Quantifying the efficiency and biases of forest Saccharomyces sampling strategies

Saccharomyces yeasts are emerging as model organisms for ecology and evolution, and researchers need environmental Saccharomyces isolates to test ecological and evolutionary hypotheses. However, methods for isolating Saccharomyces from nature have not been standardized, and isolation methods may influence the genotypes and phenotypes of studied strains. We compared the effectiveness and potential biases of an established enrichment culturing method against a newly developed direct plating method for isolating forest floor Saccharomyces spp. In a European forest, enrichment culturing was both less successful at isolating Saccharomyces paradoxus per sample collected and less labour intensive per isolated S. paradoxus colony than direct isolation. The two methods sampled similar S. paradoxus diversity: The number of unique genotypes sampled (i.e., genotypic diversity) per S. paradoxus isolate and average growth rates of S. paradoxus isolates did not differ between the two methods, and growth rate variances (i.e., phenotypic diversity) only differed in one of three tested environments. However, enrichment culturing did detect rare Saccharomyces cerevisiae in the forest habitat and also found two S. paradoxus isolates with outlier phenotypes. Our results validate the historically common method of using enrichment culturing to isolate representative collections of environmental Saccharomyces. We recommend that researchers choose a Saccharomyces sampling method based on resources available for sampling and isolate screening. Researchers interested in discovering new Saccharomyces phenotypes or rare Saccharomyces species from natural environments may also have more success using enrichment culturing. We include step‐by‐step sampling protocols in the supplemental materials.

Enrichment culturing is a reliable and frequently used method for isolating difficult-to-culture bacteria, archaea, and eukaryotic microbes, including Saccharomyces, from natural environments (Korzhenkov et al., 2019;Li, Podar, & Morgan-Kiss, 2016;Schlegel & Jannasch, 1967;Sniegowski, Dombrowski, & Fingerman, 2002;Figure 1a). Microbiologists have been relying on enrichment cultures for over a century (Beijernick, 1961) and have used them to isolate many of the model Saccharomyces strains commonly used in laboratory studies (Johnson et al., 2004;Liti et al., 2009;Sniegowski et al., 2002). To isolate a microbe using enrichment culturing, a researcher adds a small amount of natural material to a growth medium designed to be hospitable to the target microbe and inhospitable to other microbes (Liti, Warringer, & Blomberg, 2017;Schlegel & Jannasch, 1967). If the enrichment medium is well designed, the target microbe is expected to grow in abundance, and after some incubation time, this enrichment culture can be streaked to a solid medium and colonies of the target microbe can be easily isolated. An alternative to enrichment culturing is to spread a microbial substrate directly onto a selective solid medium, with or without dilution, and to pick colonies that morphologically resemble the target microbe (Glushakova et al., 2007;Stefanini et al., 2012;Figure 1b).
Because it can be difficult to isolate Saccharomyces from natural substrates, many investigations of wild Saccharomyces rely on enrichment culturing, usually in high-sugar, acidic media (Charron, Leducq, Bertin, Dube, & Landry, 2014;Robinson, Pinharanda, & Bensasson, 2016;Sniegowski et al., 2002;Sweeney, Kuehne, & Sniegowski, 2004). Comparative studies of Saccharomyces genomes have been carried out using collections of Saccharomyces strains isolated using various strategies, including both enrichment and direct culturing (Liti et al., 2009;Peter et al., 2018). However, isolation strategy can influence the genotypes and phenotypes of isolated microbes: Previous studies have documented higher genotypic diversity among bacteria isolated using direct plating compared with enrichment culturing, and FIGURE 1 Schematic illustration of sampling strategies used to isolate Saccharomyces for this project. (a) Enrichment culturing and (b) direct plating. Photo: Doreen Landermann the authors attributed these differences to selection for fast-growing phenotypes during enrichment (Dunbar, White, & Forney, 1997;Oda et al., 2008). We were concerned about the biases that might be introduced during enrichment culturing of Saccharomyces yeasts. For example, enrichment culturing might select for individuals with high relative fitness in the enrichment medium. Such potential biases in sampled yeast phenotypes are likely to lead to biases in sampled genotypes because genetic information is responsible for expressed phenotypes.
Isolation biases have also been suggested as potential explanations for differences between results of culture-dependent and cultureindependent studies of environmental Saccharomyces (Alsammar et al., 2018).
This study's goals were to compare isolation success between enrichment culturing and a direct culturing strategy and to quantify biases in Saccharomyces phenotype and genotype diversity that might be introduced when sampling a forest environment. We tested the assumption that it is easier to sample Saccharomyces from forest substrates using enrichment cultures than direct plating. We also investigated potential biases introduced by enrichment culturing by comparing growth rates and sampled genotype diversity between S. paradoxus (the wild sister species of the model laboratory yeast S. cerevisiae) colonies isolated using enrichment and direct strategies. Enrichment culturing might decrease or increase sampled S. paradoxus diversity compared with direct plating, thereby decreasing or increasing number of genotypes sampled, variance among growth rates, or both. For example, the number of unique genotypes sampled and the variance among growth rates would be low (and average growth rates high) among S. paradoxus isolated using enrichment cultures if the enrichment conditions select for the fastest growing S. paradoxus genotype present in every sample. Conversely, genotype diversity and variance among growth rates would be high among S. paradoxus isolated using enrichment cultures if diversity in the non-Saccharomyces microbial communities present on sampled substrates selects for diverse S. paradoxus among samples. S. paradoxus reproduction during enrichment may also influence sampled genotype diversity: Diversity within individual enrichment cultures may be low if a fast-growing genotype makes many asexual copies of itself or unique genotypes may be produced during enrichment culturing if S. paradoxus individuals sexually outcross with one another.
To test these predictions, we compared Saccharomyces sampling success and phenotype and genotype diversity among soil and leaf litter samples from a well-studied northern German forest (Kowallik & Greig, 2016;Kowallik, Miller, & Greig, 2015). A previous study showed that S. paradoxus is readily isolated using enrichment cultures from oak leaf litter in this forest (Kowallik & Greig, 2016). We were also previously able to isolate S. paradoxus directly from these forest substrates without enrichment (Kowallik, 2015). For the current study, we adapted a frequently used published enrichment method, which includes an enrichment step and two selective media, to design a direct plating method that included no enrichment steps and only one selective medium (Figure 1; Kowallik & Greig, 2016;Sniegowski et al., 2002). We aimed to remove as many potentially bias-inducing steps for the direct plating method, while still being able to isolate Saccharomyces spp., to understand whether these commonly used selective steps bias environmental Saccharomyces sampling.

