Abstract
Over the past few years, symptoms akin to late blight disease have been reported on a variety of crop plants in South America. Despite the economic importance of these crops, the causal agents of the diseases belonging to the genus Phytophthora have not been completely characterized. In this study, we used an integrative approach that leveraged morphological, ecological, and genetic approaches to explore cryptic speciation within P. infestans sensu lato. We described a new Phytophthora species collected in Colombia from tree tomato (Solanum betaceum), a semi-domesticated fruit. All morphological traits and population genetic analyses, using microsatellite data and a reduced representation of single nucleotide polymorphism (SNP) data, support the description of the new species, Phytophthora betacei sp. nov. We have demonstrated that ecological differences are important in the persistence of P. infestans and P. betacei as genetically isolated units across an overlapping area in the northern Andes.
Introduction
Oomycetes represent an opportunity to understand microbial speciation (Restrepo et al. 2014). Their ecological characteristics have been extensively studied to reveal a wide diversity of ecological niches (e.g Soanes et al. 2007). Particular emphasis has been placed on the study of plant pathogens for which ecological speciation seems to be a common process due to specialization to particular host species (Harrington et al. 2002; Tellier et al. 2010). Furthermore, oomycetes show an unprecedented plasticity in terms of genome size and ploidy (Haas et al. 2009a; Yoshida et al. 2013), which could influence rates of speciation and extinction (Santini et al. 2009; Wood et al. 2009; Muir & Hahn 2015; Puttick et al. 2015). Despite the hypothesized species richness of the group (Restrepo et al. 2014), no analytical framework currently delimits species boundaries in oomycetes.
Molecular taxonomy can use DNA sequences to identify and delimitate species that are not amenable to genetic crosses (Roe et al. 2010). The premise of these approaches is to identify discrete genetic groups that have ceased genetic exchanges with other groups. The number of studies defining species in this way has increased recently, mainly due to the ease of obtaining information on population-level DNA variation (Roe et al. 2010; Singh et al. 2015). However, this approach has inherent limitations. Gene genealogies tend to overestimate the number of species as the population structure within a species may be mistaken for species boundaries (Dettman et al. 2003). Furthermore, multi-locus species delimitation, relying on reciprocal monophyly and strict genealogical congruence, may fail to differentiate among recently diverged lineages (Hickerson et al. 2006; Knowles & Carstens 2007; Shaffer & Thomson 2007).
The most notable genus within oomycetes, Phytophthora, includes pathogens that infect a broad range of hosts in both agricultural and natural environments, causing adverse economic consequences (Erwin & Ribeiro 1996; Duncan 1999; Fry 2008; Forbes et al. 2013). To date, the genus Phytophthora comprises more than 150 recognized species, classified into 10 phylogenetic clades that are also supported by morphological and physiological characteristics (Blair et al. 2008; Kroon et al. 2012; Martin et al. 2014). Over the past few decades, the numbers of recognized species within most divisions of the genus Phytophthora have nearly doubled (Cooke et al. 2000; Kroon et al. 2012; Martin et al. 2012; Forbes et al. 2013). However, defining clear and objective species boundaries, as is the case for most oomycetes, remain a challenge in all Phytophthora clades.
Within the genus, P. infestans has become a “model system” because of its undoubted economic impact. This pathogen affects important crops, such as potato (Solanum tuberosum) and tomato (Solanum lycopersicum) (Haverkort et al. 2008; Visser et al. 2009), making it one of the most threatening plant disease agents in the world. Although P. infestans was once considered a single species (henceforth referred to as P. infestans sensu lato), it has been shown to be a species complex (Forbes et al. 2013). Four other species related to P. infestans sensu stricto have been identified over the last 35 years. Phytophthora mirabilis has been found in Central America (Galindo & Hohl 1985) infecting only Mirabilis jalapa, an ornamental and medicinal plant in the region. Phytophthora ipomoeae (Flier et al. 2002) infects two morning glory species endemic to the highlands of central Mexico, Ipomoea longipedunculata and I. purpurea (Flier et al. 2002; Badillo-Ponce et al. 2004). Phytophthora phaseoli, initially classified as P. infestans (Thaxter 1889), is distributed globally but infects only lima beans (Phaseolus lunatus). Due to host-preference studies, P. phaseoli was described as a different species and for over 60 years thought to be the closest relative of P. infestans. Genetic comparisons have also revealed the existence of a separate group composed of strains from Ecuador and Peru that are collectively called P. andina (Oliva et al. 2010). Phylogenetic hypothesis with both nuclear and mitochondrial markers, reveal this species as polyphyletic, suggesting that it might be a species complex (Adler et al. 2002, 2006; Kroon et al. 2004; Oliva ei al. 2010; Cárdenas et al. 2012; Forbes ei al. 2013, 2016; Goss et al. 2014; Lassiter et al. 2015). To date, P. andina is composed of the nuclear lineages EC-2 with two mitochondrial lineages, Ia and Ic; and EC3 (with one mitochondrial haplotype, Ia) (Oliva et al. 2010). A clear definition of the P. andina species is needed.
In this study, we used an integrative approach that leveraged morphological, ecological, and genetic approaches to explore cryptic speciation within P. infestans sensu lato. We found and defined a new species of Phytophthora infecting tree tomato (Solanum betaceum) in southern Colombia. Furthermore, we investigated the importance of host specificity in maintaining species boundaries within the P. infestans sensu lato species complex. Finally, we formally describe this new species as Phytophthora betacei sp. nov. For convenience, we refer to the species by using our proposed name for it throughout this article.
