COLLECTIONS AND ANALYSIS OF PHENOTYPE
The isolates originated from three natural sources: isolations from ascospores of Hypocrea specimens, direct isolations by a variety of means from soil or dead herbaceous tissue, and as isolations as endophytes from living stems of Theobroma and related tree species as reported by Evans et al. (2003). A smaller number of isolates was obtained from the American Type Culture Collection (ATCC), Centraalbureau voor Schimmelcultures (CBS) and colleagues. Cultures derived from single part-ascospores that were germinated on cornmeal agar with 2 % dextrose (CMD, Difco cornmeal agar + 2 % dextrose w/v) and isolated using a micromanipulator; usually two or more single-spore cultures were combined in a single stock culture and polyspore cultures were used in all subsequent analyses. Representative cultures are deposited in ATCC and CBS. Kornerup & Wanscher (1978) was used as the colour standard. Isolates and their GenBank numbers are listed in Table 1. The name of the most commonly cited collectors are abbreviated as G.J.S. (G.J. Samuels) and C.T.R. (C.T. Rogerson).
Cultures used for study of anamorph micromorphology were grown on CMD or, less frequently SNA (without filter paper, Nirenberg 1976), at 20 or 25 °C for 7–10 d under alternating 12 h cool white fluorescent light and 12 h darkness; in the descriptions that follow, these alternating light conditions are referred to when the word "light" is used. Approximately 20 mL of agar was poured into Petri dishes.
We did not observe any difference in anamorph morphology between CMD and SNA but there was a tendency for more reliable conidial production on SNA than on CMD. Conidial pustules of Trichoderma isolates grown on these two media appeared to be more similar to how they appear in nature than conidia formed on other commonly-used media, including potato-dextrose agar, malt agar and oatmeal agar (Gams et al. 1998).
Morphological analysis of microscopic characters was undertaken from material that was first hydrated in the case of herbarium material, or wetted in the case of living cultures, in 3 % KOH. The KOH was subsequently replaced by distilled water. Measurements were made from KOH or water; we did not observe any differences between the two reagents. Where possible, 30 units of each parameter were measured for each collection. Ninety-five percent confidence intervals of the means (CI) are provided; this figure represents the interval within which 95 % of the individuals of the parameter will be found. The parameters used for analysis are listed in Table 3. Chlamydospores were measured by inverting a 7–10 d old CMD culture on the stage of a compound microscope and observing with a 40 x objective. Data were gathered using a Nikon DXM1200 digital camera and Nikon ACT 1 software and measured using Scion Image (release Beta 4.0.2; Scioncorp, Frederick, MD).
Five types of microscopy were used, viz. stereo microscopy (stereo), bright field (BF), phase contrast (PC), Nomarski differential interference contrast (DIC) and epifluorescence (FL). The fluorescent brightener calcofluor (Sigma Fluorescent Brightener 28 C.I. 40622 Calcofluor white M2R in 2 molar phosphate buffer at pH 8.00) was used for FL.
Sections of Hypocrea stromata were prepared by rehydrating small blocks of substratum supporting stromata in 3 % KOH. The blocks were supported by Tissue Tek O.C.T. embedding medium 4583 (Miles, Inc., Elkhart, IN) and sectioned at about 15 µm on a Microtome-Cryostat (International Equipment Co., Needham Heights, MA). The sections were first floated in water and then placed on slides to make semi-permanent preparations following Volkmann-Kohlmeyer & Kohlmeyer (1996). Slides are deposited with the specimens.
Growth rate trials were performed in darkness on potato-dextrose agar (PDA, Difco or Sigma) and SNA following the procedure described by Samuels et al. (2002) with the addition that cultures were also grown at 25 °C under 12 h darkness/12 h cool white fluorescent light for 96–120 h. Each growth-rate trial was repeated three times and the results of the three were averaged.