| Field sampling and yeast isolation
All isolates were sampled from a mixed hardwood and conifer forest in Nehmten, Schleswig-Holstein, northern Germany (Nehmtener Forst).
We sampled approximately seven compressed ml total of each of leaf litter and soil from close to the bases of 10 oak trees at four sampling dates (Table 1), although not all trees were sampled at every date.
Trees were between 12 and 744 m from one another. At each date, samples were collected from leaf litter and the top organic layer of soil within 1 m of the base of each tree. Paired leaf litter and soil samples were collected on the north, south, east, and west side of each tree at all collection dates except April 7, when samples were collected at an arbitrary two of the four cardinal directions.
Material was collected simultaneously for the direct plating and enrichment collections at each sampling point ( Figure 1). First, leaf litter was collected by aseptically transferring litter into sterile collection tubes: approximately 5 ml of compressed leaf litter was collected for the direct plating method and approximately 2 ml for the enrichment method. Then, the remaining leaf litter was removed from the soil surface and the top approximate 2 cm of soil (mostly composed of soil organic layer) was aseptically transferred into sterile collection tubes.
As for leaf litter, approximately 5 ml of compressed soil was collected for the direct plating method and approximately 2 ml for the enrichment method. Instruments were sterilized between samples using 70% ethanol. Samples were transported between the field and lab at ambient temperature and processed within 4 hr of collection.
For direct plating (Figure 1a), material was mixed with 20-ml sterile water in a sterile 50-ml tube, the mixture was vigorously mixed for at least 10 s with a vortex mixer on its highest setting, and 0.2 ml of the resulting dirty liquid was pipetted on each of two plates containing the solid modified selective medium PIM1 (3-g yeast extract, 5-g peptone, 10-g sucrose, 3-g malt extract, 1-mg chloramphenicol, 80-ml ethanol, 5.2-ml 1 M HCl, and 20-g agar per litre; Kowallik & Greig, 2016;Sniegowski et al., 2002). Liquid was spread on plates using sterile glass beads, and plates were left open in a laminar flow hood until dry.
Plates were incubated for 3 days at 30°C before colonies were picked.
For enrichments (Figure 1b), material was mixed with 10 ml of the liquid selective medium PIM1 (composition as for solid PIM1 but without agar) in a 15-ml sterile tube, mixtures were inverted, and tubes were incubated, slightly open and without shaking, at 30°C. After 10 days, a sterile wooden stick was inserted into each enrichment tube and a small amount of the liquid (approximately 50 μl) was streaked onto a single plate containing the solid selective medium PIM2 (20-g methyl-(alpha)-D-glucopyranoside, 1-ml 5% Antifoam Y-30 emulsion, 6.7-g yeast nitrogen base without amino acids, 4-ml 1 M HCl, and 20-g agar per litre; Kowallik & Greig, 2016;Sniegowski et al., 2002), and plates were incubated 4 days at 30°C before colonies were picked.
We include these procedures as step-by-step protocols for the convenience of future researchers in Data S1.