Materials and methods
Disease occurrence and collection of isolates
All P. betacei samples were collected in southern Colombia between 2008 and 2009 (Figure S1). We sampled three to four leaves from 10 randomly selected tree tomato plants per plantation that showed symptoms akin to late blight. In total, we sampled 34 locations from two Colombian states, Nariño and Putumayo (Figure S1). The initial collections comprised over 970 putatively infected leaves. One to three lesions per leaf were excised (~ 0.5 to 1 cm2) from the margin between necrotic and healthy tissues. Excised leaf pieces were surface-sterilized by submerging them in 70% ethanol for 20 to 30 s and then washed with sterile distilled water to remove excess ethanol (~ 10 sec). The leaf pieces were dried on a sterile paper towel and subsequently transferred to a selective medium prepared with tree tomato fruit (0.25 g of CaCO3, 0.5 g of yeast extract, 25 g of sucrose, 15 g of agar, and 100 ml of tree tomato extract, composed of 550 g of tree tomato fruit per liter of water). Subsequently, single zoospores were isolated from sporangia washed from the tree tomato medium with sterile distilled water. The sporangial suspension was adjusted to 2.0 × 103 sporangia per ml, using a haemocytometer, and maintained at 4°C for 4 h to induce zoospore release before spreading 10 μl of the suspension onto 100-mm petri plates containing 10 ml of tree tomato medium. To better visualize zoospore germination, the medium was centrifuged at 8,000 rpm, and only the supernatant was used. The plates were incubated at 18°C for approximately 24 h before individual zoospores, observed through a stereoscope, were picked using a sterile syringe and placed onto fresh tree tomato medium.
In total, we successfully isolated 128 P. betacei strains. All single zoospore isolates were cultured in a tree tomato medium for 7 to 15 days at 18°C and then stored in the Phytophthora collection of the Museum of Natural History at Universidad de los Andes. Isolate P8084 was selected as the P. betacei holotype. This strain is maintained in culture, as well as cryopreserved with 1% glycerol, and deposited in the Museum of Natural History under accession number Andes-F 1172. All the other P. betacei isolates collected in this study are deposited in the same museum under accession numbers Andes-F 1081 to Andes-F 1207 (Table S1).
Phylogenetic analyses and population genetics
DNA extraction and sequencing
The mycelia of each Phytophthora strain were grown in a liquid Plich medium (Van der Lee T et al. 1997) for 15 days at 20°C. Subsequently, mycelia were washed with sterile distilled water and macerated thoroughly in liquid nitrogen, using a cooled pestle and mortar. The macerated mycelia (0.1 g) were immediately transferred into a micro-centrifuge tube (1.5 ml) and DNA was extracted using the DNA kit OmniPrep (G-Biosciences) and following the manufacturer’s instructions. The DNA was suspended in Tris-EDTA buffer (pH 8.0) and was treated with RNAse. The DNA quality and quantity were scored, using NanoDrop ND-1000.
Restriction fragment length polymorphism analysis using mitochondrial haplotyping and probe RG57
Phytophthora lineages are conventionally characterized by using mitochondrial haplotyping (Carter et al. 1990; Griffith & Shaw 1998) and restriction fragment length polymorphism (RFLP) analysis with the highly polymorphic probe RG57 (Goodwin et al. 1992). We used the same approach to compare P. betacei with previously reported lineages within the Phytophthora genus. The mitochondrial haplotype was determined using the PCR-RFLP method with reference strains US-1 and US-8 included as positive controls (Griffith & Shaw 1998; Ordonez et al. 2000; Adler et al. 2002; Gavino & Fry 2002).
Strain typing using microsatellite markers
To assess the population differentiation among P. infestans, P. betacei, and P. andina, simple sequence repeats (SSRs) were analyzed, using the protocols developed previously by Lees et al. (2006) and described in the Eucablight Network’s Protocol section dated March 2008 (www.eucablight.org) (Table S2). A total of 116 P. betacei isolates obtained in this study (Table S1), 117 P. infestans and 17 P. andina isolates reported in Goss et al. (2014) were included in our analyses. Among the 117 P. infestans isolates, there were 17 distinct clonal lineages, as well as genotypically diverse isolates from Mexico and Northern Europe.
Population genetic analyses using microsatellite data
We used a principal component analysis (PCA) for the combined P. betacei, P. infestans, and P. andina microsatellite data. The allele frequencies at bi-allelic sites for the triploid P. infestans isolates (1/3 or 2/3) were unknown. To account for this uncertainty, we subsampled the alleles at each locus for each isolate. Since adegenet treats ploidy as a global parameter, we generated resampled datasets for the strains from all species, assuming that all the individuals across species had the same ploidy. To account for both the uncertainty in allele frequencies at bi-allelic sites in triploid P. infestans isolates and the fact that adegenet would require all samples to have the same ploidy, we generated 100 independent diploid and 100 independent triploid resampled datasets for the PCA (i.e. within each subsampled dataset, all individuals were diploids or triploids). One hundred diploid and one hundred triploid resampled datasets were created, and adegenet was run independently on each of them.
To estimate the number of populations that would best explain the genetic variance in the group of isolates studied, we used the Bayesian model-based clustering program STRUCTURE v2.3 (Pritchard et al. 2000). To account for allele frequency uncertainty at the bi-allelic triploid P. infestans loci and because ploidy is a global parameter in STRUCTURE, we used the same 200 resampled datasets described for the PCA. We ran STRUCTURE a total of 32,000 times: (2 ploidies) × (100 resampled datasets) × (8 populations, K = 1 to 8) × (20 repetitions for K selection). Each run involved 1,000,000 MCMC steps with a burn-in of 100,000 and used the following parameters: NOADMIX = 0, LINKAGE = 0, INFERALPHA = 1, ALPHA = 1.0, UNIFPRIORALPHA = 1, ALPHAMAX = 10.0, and FREQSCORR = 0. The ΔK method (Evanno et al. 2005) was used to infer the most likely number of clusters by evaluating the rate of change in the log probability of data between successive K values for each resampled dataset.
