Computational protocol: Eucalyptus obliqua seedling growth in organic vs. mineral soil horizons

Similar protocols

Protocol publication

[…] We collected E. obliqua seeds from four parent trees approximately 1 km apart along West Picton Road, near where we had collected soil. The seeds of all four trees were pooled, mixed, and surface-disinfested by 5 min immersion in 2% sodium hypochlorite. Seeds were germinated in trays of a vermiculite-based germination mix which had been autoclaved (121°C, 1.27 kg cm−2) for 1 h, rested for 24 h, and then autoclaved again for 1 h. Seeds were allowed to germinate for up to 120 days after sowing before transplant. To begin the experiment, we carefully removed seedlings from the germination trays and rank-ordered them by size while keeping their bare roots moist. In order, we transplanted one seedling per pot to one pot of each of the five treatments in continuing rotation until all 20 pots of each treatment thereby had received seedlings with similar ranges of initial sizes. Any seedlings that died were replaced during the 14 d between transplant and the initial measurement on 2 June, 2010.Plants were grown in a controlled environment glasshouse in which the temperature was maintained at 20°C day and 10°C overnight with supplementary lighting used to maintain a12/12 h day/night cycle. Positions of all treatments' pots were fully randomized, and the pots were spaced 30–40 cm apart on glasshouse benches. Pot positions were changed weekly. Plants were watered by an automated spray mist system which operated for 10 min at 7:00 h and 10 min at 16:00 h. No pest control was required during the experiment.Beginning 14 DAT, and following at varied intervals thereafter until the final harvest at 357 DAT, we measured plant height from the soil to the end of highest petiole and, with digital calipers, stem diameter 1 cm above the cotyledons. We counted the number of leaves (not including cotyledons) and measured the length of the longest leaf from tip to base (i.e., excluding the petiole).In order to assess mycorrhizas with the best chance of detecting any colonization that might have taken place before iron fertilization was switched to phosphorus fertilization, the three visibly largest plants of each treatment were harvested at 149 DAT. All surviving plants among the 17 remaining per treatment were harvested at 357 DAT, almost a full year after transplant. We concluded the experiment at 357 DAT in order to pre-empt pot limitation; the plants in organic soil continued to increase in height, stem diameter, and number of leaves until that time.At both harvests, shoots were cut at the soil surface and the blades of all leaves were separated from petioles and stems. The three largest leaves from each plant were scanned immediately with a desktop scanner, and their area was determined with Image J (http://imagej.nih.gov/ij/index.html). All leaves, stems, and petioles were dried to constant weight at 40°C in a fan-forced oven, with scanned leaves dried separately to enable calculation of leaf specific area. Root systems were extracted from pots over a 212 μm sieve under a gentle stream of water before storage in 60% ethanol in anticipation of sub-sampling for mycorrhiza assessment. After sub-sampling, root systems were oven dried at 40°C to constant weight. We determined the dry weights of all harvested plant parts separately.We ground the dried leaves of each plant in a ball grinder (Retsch MM200) for 2 min before sending the ground tissue to CSBP Limited, Soil and Plant Analysis Laboratory, Western Australia for analysis. Foliar N, P and other elements were determined after complete acid digestion of the plant material followed by inductively coupled plasma–atomic emission spectrometry (ICP-AES).At the final harvest, we used a 0.75 cm inside-diameter sharpened copper tube to extract a soil core from each pot to a depth of 5.0 cm. The cores were dried separately to constant weight before weighing to approximate bulk density. After extracting root systems from each pot, we collected ca. 100 g of soil from each, pooled and thoroughly mixed it by treatment before air-drying for 2 week. For each treatment, a sub-sample of 500 g soil was analyzed by CSBP Limited. Ammonium and nitrate were extracted in KCl and quantified with a Lachat Flow Injection Analyser. P and K (Colwell) were extracted in sodium bicarbonate, Cu, Fe, Mn, and Zn in DTPA, and both extracts analyzed by atomic absorption spectrometry. Al, Ca, Mg, K, and Na were extracted in ammonium chloride and analyzed with ICP-AES. Soil pH and electrical conductivity were determined in a 1:5 soil:water suspension, and pH was determined also in a calcium chloride suspension.On 9 November, 2014, in order to approximate “beginning” soil attributes, we collected additional soil samples from the same two locations as previously. We manually removed woody debris and roots, then composited, thoroughly mixed, and oven-dried the organic and mineral soil layers separately before having them analyzed by CSBP