SUPPLEMENTARY METHODS DNA extraction. The DNA yield was typically from 2.0 to 2.5 g in a volume of 160 l. DNA was extracted from fungal cultures and other reference samples (below) using the same basic protocol. The ability to recover DNA from fungal biomass may vary according to the cytoplasmic volume and developmental state (e.g. mycelium vs spore). The protocol is well-established to give efficient DNA recovery from vegetative fungal cells, the efficiency of recovery from spores is less certain. The fungal DNA yield from these sites may have been affected by the prolonged period of no rainfall prior to sampling resulting in low active fungal biomass.
PCR conditions were developed using a a range of soil, bacterial and eukaryote DNA samples.
Fungi used as positive controls were strains of Alternaria alternata, Aspergillus terreus, Fusarium oxysporum, Usnea sp., Agaricus bisporus, Saccharomyces cerevisiae, Trichoderma harzianum, Candida albicans, Botrytis cinerea, and Penicillium brevicompactum. DNA samples used as negative controls included a nematode (Radopholus similans), a plant (Oryza sativa), and a mite (Ambleyseius sullivani). Representative target DNAs (1 l; approximately 50 ng) were mixed with 4 l of Genereleaser (Bioventures Inc.) in a 0.2 ml PCR tube, and covered with one drop of sterile paraffin oil. Fungi were obtained from the NSW Agriculture Herbarium (DAR), Macquarie University Herbarium (MQ), or were collected and identified by the authors. Mite, nematode and rice specimens were donated by J. MacDonald, R. Holland and B. Atwell respectively. Tubes were microwaved for 7 min (4550 W min) and then incubated at 80oC for 10 min prior to the addition of 20 l of PCR master mix. Final concentrations of reagents in the PCR were 20 mM (NH4)2SO4, 75 mM Tris-HCl (pH 9.0), 0.01% (w/v) Tween, 2 mM MgCl2, 1 M of each primer, 0.2 mM of each deoxyribonucleotide triphosphate, and 1 U Red Hot DNA Polymerase (Advanced Biotechnologies, Surrey, UK). PCR products were detected using electrophoresis on 2% agarose gels stained with ethidium bromide.
Optimum annealing temperatures were determined using 2 °C increments across a 50 °C to 72 °C annealing temperature gradient in a Stratagene Robocycler. The following thermal cycle was performed: 94 C 3 min (1 cycle); 94 °C 30 sec, 50 °C to 72 °C 30 sec, 72 °C 90 sec (35 cycles), 72 C 5 min (1 cycle). A product of the expected size was amplified from all fungal specimens using an annealing temperature of 62 C. One specimen, Rhizopus stolonifer, gave 2 bands. At this temperature no amplification products were seen for any of the non-fungal controls. Subsequent amplifications with the primer set SSU1758 and 58S1 were performed as described in the manuscript methods.
Validation of ARISAwithsoil community DNA samples.
DNA fragments in the expected size range for fungi were readily amplified from most total soil DNA samples using the conditions described above. The sandy soils (Rodges and Corner) yielded DNA that had a relatively high incidence of failure to amplify in PCR. Based on sequencing data obtained using this primer set with soil samples from other Australian locations and the predicted specificity of the primers the fragments were assumed to be principally ITS1 fragments originating from unidentified ascomycetes. Analysis of identity by cloning and sequencing was not performed with the samples from Sturt National Park.
Potential sources of error include reproducibility of band size determinations in electrophoresis, reproducibility of band profiles in PCR, and reproducibility of DNA extraction. As noted in the main text the accuracy of DNA length determination was +/- 1%. Therefore when replicate PCR samples were run on separate electrophoresis gels 100% reproducibility was only seen if gels were differentiated into base pair lengths resolved at the 2% level. The intensity of bands also varied between gels creating an issue of resolving ‘true’ bands from background noise. The RFLPscan software converts the fluorescence data into an electropherogram where each peak is equivalent to a band on a gel. The sum of the areas under the peaks is a measure of the total fluorescence detected. We adopted a threshold for detection where a band was recorded if its intensity was greater than 0.01% of the integrated optical density for the lane. This was empirically determined to be the threshold for reproducibility of PCR. In a trial experiment using 10 representative Sturt N.P. soil DNA samples, each yielded a different, complex profile of PCR product sizes. The banding profile from individual soil samples was reproducible (Ssm >95%) when either triplicate DNA extractions were performed on the same soil sample, or triplicate PCRs were performed on the same DNA sample. Increasing the sensitivity of detection did result in scoring of additional bands, but these were not reproducibly detected and thus were considered as noise.
To test the potential contribution of PCR bias to the size distribution of PCR amplicons we prepared a range of mixed template PCRs. Known DNA mixtures were prepared by adding equal quantities of each reference organism DNA to a microfuge tube. After mixing, aliquots containing ca 50 ng of the DNA mixture were used as template for PCR as described above. In the defined mixtures all organisms were detected. The large Pen. brevicompactum product (690 bp) was significantly less abundant in mixtures that included bands in the size range of 200 – 300bp. These are crude tests since the molar quantity of primer targets in each sample is unknown. These results suggest that the relative abundance of bands in the 200 – 350 bp size range could be reflective of PCR bias rather than natural distribution, but that the impact on detecting presence/absence of a band is less severe. This was further tested by ‘spiking’ soil DNA templates with an excess of DNA template from laboratory specimens known to yield different sized amplicons. Whilst the control specimen was detected in all cases, the effects on the soil fungal community signal differed for the three taxa investigated. Soil DNA samples spiked with Pen. brevicompactum DNA showed identical patterns to unspiked samples with an additional band at 690bp. In soil DNA samples spiked with S. cerevisiae (480bp), B. cinerea (230bp), or all three taxa, most soil-associated bands were still visible but there was a decrease in their intensity. From these data we concluded that although PCR bias towards shorter amplicons could affect detection of DNA species present at very low relative abundance this effect was only observed when template was added at great excess. It could be a contributing factor to the poor reproducibility of rare bands (ie <0.01% relative abundance). PCR bias is unlikely to be a major factor in the pairwise comparisons of presence-absence data performed here. It is noteworthy that no single band size dominated across the data set (or any subset of the data such as within a landsystem) as may have been expected if there was an undue influence of PCR bias.