible levels by c. 9 h after initiation of the light period. In comparison, more detailed studies for Hordelymus europaeus showed a distinct peak c. 5.5 h after the start of the light period (Gries & Runge, 1992).

Some research has investigated whether the external stimulus for PS release is light- and/or temperature-mediated. In a book chapter, Nomoto et al. (1987) stated that at least 1–2 h of darkness is required to induce the pattern of PS release. Both continuous light (Nomoto et al., 1987) and continuous darkness (another book chapter: Mori, 1994) appear to reduce peak size and eventually result in cessation of PS production, indicating that it is the cycling between light and darkness that is important for initiating PS release. In addition, it appears that temperature may have an effect on PS production. The traditional peak in production was found under continuous light, but with a day : night temperature regime (Mori, 1994). Standard diurnal light fluctuations, but with inverted night/day temperature regimes, resulted in switching of peak PS release from the morning to the evening (Mori, 1994). The size of the peak (i.e. the difference between background and peak exudation rates) can be increased by increasing the temperature differentiation between night and day (Mori, 1994). Thus, Mori (1994) suggested a diurnal pattern of release regulated solely by temperature. However, other research has shown that increased light intensity differentially affected PS production in Fe-efficient and -inefficient wheat cultivars (Cakmak et al., 1998). The efficient cultivar showed increased maximum PS production at high light intensity, while the inefficient cultivar was little affected. However, these authors did not control temperature independently, and the results could be explained by the greater diurnal temperature differential in the high-irradiance treatment.

Thus, there is still some uncertainty as to how the diurnal pattern of PS release is mediated, and much of the information has been presented in secondary sources, that is, book chapters and other publications lacking widespread circulation. Our aim was to probe further the hypotheses of temperature- vs light-mediated diurnal PS release with primary data. In addition, all research to date into the diurnal pattern has involved traditional, square-wave light/temperature regimes, and we have evaluated the effects of a more natural, ramped regime on PS production.

Materials and Methods

Wheat (Triticum aestivum L. cv. Yecora Rojo) was used as the test species. Seeds were surface-sterilized in 0.4 m NaOCl for 15 min and then germinated on moistened germination towels in a growth chamber with 24 h light and 12 h, 13°C : 12 h, 23°C temperature regime (day 0). On day 3, plants were transferred to 3 l containers with mesh-covered lids and the light regime changed to 12 h light : 12 h dark. The nutrient solutions were aerated by use of airstones (Kordon, Novalek Inc, Hayward, CA, USA) linked to polytetrafluoroethylene tubing connected to an aeration line, run from an aquarium pump. Seedlings were placed with roots below the mesh and shoots above, and the mesh was covered with Teflon beads to prevent light reaching the nutrient solution below. Initially, 25 seedlings were placed in each container, and this was thinned to 20 seedlings on day 8.

The experiment was undertaken using a pair of environmentally controlled growth chambers. The two chambers were used to provide differing light/temperature regimes (to be described later). In each 24 h period, plants received 12 h of light and 12 h of darkness (unless otherwise specified), and the relative humidity was maintained at 65% throughout.

A chelator-buffered nutrient solution was used. Basal nutrients were supplied as (µm): NO3, 4750; NH4+, 250; P, 80; K, 1080; Ca, 2000; Mg, 500; S, 500; Mn, 0.6; Zn, 8; Cu, 2; B, 10; Mo, 0.1; Ni, 0.1; and Cl, 21.4 (Pedler et al., 2000). For controlling trace-metal availability the solutions contained 35.7 µm HEDTA, that is, a 25 µm excess above the sum of the Mn, Cu, Zn and Ni concentrations. Nutrient solutions were changed on days 8, 13, 18 and 22. To buffer the nutrient solution at pH 6, the solution contained 1 mm MES and 0.5 mm NaOH. Daily measurements of pH were made and pH adjusted as necessary with 0.1 m NaOH or HCl. In the control treatment (pFe = 16.5), Fe was supplied at 75 µm as FeCl3 (plus 75 µm HEDTA). In the Fe-deficient treatment (pFe = 17.8), Fe was supplied as 3.75 µm FeCl3 (plus 3.75 µm HEDTA). GEOCHEM-PC was used to calculate the pFe of each treatment (Parker et al., 1995) where the pFe of a solution is defined as –log10(Fe3+ activity). Each treatment was replicated four times per growth chamber but, for root exudate collection, these replicates were divided into two sets of two replicates, with each set used to collect exudates during alternating collections periods during any given collection day.

