The global genetic structure of the wheat pathogen Mycosphaerella graminicola is characterized by high nuclear diversity, low mitochondrial diversity, regular recombination, and gene flow

A total of 1673 Mycosphaerella graminicola strains were assayed for DNA fingerprints and restriction fragment length polymorphism (RFLP) markers in the nuclear and mitochondrial genomes. The isolates were collected from 17 wheat fields located in 11 countries on five continents over a six year period (1989-1995). Our results indicate that genetic diversity in the nuclear genome of this fungus was high for all but three of the field populations surveyed and that populations sampled from different continents had similar frequencies for the most common RFLP alleles. Hierarchical analysis revealed that more than 90% of global gene diversity was distributed within a wheat field, while approximately 5% of gene diversity was distributed among fields within regions and approximately 3% was distributed among regions on different continents. These findings suggest that gene flow has occurred on a global scale. On average, each leaf was colonized by a different nuclear genotype. In contrast, only seven mtDNA haplotypes were detected among the 1673 isolates and the two most common mtDNA haplotypes represented approximately 93% of the world population, consistent with a selective sweep. Analysis of multilocus associations indicated that all field populations were in gametic equilibrium, suggesting that sexual recombination is a regular occurrence globally.

Population Structure of Mycosphaerella graminicola: From Lesions to Continents

ABSTRACT The genetic structure of field populations of Mycosphaerella graminicola was determined across a hierarchy of spatial scales using restriction fragment length polymorphism markers. The hierarchical gene diversity analysis included 1,098 isolates from seven field populations. Spatial scales ranged from millimeters to thousands of kilometers, including comparisons within and among lesions, within and among fields, and within and among regions and continents. At the smallest spatial scale, microtransect sampling was used to determine the spatial distribution of 15 genotypes found among 158 isolates sampled from five individual lesions. Each lesion had two to six different genotypes including both mating types in four of the five lesions, but in most cases a lesion was composed of one or two genotypes that occupied the majority of the lesion, with other rare genotypes interspersed among the common genotypes. The majority (77%) of gene diversity was distributed within plots ranging from approximately 1 to 9 m(2) in size. Genotype diversity (G / N) within fields for the Swiss, Texas, and Israeli fields was high, ranging from 79 to 100% of maximum possible values. Low population differentiation was indicated by the low G(ST) values among populations, suggesting a corresponding high degree of gene flow among these populations. At the largest spatial scale, populations from Switzerland, Israel, Oregon, and Texas were compared. Population differentiation among these populations was low (G(ST) = 0.05), and genetic identity between populations was high. A low but significant correlation between genetic and geographic distance among populations was found (r = -0.47, P = 0.012), suggesting that these populations probably have not reached an equilibrium between gene flow and genetic drift. Gene flow on a regional level can be reduced by implementing strategies, such as improved stubble management that minimize the production of ascospores. The possibility of high levels of gene flow on a regional level indicates a significant potential risk for the regional spread of mutant alleles that enable fungicide resistance or the breakdown of resistance genes.

Variation for neutral markers is correlated with variation for quantitative traits in the plant pathogenic fungus Mycosphaerella graminicola

We compared genetic variation and population differentiation at RFLP marker loci with seven quantitative characters including fungicide resistance, temperature sensitivity, pycnidial size, pycnidial density, colony size, percentage of leaves covered by pycnidia (PLACP) and percentage of leaves covered by lesions (PLACL) in Mycosphaerella graminicola populations sampled from four regions. Wide variation in population differentiation was found across the quantitative traits assayed. Fungicide resistance, temperature sensitivity, and PLACP displayed a significantly higher Q(ST) than G(ST), consistent with selection for local adaptation, while pycnidial size, pycnidial density and colony size displayed a lower or significantly lower Q(ST) than G(ST), consistent with constraining selection. There was not a statistical difference between Q(ST) and G(ST) in PLACL. We also found a positive and significant correlation between genetic variation in molecular marker loci and quantitative traits at the multitrait scale, suggesting that estimates of overall genetic variation for quantitative traits in M. graminicola could be derived from analysis of the molecular genetic markers.

