Genetic Admixture Increases Phenotypic Diversity in the Nectar Yeast Metschnikowia Reukauﬁi � * Sergio Alvarez-Perez� A, B, 1, Manpreet K

Fungal Ecology 49 (2021) 101016

Contents lists available at ScienceDirect

Fungal Ecology

journal homepage: www.elsevier.com/locate/funeco

Genetic admixture increases phenotypic diversity in the nectar yeast Metschnikowia reukauﬁi * Sergio Alvarez-Perez a, b, 1, Manpreet K. Dhami c, d, , 1, María I. Pozo e, Sam Crauwels a, ** Kevin J. Verstrepen f, Carlos M. Herrera g, Bart Lievens a, Hans Jacquemyn e, a Laboratory for Process Microbial Ecology and Bioinspirational Management, Department of Microbial and Molecular Systems, KU Leuven, Leuven, Belgium b Department of Animal Health, Faculty of Veterinary Medicine, Universidad Complutense de Madrid, Madrid, Spain c Biocontrol & Molecular Ecology, Manaaki Whenua Landcare Research, Lincoln, New Zealand d Department of Biology, Stanford University, Stanford, CA, USA e Laboratory of Plant Conversation and Population Biology, Biology Department, KU Leuven, Leuven, Belgium f VIB e KU Leuven Center for Microbiology & CMPG Laboratory of Genetics and Genomics, Department of Microbial and Molecular Systems, KU Leuven, Leuven, Belgium g Estacion Biologica de Donana,~ CSIC, Sevilla, Spain article info abstract

Article history: Understanding the relationship between population genetic structure and phenotypic diversity is a Received 24 February 2020 fundamental question in evolutionary biology. Yeasts display wide genetic diversity and exhibit Received in revised form remarkably diverse heterotrophic metabolisms that allow a variety of niche occupations. However, little 5 October 2020 is known about how intra-species genetic population structure is related to trait diversity in yeasts. In Accepted 23 October 2020 this study, we investigated the link between intra-species genetic population structure and trait diversity Available online 11 November 2020 in the floral nectar-inhabiting yeast Metschnikowia reukaufii (Ascomycota). A total of 73 strains obtained Corresponding editor: Primrose Boynton from 11 plant species were genotyped by whole genome sequencing, followed by single nucleotide polymorphism (SNP) calling, and phenotyped using a robot-assisted high-throughput screening plat- Keywords: form. Analysis of the population structure estimated the number of ancestral populations to be K ¼ 5, Genetic mosaics each one including strains from different locations and host plants, and 26% of strains showed significant Metschnikowia reukaufii genetic admixture (<80% ancestry from a single population). These mosaic strains were scattered Phenotyping throughout a maximum-likelihood phylogenetic tree built from SNP data, and differed widely in their Phylogenetic signal ancestry. While yeast strains varied in nutrient assimilation and tolerance to inhibitors, trait differen- Population genetics tiation among genetic lineages was in most cases negligible. Notably, outlier phenotypes largely corresponded to the mosaic strains, and removal of these from the data had a dramatic effect on the intra- species phylogenetic signal of studied phenotypes and patterns of trait evolution. Overall, these results suggest that genetic mosaicism broadens the phenotypic landscape explored by M. reukaufii and may allow adaptation to highly variable nectar environments. © 2020 Published by Elsevier Ltd.

1. Introduction cornerstone of evolutionary biology. Yeast biologists are increas- ingly paying attention to this issue, as demonstrated by recent The rules that govern the relationship between genetic archi- publications (Bergstrom€ et al., 2014; Paleo-Lopez et al., 2016; Peter tecture, phenotypic expression and ecological success form a and Schacherer, 2016; Shen et al., 2018; Opulente et al., 2018; Kessi- Perez et al., 2020; Wang and He, 2020). In general, the results of these and other studies have revealed that yeast genomes are * Corresponding author. Manaaki Whenua Landcare Research, Lincoln, New extremely plastic in their structure, gene content and regulation, Zealand. which allows them to adapt rapidly to changing environments and ** Corresponding author. Laboratory of Plant Conservation and Population occupy challenging habitats (Henault et al., 2017). At the same time, Biology, Biology Department, KU Leuven, Leuven, Belgium. phenotypic heterogeneity in yeasts can evolve rapidly as a direct E-mail addresses: [email protected] (M.K. Dhami), hans. [email protected] (H. Jacquemyn). response to stress and facilitates evolutionary rescue from deteri- 1 These authors contributed equally. orating environmental conditions by generating individuals with https://doi.org/10.1016/j.funeco.2020.101016 1754-5048/© 2020 Published by Elsevier Ltd. S. Alvarez-Perez, M.K. Dhami, M.I. Pozo et al. Fungal Ecology 49 (2021) 101016 exceptionally high fitness (Bodi et al., 2017). The extensive genetic whether host-specific variation in nectar chemistry influences its and metabolic diversity displayed by yeasts has allowed them to genetic and phenotypic diversity, including a propensity for mosaic thrive in virtually every free-living and host-associated habitat strains, remains unknown. In particular, we tested the hypotheses (Buzzini et al., 2017; Kurtzman and Boekhout, 2017; Yurkov, 2017; that (1) the chemical properties of nectar had a significant effect on Opulente et al., 2018). However, while some yeast species are the genetic and phenotypic diversity of M. reukaufii and (2) genetic ubiquitous generalists, others have a restricted geographic and/or mosaicism contributes to the niche breadth exhibited by this spe- habitat distributions (Kurtzman and Boekhout, 2017; Peter et al., cies. To test these hypotheses, we determined the prevalence of 2017). Furthermore, within a given yeast species, some genetic genetic mosaicism in the collection of strains under study and lineages and/or populations can display trait combinations that assessed its impact on the phenotypic landscape of M. reukaufii and conspecific organisms lack and such phenotypes may allow them to the phylogenetic signal and evolutionary model of measured traits. dwell in specific environments (Crauwels et al., 2017; Gallone et al., 2016). 2. Materials and methods The current availability of population-scale genetic data and high-throughput phenotyping technologies holds promise for 2.1. Strain isolation and identification investigating how genetic diversity in yeasts is linked to phenotypic variation and habitat adaptation at different taxonomic levels A total of 73 M. reukaufii strains were analyzed in this study (Peter and Schacherer, 2016; Vervoort et al., 2017; Shen et al., 2018). (Table S1). These strains were isolated in 2008 from nectar samples However, most research efforts in this field have focused so far on collected from 11 plant species in the Sierra de Cazorla-Segura-Las model yeasts such as Saccharomyces spp. (Gallone et al., 2016; Peter Villas Natural Park, a well-preserved montane forest in SE Spain. et al., 2018) and Schizosaccharomyces pombe (Jeffares et al., 2015), The nectar properties of the sampled plant species differed widely whereas the analysis of genotype-phenotype-habitat associations in terms of total sugar concentration, sugar composition and pH of in non-model yeasts has traditionally lagged behind (Shen et al., pristine (i.e. microbe-free) nectar, as determined in a previous 2018). study (Pozo et al., 2015b; Table S2). Collection of floral nectar Metschnikowia reukaufii (Metschnikowiaceae, Ascomycota) is a samples and isolation of yeasts from these were performed ac- yeast species closely associated with floral surfaces, floral nectar, cording to the methods described in Pozo et al. (2011). Metschni- and flower-visiting animals in diverse ecosystems, where it can kowia reukaufii strains were identified on the basis of both mediate plant-animal interactions (Brysch-Herzberg, 2004; Belisle morphological characteristics and two-way sequencing of the var- et al., 2012; Herrera et al., 2013; Schaeffer and Irwin, 2014; iable D1/D2 domain of the large subunit (26S) ribosomal RNA gene Schaeffer et al., 2014, 2017; Mittelbach et al., 2015; Chappell and using primers NL1 and NL4 (Kurtzman and Robnett, 1998). All Fukami, 2018; Sobhy et al., 2018; Klaps et al., 2020). As a nectar- strains were stored in Microbank™ (ProLab Diagnostics, Richmond dwelling specialist, the survival of M. reukaufii depends on its ca- Hill, ON, Canada) preservation system vials at 80 C and then pacity to exploit limited nitrogenous nutrients in nectar and transferred to 96-well plates with 40% glycerol in sterile deionized withstand high osmotic pressures, environmental variability and water (Pozo et al., 2015b, 2016). the presence of diverse toxins of plant origin (Herrera et al., 2010; Pozo et al., 2012, 2015a; Tucker and Fukami, 2014; Lievens et al., 2.2. Genotyping by whole genome sequencing 2015; Herrera, 2017; Letten et al., 2018; Pozo and Jacquemyn, 2019; Alvarez-Perez et al., 2019). Therefore, it has been hypothe- Genome sequencing of the studied strains was performed sized that the extensive natural variation in the chemical properties following the methods described in Dhami et al. (2018). Briefly, c. of floral nectar (Nicolson et al., 2007) imposes strong selective 106 yeast cells were harvested by centrifugation from overnight forces on M. reukaufii and other nectarivorous yeasts (Herrera, cultures from single colonies grown in yeast malt broth (YMB; 1.0% 2014; Pozo et al., 2015b, 2016). In particular, variability in the w/v glucose, 0.5% peptone, 0.3% malt extract and 0.3% yeast extract) overall sugar concentration and composition of nectar among host at 25 C. The pellets were washed twice with sterile distilled water plants seems to be a key driver of phenotypic variation in and used for genomic DNA (gDNA) extraction with the DNeasy M. reukaufii and the closely-related species Metschnikowia gruessii, Blood and Tissue 96 sample kit (Qiagen, Germantown, MD, USA). which often co-occur in floral nectar but display differences in their gDNA was quantified using the Qubit HS kit (ThermoFisher Scien- responses to nectar resources and stressors (Pozo et al., 2015b, tific, Waltham, MA, USA) and diluted to 2.5 ng/mL in sterile, 2016). In accordance with these expectations, it has been demon- nuclease-free water. Genomic libraries were prepared as described strated using amplified fragment length polymorphism (AFLP) in Baym et al. (2015), dual-labelled using the Nextera index kit markers that the genetic diversity of M. reukaufii exhibits a strong (Illumina, San Diego, CA, USA), pooled for sequencing and run on an host-mediated component, with genotypes being non-randomly Illumina HiSeq 3000 sequencer with 150 2 paired-end mode. distributed among flowers of different plant species (Herrera Genome assembly from Illumina reads and identification of et al., 2014). single nucleotide polymorphisms (SNPs) and insertions and de- In this study, we investigated the link between intra-species letions (InDels) was carried out as described previously (Dhami genetic population structure and trait diversity in M. reukaufii by et al., 2016, 2018). The genotyping pipeline included the combining data obtained through whole genome sequencing and following steps: (i) mapping to M. reukaufii MR1 diploid reference high-throughput phenotyping. Genome research of nectar yeasts is genome (version MR1_a10, length ¼ 19 Mbp, N50 ¼ 1,244,334 bp, still in its infancy, and only recently was the first high-quality draft 122 scaffolds) using BWA aligner (bwa-mem; Li and Durbin, 2009) genome of M. reukaufii presented (Dhami et al., 2016). Detailed following read error correction using karect analysis of the link between genetic diversity and trait variability in (matchtype ¼ hamming, celltype ¼ diploid; Allam et al., 2015), (ii) this yeast species showed that populations inhabiting the floral realignment around InDels (GATK) and (iii) variant discovery to nectar of Diplacus (Mimulus) aurantiacus (Phrymaceae, ‘sticky yield a genome-wide population variant (SNP) dataset. Briefly, we monkey-flower’) displayed wide genetic and phenotypic diversity used the g.vcf haplotype caller function (GATK variant calling and clustered into metabolically but not geographically distinct pipeline v3.4), in joint genotyping mode to call polymorphisms genetic lineages (Dhami et al., 2018). However, given the diversity with relaxed settings for each mapped strain genome individually, of floral hosts and a cosmopolitan distribution of M. reukaufii, after duplicate removal. SNP discovery, insertions and deletions