Six different light/temperature regimes were tested (i.e. three per growth chamber) and descriptions of each regime are given in Table 1. Maximum irradiance differed between square-wave and ramped regimes to ensure that all plants were exposed to the same total light quanta each day. Ramped regimes were designed to be representative of early growing conditions for spring wheat and were based on data for Thornton Weather Station, San Joaquin County, CA, USA during 20–22 March 2001 (http://www.ipm.ucdavis.edu). Irradiance was quantified as photosynthetically active radiation (400–700 nm) using a LI-185B photometer (Li-Cor BioSciences, Lincoln, NE, USA) equipped with a Quantum sensor.

Table 1.  Descriptions of the light and temperature regimes used
RegimeDescription (night/day)a
Irradiance (µmol m−2 s−1)Temperature
  • a

    12 h of day and night were used throughout the experiment.

1 (Standard square-wave)0/29413/23°C
2 (Standard ramped)0/gradual ramp to/from 52313/gradual ramp to/from 23°C
3 (Square-wave light/constant temperature)0/29423/23°C
4 (Ramped light/constant temperature)0/gradual ramp to/from 52323/23°C
5 (Dark/square-wave temperature)0/013/23°C
6 (Dark/constant temperature)0/023/23°C

On day 3, both growth chambers were running Regime 1. Regimes were changed on days 14, 18 and 22 to Regime 1/Regime 2 (corresponding to Growth chamber 1/Growth chamber 2), Regime 3/Regime 4 and Regime 5/Regime 6, respectively. After each change of light/temperature regime, the plants were given 1–2 d to acclimatize to the new regime. Thus, exudates were collected on days 16, 20, 23 and 24. The extra collection on day 23 was to ensure capture of the diurnal pattern before loss of photosynthetic activity in the darkness of Regimes 5 and 6.

To collect exudates, the plants in each pot were lifted from the nutrient solution in the mesh lid, rinsed thoroughly with deionized water and transferred to a 200 ml collection pot filled with 200 µm CaCl2. The plants, in their collection pot, were returned to their respective growth chambers for 1 h and then returned to the nutrient solution. Collections were hourly from 07:00 h (1 h before start of the ‘day’) to 16:00 h and then every other hour, with the last collection starting at 20:00 h (i.e. finishing 1 h after the start of the ‘night’). Exudate collections on days 23 and 24 were undertaken in a darkened room with light maintained at ≤ 0.2 µmol m−2 s−1 at all times. Each set of two replicates per growth cabinet were alternated for collection to ensure that plants could continue acquiring Fe despite the time spent in the Fe-free exudate collection medium. The smaller number of replicates per sampling time was necessary to ensure that collections could occur accurately and precisely with respect to time.

To prevent microbial degradation of the PS, a 1 ml aliquot of 10 mg l−1 Micropur was added to the collected exudates upon removal of plants; the exudates were then immediately vacuum-filtered through a 0.45 µm filter and frozen until further analysis. Phytosiderophores in the root exudates were quantified by the revised Fe-binding assay (Reichman & Parker, 2006). Exudates were also analyzed by ion chromatography to ensure the absence of other organic acids which could confound the revised Fe-binding assay, and were found to contain < 0.5 µm total of common organic acids that could also bind Fe strongly (e.g. citrate, malate, oxalate) (data not presented).

Plants were harvested on day 25. All plant parts were rinsed four times in double deionized water, separated into roots and shoots and dried for 24 h at 65°C. Data from dry weights of plants thinned on day 8 and plants harvested at day 25 were interpolated to estimate root dry weights throughout the experiment (data not shown). Shoots were ground and microwave-digested in a mixture of 70% HNO3 and 9 m H2O2 (Sah & Miller, 1992) before analysis for Fe by atomic absorption spectrophotometry.