| Yeast identification
After incubation, we streaked colonies with yeast-like morphology to fresh YPD medium (10-g yeast extract, 20-g peptone, 20-g dextrose, and 25-g agar per litre). For each method, up to six (March and April sampling days) or 12 (June and July sampling days) colonies per sample were selected. After 1 day of growth on YPD at 30°C, cultures were frozen at −80°C in 20% glycerol, and a small amount of each culture was transferred to sporulation medium (20-g potassium acetate, 2.2g yeast extract, 0.5-g dextrose, 870-mg complete amino acid mixture, and 25-g agar per litre). Any cultures with bacteria-like morphology on YPD medium (slimy culture and/or cells smaller than 1 micron across) were not frozen and were discarded. Sporulation cultures were incubated for at least 3 days at room temperature before being screened under a compound microscope for Saccharomyces-like asci (tetrads).
All cultures producing tetrads were identified using sequencing of the internal transcribed sequence (ITS), a region neighbouring rRNAcoding DNA (Schoch et al., 2012). We sequenced every strain using the ITS1/ITS4 primer pair (White, Bruns, Lee, & Taylor, 1990 (Zhang, Schwartz, Wagner, & Miller, 2000). If the sequence aligned with Saccharomyces sequences but had more than one base pair different from its closest match, we supplemented ITS sequences with sequences from the gene for translation elongation factor 1 using primers EF1-983F and EF1-2212R (Rehner & Buckley, 2005) using the protocols above, but with a PCR annealing temperature of 57°C. In some cases, cultures originating from apparent single colonies were in fact mixtures of two yeast species. We counted these colonies as Saccharomyces if sequences from one of the species was Saccharomyces.

| Growth rates
We compared the distributions of maximum growth rates between the two groups of S. paradoxus strains (strains collected using enrichment culturing and strains collected using direct plating) in three liquid media. The media were liquid PIM1, a minimal yeast medium (1.7-g yeast nitrogen base without amino acids and ammonium, 5-g ammonium sulphate, and 2.5-g dextrose per litre), and liquid YPD (composition as for solid YPD, but without agar

| Genotyping
Nine microsatellite loci were identified by searching for common S. cerevisiae repeats in the reference genome of S. paradoxus strain CBS432 (Liti et al., 2009;Young, Sloan, & Van Riper, 2000)  repeats; one locus was two-nucleotide repeats; and one locus was four-nucleotide repeats. All loci are described in Table 2. Some loci were complex, including repeats with different sequences; when analysing data, we assumed that alleles of these loci with the same length had the same sequence.
All S. paradoxus strains for which growth rates were measured (see above) were genotyped. We amplified microsatellite regions as previously described (Babiker & Tautz, 2015;Hardouin et al., 2015), with slight modifications. Reactions were carried out in 5-μl PCR mixes containing one colony of each S. paradoxus isolate, 2.5-μl 2× Qiagen Multiplex PCR master mix, and 0.2-μM each primer. Forward primers were labelled with either FAM, HEX, or NED at the five-prime end, and we multiplexed 4-5 primer pairs in each reaction. PCR cycling, dilution, and denaturation were carried out as previously described (Babiker & Tautz, 2015;Hardouin et al., 2015); fragments were run on an ABI 3730 DNA analyser and were analysed using Geneious 8.1.8 with microsatellite plugin version 1.4.4. Genescan ROX-500 (ThermoFisher Scientific) was used as a size standard. All nine microsatellite loci showed variation in the collection of S. paradoxus isolates: The lowest number of length polymorphisms detected for any locus was two and the maximum was 10.