2 S. Alvarez-Perez, M.K. Dhami, M.I. Pozo et al. Fungal Ecology 49 (2021) 101016

(InDels) discovery and genotyping were performed across all yeast growth in floral nectar (Pozo et al., 2012, 2015b, 2016). strains simultaneously resulting in a gvcf variant matrix. Hard Preparation of culture media and test conditions was performed as filtering of initial SNPs was performed using the GATK variant detailed in Pozo et al. (2015b, 2016). In order to maximize the filtering tool and VCFtools (v 1.5) with the following parameters: throughput and reproducibility of the phenotyping scheme, a high- minimum base quality ¼ 30, quality by depth (QD) ¼ 2.0, mapping density array robot (Singer ROTOR HDA, Singer Instruments, Som- quality (MAPQ) ¼ 30, Fisher strand bias ¼ 60, mapping quality erset, UK) was used to produce and replicate the selected strains on ranksum ¼12.5, and ReadPosRankSum ¼8.0. The resulting the different test media. More specifically, the 96-well plates con- matrix was filtered for InDels, non-bi-allelic SNPs, SNPs with minor taining the stored isolates (see above) were thawed, spotted on allele frequency below 0.05 and SNP positions with >50% missing yeast extract glucose chloramphenicol agar (YGC; Sigma-Aldrich, data. We further filtered sites which were homozygous with the Overijse, Belgium), and incubated for 48 h at 24 C. Next, 96-well reference genome using the VcfFilterJS (jvarkit; Lindebaum and plates containing 150 mL of YMB per well were inoculated with Redon 2017). A sliding window size of 50 SNPs advanced by 5 the strains using the robot and incubated overnight at 24 C and SNPs at a time, with an r threshold of 0.5, was employed to resolve 900 rpm on a microplate shaking platform (Heidolph Instruments, SNPs in linkage, resulting in the final dataset of high confidence Schwabach, Germany). The optical density at 600 nm of the over- SNPs. night cultures was measured using a microplate reader (Molecular Devices, San Jose, CA, USA) and manually adjusted to 1.0 in a second 2.3. Population genetics 96-well microtiter plate by diluting in sterile deionized water. This plate was used as the source plate for spotting those cell suspen- Population structure was assessed using fastSTRUCTURE version sions immediately onto the test solid media with the robot. Spotted 1.0 (Raj et al., 2014), which uses variational Bayesian Inference agar plates were sealed with Parafilm (Pechiney Plastic Packaging under a model assuming Hardy-Weinberg equilibrium and linkage Company, Chicago, IL, USA) and incubated for 48 h (or up to 12 days, equilibrium. We tested models with both simple and logistic priors for those plates testing for osmotolerance; see Table S3)at24C. keeping all other parameters at default values. A filtered set of Finally, all plates were scanned using a high-definition scanner 53,808 biallelic SNPs was parsed through the fastSTRUCTURE (Seiko Epson, Nagano, Japan), processed with ImageJ (Abramoff clustering algorithm, with varying model complexity (K ¼ 1to10) et al., 2004) combined with the ScreenMill software (Dittmar independently, and each value was cross validated 1000 times. For et al., 2010), and the colony area of each strain on each test con- each iteration, the convergence criterion was set as the change in dition was measured. the per-genotype log-marginal likelihood being <10 8. The optimal Performance of each strain in a specific test condition was number of genetically homogeneous clusters (ancestral pop- determined by dividing the colony area on each test condition by ulations or lineages, K) of strains was identified using Evanno et al.’s the area displayed in positive reference plates (yeast malt agar (2005) DK approach. For K values selected in this way, the run with without any inhibitor, yeast nitrogen base þ 1% glucose or yeast the highest likelihood value was used to assign posterior mem- carbon base without nitrogen supplementation; see details in Pozo bership coefficients to strains, and strains with a probability of et al. (2015b; 2016)). All trait values were further processed by membership to any inferred cluster 80% were considered to be conversion into Z-scores (i.e. relative growth of a particular strain representative of that lineage. In contrast, strains that possessed under certain condition minus average value of all strains for that <80% ancestry from a single population were regarded as mosaic same test, divided by the corresponding standard deviation; strains (Gallone et al., 2016). Mukherjee et al., 2014). Furthermore, we calculated two indices of overall performance of the strains on the phenotypic tests: (i) 2.4. Phylogenetic reconstruction nutrient assimilation (NA) index, defined as the sum of the Z-scores obtained for the assimilation of all carbon and nitrogen sources; The R package ‘vcfR’ (Knaus and Grünwald, 2017) was used to and (ii) tolerance to inhibitors (TI) index, corresponding to the sum retain SNPs and concatenate these into supermatrices using IUPAC of the Z-scores obtained in the assays testing for osmotolerance and ambiguity codes for heterozygous loci. Sites considered potentially growth under other inhibitory conditions. invariant by RAxML were then removed using the R package Phenotyping results were compared between groups of strains ‘Phrynomics’ (Banbury and Leache, 2014). After these filtering steps by ANOVA or the non-parametric Kruskal-Wallis test with Dunn's 317,618 SNPs remained in the dataset. A maximum-likelihood (ML) post hoc pairwise comparison, and visualized by beanplots tree was obtained in RAxML (Stamatakis, 2014), as implemented at (Kampstra, 2008) and/or scatterplots generated using R v. 3.6.0 (R the CIPRES Science Gateway (Miller et al., 2010), using the Core Team, 2019). The normal distribution of Z-scores was assessed GTRGAMMA model of sequence evolution and support for each by the Shapiro-Wilk test, using the shapiro.test() function of R node was assessed by rapid bootstrapping (100 replicates). As package ‘stats’. Additionally, visualization of selected phenotypes on concatenation of variable SNPs artificially inflates branch lengths the phylogenetic tree built from SNPs (see above) was achieved by (Leache et al., 2015; Leache and Oaks, 2017), the acquisition bias using the traitgram() and contMap() functions of the R package correction implemented in RAxML was applied to generate the final ‘phytools’ (Revell, 2012). A traitgram is basically a projection of a ML tree topology. This final tree was visualized and annotated using phylogenetic tree in a space defined by phenotype (on the y axis) and the Interactive Tree Of Life (iTOL) server (Letunic and Bork, 2019). time (on the x), whereas in the contMap method a color gradient is used to map observed and reconstructed trait values onto the edges 2.5. Phenotypic profiling of the phylogeny (Revell, 2013; Revell et al., 2018). Moreover, un- supervised classification of yeast phenotypes was performed by Phenotypic profiling of the collection of M. reukaufii strains hierarchical clustering, using Manhattan distances and the complete analyzed in this study comprised a total of 47 tests on agar plates linkage method, as implemented in the R package ‘stats’. related to assimilation of carbon and nitrogen sources (27 and three Additionally, pairwise correlations between the Z-scores ob- conditions, respectively), tolerance to different inhibitors (13 con- tained for different phenotypic traits and between the overall ditions) and other traits (hydrolysis of casein, hydrolysis of Tween performance scores (NA and TI) were determined by the conser- 80, growth on vitamin-free medium and growth on alkaline con- vative, non-parametric Spearman rank test. In order to control for ditions) (Table S3), which are among the main limiting factors of the non-phylogenetic independence of the strains analyzed,