Strain typing using genotyping-by-sequencing
Genomic DNA was isolated with the DNeasy® Plant Mini Kit (QIAGEN, Germany). Genotyping-by-sequencing (GBS) was performed (as described by Elshire et al. 2011) at the Institute of Genomic Diversity (Cornell University) for a total of 70 Phytophthora isolates. Among these, there were 12 P. infestans (10 from Colombia and two reference strains from the United States, US-8 and US-12), 35 P. betacei (clonal lineage EC-3), one P. andina (clonal lineage EC-3), five P. andina (clonal lineage EC-2), three P. andina isolates of unknown clonal lineage, five P. mirabilis, eight P. ipomoeae, and one P. phaseoli isolate (Table 1). Briefly, genome complexity was reduced by digesting the total genomic DNA from individual samples with the type II restriction endonuclease ApeKI, which recognizes a degenerate 5-bp sequence (GCWGC, where W is A or T) and creates a 5’ overhang (3 bp). The digested products were then ligated to adapter pairs with enzyme-compatible overhangs; one adapter contained the barcode sequence and the other the binding site for the Illumina sequencing primer. The GBS library fragment-size distributions were checked on a BioAnalyzer (Agilent Technologies, Inc., USA). The PCR products were quantified and diluted for sequencing on an Illumina HiSeq 2500 sequencer (Illumina Inc., USA). A 96-well plate, comprising 70 samples and one blank, was multiplexed on a single Illumina flow cell lane.
To sort each of the GBS barcode samples into separate fastq files, Phytophthora samples were demultiplexed using sabre (https://github.com/najoshi/sabre), allowing no mismatches within the barcode. In total, 1,992,701 tags were analyzed and mapped against the P. infestans T30-4 reference genome (Haas et al. 2009a), using Bowtie v2.2.3 (Langmead 2010). Out of this total number of reads, 917,890 (46.1%) were aligned to unique positions, 573,880 (28.8%) were aligned to multiple positions, and 500,931 (25.1%) could not be aligned to the reference genome. For SNP calling, each SAM sample alignment file was converted into a BAM file, followed by sorting and indexing, using SAMtools. SNPs and indels were called simultaneously, using the variant caller GATK v4.3.10 (McKenna et al. 2010a). The final dataset consisted of 70 samples with 23,480 SNPs obtained from GATK. To discard the presence of sequencing errors in our data, all samples that did not fulfill the following criteria were filtered out: mapping QUAL > 30, an overall coverage between 8 and 32X (cutoff values between 5% and 95% coverage), and a minor allele frequency (MAF) > 0.05.
Maximum likelihood phylogenetic analyses of SNP data
To infer the phylogenetic relationships of the Phytophthora 1c clade species (Table 1), we created a matrix where all high-quality SNP loci obtained from our GBS analyses were concatenated into a single alignment. We generated a maximum likelihood (ML) phylogenetic tree, using RAxML (Stamatakis 2006) under the general reversible nucleotide substitution model (GTR) with 1,000 bootstrap replicates to quantify branch support. The software jModelTest v. 2.1.7 was used to select the best-fit substitution model. P. phaseoli was used as an outgroup. The phylogenetic tree was drawn using Figtree v1.4.2 (http://tree.bio.ed.ac.uk/software/figtree/) (Rambaut 2009).
Population genetics analyses using genotyping-by-sequencing data
To corroborate the population structure analyses obtained by using 11 microsatellite loci, a PCA was conducted based on the 23,480 high-quality SNP markers obtained from our GBS analyses. High quality was defined as SNPs with MAF > 0.05 and with less than 20% missing data, that is, SNPs that were present in at least 80% of the strains assessed. Genetic structure was also estimated using the Bayesian assignment test implemented in the program STRUCTURE v2.3 (Pritchard et al. 2000) for high-quality SNP markers. These are defined as SNPs with MAF > 0.05 and with less than 10% missing data. We used a total of 48 samples: 12 for P. infestans, 29 for P. betacei and 7 for P. andina (EC-2). Run parameters were as follows: 24 runs with 4 repetitions with 100,000 MCMC steps and a burn-in period of 10,000 for 6 populations (K = 1 to 6), under the NOADMIX ancestry model and allele frequencies correlated. The ΔK of Evanno (Evanno et al. 2005) was calculated using the application Structure Harvester v. 0.6.94 (Earl & vonHoldt 2012) to infer the most likely number of clusters.
Whole genome sequencing and mitochondrial genome assembly
Phytophthora betacei strains were grown on liquid Plich medium for 10 to 15 days at 20°C for subsequent genomic DNA extraction as described above. Two P. betacei (P8084 and N9022) were sequenced using Illumina sequencing. A standard shotgun library (1x200bp) was constructed and sequenced by Beijing Genomic institute - BGI (HongKong, China) on an Illumina Hiseq2000 platform using paired-ends chemistry and 100 cycles. We generated 40 Gb of 96–100 bp paired-end reads from 2 libraries with insert lengths of 200 bp. We also generated 22 Gb of Illumina mate-pair libraries (6 kb insert size) for each of the P. betacei isolates. Read mapping was done with BWA-MEM 0.7.12 (Li 2013) with parameter k=10 using P. infestans as a reference (Haas et al. 2009). Variants were called with GATK 3.2-2 (McKenna et al. 2010b) using default parameters.
Host pathogenicity assays
Evaluation of host preference
To estimate the effect of host specialization as a reproductive isolating barrier between P. betacei and P. infestans, we compared the fitness among these and P. andina, on different plant hosts. We used two isolates of each species for all infections assays: strains P8084 and N9022 for P. betacei, strains Z3-2 and RB005 for P. infestans, and strains EC3510 (EC-3, Ia) and EC3836 (EC-3, Ia) for P. andina all isolated from tree tomato (Table 1). We also included P. andina strains EC3399 (EC-2, Ia) and EC3818 (EC-2, Ia), isolated from hosts in the Anarrichomenum complex (Table 1). Each Phytophthora isolate was inoculated onto three different Solanum host species: Solanum tuberosum group phureja (yellow potato), Solanum lycopersicum (tomato), and S. betaceum variety Común (tree tomato). For isolates EC3399 and EC3818, no symptoms of infection were detected in inoculations done on S. tuberosum, S. betaceum, or S. lycopersicum. Thus, these isolates were excluded from the final analysis.