| Statistical analyses
We compared sampling success across substrates (leaf litter or soil) and methods (direct plating or enrichment) using a generalized linear mixed-effects model with probability of isolating Saccharomyces We compared growth rate distributions by first comparing variances using Levene's test for homogeneity of variance (Levene, 1960) and then comparing medians using paired Wilcoxon signed rank tests. We visualized relationships among genotypes using a neighbourjoining tree of Edwards' genetic distance (Edwards, 1971 (Domizio et al., 2011;Gschaedler, 2017).
Although the direct plating method was more successful than the enrichment method, it was also more labour intensive (Table 5). We found more colonies with S. paradoxus-like morphology, including colonies that belonged to non-Saccharomyces genera, using the direct plating method (969) than using the enrichment method (284), and we screened all of these colonies for tetrad formation. As a result, we screened more than three times as many colonies for tetrads when using the direct plating method than we did using the enrichment method. After screening for tetrads and ITS sequencing, only 32% of the total isolated direct plating colonies were S. paradoxus, compared with 74% of enrichment colonies.
Both methods isolated Saccharomyces colonies from both substrates, most trees, and all timepoints ( Figure 2). We had significantly more sampling success on soil than leaf litter substrates (z = 5.7, p < .001, Table 4), but other relationships among sampling success, sampling method, and sampling environments were idiosyncratic. For example, direct plating did not produce any Saccharomyces isolates from Tree 6, whereas three enrichment samples from this tree isolated S. paradoxus, and enrichments produced more Saccharomyces isolates in March than direct plating did (Figure 2). Because our sampling effort was not the same for all trees at all months, we did not model tree habitat or sampling month as fixed effects; instead, we modelled these parameters as random effects and found that models including tree and month fit the data better than models without tree and month (Table 3).

| Phenotype diversity of sampled Saccharomyces paradoxus
Growth rate distributions did not differ between the two methods in PIM1 and the minimal medium and differed slightly in variance in the YPD medium ( Figure 3 and Tables 6, 7). Median growth rates did not differ significantly between the two methods in any of the three tested media (Table 6), and variances in growth rate only differed significantly in YPD (Levene's test F (1, 108) = 5.42, p = .022, Table 7), with enrichment cultures isolating a wider variance of S. paradoxus growth rates in YPD than direct plating (Figure 3c). When two outlier strains were removed (Figure 3c), this difference disappeared, F (1, 106) = 3.59, p = .06.

| Genotype diversity of sampled Saccharomyces paradoxus
The two isolation methods sampled equivalent genotype diversities, both across and within samples. In total, we found 21 unique clonal genotypes (Figure 4). The minimum number of clones per genotype was one and the maximum was 55 ( Figure S1). The enrichment method discovered 17 genotypes (95% confidence interval [12.1, 21.9]), and the direct plating method discovered 12 genotypes (95% confidence interval [8.8, 15.2]), but this difference was not significant   (Kowallik & Greig, 2016;Naumov, Naumova, & Sniegowski, 1998;Sniegowski et al., 2002). We expect reliable Saccharomyces isolation from this forest using direct plating to be a result of high S. paradoxus abundance on forest floor substrates. Indeed, a previous study determined that hundreds to tens of thousands of S. paradoxus cells can occupy a gram of leaf litter near the bases of oak trees in this forest (Kowallik & Greig, 2016). These quantitative observations were made by serially diluting enrichment cultures and estimating the number of S. paradoxus cells per gram of leaf litter based on the highest dilution in which S. paradoxus could be found. We expect direct plating to be less successful in environments in which Saccharomyces are rarer, and note that enrichment culturing is frequently used to isolate Saccharomyces from tree bark, which may be a habitat with lower Saccharomyces density than the forest floor habitats we sampled Sniegowski et al., 2002). S. paradoxus abundance can also vary over time, with spikes after environmental changes such as rain events (Anderson et al., 2018;Glushakova et al., 2007). It is possible that environmental conditions at other locations, or characteristics of non-European S. paradoxus populations, would result in different sampling successes using these two methods from that reported here.
It was not possible to completely standardize quantities of sampled natural material when comparing direct and enrichment-based sampling methods. We collected a larger volume of material for direct cultures (~5 ml) than for enrichment cultures (~2 ml), but the proportion of the original enrichment sample ultimately screened for Saccharomyces colonies depends on processes occurring during enrichment. For direct plating, we screened 400 μl of the 25-ml total suspension of soil