Determination of diclofenac sodium by capillary zone electrophoresis with electrochemical detection

Capillary zone electrophoresis was employed for the determination of diclofenac sodium using an end-column amperometric detection with a carbon fiber microelectrode, at a constant potential of 0.83 V vs. saturated calomel electrode. The optimum conditions of separation and detection are 4.90 x 10(-3) mol/l Na2HPO4-3.10 x 10(-3) mol/l NaH2PO4 (pH 7.0) for the buffer solution, 10 kV for the separation voltage, 5 kV and 10 s for the injection voltage and the injection time, respectively. The limit of detection is 2.5 x 10(-6) mol/l or 5.2 fmol (S/N=2). The relative standard deviation is 0.8% for the migration time and 4.7% for the electrophoretic peak current. The method was applied to the determination of diclofenac sodium in human urine.

A comparison of the effect of total intravenous anaesthesia with propofol and remifentanil and inhalational anaesthesia with isoflurane on the release of pro- and anti-inflammatory cytokines in patients undergoing open cholecystectomy

The aim of the study was to investigate the effects of two anaesthetic techniques (total intravenous technique vs. inhalational technique) on changes in pro- and anti-inflammatory cytokine levels during open cholecystectomy. Forty ASA PS I-II patients undergoing open cholecystectomy were randomly assigned to two groups. Group R received total intravenous anaesthesia with propofol and remifentanil and group F received balanced inhalational anaesthesia with isoflurane. The plasma levels of tumour necrosis factor-alpha (TNF-alpha), interleukin IL-6 and interleukin IL-10 were measured during and after surgery. The pro-inflammatory cytokine levels (TNF-alpha and IL-6) and the anti-inflammatory cytokine (IL-10) showed a significant increase in their concentrations compared with pre-induction levels in both groups (P < 0.05). By the end of anaesthesia and surgery, TNF-alpha and IL-6 were significantly lower in group R than in group F (P < 0.05). At the end of anaesthesia and 12 hours postoperatively, IL-10 levels in group R were higher than in group F (P < 0.05). These findings suggest that total intravenous anaesthesia using propofol and remifentanil suppresses the inflammatory response caused by surgery to a greater extent than a balanced inhalation technique using isoflurane.

Distribution of mating type alleles in the wheat pathogen Mycosphaerella graminicola over spatial scales from lesions to continents

A total of 2035 Mycosphaerella graminicola strains collected from 16 geographic locations on four continents were assayed for the mating type locus. RFLP fingerprints were used to identify clones in each population. At the smallest spatial scale analyzed, both mating types were found among fungal strains sampled from different lesions of the same leaf as well as from different pycnidia in the same lesion. At larger spatial scales, the two mating types were found at equal frequencies across spatial scales ranging from several square meters to several thousand square kilometers. Though the absolute frequencies of the two mating types sometimes varied for different sampling units within the same spatial scale in the hierarchy (plots within a field, fields within a country, or different continents of the world), none of the differences were statistically significant from the null hypothesis of equal frequencies for the two mating types. The evolutionary forces likely to maintain the even distribution of the two mating types in this pathogen were discussed.

Measuring Immigration and Sexual Reproduction in Field Populations of Mycosphaerella graminicola

ABSTRACT A field experiment was conducted to determine the relative contributions of immigration and sexual reproduction to the genetic structure of Mycosphaerella graminicola populations during the course of an epidemic. The genetic structure of M. graminicola populations sampled from wheat plots inoculated artificially with 10 isolates was compared with control plots infected naturally by airborne ascospores. Restriction fragment length polymorphisms (RFLPs) were used to test the randomness of associations among loci, and DNA fingerprints were used to identify clones. All isolates in the control plots had unique genotypes and RFLP loci were at gametic equilibrium, findings consistent with random mating. The proportion of isolates in the inoculated plots with DNA fingerprints that differed from the 10 inoculated isolates increased from 3% in the early to 39 and 34% in the mid- and late season, respectively. The degree of gametic disequilibrium was higher in the mid-season than in the late-season population. By the end of the growing season, we estimate that 66% of the isolates in the inoculated plots were asexual progeny of the 10 inoculated isolates, 10% were immigrants, and 24% were sexual recombinants. The proportion of infections caused by ascospores increased over the growing season.