Statistical analyses, such as descriptive statistics and anova, were performed using SPSS (SPSS, 2003). Restricted maximum likelihood (REML) methodology was performed using the GENSTAT package (GENSTAT 8 Committee, 2005) to compare treatment effects. REML is preferable to anova for unbalanced and nonorthogonal data and avoids the biased variance component estimates that are produced by ordinary maximum likelihood estimation (GENSTAT 8 Committee, 2005). The data were fourth-root-transformed (R. Sedcole, pers. comm.) to stabilize the variance of the residuals and significance based on analysis of the transformed data. The statistical significance of fixed terms was calculated by comparing the Wald statistic with critical chi-squared values. Where main effects were found to be significant (P < 0.05), further investigations were made of higher-order interactions between terms. In REML analysis, the order in which terms are added to the model can sometimes affect the significance of terms; thus each REML analysis was performed multiple times by adding the main effect terms in a variety of combinations. The order of addition of main effect terms had no impact on whether terms were significant or not, and thus only one REML model is presented for each data set in the Results section. The statistical significance of within-term differences was determined from t-tests (two-tailed, stratum variance degrees of freedom) for individual data pairs of interest.

Results

Reuter & Robinson (1997) defined the threshold for Fe deficiency in shoots as being 25 µg g−1, and thus the Fe-deficient and Fe-sufficient treatments produced plants below and above the Fe-deficiency threshold, respectively (Table 2). Comparison of shoot Fe concentrations (Table 2) via a two-way anova demonstrated that Fe-sufficient plants had significantly higher foliar Fe concentrations than did Fe-deficient plants (F = 10.23, d.f. = 1, P = 0.008) and there was no significant effect of growth chamber (F = 0.111, d.f. = 1, P = 0.738) on foliar Fe concentration. Thus, when we compare treatments across growth chambers, the plants are of equivalent Fe status.

Table 2.  Comparison of shoot iron (Fe) concentrations for Fe-deficient and Fe-sufficient wheat (Triticum aestivum cv. Yecora Rojo) grown in two Conviron growth chambers
 Fe concentration (µg g−1)a
Growth chamber 1Growth chamber 2
  • a

    Values are means (n = 4) ± SE.

Fe-deficient16.4 ± 3.916.1 ± 4.2
Fe-sufficient30.2 ± 4.627.8 ± 3.1

Iron-sufficient treatments in both square-wave (Fig. 1a) and ramped (Fig. 1c) light/temperature regimes had low to minimal pulses in PS release. In the Fe-deficient treatments, both the standard square-wave (Fig. 1b) and the more natural ramped light/temperature regimes (Fig. 1d) gave similar patterns of PS release, with a distinct maximum release occurring 2–3 h after the commencement of the light period, with the bulk of PS pulse in release occurring for a 3–4 h period before returning to background release rates (Figs 1b,d, 2b,d). When temperature was held constant but light was varied as either a square-wave or ramped regime (Fig. 2), patterns of PS release were similar to those observed when temperature followed a diurnal pattern (Fig. 1). The absence of light resulted in minimal PS release after an extended dark period of 24 h (day 23) for both Fe treatments, irrespective of whether temperature was varied in a square-wave regime or held constant at 23°C (Fig. 3). Minimal PS release persisted in all darkness treatments 24 h later (day 24) (data not shown). On days 23 and 24, PS release in all darkness treatments at all collection periods was below the method detection limit (1.5 µm) for the revised Fe-binding assay (Reichman & Parker, 2006). The combinations of light and temperature regimes tested did not fit into a factorial or other commonly used experimental design, and thus the data were split into two sets for REML analysis as described below.

Figure 1.

Phytosiderophore (PS) release patterns from wheat (Triticum aestivum cv. Yecora Rojo) seedlings grown in solution culture with traditional square-wave or ramped light and temperature regimes for 48 h. The vertical bars represent PS release ± 1 SE; solid line, light; dashed line, temperature. DW, dry weight.

Figure 2.

Phytosiderophore (PS) release patterns from wheat (Triticum aestivum cv. Yecora Rojo) seedlings grown in solution culture at a constant temperature and with a square-wave or ramped light regime for 48 h, The asterisk (*) in panel (b) represents missing data. The vertical bars represent PS release ± 1 SE; solid line, light; dashed line, temperature. DW, dry weight.

Figure 3.

Phytosiderophore (PS) release patterns from wheat (Triticum aestivum cv. Yecora Rojo) seedlings grown in solution culture in extended darkness (24 h) with a square-wave or constant-temperature regime. The vertical bars represent PS release ± 1 SE; dashed line, temperature. DW, dry weight.

