METHODOLOGICAL INFORMATION
1. Description of methods used for collection/generation of data: 

These methods are published in PLOS ONE:
Schilling, JS, Kaffenberger, JT, Liew, FJ, Song, Z. (2015.) Signature Wood Modifications Reveal Decomposer Community History. PLoS ONE 10(3): e0120679. doi: 10.1371/journal.pone.0120679. 

Field research on private forest land in Waingaro Forest, New Zealand grant through the P.F. Olsen Company by W.-Y. Wang. 

Isolate survey
Isolate selection
We tested 29 wood-degrading fungal isolates on white birch (Betula papyrifera) and southern yellow pine (Pinus spp.) substrates to test the strength of association between rot type, L:D, and DAS. This approach is similar to Worrall et al. [4] who focused primarily on brown rot isolates from Dacrymycetales (n=21), plus 10 isolates from 3 other Genera (Coniophora, Gloeophyllum, and Postia), totaling 4 brown rot clades [31]. We tested a broader group of ancestral lineages (Table 1), and specifically included five exoglucanase-producing brown rot Boletales fungi [32], and wood-degrading fungi with unusual nutritional modes, specifically Schizophyllum commune and four soft rot isolates [1,5]. All isolates were maintained on the media type (e.g. potato dextrose agar) recommended from their culture collection sources prior to inoculation as plugs into agar-block microcosms. Culture collection availability is listed in Table 1.

Microcosm set-up
In agar-block microcosms, petri plates containing 1% malt agar were inoculated with a single 7-mm plug and allowed to develop until the mycelial edge met the plate margin. Birch (Betula spp.) and pine (‘southern yellow,’ likely Pinus taeda) wafers to be added to these microcosms were 2-mm thick and longitudinally cut. The wafers were oven-dried (100oC, 24 hrs), weighed, sterilized (121oC, 16 psi, 1 hr), and then added without wetting atop this developed ‘lawn’ of hyphae. Two wafers were stacked when added, with 3 stacks per plate and per wood type. Stacking offset excess moisture wicking or drying, and weight loss by week 12 exceeded 50% in many cases, verifying design. Microcosms were in triplicate per fungal isolate and wood type. Plates were incubated at room temperature in the dark, and harvests were made at weeks 3, 6, and 12, using aseptic technique for the first two harvests and a destructive final harvest.

Sample processing
At the time of harvest, top wafers were extracted and oven-dried as before to determine mass loss. Wafers were then ground in a Wiley mill to 20-mesh and processed for 1) Klason lignin analysis using standard 72% H2SO4 hydrolysis of 300 mg wood and gravimetric determination of acid-insoluble lignin [41] and 2) DAS (wt%) following Shortle et al. [40]. Although lignin analysis is straightforward, we will note that lignin time-zero contents were measured in non-degraded wood to calculate lignin loss rather than inferred from published values. For the sake of explaining DAS methodology in more detail here, we suspended 101-110 mg of milled, oven-dried (100oC, 24 hr) wood in 5 ml of 0.2 NaOH in glass scintillation vials, loosely capped the vials, and then autoclaved them for 15 min at 121oC. Cooled residue mixture was then filtered using pre-weighed (oven-dry wt) fritted glass crucibles (porosity C), using 3 x 10 ml deionized water to rinse the residue on the filter. This was rinsed in 2 x 10 ml aliquots of 0.1 M HNO3, and then in 3 x 10 ml deionized water. The crucibles were again oven-dried for 24 hrs and weighed, allowing gravimetric calculation of loss on extraction as DAS (wt%). 

2. Methods for processing the data: 

Data analysis 
Lignin data analyses were straightforward as L:D data were continuous along a scale of lignin selectivity (0.0 – 5.0, with an approximate threshold where brown rot <0.8 and white rot >0.8), although there remained a density loss minimum due to variability in the earliest decay stages (set at 5%). The DAS on the other hand involved making a binary decision on rot type, so correlations with mass loss for fungi on each wood type were used to assess three variables affecting the design of a field trial: 1) the threshold of density loss at which brown rot became statistically discernable from white, soft, and no rot, 2) the relationship between the number of sample replicates and this density threshold, and 3) the DAS (wt%) threshold at a given density above which one could call the rot type ‘brown rot.’ Interactions between decay type and mass loss with DAS were analyzed to assess these threshold values with the software package Minitab 16 (Minitab Inc., State College, PA). Since the extent of decay in pine was low in our study, pine mass loss data were supplemented with those provided in Worrall et al. [4].

For determining a threshold of density loss above which DAS could be used to distinguish brown rot, we independently fitted DAS for each of the four decay type/substrate groups to linear, quadratic, and cubic regression equations using mass loss as the predictor variable. Since these models differ in complexity, an F test described by Motulsky and Ransnas [42] was used to determine if SSE reduction associated with the additional model parameters justified reduced degrees of freedom.  For each group, the simplest model whose SSE was not significantly different from the model with the lowest SSE was selected. Residual plots were used to verify that data were randomly and uniformly scattered in each model fit and the normality of the residuals was verified with Anderson-Darling tests. For each group, the corresponding model equation was used to predict solubility values and associated 95% prediction intervals for specific mass loss values ranging from 0 – 55%. For each substrate, the fits and prediction intervals for the model equations of each decay type were plotted. Since the prediction interval provides a range in which the next observed solubility is likely to fall, the mass loss at which the prediction intervals of the two groups no longer overlap is the point at which we can be 95% certain that the next observation will not fall in a region where either group is likely to be observed.  Above that mass loss, the next observed solubility is 95% likely to belong to the group in whose prediction interval the observation falls.  For this reason, the mass loss at which the upper prediction limit of the white rot fit and the lower limit of the brown rot fit intersect was taken as the mass loss above which brown rot is distinguishable.  