| Both isolation methods sampled similar Saccharomyces paradoxus diversities from forest substrates
Overall, enrichment culturing and direct plating collected similar phenotypic and genotypic diversity (Figures 3 and 4). We found no evidence that enrichment culturing selected for fitter individuals in the enrichment medium than direct plating (Figure 3a). Although we genotyped many representatives of the same clonal genotypes ( Figure S1), clonal reproduction inside of enrichment cultures did not decrease sampled diversity. High clonality in a local area is common for wild S. paradoxus populations and is most likely a result of extensive asexual reproduction in natural habitats (Tsai, Bensasson, Burt, & Koufopanou, 2008;Xia et al., 2017). We also found no evidence for sexual outcrossing in the enrichment cultures themselves. If outcrossing had occurred during enrichment, we would expect to have seen heterozygous F1 offspring among the genotypes isolated using enrichment. Instead, the only heterozygous genotype in our collection was isolated from a single environment using both enrichment and plating methods and was unlikely to have arisen during enrichment.
The enrichment method did isolate some outlier S. paradoxus phenotypes and S. cerevisiae that the direct plating method did not (Figures 2 and 3c). We did not find many of these outliers, but we speculate that diverse interactions with microbes in enrichments may have led to isolation of outlier phenotypes and Saccharomyces spp. For example, the isolated outlier Saccharomyces may have come from enrichments containing bacteria that promoted outlier S. paradoxus or S. cerevisiae growth at the expense of other S. paradoxus genotypes. Microbial diversity across enrichment cultures may similarly explain our idiosyncratic sampling success across months and trees (Figure 2). For example, it is possible that a bacterium that inhibits S. paradoxus growth in the enrichment medium was more common in summer than spring months, resulting in lower enrichment sampling success in summer.

| Recommendations for future yeast sampling
Our results identified a trade-off between resources spent on sampling and resources spent on sequencing: Enrichment culturing was less successful than direct plating at finding Saccharomyces per sample collected, but more successful per ITS region sequenced ( Figure 2 and Although, on average, both methods sampled similar phenotypic and genotypic diversity, our isolation of outlier isolates using enrichments suggests that researchers targeting outliers may also prefer enrichment culturing. For example, researchers sampling environments to find unusual Saccharomyces phenotypes for applied biotechnology (e.g., food microbiology and drug discovery) may uncover more diversity using enrichment culturing. Researchers interested in detecting rare Saccharomyces species in an environment (e.g., S. cerevisiae from our study forest and S. mikatae and Saccharomyces eubayanus from European forests; Alsammar et al., 2018) may also have more success using enrichment culturing.

FIGURE 4
Genotype rarefaction curves of genotypes detected. Thick lines represent average genotypes observed as a function of isolates sampled and shaded areas represent 95% standard errors

| CONCLUSIONS
Our results validated use of enrichment culturing for isolating diverse and representative collections of S. paradoxus from natural material.
We found no evidence that processes during enrichment culturing decrease the diversity of sampled Saccharomyces spp. and weak evidence that these processes may in fact increase sampled diversity.
Although it is generally a good idea to standardize sampling methods within a study as much as possible, conclusions from studies comparing Saccharomyces genotype and phenotype diversity from a variety of sources, including culture collections, are likely to be reliable (Strope et al., 2015;Warringer et al., 2011) and the diversity found in culture collections is likely to be representative of natural Saccharomyces diversity in sampled environments. In addition to validating the frequently used enrichment method for isolating Saccharomyces spp., this study provides a reliable direct method for isolating Saccharomyces spp. and describes a set of microsatellite markers that can be used to conveniently identify S. paradoxus genotype diversity. The utility of Saccharomyces as an ecology and evolutionary model relies on our understanding of its natural history, and we hope that these and other improvements in field sampling methods will empower researchers to explore the environmental contexts of these exciting microbial model organisms.

ACKNOWLEDGEMENTS
We would like to thank Danielle Stevens and Tjorben Nawroth for help in the field, Jenna Gallie and the Gallie lab for help with growth curves, Michael Habig for advice on statistics, and Amine Hassani for helpful conversations on microbial diversity in enrichment cultures.
Thank you to Christoph Freiherr von Fürstenberg-Plessen for permission to work in the Nehmten forest. This work was supported through a Max Planck Fellowship to Eva H. Stukenbrock.

DATA ACCESSIBILITY
All data for this project, including a list of identified yeasts, sampling success data for each soil and leaf litter sample, isolate growth rates, and isolate genotype data, have been deposited in the Edmond repository (doi:10.17617/3.2i).