ORIGINAL ARTICLE
doi:10.1111/j.1558-5646.2008.00390.x
RAPID SPECIATION FOLLOWING RECENT HOST SHIFTS IN THE PLANT PATHOGENIC FUNGUS RHYNCHOSPORIUM
Pascal L. Zaffarano,1,2,3 Bruce A. McDonald,1 and Celeste C. Linde4 1Plant Pathology, Institute of Integrative Biology, ETH-Zurich, LFW, CH-8092 Zurich,¨ Switzerland 2E-mail: [email protected] 4School of Botany and Zoology, Building 116, Daley Rd, Australian National University, Canberra ACT 0200, Australia
Received July 6, 2007 Accepted March 11, 2008
Agriculture played a significant role in increasing the number of pathogen species and in expanding their geographic range during the last 10,000 years. We tested the hypothesis that a fungal pathogen of cereals and grasses emerged at the time of domestication of cereals in the Fertile Crescent and subsequently speciated after adaptation to its hosts. Rhynchosporium secalis, originally described from rye, causes an important disease on barley called scald, although it also infects other species of Hordeum and Agropyron. Phylogenetic analyses based on four DNA sequence loci identified three host-associated lineages that were confirmed by cross-pathogenicity tests. Bayesian analyses of divergence time suggested that the three lineages emerged between ∼1200 to 3600 years before present (B.P.) with a 95% highest posterior density ranging from 100 to 12,000 years B.P. depending on the implemented clock models. The coalescent inference of demographic history revealed a very recent population expansion for all three pathogens. We propose that Rhynchosporium on barley, rye, and Agropyron host species represent three cryptic pathogen species that underwent independent evolution and ecological divergence by host-specialization. We postulate that the recent emergence of these pathogens followed host shifts. The subsequent population expansions followed the expansion of the cultivated host populations and accompanying expansion of the weedy Agropyron spp. found in fields of cultivated cereals. Hence, agriculture played a major role in the emergence of the scald diseases, the adaptation of the pathogens to new hosts and their worldwide dissemination.
KEY WORDS: Barley, coevolution, crop domestication, host shift, plant pathogens, TMRCA.
Agriculture began with the domestication of the plants and ani- by the corresponding increase in human population density that mals that enabled the rapid human population expansion of the allowed (1) the maintenance of stable pathogen populations, (2) last 10,000 years (Cavalli-Sforza et al. 1994; Diamond 1997). the increase of interspecies transmission from domesticated ani- As agriculture spread, populations of pathogens on humans and mals, and (3) the expansion of human populations into novel en- their domesticated animals and plants expanded. Agriculture may vironments and resulting exposure to novel pathogens (Diamond also have contributed to the number of pathogen species and their 2002; Armelagos and Harper 2005). A similar process affected the current geographic ranges through the anthropogenic modifica- pathogens colonizing agricultural crops. The high host densities tion of the environment (Schrag and Wiener 1995; Kolar and and genetic uniformity of host populations coupled with cultiva- Lodge 2001; Diamond 2002; Anderson et al. 2004; Armelagos tion practices and trade created more uniform environments that and Harper 2005). The expansion of human pathogens follow- maintained stable pathogen populations and were conducive for ing the shift of human societies to agriculture has been explained disease development and transmission. The movement of domes- ticated plants and their respective pathogens into new areas could 3Corresponding author. simultaneously introduce “domesticated” pathogens into new
C 2008 The Author(s) . Journal compilation C 2008 The Society for the Study of Evolution. 1418 Evolution 62-6: 1418–1436 RECENT PATHOGEN ORIGINS
areas in which they could colonize “wild” hosts and expose the was not supported for fungal pathogens of the genera Ustilago and domesticated crops to new pathogens that could shift from wild Sporisorium causing smut on Poaceae such as maize, sorghum, to domesticated hosts. Anthropogenic influences on the environ- and sugarcane. Divergence time estimates for these species were ment such as intensification of crop production and global trade millions of years, indicating that speciation occurred long be- or other factors such as climate change are thought to promote fore their hosts were domesticated less than 10,000 years ago the emergence of new plant diseases (Anderson et al. 2004; Slip- (Munkacsi et al. 2007). pers et al. 2005; Woolhouse et al. 2005; Money 2007). However, Rhynchosporium secalis (Oudem.) J.J. Davis. causes an im- the history and the processes that led wild pathogens to become portant disease called scald on barley, rye, and other grasses. domesticated and vice versa have been poorly studied. Analyses of nucleotide sequences of NIP1, a gene encoding a Many human pathogens may have originated since the rise toxin involved in