The slope of the growth curve, which reflects rate of growth per hour, is determined by linear regression. Regression is used to characterize the manner in which the colony radius changes (x's) with the time (y's) when measurements of colony radius are made. By revealing how the mean of the y measures changes as the various x measures change, the regression line is understood to describe the regression of y (colony radius) over x (time of measurement). This regression line is the slope of the growth curve; it is the predicted value of each colony radius for each time of measurement and essentially reports growth per hour (see http://www.animatedsoftware.com/statglos/sgregres.htm).
Principal Components Analysis (PCA), a multivariate analysis (Multivariate Statistical Package, version 1.131; Kovach Computing Services, U.K.), was utilized to determine patterns of variation of phenotype within phylogenetically defined groups. The eigenanalysis is shown in Table 2 and graphical output is shown in Fig. 4. The standardized data used in PCA, and other data analyses, were obtained using Systat version 10 (SPSS Inc., Chicago, IL, U.S.A.).
Dry cultures of Trichoderma species were prepared by placing all or part of a culture growing in 9-cm-diam Petri dish in a cardboard two-slide micro-slide holder (e.g. VWR Scientific, West Chester, PA, U.S.A.) and drying them for ca. 2 h over low heat of a fruit dryer. Dry cultures were prepared so as to preserve essential characters of conidiophore branching and phialides.
DNA EXTRACTION, AMPLIFICATION AND SEQUENCING
The extraction of genomic DNA was performed as reported previously (Dodd et al. 2002).
The PCR for amplification of the internal transcribed spacers 1 and 2 of the rDNA gene cluster (ITS1 and 2 including the 5.8S RNA gene) was performed in a 50 µL reaction volume using 5 µL of 10 x PCR buffer (Applied Biosystems), 200 µM dNTPs, 25 pmole of each primer (ITS1 and ITS4), 1.25 units AmpliTaq Gold (Applied Biosystems), and about 10–50 ng of template DNA. The reaction mixture was placed in a 0.2 mL PCR tube. The PCR was carried out on a PT-200 PCR system (MJ Research, Waltham, MA, U.S.A.) according to the following protocol: initial activation of AmpliTaq Gold at 95 °C for 10 min; 30 cycles of denaturation at 95 °C for 30 s, annealing at 55 °C for 30 s, extension at 72 °C for 1 min; and a final extension period at 72 °C for 10 min. Five µL of the PCR product was analyzed on 1 % agarose gel in TAE buffer. The positive PCR reactions were purified using the Qiagen QIAquick PCR purification kit (Qiagen, California, U.S.A.) following the manufacturer's instructions. The concentration of the PCR products in ng/µL was determined on 1 % agarose gel electrophoresis in TAE buffer with Lambda Hind III DNA as a marker.
Similarly, a portion of translation elongation factor 1 alpha (tef1) was amplified using the primers EF1-728F (Carbone & Kohn 1999) and TEF1 rev (Samuels et al. 2002), which resulted in a PCR product of approximately 600 bp, and was sequenced in both directions. The primers for amplification of the calmodulin-encoding gene (cal) were CAL-228F and CAL-737R (Carbone & Kohn 1999). Initially a fragment of actin gene (act) was amplified using the primers Fung.ACT.F1 and Fung.ACT.R1 and the conditions described by Wirsel et al. (2002). Based on the sequences obtained, two Trichoderma-specific act primers were designed Tact1 (5'-TGGCACCACACCTTCTACAATGA) and Tact2 (5'-TCTCCTTCTGCATACGGTCGGA). These two primers were used for amplification of act for all the isolates in this study. Additionally two sequencing primers for act were designed called Tact500F (5'-ATTCCGTGCTCCTGAG) and Tact511R (5'-CTCAGGAGCACGGAAT) and were used for sequencing reactions.
The portion of the RNA polymerase subunit B 2 (rpb2) gene was amplified and sequenced as described by Chaverri & Samuels (2004) using fRPB2-5F and fRPB2-7cR (Liu et al. 1999) as forward and reverse primers, respectively.