3 S. Alvarez-Perez, M.K. Dhami, M.I. Pozo et al. Fungal Ecology 49 (2021) 101016 correlation analyses were repeated using phylogenetically inde- of 47.9 (range 23e116 ) with >85% similarity to reference pendent contrasts (PICs; Felsenstein, 1985), which were calculated (Fig. S1). After variant calling, the data consisted of 396,789 high using the pic() function in the R package ‘ape’ (Paradis et al., 2004). coverage SNPs (Fig. S2) of which 53,808 SNPs were selected for Spearman rank correlation tests were also used to explore possible population structure analysis, following pruning for linkage relationships between nectar's key features of studied plants (mean disequilibrium, retaining only biallelic SNPs, and removing alleles sugar concentration, mean concentration of sucrose, fructose and with low minor allele frequency (MAF<0.05) and missing data. glucose, and mean pH; Table S2) and the NA and TI indices. All P- Studied strains clustered in two highly supported main clades in values were adjusted for the number of simultaneous comparisons the SNPs tree, which were further split into different subclades using the Bonferroni correction. (Fig. 1, Fig. S3). Analysis of the population structure using fast- STRUCTURE yielded an estimated number of ancestral populations 2.6. Phylogenetic signal measurement K ¼ 5 (P1 to P5; Fig. 1), each including strains from different sampling locations and host plants. Distribution of strains across the Intra-species phylogenetic dependence of the phenotypic traits populations identified by fastSTRUCTURE was mostly consistent and overall performance scores (NA and TI indices) determined in with the lineages defined in the SNP tree. this study was evaluated by calculating Blomberg's K and Pagel's l Significant admixture was apparent in 19 (26% of total) strains. (Pagel, 1999; Blomberg et al., 2003; Münkemüller et al., 2012), as Mosaic strains were scattered throughout the phylogenetic tree implemented in the R packages ‘picante’ (multiPhylosignal() func- (Fig. 1) and differed in the contribution of each parent population to tion; Kembel et al. (2010) and ‘phytools’ (phylosig() function), their genome structure. Additionally, six other strains (one from respectively. These metrics assume a Brownian motion model of population 2, two from population 4 and three from population 5) trait evolution, and the closer the value to zero the more phylo- displayed >80% but less than 100% ancestry from any single pop- genetically independent is a trait; a value of 1 corresponds to a ulation. Among the five populations delineated by fastSTRUCTURE, Brownian motion expectation, and values > 1 mean that close population P2 was the main parent (i.e. highest mean Q) in most relatives are more similar than expected under Brownian motion cases (52.6% of mosaic strains), followed by populations P1 (26.3%), (Münkemüller et al., 2012; Kamilar and Cooper, 2013; Molina- P3 and P5 (10.5% each), whereas population P4 was not the major Venegas and Rodríguez, 2017). The phylogeny was built from parent for any mosaic genotype. Mosaic strains came from all plant SNPs markers and converted into an ultrametric tree, and the data species sampled except Gladiolus illyricus (Iridaceae) and eight out set of Z-scores obtained for phenotypic traits and indices of overall of the twelve sampling sites. Additionally, none of the nectar fea- performance were used as inputs in these analyses. Statistical sig- tures (mean sugar concentration, mean percentage of sucrose, nificance was tested in all cases by randomization with 1000 rep- fructose and glucose, and mean pH; Table S2) were correlated with etitions and Bonferroni-corrected P values < 0.05 were considered the per-species percentage of strains belonging to each parent significant. population (Spearman's r ranged from 0.65 to 0.65) or displaying genetic mosaicism (r ¼0.41e0.5) (Table S4). 2.7. Model fitting 3.2. Phenotypic diversity Determination of the evolutionary models best fitting the phenotypic data was performed using the R package ‘geiger’ The Z-scores obtained for most phenotypic traits did not follow (Harmon et al., 2008). In particular, we focused on whether the a Gaussian distribution (Bonferroni-corrected P-values were >0.5 Brownian motion model provided a good fit to the phenotypic data only for three of the 45 traits analyzed, namely tolerance to digi- or alternative models better explained trait evolution in M. reukaufii tonin, cetyltrimethylammonium bromide [CTAB, 10 ppm] and 5- and, therefore, Blomberg's K and Pagel's l results are not mean- fluorocytosine). Studied strains were found to be phenotypically ingful (Narwani et al., 2015). The relative likelihoods of Brownian similar in most cases, except for a few ones that displayed higher or motion and eight alternative models (Ornstein-Uhlenbeck, Early- lower Z-scores for some traits (see heatmap in Fig. S4). Traitgrams burst, trend, lambda, kappa, d tree-transformation, drift and plotted for the NA and the TI indices showed that, in both cases, white noise) were assessed by calculating their Akaike's informa- while most lineages were divergent, only a few strains were notably tion criterion (AIC) and Akaike weights (Akaike, 1974). When no displaced from the rest (Fig. 2). For the NA index these outliers evolutionary model showed an AIC weight 0.5, it was concluded mostly corresponded to mosaic strains (ST08_16_56, ST08_16_63, that none of them performed substantially better than the others ST08_16_73 and ST08_16_99), although some non-admixed strains (Narwani et al., 2015; Van Assche et al., 2017). from population P4 (two strains: ST08_16_50 and ST08_16_65) and P2 (strain ST08_16_112) were also rather distant from the other 2.8. Impact of genetic mosaicism on the phenotypic landscape of strains in the traitgram (Fig. 2). In contrast, the outliers for the TI nectar yeasts index were more evenly distributed among the different populations (P1: ST08_16_124 and ST08_16_125; P2: ST08_16_71; P3: In order to assess the impact of admixture on the phenotypic ST08_16_100; P4: ST08_16_120; and the mosaic strain ST08_16_44) landscape of M. reukaufii, some of the aforementioned analyses (e.g. (Fig. 2). The outlier nature of some strains for the NA index was also phylogenetic signal measurement, evolutionary model fitting and apparent in the beanplots of overall performance indices (Fig. 3), correlation tests) were repeated after removing from the data set all even when there was no significant difference in the distribution of strains with mixed ancestry (i.e. those displaying <100% ancestry NA index values between the mosaic group and any population from a single population in the fastSTRUCTURE results; see above). (Dunn's Z ¼1.78-1.96, P ¼ 0.373e1). In contrast, significant differences were found for the distribution of NA index values be- 3. Results tween populations P1 and P3 (Z ¼3.23, P ¼ 0.009) and P1 and P4 (Z ¼2.94, P ¼ 0.025). TI index values were more evenly distrib- 3.1. Genotypic diversity and population structure uted across the different populations and the few outliers had different ancestries (Figs. 2 and 3). The distribution of NA and TI Whole genome mapping of M. reukaufii strains to a diploid index values in the SNPs tree did not follow any clear pattern and reference was performed for all strains resulting in mean coverage outlier phenotypes were scattered across different clades (Fig. 4).

4 S. Alvarez-Perez, M.K. Dhami, M.I. Pozo et al. Fungal Ecology 49 (2021) 101016

Fig. 1. Maximum likelihood (ML) phylogenetic tree of 73 nectar-inhabiting strains of Metschnikowia reukaufii sequenced in this study. The bar chart on the right side shows the genomic population structure identified using fastSTRUCTURE, where the horizontal axis depicts the fractional representation (i.e. estimated ancestry, Q) of resolved genotypes within each strain for K ¼ 5 assumed ancestral populations (P1 to P5). Strains showing significant genome admixture (<80% ancestry from a single population) are indicated by violet filled circles on the tip of branches.