Comparing the effects of light vs temperature on phytosiderophore release

To compare the effects of light vs temperature on PS release, REML analysis was performed on the subset of data including all light and temperature treatments, except the ramped regimes (Table 3). This created a 2 × 2 factorial design for light by temperature, that is, square-wave light and dark by square-wave light and constant temperature. Fe-sufficient plants released significantly less PS than Fe-deficient plants and the time of day had a significant effect on the amount of PS collected (Figs 1a,b, 2a,b, 3; Table 3). As a main effect, PS release followed a classic diurnal rhythm with a distinct peak in PS production between 07:00 and 12:00 h, with maximum PS release from 09:00 to 10:00 h falling to a minimum between 18:00 and 21:00 h (P < 0.01). While the light regime had a significant effect on the amount of PS release (greater in the square-wave light treatments than in the dark treatments) there was no significant effect of temperature regime on the amount of PS released (Figs 1a,b, 2a,b, 3; Table 3).

Table 3.  Residual maximum likelihood analysis of fourth-root-transformed phytosiderophore release collected between 07:00 and 21:00 h from the roots of Fe-deficient and Fe-sufficient wheat (Triticum aestivum cv. Yecora Rojo) grown under square-wave or constant light and temperature regimes
TermWald statisticDegrees of freedomChi-squared probability
Fe treatment (Fe)49.00 1< 0.001
Time of day (time)598.5811< 0.001
Light regime (light)232.68 1< 0.001
Temperature regime (temp)< 0.01 10.967
Fe × time103.7711< 0.001
Fe × light28.45 1< 0.001
Time × light452.5211< 0.001
Fe × time × light57.7011< 0.001
Stratum variance 83 

Examination of the significant two–way interactions (Table 3) revealed that there was a significant effect of Fe treatment according to time of day. Thus, both Fe-deficient and Fe-sufficient treatments yielded a significant pulse in PS release but, with the Fe-deficient treatments, the pulse occurred between 07:00 and 13:00 h and was of significantly greater magnitude than the pulse occurring between 08:00 and 11:00 h under Fe sufficiency (P < 0.05, Figs 1a,b, 2a,b, 3). In addition, there was a significant interaction between Fe treatment and light regime (Table 3). Under square-wave light regimes, the amount of PS release was significantly greater in the Fe-deficient than in the Fe-sufficient plants (P < 0.01, Figs 1a,b, 2a,b), whereas in the dark treatments (Fig. 3) the PS release was minimal (P < 0.01) and there was no significant difference between Fe-sufficient and -deficient plants (P > 0.05, Fig. 3). There was also a significant interaction between time of day and light regime (Table 3). Statistical comparisons showed a distinct pulse in PS release for the square-wave light regime between 07:00 and 12:00 h (P < 0.05). By contrast, the quantities of PS released in the dark were not significantly different from those produced during the afternoon under the square-wave light regimes (e.g. at 16:00 h) (P > 0.05).

There was a significant interaction between the three significant main effects, that is, Fe treatment by time of day by light regime (Table 3). Fe-deficient plants grown in the square-wave light regime exhibited the largest pulse in PS release, in terms of both duration (07:00–13:00 h) and maximum release rate (Figs 1b, 2b). Fe-sufficient plants grown under the square-wave light regime had a shorter pulse in PS release (08:00–11:00 h), and a significantly lower maximum release rate of PS (P < 0.01). In comparison, neither the Fe-deficient nor the Fe-sufficient plants grown in the dark regime had a significant diurnal pulse in PS release (P > 0.05).

Comparing the effects of square-wave vs ramped light/temperature regimes on phytosiderophore release

To investigate the impact of square-wave vs ramped light and temperature regimes on PS release, all treatments were included in the data set except the dark regimes. This created four combined light/temperature treatments: standard square wave, standard ramped, square-wave light/constant temperature and ramped light/constant temperature. While the main effects of Fe treatment and time of day were found to be significant there was no significant effect of the light/temperature regime on PS release (Table 4, Figs 1, 2). Thus, the square-wave light/temperature regimes favored by researchers using growth cabinets did not yield a significantly different pattern (P > 0.05) of PS release than that obtained with the more natural ramped light and temperature regimes. The effects of Fe treatment and time of day on PS release were similar to those described earlier in the Results section and are not discussed further.