Plants were grown in the greenhouse (17 - 19°C), and leaves or leaflets were harvested after 8 to 10 weeks. Detached leaves were placed, abaxial side up, on the base of 90-mm petri plates containing moist paper towels. Three leaves were used per isolate as technical replicates. Each leaf was inoculated at four points with two 20-μl droplets of a sporangial suspension (3.5 × 104 sporangia ml−1) on each side of the main vein. The petri plates were sealed with parafilm and incubated at 15 °C with a 16-h light period. Each experiment consisted of 4 hosts, 3 genotypes, 2 isolates per genotype, 3 leaflets per isolate, and 4 inoculation points per leaflet. The whole experiment was repeated three times.
The latent period, total lesion area, and number of sporangia produced were documented by taking daily pictures of the inoculated leaves from day 1 to day 9. The latent period was scored as the number of days it took from inoculations until sporangia were observed. The lesion area was scored as the necrotic area around the inoculation site, 9 days post inoculation (dpi), and was measured using Image J (rsb.info.nih.gov/ij/). The number of sporangia produced on each leaf 9 dpi was assessed by excising individual lesions and pooling them into 15-ml disposable polypropylene culture tubes with 3 ml of sterile distilled water. After vortexing for 10 sec sporangial numbers were counted at least twice using a haemocytometer. The total number of sporangia was calculated by averaging the sporangia counts per aliquot and then multiplying it by the dilution factor.
The total number of sporangia per day were calculated by dividing the total number of sporangia produced after 9 days by the number of days when sporangia were visible (9 days – latent period). The number of sporangia produced was calculated by subtracting the total number of sporangia produced 9 dpi from the sporangial concentration in the original inoculum suspension (2,800 sporangia per leaflet). We calculated fitness parameter for each replicate, as the reproductive rate of each genotype on each host as follows:
To quantify the variation in fitness, we fitted a full factorial, linear mixed model with the R package ‘nlme’ (function ‘lme’; Pinheiro et al. 2013). In the linear model, fitness was the response variable, genotype (P. infestans, P. betacei, or P. andina) and host (capiro potato, yellow potato, tomato, or tree tomato) were fixed effects, and strain (two independent isolates per genotype) was a random effect nested within a genotype. The significance of all interactions was assessed with Crawley’s (1993, 2002) ML approach, in which the full model containing all factors and interactions was fitted and then simplified by a series of stepwise deletions, starting with the fixed-effect interaction and progressing to the interaction terms. The critical probabilities for retaining factors and determining whether effects or interactions were significant were 5% for main effects and 1% for the two-way interactions. The linear model followed the formula:
Because the residuals of this linear model were not normally distributed (Shapiro-Wilk normality test, W = 0.8342; P < 1 × 10−15) they were analyzed in a nonparametric framework. Fitness calculations on each host were compared among genotypes, using a Kruskal-Wallis test with the R package ‘stats’ (R Core team, 2013), followed by pairwise comparisons, using a Nemenyi test with a Tukey-Dist approximation for independent samples, using the R package ‘pmcmr’ (Pohlert 2014).
Phylogenetic analysis using mitochondrial genomes
We compared two newly assembled mitochondrial genomes of P. betacei sp. nov. (data taken from the whole genome sequences) and compared to data from Martin et al. (2016) in order to infer their phylogenetic position within clade 1c. MITObim (Hahn et al. 2013) was used to assembly the mitochondria of each genome. The process included 30 iterations using the quick approach with T30-4 haplotype Ia as the reference genome. Individual gene regions were aligned, using MAFFT v. 7.187 (Katoh & Toh 2010). Next, ML analyses were performed, using RAxML v.7.6.3 (Stamatakis 2006) as implemented on the CIPRES portal (Miller et al. 2010). The sequence alignment was partitioned into five subsets (rRNA genes, tRNA genes, first and second codon positions, third codon positions, and intergenic regions), similar to the work of Martin et al. (2016). The GTRGAMMA (GTR + G) model for nucleotide substitution was used but allowed the estimation of different shapes, GTR rates, and base frequencies for each partition. The majority rule criterion implemented in RAxML (-autoMRE) was used to assess clade support.
Physiological and morphological characterization of P. betacei
Effect of temperature and culture media on colony morphology and mycelial radial growth
To assess the effect of temperature and culture media on colony morphology and on mycelial radial growth of P. betacei isolates (Table 1), we evaluated four distinct media and three different incubation temperatures. The four culture media tested were: V8 juice agar (V8), Potato Dextrose Agar (PDA, Oxoid Ltda, UK), Corn Meal Agar (CMA), and Tree Tomato Agar (TTA). For each medium, one agar plug (~ 5 mm diameter) of each actively growing culture was placed in the center of each petri plate (90 mm diameter). Colony morphology and mycelial radial growth for each isolate-media combination was evaluated 15 days post inoculation (dpi) by taking pictures using a Canon Digital EOS Rebel T3i / 600D camera (Tokyo, Japan). Colony morphology was described according to Erwin & Ribeiro (1996) and Gallegly & ChuanXue (2008). Radial growth was calculated by measuring the total mycelial growth area using ImageJ (rsb.info.nih.gov/ij/). To evaluate the optimum temperature for mycelial growth, all isolate-medium combinations previously described, were incubated at 4, 18, and 25°C in a dark chamber with constant humidity.
We followed the same procedure to evaluate colony morphology and mycelial radial growth in P. andina and P. infestans. Three isolates of P. andina EC3510 (EC-3; Ia), EC3399 (EC-2; Ia), and EC3818 (EC-2; Ia) and three isolates of P. infestans Z3-2 (EC-1), US040009 (US-8), and US970001 (US-17) (Table 1) were used. All combinations of isolates and media were tested in two independent blocks with two technical replicates per combination. Given the absence of mycelial growth at 4°C and 25°C, statistical analyses for mycelial radial growth were only conducted for isolates incubated at 18°C. For this analysis we compared data for P. betacei, P. andina (clonal lineage EC-2 and EC-3), and P. infestans for a total of 240 data points (1 temperature (18°C) × 15 isolates tested × 4 media × 2 technical replicates × 2 blocks). We assessed normality of the residuals of the linear models for each trait measured. In all cases, they were not normally distributed (Shapiro-Wilk test; P < 0.05; Table S2) and thus, we assessed whether there were differences in mycelial radial growth of the three species at different temperatures and different media.