The interaction among evolutionary forces in the pathogenic fungus Mycosphaerella graminicola

The population genetic dynamic of a species is driven by interactions among mutation, migration, drift, mating system, and selection, but it is rare to have sufficient empirical data to estimate values for all of these forces and to allow comparison of the relative magnitudes of these evolutionary forces. We combined data from a mark-release-recapture experiment, extensive population surveys, and computer simulations to evaluate interactions among these evolutionary forces in the pathogenic fungus Mycosphaerella graminicola. The results from these studies showed that, on average, the immigration rate was 0.027, the fraction of outcrossing individuals was 0.035, and the selection coefficient associated with immigrants was 0.106 each generation. We also estimated that effective population sizes for this fungus were larger than 24,000 and the mutation rate for the RFLP markers used in surveys and field experiments was approximately 4 x 10(-5). Computer simulations based on these estimates indicate that, on average, the global population of M. graminicola has reached equilibrium. Population genetic parameters including number of alleles, gene diversity, and population subdivision estimated from the computer simulations were surprisingly close to empirical estimates. Simulations also revealed that random drift is the major evolutionary force decreasing genetic variation in this fungus, followed by natural selection. The major force adding to genetic variation was mutation, followed by gene flow and sexual recombination. Gene flow played the leading role in decreasing population subdivision while natural selection was the major factor increasing population subdivision.

Genetic Structure of Rhynchosporium secalis in Australia

ABSTRACT Restriction fragment length polymorphism (RFLP) markers were used to determine the genetic structure of Australian field populations of the barley scald pathogen Rhynchosporium secalis. Fungal isolates were collected by hierarchical sampling from five naturally infected barley fields in different geographic locations during a single growing season. Genetic variation was high in Australian R. secalis populations. Among the 265 fungal isolates analyzed, 214 distinct genotypes were identified. Average genotype diversity within a field population was 65% of its theoretical maximum. Nei's average gene diversity across seven RFLP loci was 0.54. The majority (76%) of gene diversity was distributed within sampling site areas measuring 1 m(2); 19% of gene diversity was distributed among sampling sites within fields; and 5% of gene diversity was distributed among fields. Fungal populations from different locations differed significantly both in allele frequencies and genotype diversities. The degree of genetic differentiation was significantly correlated with geographic distance between populations. Our results suggest that the R. secalis population in Western Australia has a different genetic structure than populations in Victoria and South Australia.

The Genetic Structure of Field Populations of Rhynchosporium secalis from Three Continents Suggests Moderate Gene Flow and Regular Recombination

ABSTRACT Restriction fragment length polymorphism (RFLP) markers were used to compare the genetic structure of field populations of Rhynchosporium secalis from barley. A total of 543 isolates representing 8 field populations were sampled from Australia, California, Finland, and Norway. Gene and genotype diversity were high in all populations. Nei's average gene diversity across seven RFLP loci was 0.513. Hierarchical gene diversity analysis showed that 9% of the total genetic variability was distributed among continents, 4% was distributed among fields within continents, and 13% was distributed among collection stations within a field. The majority (74%) of genetic variability was distributed within collection areas of approximately 1 m(2) within fields. Gene flow appears to be significant on a regional scale but more restricted among continents. Allele frequencies were significantly different at several RFLP loci. Genetic distances were small among populations within regions and large between regions. Pairwise comparisons of genotype diversity in the populations revealed significant differences among populations that were related mainly to differences in sampling strategies. Isolates from Norway and Finland showed a lower copy hybridization pattern with probe pRS26. This probe functioned as a fingerprint probe for the California and Australian isolates. Seven out of the eight populations studied were at gametic equilibrium for RFLP loci, suggesting that R. secalis populations in Norway, Finland, and Australia undergo regular recombination, although a teleomorph has not yet been recognized.

Sexual reproduction facilitates the adaptation of parasites to antagonistic host environments: Evidence from empirical study in the wheat-Mycosphaerella graminicola system