Similarly, hierarchical clustering of yeast phenotypes showed no n ¼ 990), respectively (Fig. S7, Table S5). Most pairwise correlations clear grouping of strains according to their genetic populations between traits related to assimilation of nutrients and tolerance to (Fig. S5). inhibitors were weak and non-signiﬁcant.

3.3. Trait correlations 3.4. Phylogenetic signal and evolutionary modelling of traits

No significant correlation was found between the NA and the TI Significant phylogenetic signal was found for assimilation of 30% indices in either conventional or PIC-based analyses (n ¼ 73 strains; and 63.3% of nutrient sources when tested by Blomberg's and Spearman's r ¼0.14 and 0.17, respectively, P 0.17 in both Pagel's method, respectively (Tables 1 and S6). Significant phylo- cases) (Figs. S6A and B). Moreover, strains from different origins genetic signal was also found using both methods for the NA index, and genetic lineages were intermixed in the scatter plot displaying but not for the TI index and most traits related to tolerance to in- the distribution values for these scores (Fig. S6A). However, analysis hibitors (with the exception of resistance to the antifungal pyrim- of the Z-scores obtained for different phenotypic traits and the idine analogue 5-fluorocytosine; Tables 1 and S6). K values were in corresponding PICs revealed 112 and 126 significant correlations most cases low indicating that the strength of phylogenetic signal (11.3% and 12.7% of the total number of pairwise comparisons, was weaker than would be expected under a Brownian motion

5 S. Alvarez-Perez, M.K. Dhami, M.I. Pozo et al. Fungal Ecology 49 (2021) 101016

Fig. 2. Traitgrams obtained for the indices of overall performance: (A) nutrient assimilation (NA) and (B) tolerance to inhibitors (TI). Colors on the tip of branches refer to different genomic populations (P1eP5, see legend in Fig. 1) or indicate mosaic strains (black). Most strain names have been removed from the tips of branches to facilitate interpretation of the images.

Fig. 3. Beanplots illustrating the distribution of (A) NA index and (B) TI index values for different genomic groups of Metschnikowia reukaufii (populations P1eP5 and mosaic [M] strains). The shape of the bean represents the distribution of index values. Black thick lines represent the average index value for each group, the dotted line that among all groups. model. Only one trait (assimilation of arbutin) displayed a K Evolutionary model fitting indicated that the white noise model value > 1, which was nevertheless non-significant (P ¼ 0.404). In was the most commonly supported for those traits related to contrast, high l values (0.99) were obtained for 27.7% of traits, all tolerance to inhibitors (72.7% of traits) and, accordingly, the TI in- of them related to nutrient assimilation. dex (Tables 2 and S7). In contrast, for 53.3% of traits related to

6 S. Alvarez-Perez, M.K. Dhami, M.I. Pozo et al. Fungal Ecology 49 (2021) 101016

Fig. 4. ContMap trees obtained for the indices of overall phenotypic performance: (A) nutrient assimilation (NA) and (B) tolerance to inhibitors (TI). Colors on the tip of branches refer to different genomic populations (see legend in Fig. 1).

nutrient assimilation and the NA index no model performed proportion of traits explained by the lambda model and a significantly better than the others (i.e. Akaike weights were for all concomitant decrease in the relative frequency of other models models <0.5). None of the traits considered seemed to fit the (Tables 2 and S7). In contrast, removal of admixed strains did not Brownian motion, Ornstein-Uhlenbeck, kappa or drift models of substantially alter pairwise trait correlations (Figs. S6 and S7, evolution. Table S5) or the correlation of nectar's chemical properties with the average per-host NA and TI indices of strains (r ¼0.33e0.47 and P 0.17, and r ¼0.15e0.09 and P 0.68, respectively). 3.5. Impact of genetic mosaicism on phylogenetic dependence of traits 4. Discussion Removal of the 25 admixed strains (including the 19 mosaic strains and six additional strains with >80% but <100% ancestry In this study, we analyzed the relationship between population from a single population) from the SNP tree and the phenotypic structure and trait diversity in the nectarivorous yeast M. reukaufii. data had a dramatic effect on phylogenetic signal results: K was Nectar-inhabiting microbes are naturally able to endure diverse then <0.15 and l < 0.68 in all cases, and only one trait (resistance to growth limiting factors of abiotic and biotic origin (Lievens et al., 5-fluorocytosine, as observed for the whole data set) displayed 2015; Pozo et al., 2015a; Chappell and Fukami, 2018; Alvarez- significant phylogenetic signal for both metrics (Tables 1 and S6). Perez et al., 2019; Klaps et al., 2020) and, therefore, they are an Similarly, elimination of admixed strains from the SNPs tree and ideal system to analyze the patterns of genetic and phenotypic phenotypic data had a remarkable impact on the distribution of variation in stressful environments. Taken together, our results models of trait evolution, especially for traits related to nutrient revealed that: (i) the studied yeast strains are both genetically and assimilation (Tables 2 and S7). In particular, 63.3% of traits in that phenotypically diverse; (ii) host plant identity and the chemical latter category and the NA index were then better explained by the properties of nectar have only a limited influence on the intra- white noise model, and there was an increase in the overall species population structure and phenotypic traits of M. reukaufii;

7 S. Alvarez-Perez, M.K. Dhami, M.I. Pozo et al. Fungal Ecology 49 (2021) 101016

Table 1 Overview of the results obtained in tests of phylogenetic signal for different phenotypic traits of Metschnikowia reukauﬁi.a

Whole data set (N ¼ 73) After removal of admixed strains (N ¼ 48)

Trait category (n) Index valuesb K l K l

Nutrient assimilation (30) Range 0.059e1.736 6.670 $ 10 5 e 1.009 0.038e0.151 6.663 $ 10 5 e 0.349 Mean 0.357 0.763 0.101 0.082 Significant traits 9 (30%) 19 (63.3%) 0 (0%) 0 (0%) Tolerance to inhibitors (11) Range 0.030e0.150 6.670 $ 10 5 e 0.450 0.026e0.137 6.663 $ 10 5 e 0.682 Mean 0.077 0.125 0.069 0.234 Significant traits 0 (0%) 1 (9.1%) 1 (9.1%) 1 (9.1%) Overall (45) Range 0.030e1.736 6.670 $ 10 5 e 1.009 0.026e0.151 6.663 $ 10 5 e 0.682 Mean 0.264 0.544 0.090 0.132 Significant traits 9 (20%) 20 (44.4%) 1 (2.2%) 1 (2.2%)

a The detailed results obtained in these analyses can be found in Table S6. b For each phylogenetic signal metric and trait category, the range of values, the mean value and the number (and percentage) of traits for which the corresponding phylogenetic signal index was statistically signiﬁcant (P < 0.05) are indicated.

Table 2 Overview of model ﬁtting results obtained for phenotypic traits of Metschnikowia reukauﬁi.a

Evolutionary modelb

White noise Lambda Delta Trend Early burst None

Whole data set (N ¼ 73)

Nutrient assimilation (30c) 6 (20%) 2 (6.7%) 3 (10%) 2 (6.7%) 1 (3.3%) 16 (53.3%) Tolerance to inhibitors (11) 8 (72.7%) 3 (27.3%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) Overall (45) 17 (37.8%) 6 (13.3%) 3 (6.7%) 2 (4.4%) 1 (2.2%) 16 (35.6%)

After removal of admixed strains (N ¼ 48)

Nutrient assimilation (30) 19 (63.3%) 5 (16.7%) 0 (0%) 0 (0%) 0 (0%) 6 (20%) Tolerance to inhibitors (11) 7 (63.6%) 4 (36.4%) 0 (0%) 0 (0%) 0 (0%) 0 (0%) Overall (45) 29 (64.4%) 10 (22.2%) 0 (0%) 0 (0%) 0 (0%) 6 (13.3%)

a The detailed results obtained in these analyses can be found in Table S7. b Number (and percentage) of traits for which each evolutionary model was selected as the best ﬁt to the data. c Number of traits analyzed.