Limited. [...] Because of our unbalanced experiment design, our general approach to statistical analyses was to separately compare seedling performance in mineral soil to that in only ambient organic soil (i.e., both not-treated soils) by One-Way analysis of variance (ANOVA), and then to compare the effects of fumigation and fertilization on organic soil alone by Two-Way, factorial ANOVA. Levene's test was used to check homogeneity of variances, and except where otherwise noted, analyses of multiple response variables were Bonferroni-corrected for the number of variables analyzed. All statistical analyses were conducted with Statistix 10.0 except for repeated-measures analyses which used REML estimation in JMP Pro 10.0.0.Levene's test showed soil bulk densities to be heteroscedastic, but a log10 transformation homogenized variances so that all five treatments could be compared in a One-Way ANOVA. Because we used single, composited samples to represent each treatment for other soil attributes, we performed separate, Two-Way, factorial ANOVAs for each attribute for only the four organic soil treatments by using the interaction term as an error estimate. Because of the weakness of these analyses, we did not Bonferroni-correct for the number of attributes analyzed, and we accepted P ≤ 0.09 as suggestive of differences among treatment means. Moreover, because they represented collections at different times, we did not statistically compare “beginning” soil attributes to those of soils taken from pots at the end of the experiment.Percentage ectomycorrhizal root tips were analyzed separately for each harvest because we intentionally had biased the initial harvest to the largest plants of each treatment. We analyzed arcsine-square-root-transformed percentages for each harvest by One-Way ANOVA because the variances were homogenous among all five treatments. We used Pearson's correlation coefficients to examine associations between whole plant dry weight and percentage ectomycorrhizal root tips.Using our general approach, we examined four morphological response variables (height, longest leaf length, number of leaves, and stem diameter) with repeated-measures ANOVAs and Bonferroni-corrected probabilities (P ≤ 0.05/4 = 0.0125). For the repeated-measures analyses, however, we separated the time intervals during which we fertilized with Fe (0–121 DAT) and P (121–357 DAT). Because the three largest plants of each treatment were harvested at 149 DAT, we excluded those 15 plants from the repeated-measures analyses for both intervals. For the second time interval, all individual plant measurements were relativized by deduction of their values at the beginning of the interval in order to reflect changes in size during the interval. In comparing the mineral vs. ambient organic soil, leaf length was log10-transformed for the first interval, but no parametric transformation homogenized diameter variances, so diameter was rank-transformed. For the second interval, height change only was log10-transformed. Number of leaves change was marginally heteroscedastic according to Levene's test (P = 0.0458), so was not transformed. All organic soil treatments were homoscedastic for both intervals.When we considered the final harvest dry weights of leaves, stems, roots, total plant, and root-to-shoot ratios for mineral vs. ambient organic soil, leaf and stem weights were heteroscedastic, so we used Welch's test for mean differences (Welch, ) for all five comparisons with a Bonferroni-corrected probability of P ≤ 0.0100. We also used Welch's test to compare mean leaf specific areas for these two treatments. When only organic soil treatments were compared, all final harvest response variables were homoscedastic, and so we used Two-Way ANOVAs with effects Bonferroni-corrected as for the mineral vs. ambient organic soil treatment comparisons followed by Tukey's honestly significant difference tests to separate means.Mineral nutrient concentrations in leaf tissue of plants grown in mineral soil vs. ambient organic soil were heteroscedastic for Mg, Mn, Na, and N:P (Levene's test P ≤ 0.0245), so we compared all elements by using Welch's test with a Bonferroni-corrected probability of P ≤ 0.0038 (12 elements and one element ratio). Mineral nutrient contents (= concentration × leaf dry weight) were homoscedastic only for Fe and Cu, so we compared all elements by using Welch's test with a Bonferroni-corrected probability of P ≤ 0.0042. For plants in all organic soil treatments, P and Fe concentrations were heteroscedastic (Levene's test P ≤ 0.0004) as were P, Fe, and Zn contents (Levene's test P ≤ 0.0326), but in all those instances log10 transformations resulted in homoscdasticity. We compared foliar element concentrations and contents among plants grown in all organic soil treatments by Two-Way ANOVAs with Bonferroni corrections followed by Tukey's tests. […]

Pipeline specifications

Software tools ImageJ, JMP Pro
Applications Miscellaneous, Microscopic phenotype analysis
Organisms Eucalyptus obliqua
Chemicals Iron, Phosphorus