We excluded the observations on PDA because P. betacei did not grow on this medium. We pooled the observations obtained from all other media and fitted a linear model where colony area was the response and the interaction between temperature and species was the only effect of the model. Pairwise comparisons were done using Tukey’s HSD (honest significant difference) test with the R library ‘multcomp’ (function ‘glht’).
Morphological characterization
We investigated whether isolates of P. betacei presented morphological differences with respect to isolates of P. andina and P. infestans (Table 1). We examined three morphological traits: i) sporangial morphology, ii) presence of hyphal swelling and chlamydospores, and iii) mycelia morphology as follows:
i Asexual reproductive structure (sporangia) morphology
We recorded the shape (length (μm), width (μm), and area (μm)), position, and caducity of sporangia on each isolate-medium combination described above, at 18°C (optimum growth temperature for all isolates; see results). Sporangia morphology was scored by measuring length (μm), width (μm), and area (μm) of sporangia for each isolate-medium combination. Mycelia from 15-day-old actively growing colony margins on each medium was excised and immersed directly in ~ 1 ml of sterile distilled water. From 10 to 30 sporangia were measured using a 60X oil objective and the FluoView FV1000 4.0 software (Olympus PlanApo 60X, 1.42NA) implemented in an Olympus IX81 microscope, for each isolate-medium combination. Pictures of sporangia were further analyzed and processed using the ImageJ software. Replicates were conducted for each isolate-medium combination in two separate blocks.
ii Presence of hyphal swelling and chlamydospores
Mycelia from 15-day-old actively growing colony margins on CMA medium (Difco), at 18°C, was excised and immersed directly in ~ 1 ml of sterile distilled water to score the presence of hyphal swelling and chlamydospores using a 60X oil objective in an Olympus IX81 microscope.
iii Mycelia morphology
The hyphal width (μm) of each of the three species on each of the four tested media (CMA, V8, PDA, and TTA) at 18°C were measured using mycelia from 15-day-old actively growing colony margins collected using a scalpel and immediately suspended in a drop (~ 50 μl) of sterile distilled water. Twenty randomly selected hyphae of each isolate-medium combination were measured using a 60X oil objective and the FluoView FV1000 4.0 software implemented in an Olympus IX81 microscope. Pictures were further analyzed and processed using the ImageJ software. Replicates were included for each combination in two separate blocks.
To assess heterogeneity among species in each of the studied traits, we used linear models where the measurements were the response variable and the species was the only fixed effect. We assessed the normality of the residuals of each linear model using a Shapiro-Wilk test (function ‘shapiro.test’, package ‘stats’; R Core team, 2013). Based on the Shapiro-Wilk test, residuals were not normally distributed in any of the linear models (Table S3, P< 0.05). Thus, we used the non-parametric Kruskal-Wallis test. To identify which group of isolates differed from each other, we performed multiple comparisons using non-parametric Nemenyi post hoc tests for Kruskal-Wallis. We performed all Kruskal-Wallis test using the R package ‘stats’ (R Core team, 2013) and the Nemenyi and Tukey post hoc tests using the Pairwise Multiple Comparison of Mean Ranks Package (‘pmcmr’) implemented in R (Pohlert 2014).
Discriminant analysis of morphological and physiological traits
Next, it was established whether the morphology of P. betacei and other Phytophthora species differed by visualizing all the morphological traits in a bidimensional plane using a discriminant function analysis (DA) based on the linear combination of morphological variables. To this end, a matrix with six traits (mycelial growth, hyphal width, sporangia length, sporangia width, sporangia area, and the sporangia length:width ratio) and a total of 15 individuals (five isolates for P. betacei, seven isolates for P. infestans, one isolate for P. andina clonal lineage EC-3, and two isolates for P. andina clonal lineage EC-2) was generated. Analyses were conducted using the “lda” function from the package ‘mass’ in R (Venables & Ripley 2002).
Molecular diagnosis of P. betacei based on SNP data
To distinguish P. betacei, P. andina (EC-2) and P. infestans, a set of 22,788 SNPs obtained from GBS data were analyzed for a total of 55 Phytophthora samples (12 P. infestans, 35 P. betacei, and 8 P. andina (EC-2); Table 1). Potentially diagnostic SNPs were selected calculating the allele frequencies and allele counts of each SNP for the entire dataset (55 samples and 22,788 SNPs). Major and minor alleles were obtained in each position. Samples belonging to each species were separated into three different files (P. betacei, P. andina (EC-2) and P. infestans) and allele counts were calculated for each dataset. SNPs with changes in the major allele in P. betacei were selected as candidates of differentiations relative to P. infestans and P. andina (EC-2) samples.
Results
Disease occurrence and P. betacei symptoms in the field
In 2008 and 2009, we identified a disease akin to late blight on tree tomato crops in southern Colombia. Field observations indicated that this disease can lead to the complete loss of the crop five to 10 days after the first symptoms are detected. In field, the pathogen is able to completely defoliate the tree in approximately one week. The symptoms of P. betacei on tree tomato differed from those generated by P. infestans on potato in forming concentric blighted areas that produced sporangia, and covered large areas of the leaves and petioles (Figure S2). In the field, no symptoms were observed on fruits, and the disease was rarely found on stems (Figure S2).
Phylogenetic reconstruction and molecular population genetics
Mitochondrial haplotyping and RFLP analysis using probe RG57
All P. betacei isolates belonged to the Ia mitochondrial haplotype, and were assigned to the EC-3 clonal lineage based on the RG57 probe fingerprint pattern (Table 1).