pathogenicity, indicated that R. secalis started of agriculture as a result of host shifts from domestic animals to colonize barley ∼5000–7500 years after the domestication of (Pearce-Duvet 2006; Wolfe et al. 2007). Although this hypothesis barley (Brunner et al. 2007) that occurred ∼10,000 B.P. (Badr has been widely proposed for many human diseases, unequivo- et al. 2000). This led to the hypothesis that domestication and cal evidence based on phylogenetics and estimates of divergence agricultural practices affected the emergence and global spread of times are rarely presented (Pearce-Duvet 2006 for review). For the this barley pathogen. In this study we investigate how agriculture tuberculosis bacterium Mycobacterium tuberculosis, tapeworms shaped the demography of the pathogen through time and deter- of the genus Taenia, and the protozoan Plasmodium falciparum mine whether the emergence of the pathogen fits the domestication causing falciparal malaria, there is evidence that their progenitors hypothesis. The evolutionary history of R. secalis was investigated predate the rise of agriculture and may have already been hu- using nucleotide sequences of several housekeeping genes not in- man pathogens before animal domestication (Pearce-Duvet 2006 volved in pathogenicity, in contrast to NIP1. We included 316 for review). In a similar way, the progenitors of modern plant isolates from barley, rye, and uncultivated grasses from different pathogens may have been present already on the progenitors of continents. Population genetic analysis, multiple gene genealo- crop plants and diverged with them after domestication (Munkacsi gies, coalescent-based approaches, and analyses of pathogenicity et al. 2007). Alternatively, domestication might have strongly in- were combined to determine if cryptic species exist on cultivated fluenced host shifts leading to the emergence of new diseases on and wild hosts of the pathogen. We then determined whether the crops (Couch et al. 2005). Agricultural practices might have sub- fungal populations on uncultivated grasses were ancestral to pop- sequently favored host specialization, reproductive isolation, and ulations on cereal hosts or originated from the same or different speciation of plant pathogens on new hosts (Hansen 1987; Kohn ancestors. Divergence time between and within host-associated 2005). The role that agriculture has played in the emergence of populations was estimated and the demographic history of the a plant disease can be evaluated by dating the divergence of the pathogen reconstructed. We present evidence that agriculture has causal agent. For the most important group of plant pathogens, the driven the evolution of R. secalis since the Neolithic through a fungi, few studies have attempted to date divergence from their host shift from wild grasses to cultivated barley, falsifying the progenitors (Couch et al. 2005; Munkacsi et al. 2007; Stukenbrock domestication hypothesis for this plant pathogen. et al. 2007), mainly due to the lack of fossil records and large er- rors associated with molecular clocks. In contrast, the evolutionary Material and Methods history and time of domestication has been well studied for most FUNGAL ISOLATES of the important staple crops infected by these fungal pathogens. Isolates of R. secalis from nine different hosts including culti- The origins of the fungal pathogens Mycosphaerella gramini- vated barley (Hordeum vulgare), rye (Secale cereale) and triticale cola causing septoria leaf blotch on wheat and Magnaporthe (× Triticosecale Wittmack), as well as six wild grasses; Agropy- oryzae causing rice blast coincided with the domestication of their ron caninum, Agropyron repens, Bromus diandrus, Hordeum lep- current hosts (Couch et al. 2005; Stukenbrock et al. 2007), start- orinum, Hordeum murinum and Hordeum spontaneum, were in- ing ∼10,000 years before present (B.P.) for wheat in the Fertile cluded in this study. Isolates of Rhynchosporium orthosporum (the Crescent (Flannery 1973; Lev-Yadun et al. 2000; Salamini et al. only other described Rhynchosporium species) infecting Dactylis 2002) and ∼7000 years B.P. for rice in East Asia (Flannery 1973; glomerata were included as outgroup. The R. secalis isolates orig- Crawford and Shen 1998; Higman and Lu 1998). Domestication, inated from 21 countries representing five continents. Many of agricultural practices, and trade strongly influenced the pathogen’s these isolates were representatives of a collection of R. secalis that evolution and the diseases they cause on these crops (Couch et al. was characterized previously (McDermott et al. 1989; McDonald 2005; Stukenbrock et al. 2007). We refer to this very recent origin et al. 1999; Salamati et al. 2000; Linde et al. 2003; Zaffarano of the pathogens associated with the domestication of the host et al. 2006). Detailed descriptions of the isolates are found in as the “domestication hypothesis.” The domestication hypothesis Table 1. Each isolate had a unique multilocus genotype based on
EVOLUTION JUNE 2008 1419 PASCAL L. ZAFFARANO ET AL.