Most eukaryotes use sexual reproduction to transmit genetic information from generation to generation despite the advantages offered by asexual reproduction. One theory to explain the origin and maintenance of sexual reproduction hypothesises that sexual recombination generates genetic variation that allows faster adaptation to fluctuating and/or stressful environments. We used a combination of ecological, molecular genetic, statistical and experimental evolution approaches to test this hypothesis in an agricultural plant-pathogen system. We inoculated wheat hosts with 10 strains of the fungal pathogen Mycosphaerella graminicola in a field experiment and estimated the contributions of sexual reproduction, asexual reproduction and immigration to the genetic composition of fungal populations sampled from moderately resistant and susceptible hosts through the course of an epidemic cycle. We found that a significant proportion of the M. graminicola population in the late phase of the epidemic originated from sexual reproduction among isolates that had been introduced into the field plots at the beginning of the epidemic. Recombinants were recovered at a higher frequency on the moderately resistant plant host Madsen than on the susceptible host Stephens. By the end of the growing season, we estimated that approximately 13% of the strains sampled from the resistant host were recombinants, compared with 9% in the samples collected from the susceptible host. We also found that pathogen strains originating from the resistant cultivar displayed higher levels of fitness, virulence and fungicide tolerance than those originating from the susceptible cultivar. Our results provide empirical support for the hypothesis that sexual reproduction facilitates the evolution of parasites to overcome host resistance.

Using Restriction Fragment Length Polymorphisms to Assess Temporal Variation and Estimate the Number of Ascospores that Initiate Epidemics in Field Populations of Mycosphaerella graminicola

ABSTRACT Restriction fragment length polymorphisms (RFLPs) and DNA fingerprints were used to assess temporal variation and estimate the effective population size of the wheat pathogen Mycosphaerella graminicola over a 6-year period. In each year, the fungal population was founded by ascospores originating from outside the sampled fields. A total of 605 fungal isolates were included in this study. Our results indicate that the genetic structure of these M. graminicola populations were stable over the 6-year period. The common alleles at each RFLP locus were present at similar frequencies each year. More than 99% of gene diversity was distributed within populations sampled from the same year and less than 1% was attributed to differences among years. The lack of population differentiation among collections taken in different years indicated that the effective size of the source population was sufficiently large that genetic drift was insignificant in this location. It also suggests that the initial colonists from ascospore founder populations were a fair reflection of the source population. We estimate that the effective sizes of these field populations ranged from 3,400 to 700,000 individuals, depending on the size of the field sampled and assumptions about mutation rates. Estimates of the number of ascospores initiating epidemics of leaf blotch disease in each field plot and factors that contribute to the large effective population size of M. graminicola are discussed.

Nitrite reduces acute lung injury and improves survival in a rat lung transplantation model

Ischemia/reperfusion injury (IRI) is the most common cause of early mortality following lung transplantation (LTx). We hypothesized that nitrite, an endogenous source of nitric oxide (NO), may protect lung grafts from IRI. Rat lung grafts were stored in preservation solution at 4°C for 6 hours. Both grafts and recipients were treated with nitrite. Nitrite treatment was associated with significantly higher levels of tissue oxygenation, lower levels of cytokines and neutrophil/macrophage infiltration, lower myeloperoxidase activity, reduced oxidative injury and increased cGMP levels in grafts than in the controls. Treatment with either a nitric oxide scavenger or a soluble guanylyl cyclase (sGC) inhibitor diminished the beneficial effects of nitrite and decreased cGMP concentrations. These results suggest that nitric oxide, generated from nitrite, is the molecule responsible for the effects of nitrite via the nitric oxide/sGC/cGMP pathway. Allopurinol, a xanthine oxidoreductase (XOR) inhibitor, abrogated the protective effects of nitrite, suggesting that XOR is a key enzyme in the conversion of nitrite to nitric oxide. In vitro experiments demonstrated that nitrite prevented apoptosis in pulmonary endothelial cells. Nitrite also exhibits longer survival rate in recipients than control. In conclusion, nitrite inhibits lung IRI following cold preservation and had higher survival rate in LTx model.

Selection for increased cyproconazole tolerance in Mycosphaerella graminicola through local adaptation and in response to host resistance

