Supplementary Information to Ovaskainen et al. (2013): Combining high-throughput sequencing with fruit body surveys reveals contrasting life-history strategies in fungi
Measuring time delay
As we did not have time-series data, we inferred the decay class in which the species typically appeared as mycelia (or as fruit bodies) by comparing prevalences among different decay classes. We denote by and the prevalence of the species in decay classes as fruit bodies () and mycelia (). As a species cannot be present as living fruitbodies unless it is present also as mycelia, we should have for . In our data this was not always the case as a species may have been missed as mycelia. To avoid logical problems when estimating the time delay, we use here corrected mycelial prevalence , obtained simply by assuming that the species was present as mycelia in all those logs where it was present as fruit bodies. The example in Table S1 illustrates the logic of the following reasoning. The fraction of logs for which fruitbodies appear in decay class 1 is . Assuming that the species does not show extinction-recolonization dynamics during the course of decay, the fraction of logs for which fruitbodies(or mycelia) first appear in decay class can be recursively estimated as , . Here we have truncated to positive values as prevalence may decrease due to death of existing occurrences (and as our data are not a true time series but snapshot data from different decay classes creating additional noise). We denote by the fraction of logs which are colonized by mycelia at decay class and for which fruitbodies first appear in decay class , and by the fraction of logs which are colonized by mycelia at decay class and for which no fruit body has appeared until decay class . For decay class 1 we have , and . To construct these terms for remaining decay classes in a recursive fashion, we assume that fruit bodies appear on the logs in the order in which the logs are colonized by mycelia. Thus, for , and . For mycelia that appeared in decay class 2, we have,, and for , .More generally, for any, , , and for , .
For the cases that belong to the class , the time delay from mycelial to fruitbody appearance is , and hence the time delay can be computed as
DC1 / DC2 / DC3 / DC4/ 0.3 / 0.5 / 0.6 / 0.7
/ 0.1 / 0.15 / 0.4 / 0.2
/ 0.3 / 0.2 / 0.1 / 0.1
/ 0.1 / 0.05 / 0.25 / 0
/ 0.1 / 0.05 / 0.15 / 0
/ 0.2 / 0.15 / 0 / 0
/ NA / 0 / 0.1 / 0
/ NA / 0.2 / 0.1 / 0.1
/ NA / NA / 0 / 0
/ NA / NA / 0.1 / 0.1
/ NA / NA / NA / 0
/ NA / NA / NA / 0.1
Table S1. An illustration of the computation of time delay from mycelial establishment until fruiting, leading to the estimate .
Relationships between decay class and physical and chemical properties of wood
In the main text, we characterize the decay stage of each log with the nominal variable decay class, varying from 1 to 4, i.e. from little to strongly decomposed. The classification to decay classes is based on examining the hardness of the wood by knife(c.f. Hottola & Siitonen, 2008). To relate decay class to physical and chemical properties of the wood, we measured the following variables: wet and dry weights of the samples, and the mass concentrations of carbon (C) and nitrogen (N).
As shown by Table S2, the measured variables co-vary with decay class in a systematic manner. Wet weigh was much greater in decay class 4 than in the other decay classes, and concentration of nitrogen increases more than 5-fold as decay class increases from 1 to 4, while the concentration of carbon stays stable.
Variable / DC1 / DC2 / DC3 / DC4 / pWet weight / 8,172895 / 8,16629 / 8,193382 / 12,31594 / 5,07E-09
Dryweight / 6,106842 / 5,340806 / 4,494118 / 3,484688 / 1,73E-07
N,mass % / 0,094128 / 0,124789 / 0,168491 / 0,4801 / 5,82E-09
C,mass % / 48,86956 / 48,68659 / 49,23492 / 49,81764 / 0,211112
Table S2. Mean values of the measured physical and chemical parameters for each decay class. The p-value gives the significance of decay class (considered as nominal variable in the ANOVA) in explainingthe variance for each of the variables.
Quantitative PCR analyses
As explained in the main text, we diluted each sample both to 0.05 ng/µl and 0.025 ng/µl total DNA concentrations prior the qPCR reactions. Both dilutions were run in a triplicate, thus we had in total six measurements for each sample. We estimated the fraction of fungal DNA out of all DNA, denoted here by f, by comparison to three standard dilution series based on pure fungal cultures (see main text).
We first examined the technical repeatability of the qPCR results by comparing the three measurements in each triplicate against each other. Most of the data were consistent, and we removed the outliers to reduce noise due to technical problems (Figure S1). The criteria used for exclusion of a data point was that the absolute difference between the log-10 transformed value of the variablef for the median (out of the three replicates) and the focal replicate was greater than 0.1.
We then averaged over replicate data points to yield one estimate for each dilution of each sample. The two dilutions gave generally consistent results, and we again removed the outliers to reduce noise due to technical problems (Figure S2). The criteria used for exclusion of a data point was that the absolute difference between the log-10 transformed values of the variable f for two dilutions was greater than 0.5.
After averaging the log-10 transformed f values obtained for two dilutions, we were left with one estimate per wood sample for the fraction of fungal DNA (f) and for the total DNA concentration (denoted here by q). For each log, there were two woodsamples: one pooled from the drilling holes in the basal part of the log and another one pooled fromthe drilling holes the middle part of the log. Total DNA concentrations were consistent among the two samples (p<0.001 in the regression) (Fig. 3). The fraction of fungal DNA was also consistent in the sense of a highly significant (p=0.009) positive slope in the linear regression, but it varied much more among the basal and middle parts of the log than total DNA concentration (Fig. 3). The DNA concentration was somewhat higher in the middle part of the log than in the basal part of the log (Fig. S3; p=0.04 for median test), whereas with fraction of fungal DNA there was no systematic difference.
To obtain one value per log (the analyses in the main text are conducted using log as the sampling unit), we averaged the log-10 transformed values over the two sampling locations (basal and middle parts of the log) both for f and q.We added the two log-10 transformed numbers together to give an estimate of total concentration of fungal DNA in each log.
Figure S1.Technical repeatability of qPCR measurements. The data points compare the three technical replicates (of each dilution of each sample) against each other. The variable f is the estimate for fraction of fungal DNA out of all DNA. Panel a shows the raw data, panel b the cleaned data after removal of outliers. The black line depicts identity.
Figure S2. Repeatability of qPCR measurements based on comparison between twodilution levels. Symbols as in Figure S1, but here the data points compare the two dilutions (0.05 ng/µl and 0.025 ng/µl) for data averaged over the technical replicates.
Figure S3.Comparison of total DNA concentrations (panel a) and fraction of fungal DNA out of all DNA (panel b) between basal and middle parts of the log. Black lines show linear regressions, red line shows identity line. In this figure, all samples with PEG treatment have been excluded.