To determine required sample size at the threshold between decay class I and II (9.5% and 26.4% mass loss in birch and pine, respectively), mean DAS values for each decay type were predicted at each mass loss threshold using the appropriate regression model [42]. Using a 95% confidence interval, a power of 0.8, and assuming an equal sample size ratio of white and brown rot, power analysis was performed to determine the sample size required to discern decay types at each mass loss threshold. 

Lastly, to determine a DAS threshold above which one could proclaim the decay type as brown rot, binary logistic regression models were fit to experimental data for each substrate using this equation: 
logit (P(brown rot)) = B0 + B1 * mass loss + B2 * DAS + B3 * mass loss * DAS	
Pearson’s Chi-squared fit test for pine [?2 = 295.6, DF = 318, P = 0.811] and birch [?2 = 213.012, DF = 282, P = 0.999], indicated good model fits.  The fitted parameters and equation were used to predict P=0.1, P=0.5, and P=0.9 interfaces between the decay type groups for each substrate. 
Additionally, Kruskal-Wallis one-way ANOVA tests were performed to determine the significance of difference between median L:D values between decay type groups. Kruskal-Wallis was selected given the non-normality of the data, even after transformation.  


3. Environmental/experimental conditions:

Lab Microcosms
This is described, above.

Field trial
Using a paired-plot approach and drill bit extractions to sample within and among logs, we sampled the eastern and western margins of a radiata pine (Pinus radiata) plantation in Northwestern New Zealand. The forest site was located in the Waingaro Springs Forest, a private 5 km2 radiata pine plantation planted in 1994 [45] (37o41’ S, 174o59’ E, permit P.F. Olsen Co., W.-Y. Wang). The site is adjacent to the Tasman Sea where it receives prevailing winds from the west. We targeted decay class II/III P. radiata logs and analyzed replicate samples (n=5) from eight logs on the western plot and ten on the eastern plot, extracted using a cleaned 3.7 mm (3/16 Imperial) drill bit and collecting sawdust into sterile 1.5 ml microcentrifuge tubes. Samples were first processed for DAS using two approaches: 1) individually in the western plot, to test within and among log variability, and 2) with four technical replicates from a pooled sample of the eastern plot samples. Again, we had a preliminary hypothesis, in this case that the plot facing prevailing winds might have lower brown rot incidence than the plot downwind of the forest site center, where brown rot fungi and their spore loads might increase in the conifer-rich system. Our specific goals, however, were to assess within- and among-log DAS variability with the drill bit approach in the absence of a density measure and to test a pooled sawdust approach, both as space- and time-saving options for sampling deadwood. We used the DAS data as a screen for rot type, with the plan to use L:D if brown rot was present in significant amounts, but this was deemed unnecessary at the time.

4. Describe any quality-assurance procedures performed on the data:

QC was via negative controls, with technical replicates to complement experimental replication.


5. People involved with sample collection, processing, analysis and/or submission:
Jonathan S. Schilling, Justin Kaffenberger, Feng Jin Liew, Zewei Song

6. Data-Specific Information
FILNEAME: Pine and Birch Isolate Survey
FILE DESCRIPTION: Spreadsheet of Pine and Birch Isolate Data
  A. Code/symbol: ODW
   B. Definition: Oven-dry weight
   A. Code/symbol: Cruc
   B. Definition: Crucible (Gooch coarse)
   A. Code/symbol: L:M
   B. Definition: Lignin-to-Mass loss ratio
   A. Code/symbol: stdev
   B. Definition: Standard deviation
   A. Code/symbol: sterr
   B. Definition: Standard error

FILENAME:Figure 1 Data Distillations
FILE DESCRIPTION: Figures of Distillation Data
   A. Code/symbol: DAS
   B. Definition: Dilute Alkali Solubility

FILENAME: Density Bias Data
FILE DESCRIPTION: Spreadsheet of Density Bias in Sample
   A. Code/symbol: WR
   B. Definition: White rot
   A. Code/symbol: BR
   B. Definition: Brown rot
   A. Code/symbol: OD
   B. Definition: Oven-dried
   A. Code/symbol: LOE
   B. Definition: Loss on extraction

FILENAME: Legacy Effect Data
FILE DESCRIPTION: Spreadsheet of Legacy Effect Data
   A. Code/symbol: Gt
   B. Definition: Gloeophyllum trabeum
   A. Code/symbol: se
   B. Definition: Standard Error
   A. Code/symbol: ecomp
   B. Definition: east site composite sample replicates
   A. Code/symbol: e
   B. Definition: east site
   A. Code/symbol: w
   B. Definition: west sites
   A. Code/symbol: wt
   B. Definition: weight
   A. Code/symbol: OD
   B. Definition: oven-dry
   A. Code/symbol: DAS
   B. Definition: dilute alkali solubility
   A. Code/symbol: SE
   B. Definition: standard error
   A. Code/symbol: reps
   B. Definition: replicates