(iii) traits related to nutrient source assimilation and tolerance to Furthermore, no significant correlation was found between the inhibitors are only weakly phylogenetically structured in this yeast chemical properties of pristine nectar, which displayed notable population; and (iv) a high proportion of M. reukaufii strains have variation across and sometimes within host plant species, and the mixed ancestry, which is often linked to the display of ‘atypical’ (i.e. NA or TI indices. However, the limited number of isolates analyzed outlier) phenotypes. in this study and the uneven distribution of these according to their The high genetic diversity observed for our collection of strains plant origin precluded a detailed analysis of the link between confirms the results of previous studies reporting a similarly high phenotypic differentiation and host identity. The advent of afford- intra-species diversity in M. reukaufii (Herrera et al., 2014; Dhami able high-throughput technologies for genome sequencing and et al., 2018) and also in M. gruessii (Herrera et al., 2011). However, phenotype screening will probably allow a more detailed descrip- while genotyping based on AFLP markers revealed that Metschni- tion of the genetic and trait variation occurring in wild yeast pop- kowia genotypes are non-randomly distributed among floral ulations and, therefore, contribute to elucidate the evolutionary microsites (nectar, corolla, anthers and bracts), flowers of different history of different yeast clades (Peter and Schacherer, 2016; species, or flowers of conspecific individuals locally co-occurring Vervoort et al., 2017). (Herrera et al., 2011, 2014), whole genome sequencing in this Associations between phenotypic characteristics are prevalent study and Dhami et al. (2018) detected no geographical or host in nature, occur across a diverse range of taxa and traits, and can be dependence of genetic clustering. The reasons behind such con- driven by factors intrinsic or extrinsic to an organism (Opulente trasting results are currently unknown, but sampling design and et al., 2018). At the subphylum level, the Saccharomycotina, the different resolution powers of the techniques used to subtype which includes the Metschnikowiaceae family, display extensive yeast strains in different studies may explain the results. Further- positive and negative phenotype associations (i.e., traits that tend more, the distribution of yeast genotypes is determined by a to co-occur and not occur together, respectively) (Opulente et al., complex interaction between intrafloral selection and pollinator- 2018). Such associations are largely explained by the biological mediated migration/colonization (Herrera et al., 2014), so the ge- properties of the traits (e.g. overlapping biological pathways) and to netic structure of M. reukaufii strains inhabiting nectar or other some extent by the habitat of studied isolates, whereas phylogeny floral microsites might differ. plays only a limited role (Opulente et al., 2018). Due to its typically Regarding the trait diversity of Metschnikowia yeasts, Pozo et al. high sugar concentration, paucity of nitrogen sources, and antimi- (2015b, 2016) reported a wide phenotypic landscape for both crobial defenses, floral nectar exerts a strong filtering effect on M. reukaufii and M. gruessii, which was significantly related to the microbial diversity (Herrera et al., 2010; Lievens et al., 2015; Dhami host plant of origin of tested strains both at the family and species et al., 2016; Herrera, 2017; Alvarez-Perez et al., 2019). The lack of level. The results of the present study confirm the extensive correlation found in this study between the overall performance phenotypic versatility of M. reukaufii, and revealed high phenotypic indices (NA and TI), and among most individual traits related to overlap between the five genetic populations identified. assimilation of nutrients and tolerance to inhibitors, suggests that

8 S. Alvarez-Perez, M.K. Dhami, M.I. Pozo et al. Fungal Ecology 49 (2021) 101016 the nutrient assimilation capability of M. reukaufii strains is not M. reukaufii was beyond the scope of the present study, but it may linked to their survival potential under stressful conditions. be important to explain the intra- and inter-population differences Furthermore, our results suggest that there is no single recipe for in nutrient assimilation and tolerance to inhibitors and, eventually, adaptation to floral nectar, and that poor performance in some to understand the ecological success of this yeast species as a traits might be compensated by stronger manifestation of others. In nectar-inhabitant. other words, there seems to be no ‘trait syndrome’ (i.e. set of traits Although hybridization and intra-species admixture are that collectively provide a fitness advantage in a given environment important mechanisms of genome evolution in yeasts, most (Opulente et al., 2018)), specifically allowing the adaptation of research carried out so far on these has focused on species linked to M. reukaufii to nectar dwelling. Actually, this species is also biotechnological processes and other human activities (Hittinger, frequently isolated from insects and other visitors (Brysch- 2013; Gallone et al., 2016; Henault et al., 2017; Peter et al., 2018; Herzberg, 2004; Good et al., 2014; Vannette and Fukami, 2017), Sipiczki, 2018). Both hybridization and intra-species admixture and, more sporadically, from other natural environments such as seem also to occur in wild yeast populations (Barbosa et al., 2016; freshwater swamps and sea water (Hagler et al., 2017). Apart from Leducq et al., 2016), but their contribution to genetic diversity in that, niche differentiation and resource partitioning are considered nature still remains largely unknown (Henault et al., 2017). In the important driving forces of community assembly for the nectar present study, fastSTRUCTURE results revealed that 34.2% of the microbiota, which may explain, for example, the frequent co- strains analyzed had some degree of mixed ancestry (including all occurrence of M. reukaufii and M. gruessii in this habitat (Pozo strains with <100% ancestry from a single population). A similar et al., 2016). result was obtained by Dhami et al. (2018), who observed frequent In general, microbial traits are considered to be phylogenetically mosaicism (c.14.7%) among M. reukaufii strains retrieved from the conserved (i.e. closely related taxa are phenotypically more similar floral nectar of 12 D. aurantiacus populations distributed across the than those with a more ancient common ancestry), but the depth of San Francisco region in California, USA. The mosaic strains identi- their conservation varies widely from trait to trait (Martiny et al., fied by Dhami et al. (2018) and in the present study differed in their 2013, 2015). Our results revealed weak phylogenetic determina- sources of ancestry (major genetic populations of origin), which is tion of traits related to nutrient source assimilation and tolerance to consistent with multiple recombination events. Moreover, mosaic inhibitors in M. reukaufii. In general, Blomberg's K and Pagel's l strains were scattered throughout the SNPs tree, suggesting that values obtained for most traits were low (albeit still significant in genetic admixture may not be restricted to specific lineages of some cases), and model-fitting results revealed that most studied M. reukaufii. Although sexual reproduction does not seem to occur traits followed a white noise model of evolution, which is a non- naturally in the nectar-living yeast populations of southern Spain, phylogenetic model assuming that trait data come from a random where both M. reukaufii and M. gruessii are predominantly or normal distribution and species have no significant trait covariance exclusively clonal, it has been hypothesized that sexual reproduc- (Narwani et al., 2015). The non-phylogenetic pattern of trait evo- tion in nectar yeasts might take place occasionally away from lution observed in the present study for most traits was further flowers (e.g. in bumble bee nests; Herrera et al., 2011, 2014), which reinforced when admixed strains were removed from the analysis, would lead to genetic admixture. Notably, mosaic strains showed thus suggesting the intriguing hypothesis that events of genomic outlier values for different traits. Accordingly, we hypothesize that admixture might have contributed to strengthen associations be- genetic mosaicism might contribute to broadening the phenotypic tween phylogenetic affiliation and trait variation in M. reukaufii.In capacities explored by M. reukaufii, allowing adaptation to highly contrast, elimination of admixed strains from the SNPs tree and variable nectar environments (Chappell and Fukami, 2018). Genetic phenotype data set had only a limited effect on trait correlations. tools for transcriptomic analysis and genome editing could help to Nevertheless, to get a full understanding of these results, we may clarify the link between genetic diversity and phenotypic plasticity need to consider some factors that can determine trait expression in M. reukaufii and other nectar yeasts (Chappell and Fukami, 2018). and/or evolution in yeasts. For example, phylogenetic trees are not always well suited to capture relationships between organisms, and 5. Conclusions phylogenetic networks including reticulation points (i.e. nodes where distinct branches come together to create a new lineage) are In conclusion, this study corroborated the extensive phenotypic sometimes required (Bastide et al., 2016). Such reticulations can plasticity in M. reukaufii and revealed the frequent and non- represent various biological mechanisms, such as hybridization, phylogenetically structured occurrence of genetic admixture in gene flow and horizontal gene transfer (Bastide et al., 2018). In this species. Outlier phenotypes for both nutrient assimilation and addition, the analysis of trait evolution in networks needs to define tolerance to inhibitors mostly corresponded to the mosaic strains. a model for the trait expression in hybrid lineages. In some cases, Overall, these results suggest that genetic mosaicism may broaden trait expression in the hybrid is a mixture of the expression in the the phenotypic variation in M. reukaufii, allowing adaptation to parents, weighted by their relative genetic contributions (i.e. highly variable nectar environments. Further research should weighted-average merging rule), whereas in other instances trait clarify the ecological relevance of such increased phenotypic vari- expression in the hybrid genotype is clearly outside of the range of ation of nectar yeasts, including the analysis of any potential effect the parents, an effect which is known as transgressive evolution on the plant-animal interactions that these yeasts mediate. (Bastide et al., 2018). Furthermore, some traits measured in this study, and particularly those involved in tolerance to inhibitors, are Acknowledgements most likely determined by multiple genes in combination and by their interactions with the environment, so the possibility of epis- This work was supported by funding from the Strategic Science tasis should not be disregarded (Carlborg and Haley, 2004). It is also Investment Fund, New Zealand Ministry of Business, Innovation & worth noting that, in addition to genetically-based (i.e., based on Employment (to M.K.D.), the Center for Computational, Evolu- DNA sequence) responses to stressful floral environments, induc- tionary and Human Genomics at Stanford University (to M.K.D.), ible epigenetic responses based on flexible DNA methylation have the European Union program FP7 PEOPLE-2012-IEF (to M.I.P., grant been shown to play a role in the response of M. reukaufii to nectar no. 327635) and Fonds Wetenschappelijk Onderzoek (FWO) (to variations (Herrera et al., 2012). A detailed analysis of the relative M.I.P. application no. 12A0716N). S.A.-P. acknowledges a ‘Ramon y importance of all these mechanisms of trait expression/evolution in Cajal’ contract funded by the Spanish Ministry of Science and