Table 1. Origin of the Rhynchosporium isolates used in the study.
Host Geographic No. of Previous publication or source and origin isolates year of collection (in parentheses) Agropyron caninum Switzerland 1 C.C. Linde (2002) Agropyron repens Switzerland 50 P. L. Zaffarano, M. Zala, C.C. Linde (2004–2005) Barley Australia 7 McDonald et al. (1999) Barley Azerbaijan 2 A. Yahyaoui (2003) Barley Ethiopia 13 A. Yahyaoui (2003), Zaffarano et al. (2006) Barley Eritrea 6 A. Yahyaoui (2003) Barley Finland 8 Salamati et al. (2000) Barley France 5 Zaffarano et al. (2006) Barley Germany 7 Zaffarano et al. (2006) Barley Jordan 4 Zaffarano et al. (2006) Barley Kyrgistan 2 A. Yahyaoui (2003) Barley New Zealand 8 M. Cromey (2004) Barley Norway 16 Salamati et al. (2000) Barley South Africa 3 Linde et al. (2003) Barley Sweden 1 S. Salamati (1996) Barley Switzerland 45 Linde et al. (2003) Barley Syria 5 Linde et al. (2003) Barley Tunisia 3 A. Yahyaoui (2003) Barley Turkey 2 Zaffarano et al. (2006) Barley United Kingdom 3 Zaffarano et al. (2006) Barley USA 6 McDermott et al. (1989) Bromus diandrus Australia 2 McDonald et al. (1999) Dactylis glomerata Italy 7 C.C. Linde (2004) Dactylis glomerata Switzerland 6 P. L. Zaffarano, M. Zala, C.C. Linde (2004–2005) Hordeum leporinum Australia 18 McDonald et al. (1999) Hordeum murinum Switzerland 4 C.C. Linde, P. L. Zaffarano (2004) Hordeum murinum USA 3 C.C. Linde (2003) Hordeum spontaneum Syria 5 M. Abang (2003) Rye Russia 25 L. Lebedeva (2003) Rye Switzerland 27 C.C. Linde (2002), Zaffarano et al. (2006) Triticale Switzerland 6 C.C. Linde (2002) Triticale France 16 A. Bouguennec, L. Jestin (2002) Total 316
RFLP loci, RAPD fingerprints, mating types (Linde et al. 2003; 728F and EF1-986R (Carbone and Kohn 1999). Portions of - Zaffarano et al. 2006), and microsatellites (Linde et al. 2005). tubulin were amplified in three steps. The first primer set used was ATUB-25F (5-GAGAAGCTATTAGCATCAACG-3) and DNA EXTRACTION, AMPLIFICATION, AND ATUB-649R (5 -CTCCTTTCCAACAGTGTAGTGAC-3 ), the SEQUENCING second ATUB-540F (5 -TCCCTAGAACCATCTACTGCG-3 ) Isolation and culturing of fungal isolates were as described in and ATUB-1206R (5 -CTTTGGCGGCAGACAACTG-3 ), and McDonald et al. (1999). DNA was extracted from lyophilized tis- the third ATUB-1103F (5 -ACAGTTGTCTCCTCCATTACCG- sue with either the method described in McDonald et al. (1999) 3 ) and ATUB-1603R (5 -TGGACGAAGGCACGCTTAGAG- or the DNeasy Plant Mini DNA extraction kit (Qiagen GmbH, 3 ). Part of the -tubulin gene was amplified with the Hilden, Germany) according to the manufacturer’s instructions. primers BTUB-21F (5 -ATGCGTGAAATCGTACGTCAC-3 ) Four nuclear DNA loci were analyzed. Sequences of the ITS and BTUB-615R (5 -TGACCGAAAGGACCAGCACG-3 ). PCR region (ITS1, 5.8S rRNA gene, ITS2) were obtained by PCR reactions were carried out in 20 l volumes containing 2 l10× amplification of genomic DNA of isolates using the primers PCR buffer, 0.1 mM of each dNTP, 0.5 M of each primer, ITS4 and ITS5 (White et al. 1990). The translation elonga- 0.5 U of Taq polymerase (New England BioLabs, Allschwil, tion factor 1 alpha (EF-1 ) was amplified with primers EF1- Switzerland), and 5 l of genomic DNA (5–20 ng final DNA
1420 EVOLUTION JUNE 2008 RECENT PATHOGEN ORIGINS
concentration). PCR conditions for ITS included a denaturing bootstrap replicate. All characters were equally weighted and un- step at 96◦C for 2 min, followed by 35 cycles at 96◦C for 1 ordered. min, 55◦C for 1 min, and 72◦C for 1 min. Finally, a 5-min Combining several independent loci can increase the accu- PCR extension was carried out at 72◦C. PCR amplifications for racy and confidence of phylogenetic inference (Pamilo and Nei the other three genes were the same as for ITS except that an- 1988; Takahata 1989; Rosenberg 2002). To test whether the data nealing cycles were carried out at 50◦CforEF-1 ,at56◦Cfor of individual loci might be combined, that is, whether the gene -tubulin and at 65◦Cfor -tubulin. For the R. orthosporum genealogies from the four loci were significantly different from isolates the annealing temperature had to be lowered to 53◦Cto each other, the partition homogeneity test (PHT) (Farris et al. amplify -tubulin and -tubulin. 