Phylogenetic relationships of the Phytophthora 1c clade species using nuclear and mitochondrial genomes
A phylogenetic reconstruction using 23,480 nuclear SNPs showed P. betacei, P. andina, and P. infestans as more closely related to each another than to P. ipomoeae and P. mirabilis (Figure 1). Consistently, the former three species formed a monophyletic group. All P. infestans clonal lineages (EC-1, US-8, and US-12) formed a monophyletic group. Phytophthora betacei appeared as the sister group of the P. andina strains collected from wild Solanaceae. This P. andina group comprised isolates of the EC-2 clonal lineage with mitochondrial haplotypes Ia and Ic and some isolates of unknown clonal lineage. The two clades, P. betacei and P. andina (EC-2clonal lineage), were reciprocally monophyletic, providing evidence for the divergence of the two species. The only isolate of P. andina of the EC-3 clonal lineage that was included in our analysis grouped together with P. betacei (Figure 1). Interestingly, this strain was also isolated from S. betaceum.
Phylogenetic analysis of the mitochondrial genome sequences did not differentiate among the three species of the P. infestans sensu lato complex, with one notable exception: P. andina clonal lineage EC-2 with the Ic mtDNA type appeared as the sister species of P. mirabilis (Figure 2).
Population structure analyses
Microsatellite and whole-genome SNP data clearly differentiated among the populations of P. infestans, P. betacei, and P. andina (Figure 3). The results obtained from the PCA, using microsatellite data, suggested three genetic groups (Figure 3A). PC1 separated P. infestans and P. betacei (mean variance explained = 12.40%), indicating that a large proportion of the genetic variation is explained by the genetic differentiation between isolates belonging to these two species. PC2 (mean variance explained = 4.44%) showed intraspecific variation within P. infestans, which is larger than the variation within P. betacei or P. andina. Notably, the two P. andina strains of the EC-3 clonal lineage grouped closely with P. betacei. For this analysis, PCs 3, 4, and 5 primarily represent intraspecific variation within P. infestans (Figure S4). Similar results across all of the diploid and triploid resampled datasets suggested that regardless of the ploidy of the three species, the pattern of genetic differentiation is consistent (Figures S3 and S4, Table S6). A PCA on the GBS data including 23,480 SNPs supported the clustering of the SSR analysis (Figure 3B). The PCA shows strong genetic differentiation among strains of P. infestans, P. betacei, and P. andina (Figure 3B). PC1 accounted for 37.5% of the total variation and separated P. infestans and the P. andina/P. betacei clades. PC2 identified 5.8% of the variation between the strains and separated P. betacei and P. andina (Figure 3B).
We conducted two Bayesian assignment tests in STRUCTURE. For the GBS data, the most likely clustering was three populations (ΔK = 3) (Figure S5). The first genetic clusters matched with P. infestans sensu stricto and the second matched with P. betacei. The third cluster was assigned to P. andina samples (lineage EC-2) supporting the genetic differentiation between P. infestans, P. andina and P. betacei samples (Figure S5). For the SSR data, Phytophthora andina was most similar to P. betacei but contained genetic material from P. infestans (Figures S6-S8). Again, the two strains of P. andina of the EC-3 clonal lineage showed more genetic similarity to P. betacei. STRUCTURE for the SSR data revealed that the genetic variance in the sample was best explained by two genetic clusters’ populations (K = 2; Figure S6 and Tables S4 and S5). Population assignments for higher values of K are shown in Figures S6 and S7. The population assignments were robust to uncertainty in allele frequencies at bi-allelic triploid sites in P. infestans (K = 2 and equivalent individual assignments across diploid and triploid subsample datasets; Tables S4 and S5, and Figure S6 and S7). STRUCTURE analyses supports genetic differentiation between the three species.
Host pathogenicity assays
The strongest line of evidence for a scenario of ecological speciation of plant pathogens comes from their host preference. Since P. infestans, P. betacei, and P. andina (EC-2) strains were isolated from different hosts, we tested the hypothesis that the three species were host specialized or had reduced fitness on their alternate host. We included isolates of P. andina of the EC-2 and EC-3 clonal lineages to make all possible pairwise comparisons. Isolates of P. andina of the EC-2 clonal lineage did not produce any symptoms on any of the hosts tested. Phytophthora infestans had higher fitness on tomato and yellow potato compared to P. betacei and P. andina (Table 2). Phytophthora betacei could not infect either tomato or potato but showed the highest fitness on tree tomato (Table 2). The P. andina (EC-3) strains assayed here were able to infect all hosts but showed lower fitness than P. infestans on three hosts (tomato, yellow potato, and tree tomato). They also displayed lower fitness on tree tomato compared to P. betacei. All pairwise comparisons indicated that strains of P. infestans and P. betacei displayed different fitness properties on every host assessed (Figure 4).
Discriminant analysis of morphological and physiological traits
Phytophthora betacei and P. infestans were highly differentiated when all morphological and physiological traits were analyzed jointly. Two variables (Table 3) explained 99% of the variance in a discriminant analysis (Figure 5). The first function (LD1) characterized the groups based mostly on the length:width ratio of sporangia (Table 3). For the second function (LD2), both hyphal width and length:width ratio helped discriminate among these groups of isolates (Table 3). Figure 5 shows the plot of the first and the second discriminant components for P. betacei, P. infestans, and P. andina (EC-2 and EC-3 clonal lineages). Physiological and morphological characterization of P. betacei is described in supplementary file 1.
TAXONOMIC DESCRIPTION OF Phytophthora betacei
The taxonomic description of P. betacei was deposited in the MycoBank database (http://www.mycobank.org/) following standard taxonomy procedures.