9 S. Alvarez-Perez, M.K. Dhami, M.I. Pozo et al. Fungal Ecology 49 (2021) 101016

Innovation [RYC 2018-023847-I]. Tadashi Fukami (Department of Dhami, M.K., Hartwig, T., Letten, A.D., Banf, M., Fukami, T., 2018. Genomic diversity Biology, Stanford University) contributed to the conception of the of a nectar yeast clusters into metabolically, but not geographically, distinct lineages. Mol. Ecol. 27 (8), 2067e2076. https://doi.org/10.1111/mec.14535. study, drafting of the manuscript, and the financial support of the Dittmar, J.C., Reid, R.J., Rothstein, R., 2010. ScreenMill: a freely available software experiments through grants from the US National Science Foun- suite for growth measurement, analysis and visualization of high-throughput dation (nos. DEB 1149600, DEB 1555786, DEB 1737758), and a screen data. BMC Bioinf. 11, 353. https://doi.org/10.1186/1471-2105-11-353. Evanno, G., Regnaut, S., Goudet, J., 2005. Detecting the number of clusters of in- Terman Fellowship. New Zealand eScience Infrastructure (NESI) dividuals using the software STRUCTURE: a simulation study. Mol. Ecol. 14 (8), and Genomics Aotearoa provided access to computing resources, 2611e2620. https://doi.org/10.1111/j.1365-294X.2005.02553.x. and Ben Roberts (NESI/Manaaki Whenua Landcare Research) Felsenstein, J., 1985. Phylogenies and the comparative method. Am. Nat. 125 (1), 1e15. https://www.jstor.org/stable/2461605. contributed to bioinformatic analyses. We would also like to thank Gallone, B., Steensels, J., Prahl, T., Soriaga, L., Saels, V., Herrera-Malaver, B., et al., constructive comments from two anonymous reviewers. 2016. Domestication and divergence of Saccharomyces cerevisiae beer yeasts. Cell 166 (6), 1397e1410. https://doi.org/10.1016/j.cell.2016.08.020 e16. Good, A.P., Gauthier, M.P., Vannette, R.L., Fukami, T., 2014. Honey bees avoid nectar Appendix A. Supplementary data colonized by three bacterial species, but not by a yeast species, isolated from the bee gut. PLoS One 9 (1), e86494. https://doi.org/10.1371/ Supplementary data to this article can be found online at journal.pone.0086494. Hagler, A.N., Mendonça-Hagler, L.C., Pagnocca, F.C., 2017. Yeasts in aquatic ecotone https://doi.org/10.1016/j.funeco.2020.101016. habitats. In: Buzzini, P., Lachance, M.A., Yurkov, A. (Eds.), Yeasts in Natural Ecosystems: Diversity. Springer International Publishing AG, Cham, e Data availability Switzerland, pp. 63 85. Harmon, L.J., Weir, J.T., Brock, C.D., Glor, R.E., Challenger, W., 2008. GEIGER: investigating evolutionary radiations. Bioinformatics 24 (1), 129e131. https://doi.org/ Sequence data have been archived under GenBank 10.1093/bioinformatics/btm538. (PRJNA542214). SNP vcf file, phenotype raw data and Henault, M., Eberlein, C., Charron, G., Durand, E., Nielly-Thibault, L., Martin, H., e Landry, C.R., 2017. Yeast population genomics goes wild: the case of Saccharo- Supplementary Table S5 S7 have been submitted to Zenodo Data myces paradoxus. In: Polz, M.F., Rajora, O.P. (Eds.), Population Genomics: Mi- repository, https://doi.org/10.5281/zenodo.4045172. croorganisms. Springer International Publishing AG, Cham, Switzerland, pp. 207e230. Herrera, C.M., 2014. Population growth of the floricolous yeast Metschnikowia References reukaufii: effects of nectar host, yeast genotype and host x genotype interaction. FEMS (Fed. Eur. Microbiol. Soc.) Microbiol. Ecol. 88 (2), 250e257. https:// Abramoff, M.D., Magalhaes,~ P.J., Ram, S.J., 2004. Image processing with ImageJ. doi.org/10.1111/1574-6941.12284. Biophot. Int. 11 (7), 36e42. Herrera, C.M., 2017. Scavengers that fit beneath a microscope lens. Ecology 98 (10), Akaike, H., 1974. A new look at the statistical model identification. IEEE Trans. 2725e2726. https://doi.org/10.1002/ecy.1874. Automat. Contr. 19, 716e723. https://doi.org/10.1109/TAC.1974.1100705. Herrera, C.M., Canto, A., Pozo, M.I., Bazaga, P., 2010. Inhospitable sweetness: nectar Allam, A., Kalnis, P., Solovyev, V., 2015. Karect: accurate correction of substitution, filtering of pollinator-borne inocula leads to impoverished, phylogenetically insertion and deletion errors for next-generation sequencing data. Bioinfor- clustered yeast communities. Proc. Biol. Sci. 277 (1682), 747e754. https:// matics 31 (21), 3421e3428. https://doi.org/10.1093/bioinformatics/btv415. doi.org/10.1098/rspb.2009.1485. Alvarez-Perez, S., Lievens, B., Fukami, T., 2019. Yeastebacterium interactions: the Herrera, C.M., Pozo, M.I., Bazaga, P., 2011. Clonality, genetic diversity and support for next frontier in nectar research. Trends Plant Sci. 24 (5), 393e401. https:// the diversifying selection hypothesis in natural populations of a flower-living doi.org/10.1016/j.tplants.2019.01.012. yeast. Mol. Ecol. 20 (21), 4395e4407. https://doi.org/10.1111/j.1365- Banbury, B., Leache, A., 2014. Phrynomics: an SNP Toolkit. R Package Version 2.0 294X.2011.05217.x. [Software]. Retrieved from. https://github.com/bbanbury/phrynomics/issues. Herrera, C.M., Pozo, M.I., Bazaga, P., 2012. Jack of all nectars, master of most: DNA Barbosa, R., Almeida, P., Safar, S.V.B., Santos, R.O., Morais, P.B., Nielly-Thibault, L., methylation and the epigenetic basis of niche width in a flower-living yeast. et al., 2016. Evidence of natural hybridization in Brazilian wild lineages of Mol. Ecol. 21 (11), 2602e2616. https://doi.org/10.1111/j.1365- Saccharomyces cerevisiae. Genome Biol. Evol. 8 (2), 317e329. https://doi.org/ 294X.2011.05402.x. 10.1093/gbe/evv263. Herrera, C.M., Pozo, M.I., Bazaga, P., 2014. Nonrandom genotype distribution among Bastide, P., Solís-Lemus, C., Kriebel, R., Sparks, K.W., Ane, C., 2018. Phylogenetic floral hosts contributes to local and regional genetic diversity in the comparative methods on phylogenetic networks with reticulations. Syst. Biol. nectareliving yeast Metschnikowia reukaufii. FEMS (Fed. Eur. Microbiol. Soc.) 67 (5), 800e820. https://doi.org/10.1093/sysbio/syy033. Microbiol. Ecol. 87 (3), 568e575. https://doi.org/10.1111/1574-6941.12245. Baym, M., Kryazhimskiy, S., Lieberman, T.D., Chung, H., Desai, M.M., Kishony, R., Herrera, C.M., Pozo, M.I., Medrano, M., 2013. Yeasts in nectar of an early-blooming 2015. Inexpensive multiplexed library preparation for megabase-sized ge- herb: sought by bumble bees, detrimental to plant fecundity. Ecology 94 (2), nomes. PLoS One 10 (5), e0131262. https://doi.org/10.1371/ 273e279. https://doi.org/10.1890/12-0595.1. journal.pone.0128036. Hittinger, C.T., 2013. Saccharomyces diversity and evolution: a budding model genus. Belisle, M., Peay, K.G., Fukami, T., 2012. Flowers as islands: spatial distribution of Trends Genet. 29 (5), 309e317. https://doi.org/10.1016/j.tig.2013.01.002. nectar-inhabiting microfungi among plants of Mimulus aurantiacus,a Jeffares, D.C., Rallis, C., Rieux, A., Speed, D., Prevorovský, M., Mourier, T., et al., 2015. hummingbird-pollinated shrub. Microb. Ecol. 63 (4), 711e718. https://doi.org/ The genomic and phenotypic diversity of Schizosaccharomyces pombe. Nat. 10.1007/s00248-011-9975-8. Genet. 47 (3), 235e241. https://doi.org/10.1038/ng.3215. Bergstrom,€ A., Simpson, J.T., Salinas, F., Barre, B., Parts, L., Zia, A., et al., 2014. A high- Kamilar, J.M., Cooper, N., 2013. Phylogenetic signal in primate behaviour, ecology definition view of functional genetic variation from natural yeast genomes. Mol. and life history. Philos. Trans. R. Soc. Lond. Ser. B Biol. Sci. 368 (1618), 20120341 Biol. Evol. 31 (4), 872e888. https://doi.org/10.1093/molbev/msu037. https://doi.org/10.1098/rstb.2012.0341. Blomberg, S.P., Garland Jr., T., Ives, A.R., 2003. Testing for phylogenetics signal in Kampstra, P., 2008. Beanplot: a boxplot alternative for visual comparison of dis- comparative data: behavioral traits are more labile. Evolution 57 (4), 717e745. tributions. J. Stat. Software 28, 1e9. https://doi.org/10.18637/jss.v028.c01. https://doi.org/10.1111/j.0014-3820.2003.tb00285.x. Kembel, S.W., Cowan, P.D., Helmus, M.R., Cornwell, W.K., Morlon, H., Ackerly, D.D., Bodi, Z., Farkas, Z., Nevozhay, D., Kalapis, D., Laz ar, V., Csorg€ o,} B., et al., 2017. et al., 2010. Picante: R tools for integrating phylogenies and ecology. Bioinfor- Phenotypic heterogeneity promotes adaptive evolution. PLoS Biol. 15 (5), matics 26 (11), 1463e1464. https://doi.org/10.1093/bioinformatics/btq166. e2000644 https://doi.org/10.1371/journal.pbio.2000644. Kessi-Perez, E.I., Molinet, J., Martínez, C., 2020. Disentangling the genetic bases of Brysch-Herzberg, M., 2004. Ecology of yeasts in plant-bumblebee mutualism in Saccharomyces cerevisiae nitrogen consumption and adaptation to low nitrogen Central Europe. FEMS (Fed. Eur. Microbiol. Soc.) Microbiol. Ecol. 50 (2), 87e100. environments in wine fermentation. Biol. Res. 53 (1), 2. https://doi.org/10.1186/ https://doi.org/10.1016/j.femsec.2004.06.003. s40659-019-0270-3. Buzzini, P., Lachance, M.-A., Yurkov, A., 2017. Yeasts in Natural Ecosystems: Di- Klaps, J., Lievens, B., Alvarez-Perez, S., 2020. Towards a better understanding of the versity. Springer International Publishing AG, Cham, Switzerland. role of nectar-inhabiting yeasts in plant-animal interactions. Fungal Biol. Bio- Carlborg, O., Haley, C.S., 2004. Epistasis: too often neglected in complex trait technol. 7, 1. https://doi.org/10.1186/s40694-019-0091-8. studies? Nat. Rev. Genet. 5 (8), 618e625. https://doi.org/10.1038/nrg1407. Knaus, B.J., Grünwald, N.J., 2017. VCFR: a package to manipulate and visualize Chappell, C.R., Fukami, F., 2018. Nectar yeasts: a natural microcosm for ecology. variant call format data in R. Mol. Ecol. Resour. 17 (1), 44e53. https://doi.org/ Yeast 35 (6), 417e423. https://doi.org/10.1002/yea.3311. 10.1111/1755-0998.12549. Crauwels, S., Van Opstaele, F., Jaskula-Goiris, B., Steensels, J., Verreth, C., Bosmans, L., Kurtzman, C.P., Boekhout, T., 2017. Yeasts as distinct life forms of fungi. In: et al., 2017. Fermentation assays reveal differences in sugar and (off-) flavor Buzzini, P., Lachance, M.-A., Yurkov, A. (Eds.), Yeasts in Natural Ecosystems: metabolism across different Brettanomyces bruxellensis strains. FEMS Yeast Res. Ecology. Springer International Publishing AG, Cham, Switzerland, pp. 1e37. 17 (1), fow105. https://doi.org/10.1093/femsyr/fow105. Kurtzman, C.P., Robnett, C.J., 1998. Identification and phylogeny of ascomycetous Dhami, M.K., Hartwig, T., Fukami, T., 2016. Genetic basis of priority effects: insights yeasts from analysis of nuclear large subunit (26S) ribosomal DNA partial se- from nectar yeast. Proc. Biol. Sci. 283 (1840), 20161455 https://doi.org/10.1098/ quences. Antonie Leeuwenhoek 73 (4), 331e371. https://doi.org/10.1023/A: rspb.2016.1455. 1001761008817.