1995) was used. The PHT used only informative characters and Amplification products were electrophoresed on 1% agarose 100 replicates of simple stepwise-addition MP heuristic searches gels to verify the amplification of a single fragment of the appro- (TBR; maxtrees = 500). However, methods to detect conflicts priate length. Amplified products were purified using a Millipore among data partitions, such as the incongruence length difference multiscreen PCR plate (MANU 030 PCR) following the manufac- (ILD) test (implemented as the PHT in PAUP) could be poor indi- turer’s instructions. The purified PCR products were resuspended cators of dataset combinability (e.g., Cunningham 1997; Dolphin in 20 l of water and sequenced bidirectionally with ABI PRISM et al. 2000; Barker and Lutzoni 2002; Darlu and Lecointre 2002; BigDye Terminator v3.0 and 3.1 ready reaction cycle sequencing Dowton and Austin 2002). Therefore, congruence between gene kit (Applied Biosystems, Foster City, CA). The sequencing reac- phylogenies was also estimated by visual inspection of topologies tion was conducted in a final volume of 10 l containing 0.4 l and statistical support (Mason-Gamer and Kellogg 1996; Wiens ready reaction mix, 1.6 l5× reaction buffer (400 mM Tris-HCL, 1998). pH 9.0, 10 mM MgCl2), 1 M primer, and 5 l of the purified For BML analyses the optimal model selected under the AIC PCR product. The DNA samples were sequenced with an ABI- implemented in Modeltest was specified as the prior for each 3100 automated sequencer. gene. MrBayes allowed different data partitions to be modeled separately for the combined dataset (Ronquist and Huelsenbeck PHYLOGENETIC ANALYSES 2003). One cold and three incrementally heated Markov chains DNA sequences were aligned and edited manually with Se- were run simultaneously starting from random trees for 5,000,000 quencher 4.5 (Gene Codes Corporation, Ann Arbor, MI). Iso- generations for the single loci, and for 10,000,000 generations for lates were assigned to haplotypes, that is, to unique DNA the combined dataset. Trees were sampled every 500th generation sequences at each sequence locus. Haplotypes and their frequen- for the single loci and every 1000th generation for the combined cies were obtained with the program MAP (Aylor et al. 2006) dataset, resulting in 10,000 trees of which 1000 were discarded and SITES version 1.1 (Hey and Wakeley 1997) by recoding as the “burn-in.” At least two independent runs were performed insertions or deletions (indels) and removing infinite site vi- to ensure analyses were not converging on a local optimum. The olations, as implemented in the SNAP Workbench (Price and replicate runs were compared to confirm that the analyses reached Carbone 2005). Redundant sequences were removed from the stationarity at similar likelihood scores by plotting the −lnL per datasets by choosing only one individual for each sequence generation in the program Tracer 1.3 (Rambaut and Drummond haplotype. 2003). After confirming that the replicate runs reached stationarity Two tree-building methods were used, namely maximum par- at similar likelihood scores and that the topologies were similar, simony (MP) and Bayesian maximum likelihood (BML), which the remaining trees of the separate runs were pooled together and were performed in PAUP∗ version 4.0b10 (Swofford 2002) and used for calculating the posterior probabilities in the 50% majority MrBayes 3.0b4 (Ronquist and Huelsenbeck 2003), respectively. rule consensus tree in PAUP. The program Modeltest 3.7 (Posada and Crandall 1998) was used to assess which model of nucleotide substitution best fit the data COALESCENT ANALYSES of each locus for BML under the Akaike information criterion The ancestral history of the host-associated populations of the (AIC). The trees were rooted with sequences from the only other fungus was inferred by coalescent-based gene genealogies. The described Rhynchosporium species, R. orthosporum. For MP anal- analysis was conducted in the SNAP Workbench (Price and Car- ysis, gaps were treated as fifth character states in heuristic searches bone 2005) that contains a series of programs described below that were conducted following 100 replicates of random step- to reconstruct the history of haplotypes. Guidelines in Carbone wise addition and tree bisection-reconnection (TBR) for branch- et el. (2004) were followed for each analysis. Sequences were swapping. Branch support for all parsimony analyses was esti- collapsed into unique haplotypes using SNAP MAP (Aylor et al. mated by performing 1000 bootstrap replicates with a heuris- 2006) and SITES version 1.1 (Hey and Wakeley 1997) by recoding tic search consisting of 100 random-addition replicates for each indels and removing infinite site violations. Prior to the application