Phytophthora betacei sp. nov. M.F. Mideros, L.E. Lagos, et S. Restrepo, sp. nov. (Figure S9 - S11)
Mycobank Number No. MB 815748
Type material:— Holotype
Isolate of P. betacei from COLOMBIA, Putumayo, Colon, San Pedro locality, on infected leaves from Solanum betaceum (Solanaceae, Solanales), 1°13’26.9”N - 76°56’73.9”W, 24 Oct 2008, MF Mideros, (Andes-F 1172, holotype). Ex-type: LAMFU-COL-P8084
Description
Phytophthora betacei sp. nov. is an oomycete, plant pathogenic species that produced typical Phytophthora colonies that are white, smooth on V8 juice agar (V8). P. betacei grows well on V8 at 18°C. Aerial mycelia were generally abundant. Mycelial radial growth averaged 37.8 cm2 (SD = 9.6 cm2) after 15 days of incubation at 18°C. Sporangia of all isolates tested were borne terminally on the sporangiophore, and were caducous, ovoid, and semi-papillate with an average length of 36.3 μm (SD = 6.0 μm) and an average width of 17.3 μm (SD = 2.9 μm). The average length:width ratio was 2.1 (SD = 0.4). The area of sporangia was on average 436.3 μm2 (SD = 119.4 μm2). Growth of P. betacei isolates on potato dextrose agar (PDA) was limited. Mycelial radial growth was on average 0.27 cm2 (SD = 0.51 cm2) after 15 days of incubation at 18°C. No sporulation was detected on PDA. On corn meal agar (CMA), P. betacei isolates produced typical Phytophthora colonies that were white and smooth. All isolates were able to grow on CMA at 18°C. Aerial mycelium was generally abundant. Mycelial radial growth averaged 22.9 cm2 (SD ± 12.7 cm2) after 15 days of incubation at 18°C. Sporangia of all isolates were borne terminally on the sporangiophore, and were caducous, ovoid, and semi-papillate with an average length of 37.6 μm (SD = 5.4 μm) and an average width of 16.9 μm (SD = 2.3 μm). The average length:width ratio was 2.2 (SD = 0.2). The area of sporangia was on average 386.1 μm2 (SD = 33.1 μm2) on CMA. On tree tomato agar (TTA), P. betacei produced typical Phytophthora white smooth colonies. Mycelial radial growth and sporulation of P. betacei isolates was more abundant on TTA medium than on any of the other three media tested. Mycelial radial growth was on average 33.0 cm2 (SD = 6.2 cm2) after 15 days of incubation at 18°C. Sporangia of all isolates were borne terminally to the sporangiophore, and were caducous, ovoid, and semi-papillate with an average length of 39.3 μm (SD = 4.8 μm) and an average width of 15.8 μm (SD = 5.7 μm). The average length:width ratio was 2.6 (SD = 0.2). The area of sporangia was on average 311.5 μm2 (SD = 39.5 μm2). Hyphal swellings and chlamydospores were absent. Isolates were heterothallic with low oospore production (15.4 ± 10.9 oospores per mm2) and of abnormal appearance when crossed with a P. infestans strain of the A2 mating type (US040009). Oogonia were not ornamented. No self-fertile isolates were observed.
Material examined
Listed in Supplementary file 3.
Distribution
Isolates collected in southern Colombia, in the departments of Putumayo and Nariño.
Etymology
“betacei” refers to S. betaceum, the host plant from which the isolates were obtained.
Molecular diagnosis of P. betacei based on SNP data
From the GBS data, we identified a total of 22,788 SNPs from which 150 were able to discriminate P. betacei from P. andina (EC-2) and P. infestans samples. Although all 150 SNPs were classified as potentially diagnostic SNPs, this set of markers should then be validated in a larger Phytophthora betacei collection to validate their robustness to diagnose this species. All SNPs are listed in Supplementary file 3.
Discussion
Here we describe the new species P. betacei which is closely related to P. infestans and P. andina but it is ecologically distinct. The divergence is recent but the levels of host specialization are very high suggesting ecological speciation in allopatry. The cross-pathogenicity tests showed strong host specificity when isolates of P. betacei were inoculated on S. tuberosum or S. lycopersicum, the main hosts of P. infestans. Phytophthora infestans is able to infect tree tomato, but its fitness (measured as the number of sporangia produced) on this host is relatively low compared with that on its more commonly described hosts (S. tuberosum and S. lycopersicum). Thus, host specificity might be playing an important role in maintaining gene flow between P. infestans and P. betacei restricted. Our findings provide novel insights into the evolutionary history of the Irish famine pathogen P. infestans and its close relatives. Furthermore, we also refine the species boundaries within the complex of P. andina, originally described as a polyphyletic taxon.
Phytophthora betacei as a new species
We describe the new taxon, P. betacei, based on physiological, morphological, population genetic, and phylogenetic analyses, as well as differences in host specificity. All these analyses strongly support the designation of the new species P. betacei within the Phytophthora 1c clade.
The first line of evidence for the distinction between P. betacei and the other species of the Phytophthora 1c clade is the high genetic differentiation among the genetic groups. Nuclear phylogenies indicate that the triad P. infestans, P. betacei, and P. andina form a monophyletic clade whose closest known relatives are other members of the Phytophthora 1c clade (i.e., P. ipomoeae, P. phaseoli, and P. mirabilis). The three species, P. betacei, P. infestans, and P. andina (from the clonal lineage EC-2; see below) are clearly separated and P. betacei and P. andina (as defined here) are reciprocally monophyletic, suggesting that they lack recent gene flow and can be considered different species. Interestingly, the mitochondrial markers do not separate the three species. Our results are consistent with a scenario of speciation with secondary contact, and mitochondrial introgression, a phenomenon common across the tree of life (Funk & Omland 2003).
A second line of evidence for the existence of P. betacei as a separate species from P. infestans sensu stricto involves differences in allele frequencies in each of these genetic groups. All analyses using both SSR loci and SNP markers, suggest the existence of two discrete genetic clusters that correspond to P. infestans sensu stricto and P. betacei. Phytophthora andina has a less clear origin and this is discussed below. Our evidence suggests that P. infestans and P. betacei are isolated genetic groups with little detectable nuclear gene flow between them (Figure 3).
In addition to genetic variation, we determined whether P. betacei shows distinct morphological differences with P. andina and P. infestans. Four morphological (hyphal width and sporangial length, width, and length:width ratio) and one physiological (mycelial radial growth on four different media) traits were measured. The discriminant analysis clearly separated P. betacei from P. infestans in all media tested. The most striking morphological differences between P. betacei and P. infestans are the length:width ratio of sporangia, the hyphal width (μm), and the mycelial radial growth (cm2). Differences between P. andina and P. infestans or P. betacei are not as clear as the statistical differences between P. andina vs. P. infestans and P. andina vs. P. betacei and are dependent on the medium tested. Combining all the morphological variables, we show that strains of P. betacei collected in Colombia comprise a well-differentiated group of strains (Figure 5).