DNA sequences were obtained using the BigDye Terminator cycle sequencing kit (Applied Biosystems, Foster City, California). Products were analyzed directly on a 3100 DNA sequencer (Applied Biosystems). Both strands were sequenced for each gene.
ANALYSIS OF SEQUENCE DATA
Sequences were edited and assembled using Sequencher 4.1 (Gene Codes, WI). Clustal X (Thompson et al. 1997) was used to align the sequences; the alignment of each locus was manually edited using MacClade and GeneDoc 2.6 (Nicholas & Nicholas 1997). The sequences were deposited in GenBank (Table 1) and alignments were deposited in TreeBase (http://herbaria.harvard.edu/treebase/), submission number SN 1008). The multiple sequence alignment file for the tef1 locus is also available at http://www.isth.info/phylogeny/koningii.php.
The interleaved NEXUS file was formatted using PAUP* v. 4.0b10 (Sinauer Associates, Sunderland, MA) and manually formatted for the MrBayes v3.0B4 program. The Bayesian approach to phylogenetic reconstructions (Rannala & Yang 2005) was implemented using MrBayes 3.0B4 (Huelsenbeck & Ronquist 2001). The MODELTEST3-06 package (http://bioag.byu.edu/zoology/crandall_lab/modeltest.htm) was used to compare the likelihood of different nested models of DNA substitution and select the best-fit model for the investigated data set. The modelblock3. nex which is compatible with the current version of PAUP* v. 4.0b10 was downloaded from http://workshop.molecularevolution.org/software/modeltest/files/modelblock3. Both hierarchical LRT and AIC output strategies were considered, although the preference was given to the last one. The unconstrained GTR + I + G substitution model was selected for all tree loci.
Metropolis-coupled Markov chain Monte Carlo (MCMCMC) sampling was performed with four incrementally heated chains with the default heating coefficient 
= 0.2, heats for cold chains 1 and heated chains 2, 3 and 4 are 1, 0.83, 0.71 and 0.63, respectively) that were simultaneously run for 5 million generations for the tef1 alignment, which comprised more than 200 sequences. Alignments of the other two loci (cal and act), neither of which exceeded 100 sequences, were analysed using 3 million generations. To check for potentially poor mixing of MCMCMC, each analysis was repeated at least three times. The convergence of MCMCMC was monitored by examining the value of the marginal likelihood through generations. Convergence of substitution rate and rate heterogeneity model parameters were also checked. Bayesian posterior probabilities (PP) were obtained from the 50 % majority rule consensus of trees sampled every 100 generations after removing the first 2000 trees for tef1 and the first 500 for cal and act using the "burn" command. According to the protocol of Leache & Reeder (2002), PP values lower then 0.95 were not considered significant, while values below 0.9 are not shown on phylograms and radial trees. Model parameter summaries after MCMC run and burning first samples were collected. For tef1 mean substitution values were estimated as G
T =1, CT = 3.33, CG = 1.14, AT = 1.32, AG = 5.98, AC = 1.43; nucleotide frequencies were estimated as 0.19(A), 0.28(C), 0.17(G), 0.36(T); alpha parameter of gamma distribution shape was 0.23. For cal mean substitution values were estimated as GT =1, CT = 4.43, CG = 0.83, AT = 1.15, AG = 3.55, AC = 1; nucleotide frequencies were estimated as 0.26(A), 0.26(C), 0.24(G), 0.24(T); alpha parameter of gamma distribution shape was 0.1. For act mean substitution values were estimated with a high affinity to pyrimidine transitions (CT = 81.9); other transitions were GT =1, CG = 0.3, AT = 0.85, AG = 0.83, AC = 0.61; nucleotide frequencies were estimated as 0.2(A), 0.3(C), 0.24(G), 0.26(T); alpha parameter of gamma distribution shape was 0.09. The genetic distance was computed in PAUP* v. 4.0b10 under the GTR + I model.
Answers to all your Biology Questions