EVOLUTION JUNE 2008 1421 PASCAL L. ZAFFARANO ET AL.
of coalescent methods, the absence of selection and recombina- demographic history of lineages in a Bayesian coalescent-based tion had to be verified. Therefore, deviation from neutrality was framework. The Bayesian skyline plot (BSP) (Drummond et al. measured by Fu and Li’s D∗ and Fu and Li’s F∗ (Fu and Li 1993) 2005) was specified as demographic model because it can fit a and Tajima’s D (Tajima 1989) with DnaSP version 4.0 (Rozas wide range of demographic scenarios. The MCMC analyses were et al. 2003). Incompatibility matrices were generated to detect in- first performed with short runs with chain length of 106 to op- compatibility among segregating sites in SNAP Clade (Markwordt timize the scale factors of the priors. The analysis was then run et al. 2004) and SNAP Matrix (Markwordt et al. 2004). for 108 generations sampling every 1000th iteration after an ini- Migration matrices indicating the number and direction of tial burn-in of 10%. The performance of the MCMC process was migrants exchanged between populations were constructed in checked for stationarity and large effective sample sizes in Tracer. the program MIGRATE (Beerli and Felsenstein 1999, 2001). The mean and corresponding credibility intervals of the estimated MIGRATE estimates the product of effective population size and parameters and the BSP were depicted using Tracer. mutation rate, and the amount and direction of gene flow between the host-associated populations. The analysis included 20 short PATHOGENICITY ASSAYS chains with 500 sampled genealogies each and 5 long chains with To determine whether Rhynchosporium isolates from each phy- 5000 sampled genealogies each. Chain heating was adaptive, with logenetic lineage (see Results) could infect hosts of other phy- four different temperatures. The migration matrices were used as logenetic lineages, barley, rye, and H. murinum were inoculated starting backward migration matrices for coalescent analysis with with representative isolates from each lineage originating from the population subdivision in the program GENETREE version 9.0 same geographical region, in this case Switzerland. Pathogenic- (Griffiths and Tavare´ 1994; Bahlo and Griffiths 2000) as incor- ity is here defined as the ability of a fungal isolate to infect a porated in the SNAP workbench. GENETREE reconstructs the host species. In this case pathogenicity was used as a phenotypic ancestral history of haplotypes showing a coalescence tree with marker to assess species boundaries and host specialization. relative time of divergence between host-associated pathogen pop- The isolates for the pathogenicity tests were a representa- ulations. The genealogy with the highest root probability was de- tive sample including different Swiss locations, collection years, termined by performing 500,000 simulations of the coalescent and genotypes. Two trials were undertaken. In the first trial, with five different starting random number seeds. From these runs, only isolates from cultivated grasses were used. The 16 selected the tree with the highest root probability was selected showing isolates originated from fields planted to barley, rye, and triti- the distribution of mutations along the branches of the pathogen cale. The seven barley-infecting isolates 99CH2A2B, 99CH5E4A, populations. 