Our final and strongest line of evidence comes from infection assays on the native host range of the three species and from observations in nature. The host pathogenicity assays indicate that P. betacei is a tree tomato specialist unable to colonize potato and tomato (Figure 4). Conversely, P. infestans has low fitness on tree tomato, the only known host of P. betacei. These reciprocal differences in host pathogenicity represent a strong reproductive isolating mechanism between the two species. Host specificity is considered one of the most important isolating mechanisms between species of plant pathogens (reviewed in Harrington & Rizzo 1999; Coyne & Orr 2004). In asexual populations, host specialization could be associated with strong niche partition, which is common in species with asexual reproduction and strong local adaptation to the host (Poulin 2005; Halkett et al. 2006). Plant pathogens are commonly restricted to their hosts; thus, host specialization can result in a strong premating barrier (Stukenbrock 2013; Vialle et al. 2013).
Notably, we find that all isolates of P. betacei belonging to the EC-3 clonal lineage are closely related to the isolates previously described as P. andina EC-3 (Oliva et al. 2010; Goss et al. 2011, 2014; Lassiter et al. 2015; Martin et al. 2015). All EC-3 isolates form a monophyletic group differentiated from P. infestans and P. andina EC-2. We propose that all these EC-3 isolates should be considered P. betacei and all EC-2 isolates as P.andina. In this way, all species are rigorously defined as true monophyletic species.
Phytophthora andina as a polyphyletic group
In the literature, P. andina has been reported to be polyphyletic and include the following three clonal lineages: P. andina EC-2 mitochondrial haplotype Ia, P. andina EC-2 mitochondrial haplotype Ic, and P. andina EC-3 (Adler et al. 2004; Gómez-Alpizar et al. 2007a). This species has been controversial since its erection since species are expected to be monophyletic with the expectation of descent from one common ancestor. Phytophthora andina was proposed to be a hybrid based on cloning nuclear haplotypes from several loci showing that one ancestor is P. infestans while the other ancestor remains to be described (Goss et al. 2011). Later, it has been hypothesized to have arisen from hybridization based on the conflicting phylogenetic information of mitochondrial and nuclear genealogies (Martin et al. 2015). Based on the hypothesis that P. andina was hybrid and the polyphyletic mitochondrial phylogenies in P. andina, we previously argued that P. andina was not appropriately described as a new species (Cárdenas et al. 2012). The identification of P. betacei as a new species, sheds some light on the origin of P. andina.
Our results support a monophyletic grouping of the EC-2 P. andina clonal lineages of mitochondrial haplotypes Ia and Ic that are closely related and form a monophyletic group distinct from P. betacei and P. infestans. Supported by phylogenetic and population genetic analyses, we suggest P. andina EC-3 should now be considered P. betacei. Our results indicate that the initial definition of P. andina included isolates that were either P. betacei or were closely related to P. betacei, namely, the EC-3 clonal lineage. At this point P. betacei cannot be considered a lineage of P. andina because it would reinforce the polyphyletic nature of P. andina. Again, a species cannot be described and proposed as polyphyletic. Our results showed the genetic and ecological separation of these two species, P. betacei and P. andina and our scenario of three species in the northern part of South America propose the most rigorous description of species.
Generally, our results confirm previous observations that P. andina, as currently described, is a polyphyletic group that requires redefinition (Gomez-Alpizar et al. 2008; Cárdenas et al. 2012; Forbes et al. 2012). Redefining P. andina including only strains of clonal lineage EC-2 makes this group monophyletic and provides a biologically rigorous species definition. We propose using P. andina sensu lato as the proper description of P. andina EC-2.
Whether there is reciprocal host specificity between P. betacei and P. andina (EC-2) was shown here. We have demonstrated that strains of P. andina of the EC-2 clonal lineage cannot infect S. betaceum, the only known host of P. betacei. Interestingly, isolates of the EC-3 clonal lineage have always been collected from S. betaceum plants, suggesting a strong isolating mechanism between P. betacei (clonal lineage EC-3) and P. andina (EC-2) in nature. Also, several authors have documented host specificity between P. andina EC-2 and EC-3 clonal lineages in nature (Adler et al. 2004; Gómez-Alpizar et al. 2007b; Oliva et al. 2010). All known EC-2 P. andina isolates have been collected in Anarrichomenum and other wild species, thus we hypothesize that the species might be specialized on these plants.
It is important to mention that a group of strains referred to as P. andina (new lineage PE-8) has recently been reported as infecting S. betaceum in Peru (Forbes et al. 2016). Further genetic, phylogenetic and population analyses and a greater number of isolates are needed to determine the identity, host range and fitness of isolates belonging to the PE-8 clonal lineage.
Conclusions
We have provided several lines of evidence supporting the claim that P. betacei is a distinct, previously undescribed species within the Phytophthora 1c clade. Our findings and the scenario of the three species also resolve the polyphyletic nature of P. andina. The new species is ecologically distinct from its closely related species, P. andina and P. infestans showing high levels of host specialization suggesting ecological speciation in allopatry. The strong host specialization of P. infestans and P. betacei may act as premating barriers that restrict gene flow between these two species in nature. It remains unclear if host specialization facilitated or initiated the speciation process in the P. infestans sensu lato complex. However, in this report, we have demonstrated that ecological differences are important in the persistence of P. infestans and P. betacei as genetically isolated units across an overlapping area in the northern Andes. More studies are needed to further characterize the evolution of the closely related species and to understand the process of divergence in this group. In general, our results have implications for the understanding of how new plant pathogen species originate and persist. Our findings also highlight the importance of sampling plant pathogens of semi-domesticated or undomesticated hosts.
Acknowledgements
This work was supported by the Department of Biological Sciences at Universidad de los Andes. Additional funding for this research was provided by the Research Fund of the School of Sciences and the Office of the Vice President for Research from Universidad de los Andes.