Table 4.  Residual maximum likelihood analysis of fourth-root-transformed phytosiderophore release collected between 07:00 and 21:00 h from the roots of Fe-deficient and Fe-sufficient wheat (Triticum aestivum cv. Yecora Rojo) grown under square-wave or ramped light/temperature regimes
TermWald statisticDegrees of freedomChi-squared probability
Fe treatment (Fe)183.00 1< 0.001
Time of day (time)741.6711< 0.001
Light/temperature regime  2.42 30.490
Fe × time265.8011< 0.001
Stratum variance 73 

Discussion

Our findings strongly support a light-mediated pattern of PS release in wheat. We observed a classic diurnal pulse of PS release in Fe-deficient wheat when both light and temperature were varied in a traditional square-wave manner (Fig. 1). When the square-wave light regime was maintained, but the temperature was held constant for 48 h before sampling, the diurnal pulse was still fully evident in Fe-deficient wheat and clearly linked to the commencement of the light period (Fig. 2). However, once light was removed from the system, irrespective of whether temperature was maintained in a square-wave regime or held constant, only 24 h of extended darkness was needed to reduce PS release to negligible amounts in both Fe-deficient and Fe-sufficient wheat (Fig. 3). Our results are in agreement with data presented in a book chapter (Mori, 1994), showing peak PS release from Fe-deficient barley roots was approximately halved if PS release was measured during the first day of extended darkness, and was minimal by the second day of extended darkness (equivalent to our day 23 measurements, Fig. 3). However, in the previous study, no information was given on the corresponding temperature regime to which the barley plants were subjected. By contrast, Ma et al. (2003) grew Fe-deficient F. rubra in a glasshouse with a constant temperature of 20°C and natural light conditions. Half the plants were enclosed in a light-proof box just before dawn and no difference was found in the pattern of PS release between plants in the light or those in the box on the day light was removed from the system. Ma et al. (2003) made no PS measurements on the following day (again equivalent to our day 23 measurements, Fig. 3). Thus, while removal of light from the system, irrespective of the temperature regime, appears to rapidly reduce PS release to zero, implying that the trigger for production and/or release has been removed, there does appear to be some lag in this response. This may indicate that some or all of the trigger for PS release is a response to historical rather than current light conditions, or that light may be required in the preceding 24 h for PS production for the current day.

Our finding that PS release is mediated only by light but not temperature is in agreement with Nomoto et al. (1987) but not Mori (1994). Because other research into the diurnal rhythms of PS release is presented in either book chapters or other publications lacking widespread circulation, we do not have full access to the methods used. Thus, it is difficult to assess why our results and those of Nomoto et al. (1987) differ from those of Mori (1994). However, ours are the first peer-reviewed results to investigate thoroughly whether PS release is mediated by light or by temperature.

Our findings indicate that standard growth chamber setups are not oversimplifying the PS release pattern of grasses. Although a small pulse in PS release was visually discernible in Fe-sufficient plants grown under square-wave light regimes (Figs 1a, 2a), it was not significantly different from the absence of any observable PS pulse in Fe-sufficient plants grown under ramped light regimes (Figs 1c, 2c). It is difficult to compare our results with the literature, as many studies on the diurnal release of PS have not included Fe-sufficient controls (Nomoto et al., 1987; Mori, 1994) or, where included, the scale of the figures makes it difficult to discern any exudation in Fe-sufficient plants (Zhang et al., 1991). Although Ma et al. (2003) measured PS release in F. rubra under a natural light regime in a glasshouse, they did not measure the diurnal cycle of Fe-sufficient plants. To our knowledge, this is the first time that the square-wave regime has been compared with a more natural ramped regime.

In conclusion, our results with wheat strongly support the hypothesis that the daily pulse in PS release in the Poaceae is mediated by changes in light rather than temperature. Our comparison of traditional square-wave and more natural ramped light/temperature regimes suggests that the diurnal responses patterns of PS release in wheat can be confidently studied using traditional square-wave regimes, and this is likely to be the case with other Poaceae as well.

Acknowledgements

Thanks to Judith Pedler, Heather Allen and Richard Collins for advice on experimental design and to Richard Sedcombe and James Ross for statistical advice. The technical assistance of David Thomason and Peggy Resketo is gratefully acknowledged. This research was supported by the National Research Initiative of the USDA Cooperative State Research, Education and Extension Service (grant number 2002-35107-12215).

Ancillary