SUMMARY Sterol demethylation inhibitors (DMIs) represent one of the largest groups of systemic fungicides that have been used to control agriculturally important fungal pathogens. Knowledge regarding the evolution of fungicide resistance in agricultural ecosystems is fragmentary and a better understanding of the processes driving the development of DMI resistance in populations of fungal pathogens is needed by plant pathologists and the agrochemical industry. We considered some of these processes using approaches based on molecular population and quantitative genetics. Five Mycosphaerella graminicola populations sampled from unsprayed wheat fields on four continents were assayed for eight restriction fragment length polymorphism (RFLP) markers and their level of tolerance to cyproconazole. DMI fungicides such as cyproconazole inhibit the enzyme eburicol 14-alpha-demethylase. The gene encoding this target, CYP51, was sequenced for all isolates. We found unimodal, continuous variations in cyproconazole tolerance among the M. graminicola isolates sampled from individual fields, consistent with a polygenic mode of inheritance. We also found that population differentiation for cyproconazole tolerance (Q(ST)) among the five M. graminicola populations was significantly higher than the corresponding population differentiation for neutral RFLP markers (G(ST)), suggesting that selection for cyproconazole tolerance in the Swiss population has already led to local adaptation that can be seen even in an unsprayed population. The Swiss population displayed the highest level of tolerance to cyproconazole, in addition to a lower than expected quantitative variation in fungicide tolerance and a skewed distribution, indicating that selection had increased the overall tolerance of this population. Further analysis with DNA sequencing showed that the population from Switzerland was dominated by isolates with several point mutations and a 6-bp deletion in CYP51. This deletion and one of the point mutations were previously related to increased resistance in field isolates. The fungal population from Oregon sampled from an unsprayed resistant host cultivar displayed the same gene diversity in RFLP loci but higher cyproconazole tolerance and quantitative variation in tolerance than the fungal population from the same field sampled from an unsprayed susceptible host cultivar.

Machine learning-based radiomic evaluation of treatment response prediction in glioblastoma

AIM

To investigate machine learning based models combining clinical, radiomic, and molecular information to distinguish between early true progression (tPD) and pseudoprogression (psPD) in patients with glioblastoma.

MATERIALS AND METHODS

A retrospective analysis was undertaken of 76 patients (46 tPD, 30 psPD) with early enhancing disease following chemoradiotherapy for glioblastoma. Outcome was determined on follow-up until 6 months post-chemoradiotherapy. Models comprised clinical characteristics, O⁶-methylguanine-DNA methyltransferase (MGMT) promoter methylation status, and 307 quantitative imaging features extracted from enhancing disease and perilesional oedema masks on early post-chemoradiotherapy contrast-enhanced T1-weighted imaging, T2-weighted imaging (T2WI), and apparent diffusion coefficient (ADC) maps. Feature selection was performed within bootstrapped cross-validated recursive feature elimination with a random forest algorithm. Naive Bayes five-fold cross-validation was used to validate the final model.

RESULTS

Top selected features included age, MGMT promoter methylation status, two shape-based features from the enhancing disease mask, three radiomic features from the enhancing disease mask on ADC, and one radiomic feature from the perilesional oedema mask on T2WI. The final model had an area under the receiver operating characteristics curve (AUC) of 0.80, sensitivity 78.2%, specificity 66.7%, and accuracy of 73.7%.

CONCLUSION

Incorporating a machine learning-based approach using quantitative radiomic features from standard-of-care magnetic resonance imaging (MRI), in combination with clinical characteristics and MGMT promoter methylation status has a complementary effect and improves model performance for early prediction of glioblastoma treatment response.

Estimation of rates of recombination and migration in populations of plant pathogens-a reply

ABSTRACT We find that the maximum likelihood method proposed by J. K. M. Brown has deficiencies that limit its usefulness for actual data sets. We propose two alternative statistical methods based on maximum likelihood that could be used to quantify rates of recombination and immigration in fungal populations. We also show that minor modification of our original method, which was based upon posterior probabilities, leads to a result that is identical to one of the maximum likelihood methods. Our previous estimates of the relative contributions of sexual reproduction, asexual reproduction, and immigration to the genetic structure of a Mycosphaerella graminicola population did not change significantly following reanalysis of our data with these new methods.

HSV (type 1) infection of the trigeminal complex

Following corneal inoculation with herpes simplex virus (Type 1) (HSV), virus spreads to the CNS by axonal transport in the central branches of trigeminal ganglion cell neurons. Although this mode of viral entry to the CNS is rare for humans, it appears to be the principal route of entry into the CNS in animal models of herpetic corneal disease. In this study, the corneas of BALB/c mice were unilaterally inoculated with HSV, and the distribution of HSV-immunoreactive label was studied to identify the central branches of the axons of infected trigeminal ganglion cells. Virus was first noted in the brainstem trigeminal complex 4 days after corneal inoculation, when HSV-labeled afferents were found throughout the course of the descending tract of V as well as in interstitial neurons in the tract. By 5 days labeled neurons were also found not only in the n. caudalis and portions of the n. interpolaris of the trigeminal complex but also in laminae I-IV of the dorsal horn of the upper cervical levels of the spinal cord. No immunoreactivity was seen in other regions of the complex, including the n. oralis or the main sensory n. of V. By 6 days, however, the infection had spread to the main sensory division of V.