99CH5E6B, 99CH5H10A, 99CH6C3A, 99CH6E3B, and 00A1B, originated from collections of 1999 and 2000 that were de- DIVERGENCE TIME ESTIMATES AND DEMOGRAPHIC scribed previously (Linde et al. 2003; Zaffarano et al. 2006). ANALYSIS Four of the eight rye-infecting isolates (99CH1E7A, 99CH1B8, The Bayesian Markov Chain Monte Carlo (MCMC) method im- 99CH1H10B, and 99CH1D4A) were collected in 1999 as de- plemented in the program BEAST version 1.4.1 (Drummond and scribed earlier (Zaffarano et al. 2006). The remaining isolates Rambaut 2005) was used to estimate time of divergence between (02CH4-14a.1, 02CH4-9a.2, 02CH4-5a.1, and 02CH4-6a.1) were host-associated populations of the pathogen, that is the time to collected in 2002 from a rye field near Maur in the canton of the most recent common ancestor (TMRCA), and past popula- Zurich, whereas the isolate 02CH2-3c.1 was collected from triti- tion dynamics. Kasuga et al. (2002) proposed a range of mutation cale at the experimental station of Changins in the canton of Vaud rates for the Eurotiomycetes, a monophyletic class of Ascomycota. in 2002. We applied these mutation rates representing the lower end, the To produce inoculum, all isolates were grown from sil- mean, and the upper end of the range, that is, 0.9 × 10−9, 8.8 × ica gel storage onto Difco lima bean agar (Becton, Dickinson 10−9, and 16.7 × 10−9 mutations per site and per year. The anal- and Co., Sparks, MD) amended with kanamycin (50 mg/L). ysis was conducted with the multilocus dataset as the program Plates were incubated for 14 days at 18◦C in the dark. Colonies allows partitioning the combined datasets. Different evolutionary were then transferred to fresh lima bean agar plates and in- substitution models can be included for each partition and ap- cubated under the same conditions as above. After 14 days plied simultaneously. The substitution models specified for each spores were harvested by adding 2 mL of sterile water to each gene were the same as obtained under the AIC in Modeltest. Esti- plate and scraping spores off the agar surface with a steril- mates assuming a strict molecular clock were compared to those ized microscope slide. A portion of the spores from each iso- performed using the relaxed molecular clock option with uncor- late was transferred into 1.8 mL CryoTubes (Nunc Cryoline Sys- related, branch-specific rates following lognormal or exponen- tems, Roskilde, Denmark) containing anhydrous silica gel (Fluka tial distribution (Drummond et al. 2006). BEAST also infers the Chemie GmbH, Steinheim, Germany) for long-term storage at
1422 EVOLUTION JUNE 2008 RECENT PATHOGEN ORIGINS
−80◦C. The rest of the spore suspension was spread across mental station of Changins in the canton of Vaude in Switzerland, the surface of 20–50 fresh lima bean agar plates. The plates except SEG-A3.2.1.1 which was collected from A. repens plants were incubated at 18◦C and after 14 days spores were harvested bordering the road on the Sattelegg pass in canton Schwyz in as above and filtered through two layers of cheesecloth. Spore Switzerland. The isolates from H. murinum were collected from concentrations were adjusted to 2 × 105 spores/mL with a hema- weedy plants growing alongside streets in Zurich. Seed was har- cytometer (Thoma cell, 0.1-mm depth, 0.0025 mm2) in a spore vested from these plants for inoculation trials. Two previously used solution of 150 mL per isolate. isolates, one each from barley (00CHA1B) and rye (99CH1E7a), Inoculations were conducted on four barley (Chariot, Julia, were included as control isolates. The barley and rye varieties as Pasadena, and Plaisant) and four rye varieties (Avanti, Born, well as H. murinum were represented by three pots each with five Danko, and Picasso). These varieties ranged from moderately to seedlings per pot. In total 10 inoculations were applied for a total highly susceptible to scald. All host varieties were grown in pots of 150 pots. The negative control, sprayed with water and Tween, with a diameter of 13 cm filled with soil mixture Rasenerde Top was represented by 60 pots, that is, six pots containing all host va- Dressing (containing sand, compost, perlite, white peat, and min- rieties and H. murinum per inoculated isolate. The second trial was eral fertilizer; Ricoter AG, Aarberg, Switzerland). Seeds of each repeated once. Procedures for the isolate culturing, inoculation, host variety were sown separately in pots and thinned to five plants and disease assessment were the same as for the first