10 S. Alvarez-Perez, M.K. Dhami, M.I. Pozo et al. Fungal Ecology 49 (2021) 101016

Leache, A.D., Banbury, B.L., Felsenstein, J., de Oca, A.N., Stamatakis, A., 2015. Short Pozo, M., Jacquemyn, H., 2019. Addition of pollen increases growth of nectar-living tree, long tree, right tree, wrong tree: new acquisition bias corrections for yeasts. FEMS (Fed. Eur. Microbiol. Soc.) Microbiol. Lett. 366 (15), fnz191. https:// inferring SNP phylogenies. Syst. Biol. 64 (6), 1032e1047. https://doi.org/10.1093/ doi.org/10.1093/femsle/fnz191. sysbio/syv053. Pozo, M.I., Lachance, M.A., Herrera, C.M., 2012. Nectar yeasts of two southern Leache, A.D., Oaks, J.R., 2017. The utility of single nucleotide polymorphism (SNP) Spanish plants: the roles of immigration and physiological traits in community data in phylogenetics. Annu. Rev. Ecol. Evol. Syst. 48, 69e84. https://doi.org/ assembly. FEMS (Fed. Eur. Microbiol. Soc.) Microbiol. Ecol. 80 (2), 281e293. 10.1146/annurev-ecolsys-110316-022645. https://doi.org/10.1111/j.1574-6941.2011.01286.x. Leducq, J.B., Nielly-Thibault, L., Charron, G., Eberlein, C., Verta, J.P., Samani, P., et al., Pozo, M., Lievens, B., Jacquemyn, H., 2015a. Impact of microorganisms on nectar 2016. Speciation driven by hybridization and chromosomal plasticity in a wild chemistry, pollinator attraction and plant fitness. In: Peck, R.L. (Ed.), Nectar: yeast. Nat. Microbiol. 1, 15003. https://doi.org/10.1038/nmicrobiol.2015.3. Production, Chemical Composition and Benefits to Animals and Plants. Nova Letten, A.D., Dhami, M.K., Ke, P.J., Fukami, T., 2018. Species coexistence through Science Publishers Inc, New York, NY, pp. 1e40. simultaneous fluctuation-dependent mechanisms. Proc. Natl. Acad. Sci. U. S. A. Pozo, M.I., Herrera, C.M., Bazaga, P., 2011. Species richness of yeast communities in 115 (26), 6745e6750. https://doi.org/10.1073/pnas.1801846115. floral nectar of southern Spanish plants. Microb. Ecol. 61 (1), 82e91. https:// Letunic, I., Bork, P., 2019. Interactive Tree of Life (iTOL) v4: recent updates and new doi.org/10.1007/s00248-010-9682-x. developments. Nucleic Acids Res. https://doi.org/10.1093/nar/gkz239 pii: Pozo, M.I., Herrera, C.M., Lachance, M.A., Verstrepen, K., Lievens, B., Jacquemyn, H., gkz239. 2016. Species coexistence in simple microbial communities: unravelling the Li, H., Durbin, R., 2009. Fast and accurate short read alignment with phenotypic landscape of co-occurring Metschnikowia species in floral nectar. BurrowseWheeler transform. Bioinformatics 25 (14), 1754e1760. https:// Environ. Microbiol. 18 (6), 1850e1862. https://doi.org/10.1111/1462-2920.13037. doi.org/10.1093/bioinformatics/btp324. Pozo, M.I., Herrera, C.M., Van den Ende, W., Verstrepen, K., Lievens, B., Lievens, B., Hallsworth, J.E., Pozo, M.I., Belgacem, Z.B., Stevenson, A., Willems, K.A., Jacquemyn, H., 2015b. The impact of nectar chemical features on phenotypic Jacquemyn, H., 2015. Microbiology of sugar-rich environments: diversity, variation in two related nectar yeasts. FEMS (Fed. Eur. Microbiol. Soc.) Micro- ecology and system constraints. Environ. Microbiol. 17 (2), 278e298. https:// biol. Ecol. 91 (6) https://doi.org/10.1093/femsec/fiv055 fiv055. doi.org/10.1111/1462-2920.12570. R Core Team, 2019. R: A Language and Environment for Statistical Computing. R Lindebaum, P., Redon, R., 2017. bioalcidae, samjs and vcffilterjs: object-oriented Foundation for Statistical Computing, Vienna, Austria. formatters and filters for bioinformatics files. Bioinformatics 34 (7), Raj, A., Stephens, M., Pritchard, J.K., 2014. fastSTRUCTURE: variational inference of 1224e1225. https://doi.org/10.1093/bioinformatics/btx734. population structure in large SNP data sets. Genetics 197 (2), 573e589. https:// Martiny, A.C., Treseder, K., Pusch, G., 2013. Phylogenetic conservatism of functional doi.org/10.1534/genetics.114.164350. traits in microorganisms. ISME J. 7 (4), 830e838. https://doi.org/10.1038/ Revell, L.J., 2012. Phytools: an R package for phylogenetic comparative biology (and ismej.2012.160. other things). Methods Ecol. Evol. 3 (2), 217e223. https://doi.org/10.1111/ Martiny, J.B., Jones, S.E., Lennon, J.T., Martiny, A.C., 2015. Microbiomes in light of j.2041-210X.2011.00169.x. traits: a phylogenetic perspective. Science 350 (6261), aac9323. https://doi.org/ Revell, L.J., 2013. Two new graphical methods for mapping trait evolution on phy- 10.1126/science.aac9323. logenies. Methods Ecol. Evol. 4 (8), 754e759. https://doi.org/10.1111/2041- Miller, M.A., Pfeiffer, W., Schwartz, T., 2010. Creating the CIPRES Science Gateway for 210X.12066. inference of large phylogenetic trees. In: Proceedings of the Gateway Revell, L.J., Schliep, K., Valderrama, E., Richardson, J.E., 2018. Graphs in phylogenetic Computing Environments Workshop. GCE), pp. 1e8. comparative analysis: anscombe's quartet revisited. Methods Ecol. Evol. 9 (10), Mittelbach, M.A., Yurkov, A.M., Nocentini, D., et al., 2015. Nectar sugars and bird 2145e2154. https://doi.org/10.1111/2041-210X.13067. visitation define a floral niche for basidiomycetous yeast on the Canary Islands. Schaeffer, R.N., Irwin, R.E., 2014. Yeasts in nectar enhance male fitness in a montane BMC Ecol. 15 (2) https://doi.org/10.1186/s12898-015-0036-x. perennial herb. Ecology 95 (7), 1792e1798. https://doi.org/10.1890/13-1740.1. Molina-Venegas, R., Rodríguez, M.A., 2017. Revisiting phylogenetic signal; strong