Modification of ricin A chain, by addition of endoplasmic reticulum (KDEL) or Golgi (YQRL) retention sequences, enhances its cytotoxicity and translocation

A pKK expression system in Escherichia coli was used to produce recombinant ricin A chain (rRTA) and rRTA modified by addition of organelle-specific amino acid retention sequences, including KDEL (an endoplasmic reticulum, ER, lumen retention signal), KKMP (an ER membrane retention signal), YQRL (a trans-Golgi network retention signal) and KFERQ (a lysosome-targeting signal) to the C terminus of rRTA. The toxicities of these RTA mutants were assessed in Jurkat cells following fluid-phase endocytosis. rRTA-KDEL and rRTA-YQRL were significantly more cytotoxic for Jurkat cells than rRTA, rRTA-KKMP or rRTA-KFERQ. This difference did not result from signal(KDEL or YQRL)-mediated binding of these RTA mutants to the cell surface. Reconstituted ER and Golgi vesicles have been employed to assess translocation of rRTA and mutant rRTA. RTA-KDEL and RTA-YQRL respectively exhibited 6.7-fold and 6.1-fold more protection against papain digestion in reconstituted ER vesicles and 2.2-fold and 1.8-fold more protection in reconstituted Golgi vesicles, than unmodified rRTA. These mutants were reassociated with ricin B chain to form holotoxins. The mutant RTA-KDEL and RTA-YQRL holotoxins were 3.8-fold and 1.5-fold more cytotoxic for target cells, respectively, than ricin produced using unmodified rRTA. Our results suggest that both ER and the trans-Golgi network may play important roles in the intracellular trafficking and translocation of ricin A chain.

Sensitive nonradioactive dot blot/ribonuclease protection assay for quantitative determination of mRNA

We have developed a simple and sensitive method for the rapid quantitation of mRNA from cell cultures and small tissue samples. The method combines the high sensitivity and specificity of the ribonuclease protection assay with simple handling and rapid execution of dot blotting. The use of digoxygenin-labeled cRNA probes eliminates all problems associated with radioisotopes commonly used in the ribonuclease protection assay. The RNA preparation is dotted directly onto nylon membranes, and after hybridization the filters are treated with ribonuclease A, which removes the nonhybridized single-stranded RNA. The mRNA-hybrid is then visualized by the chemiluminescence technique using labeled anti-digoxigenin antibody, and the signal intensity is quantitated. Comparison with the Northern blotting ribonuclease protection assay revealed that this dot blot technique is almost ten times more sensitive and that its signals are linear over a wide range of RNA concentrations (0.01-10 micrograms/microL/dot). This method seems particularly valuable for simultaneous processing of large numbers of samples containing a wide range of RNA concentrations.

Evidence for Natural Selection in the Mitochondrial Genome of Mycosphaerella graminicola

ABSTRACT Pathogenicity assays were combined with restriction fragment length polymorphism (RFLP) markers in the mitochondrial and nuclear genomes to compare Mycosphaerella graminicola populations adapted to bread wheat (Triticum aestivum) and durum wheat (T. turgidum) in the Mediterranean Basin. The majority of isolates had unique nuclear DNA fingerprints and multilocus haplotypes. Only six mitochondrial DNA (mtDNA) haplotypes were identified among 108 isolates assayed. There were minor differences in frequencies of alleles at nuclear RFLP loci between the two host-adapted populations, but differences in the frequencies of mtDNA haplotypes were highly significant (P < 0.0001). mtDNA haplotype 1 dominated on the isolates adapted to bread wheat, and its frequency was twice as high as for the isolates adapted to durum wheat. mtDNA haplotype 4, which contained a unique approximately 3-kb insertion, was detected only in isolates showing specificity toward durum wheat and was the dominant haplotype on this species. We propose that the low mitochondrial diversity in this pathogenic fungus is due to a selective sweep and that differences in the frequencies of mtDNA haplotypes between the two host-adapted populations were due to natural selection according to host species.