inoculation per pot. trial, except that the spore concentration was 106 spores/mL for The plants were grown in a single greenhouse chamber with both the first experiment and the repetition. a photoperiod of 14 h-day at 18◦C and a 10 h-night period at 15◦C. Relative humidity was set at 60%. The seedlings were inoculated when they had two to three fully emerged leaves. Two drops of Results Tween 20 (Sigma-Aldrich, Buchs, Switzerland) were added to 150 NUCLEOTIDE SEQUENCES AND PHYLOGENETIC mL of spore suspension. For each isolate, three pots of each variety ANALYSES were inoculated. For inoculation, all 24 pots were placed onto a The ITS region (848 bp), portions of the -tubulin (1609 bp), - rotating table in a semiautomatic inoculation chamber and the tubulin (609 bp), and the EF-1 (365 bp) (sizes including gaps) leaves were sprayed with a fine mist until run-off. An additional were amplified for all isolates. Summaries of the phylogenetic in- set of 96 pots, that is, 6 pots per each inoculated isolate, containing formation for the four loci are shown in Table 2. There were 24, 20, all host varieties, was sprayed with sterilized water amended with 15, and 11 parsimony informative characters, respectively for each two drops of Tween 20 as a negative control. The inoculated pots locus (Table 2). Haplotypes representing one of each set of iden- were kept for 48 h at a relative humidity of 90–100%. tical sequences were used for analysis. The haplotypes defined by After 14 days disease was assessed on the second and third the variable sites of the datasets are reported in the online Supple- leaves following the scale described in Ali and Boyd (1973). The mentary Table S1. Haplotypes were shared only between either ratings were 0 = no visible lesion or symptoms, 1 = small lesions populations from rye and triticale, or between populations from at the tip or on the margin and base of the leaf blades, 2 = nar- barley, H. leporinum, H. murinum, and H. spontaneum (Hordeum row band of lesions extending down the length of the leaf blade, spp.) and B. diandrus for all loci. The single isolate from A. can- 3 = broad well-developed lesions covering large areas across leaf inum always grouped with the haplotypes formed by the A. repens blade, 4 = leaves wilted, no evidence of discrete lesions. Reac- isolates. Haplotype H1 occurred most frequently for each respec- tion 0 was considered highly resistant, reaction 1 was resistant, tive locus, and was represented in most of the sampled geograph- reaction 2 was intermediate, and reactions 3 and 4 susceptible ical populations originating from barley, as well as in Hordeum and highly susceptible, respectively. This inoculation trial was re- spp. and B. diandrus. Conversely, many haplotypes were unique peated one month later using a higher inoculum concentration of to one population or host species (see online Supplementary 106 spores/mL. Table S1). In the second trial, Rhynchosporium isolates from uncul- The MP and BML analyses were used to infer genealogies tivated grasses were included for pathogenicity testing. Six of the haplotypes from the four single-locus alignments (Table 2, Rhynchosporium isolates from A. repens (Danikon-1.1.1,¨ WPK- Fig. 1, and see online Supplementary Figs. S1–S4). BML analysis 5b.1A3.1, K2B-2C1.1, Brutten-3.1.2,¨ RAC-2-A9.1, and SEG of the four loci followed an optimal evolutionary model selected A3.2.1.1) and two from H. murinum (WPK-2.1 and WPK-A8.4) under the AIC in Modeltest. These models were different for each that were inoculated onto two of the previously used barley (Char- gene (Table 2). All repeated runs of the BML analyses converged iot and Pasadena) and rye varieties (Danko and Picasso), as well as on the same topology. The topological patterns were consistent on H. murinum. The isolates from A. repens were collected at the across the MP and BML estimations of the phylogeny for all loci borders of cereal fields in the canton of Zurich and at the experi- except for minor differences as described below.
EVOLUTION JUNE 2008 1423 PASCAL L. ZAFFARANO ET AL.
Table 2. Phylogenetic information of the genomic regions used in this study for the individual and combined datasets.
DNA sequence region