or Schaeffer, R.N., Mei, Y.Z., Andicoechea, J., Manson, J.S., Irwin, R.E., 2017. Conse- negligible impacts of polytomies and branch length information? BMC Evol. quences of a nectar yeast for pollinator preference and performance. Funct. Biol. 17 (1), 53. https://doi.org/10.1186/s12862-017-0898-y. Ecol. 31 (3), 613e621. https://doi.org/10.1111/1365-2435.12762. Münkemüller, T., Lavergne, S., Bzeznik, B., Dray, S., Jombart, T., Schiffers, K., Schaeffer, R.N., Phillips, C.R., Duryea, M.C., Andicoechea, J., Irwin, R.E., 2014. Nectar Thuiller, W., 2012. How to measure and test phylogenetic signal. Methods Ecol. yeasts in the tall Larkspur Delphinium barbeyi (Ranunculaceae) and effects on Evol. 3 (4), 743e756. https://doi.org/10.1111/j.2041-210X.2012.00196.x. components of pollinator foraging behavior. PLoS One 9 (10), e108214. https:// Mukherjee, V., Steensels, J., Lievens, B., Van de Voorde, I., Verplaetse, A., Aerts, G., doi.org/10.1371/journal.pone.0108214. et al., 2014. Phenotypic evaluation of natural and industrial Saccharomyces Shen, X.X., Opulente, D.A., Kominek, J., Zhou, X., Steenwyk, J.L., Buh, K.V., et al., 2018. yeasts for different traits desirable in industrial bioethanol production. Appl. Tempo and mode of genome evolution in the budding yeast subphylum. Cell Microbiol. Biotechnol. 98 (22), 9483e9498. https://doi.org/10.1007/s00253- 175 (6), 1533e1545. https://doi.org/10.1016/j.cell.2018.10.023 e20. 014-6090-z. Sipiczki, M., 2018. Interspecies hybridisation and genome chimerisation in Narwani, A., Alexandrou, M.A., Herrin, J., Vouaux, A., Zhou, C., Oakley, T.H., Saccharomyces: combining of gene pools of species and its biotechnological Cardinale, B.J., 2015. Common ancestry is a poor predictor of competitive traits perspectives. Front. Microbiol. 9, 3071. https://doi.org/10.3389/fmicb.2018.0307. in freshwater green algae. PLoS One 10 (9), e0137085. https://doi.org/10.1371/ Sobhy, I.S., Baets, D., Goelen, T., Herrera-Malaver, B., Bosmans, L., Van den Ende, W., journal.pone.0137085. et al., 2018. Sweet scents: nectar specialist yeasts enhance nectar attraction of a Nicolson, S.W., Nepi, M., Pacini, E., 2007. Nectaries and Nectar. Springer, Dordrecht, generalist aphid parasitoid without affecting survival. Front. Plant Sci. 9, 1009. The Netherlands. https://doi.org/10.3389/fpls.2018.01009. Opulente, D.A., Rollinson, E.J., Bernick-Roehr, C., Hulfachor, A.B., Rokas, A., Stamatakis, A., 2014. RAxML version 8: a tool for phylogenetic analysis and post- Kurtzman, C.P., Hittinger, C.T., 2018. Factors driving metabolic diversity in the analysis of large phylogenies. Bioinformatics 30 (9), 1312e1313. https:// budding yeast subphylum. BMC Biol. 16 (1), 26. https://doi.org/10.1186/s12915- doi.org/10.1093/bioinformatics/btu033. 018-0498-3. Tucker, C.M., Fukami, T., 2014. Environmental variability counteracts priority effects Pagel, M., 1999. Inferring the historical patterns of biological evolution. Nature 401 to facilitate species coexistence: evidence from nectar microbes. Proc. Biol. Sci. (6756), 877e884. https://doi.org/10.1038/44766. 281 (1778), 20132637 https://doi.org/10.1098/rspb.2013.2637. Paleo-Lopez, R., Quintero-Galvis, J.F., Solano-Iguaran, J.J., Sanchez-Salazar, A.M., Van Assche, A., Alvarez-Perez, S., de Breij, A., De Brabanter, J., Willems, K.A., Gaitan-Espitia, J.D., Nespolo, R.F., 2016. A phylogenetic analysis of macroevo- Dijkshoorn, L., Lievens, B., 2017. Phylogenetic signal in phenotypic traits related lutionary patterns in fermentative yeasts. Ecol. Evol. 6 (12), 3851e3861. https:// to carbon source assimilation and chemical sensitivity in Acinetobacter species. doi.org/10.1002/ece3.2097. Appl. Microbiol. Biotechnol. 101 (1), 367e379. https://doi.org/10.1007/s00253- Paradis, E., Claude, J., Strimmer, K., 2004. APE: analyses of phylogenetics and evo- 016-7866-0. lution in R language. Bioinformatics 20 (2), 289e290. https://doi.org/10.1093/ Vannette, R.L., Fukami, T., 2017. Dispersal enhances beta diversity in nectar mi- bioinformatics/btg412. crobes. Ecol. Lett. 20 (7), 901e910. https://doi.org/10.1111/ele.12787. Peter, G., Takashima, M., Cadez, N., 2017. Yeast habitats: different but global. In: Vervoort, Y., Linares, A.G., Roncoroni, M., Liu, C., Steensels, J., Verstrepen, K.J., 2017. Buzzini, P., Lachance, M.-A., Yurkov, A. (Eds.), Yeasts in Natural Ecosystems: High-throughput system-wide engineering and screening for microbial Ecology. Springer International Publishing AG, Cham, Switzerland, pp. 39e71. biotechnology. Curr. Opin. Biotechnol. 46, 120e125. https://doi.org/10.1016/ Peter, J., De Chiara, M., Friedrich, A., Yue, J.X., Pflieger, D., Bergstrom,€ A., et al., 2018. j.copbio.2017.02.011. Genome evolution across 1,011 Saccharomyces cerevisiae isolates. Nature 556 Wang, J., He, X., 2020. The completeness and stratification in yeast genotype- (7701), 339e344. https://doi.org/10.1038/s41586-018-0030-5. phenotype space. bioRxiv. https://doi.org/10.1101/2020.01.13.904607. Peter, J., Schacherer, J., 2016. Population genomics of yeasts: towards a compre- Yurkov, A., 2017. Temporal and geographic patterns in yeast distribution. In: hensive view across a broad evolutionary scale. Yeast 33 (3), 73e81. https:// Buzzini, P., Lachance, M.-A., Yurkov, A. (Eds.), Yeasts in Natural Ecosystems: doi.org/10.1002/yea.3142. Ecology. Springer International Publishing AG, Cham, Switzerland, pp. 39e71.