Dose‐dependent selection drives lineage replacement during the experimental evolution of SDHI fungicide resistance in Zymoseptoria tritici

Abstract Fungicide resistance is a constant threat to agricultural production worldwide. Molecular mechanisms of fungicide resistance have been studied extensively in the wheat pathogen Zymoseptoria tritici. However, less is known about the evolutionary processes driving resistance development. In vitro evolutionary studies give the opportunity to investigate this. Here, we examine the adaptation of Z. tritici to fluxapyroxad, a succinate dehydrogenase (Sdh) inhibitor. Replicate populations of Z. tritici derived from the sensitive isolate IPO323 were exposed to increasing concentrations of fluxapyroxad with or without UV mutagenesis. After ten increases in fungicide concentration, sensitivity had decreased dramatically, with replicate populations showing similar phenotypic trajectories. Sequencing the Sdh subunit B, C, and D encoding genes identified seven mutations associated with resistance to fluxapyroxad. Mutation frequency over time was measured with a pyrosequencing assay, revealing sequential lineage replacement in the UV‐mutagenized populations but not in the untreated populations. Repeating selection from set time‐points with different fungicide concentrations revealed that haplotype replacement of Sdh variants was driven by dose‐dependent selection as fungicide concentration changed, and was not mutation‐limited. These findings suggest that fungicide field applications may select for highly insensitive Sdh variants with higher resistance factors if the fungicide concentration is increased to achieve a better disease control. However, in the absence or presence of lower fungicide concentrations, the spread of these strains might be restricted if the underlying Sdh mutations carry fitness penalties.

of pathogens to toxicants (Cowen et al., 2000;Weinreich, Delaney, Depristo, & Hartl, 2006). Laboratory evolution studies give the opportunity to investigate the process of adaptation to new selective pressures including toxicants. Lenski, Rose, Simpson, and Tadler (1991) founded a long-term evolution experiment from a single Escherichia coli bacterium in 1988 to characterize the dynamics of adaptive evolution over long-term periods under a constant environment. Since then, 12 replicate E. coli-derived populations have been evolving for over 60,000 generations (Cooper, Rozen, & Lenski, 2003;Cooper, Schneider, Blot, & Lenski, 2001;Lenski & Travisano, 1994;Woods, Schneider, Winkworth, Riley, & Lenski, 2006). Keeping a "fossil record" in the form of glycerol stocks of intermediate generations of the E. coli-derived populations allowed the experimental evolution to be repeated from specific time-points. This enabled investigation into the repeatability of evolution under different conditions and the role of chance mutations (Blount, Borland, & Lenski, 2008).
Studies examining the evolution of resistance to toxicants have found parallel increases in resistance between replicate populations. Cowen et al. (2002) reported parallel evolution of resistance to fluconazole after 330 generations in four replicate populations of Candida albicans. Resistance was conferred by overexpression of either an ATP-binding cassette (ABC) transporter or a multidrug major facilitator (MFS) transporter encoding gene. Similarly, resistance to fluconazole in Saccharomyces cerevisiae was also found to be conferred by overexpression of two ABC transporters (pdr5 or snq2) after 400 generations in replicate populations subjected to increasing concentrations of the fungicide (Anderson et al., 2003). Toprak et al. (2012)  Zymoseptoria tritici, the causal agent of Septoria leaf blotch (SLB) in wheat, is a highly adaptable fungal pathogen (Zhan & McDonald, 2004). This plant pathogen has developed resistance or reduced sensitivity to fungicides with different mode of action, including methylbenzimidazole carbamates (MBCs) (Griffin & Fisher, 1985), quinone outside inhibitors (QoIs) (Amand et al., 2003;Fraaije et al., 2005), and sterol-demethylation inhibitors (DMIs) (Clark, 2006;. Therefore, options for the chemical control of SLB are currently limited. Since 2003, a new generation of succinate dehydrogenase inhibitor (SDHI) fungicides (e.g., boscalid, isopyrazam, bixafen, penthiopyrad, and fluxapyroxad) with strong activity against SLB has been launched in the crop protection market (Fraaije et al., 2012;Scalliet et al., 2012). SDHI fungicides target the protein succinate dehydrogenase (Sdh), also known as succinate ubiquinone oxidoreductase, that consists of four subunits. SdhA oxidises succinate to fumarate as a component of the tricarboxylic acid cycle (TCA), which is coupled with the reduction of ubiquinone to ubiquinol as part of the mitochondrial electron transport chain (Iverson, 2013). The SDHIs exert their fungicidal action by physically blocking the ubiquinone-binding pocket, formed by Sdh subunits B, C, and D (Ulrich & Mathre, 1972).
Laboratory studies using single generation exposures to different SDHIs have reported several target-site mutations conferring reduced sensitivity in mutants of Z. tritici and other plant pathogens (Fraaije et al., 2012;Scalliet et al., 2012;Sierotzki & Scalliet, 2013;Skinner et al., 1998). The SDHI sensitivity can be differentially affected by mutations. For example, SdhB-H267Y mutants of Z. tritici are insensitive to boscalid but hypersensitive to fluopyram. Other mutations of laboratory mutants, such as SdhB-H267L, SdhC-N86K, and SdhD-D129G, confer high levels of resistance to all SDHIs, including fluopyram. Due to the resistance risk, the SDHIs are used in mixtures with azoles or multisite inhibitors at recommended rates to delay fungicide resistance development in Z. tritici (FRAC, 2015). Field monitoring studies conducted since 2003 first detected sensitivity changes in 2012.
Between 2012 and 2015, five different Sdh variants, SdhB-N225T, B-T268I, C-T79N, C-W80S, and C-N86S, were reported at low frequencies in France, Germany, Ireland and the UK (FRAC, 2016). These Sdh variants display low levels of insensitivity, and control of SLB has not been affected so far. However, this may change as field strains carrying C-H152R, showing high resistance factors to SDHIs in vitro, have recently been detected in Ireland (Dooley, Shaw, Mehenni-Ciz, Spink, & Kildea, 2016) and the UK (B. A. Fraaije, unpublished), albeit at low levels.
Insensitivity to SDHIs has evolved in Z. tritici field populations as predicted from the variants found in laboratory mutational experiments. However, the extent of fitness penalties associated with mutations in the target protein conferring resistance to SDHIs, and the functional constrains affecting Sdh evolution under SDHI field selection where populations are exposed to different dose rates and spray frequencies remain unknown. In vitro evolutionary studies enabling to accelerate the evolution give the opportunity to investigate this.
In this study, we determined the course of evolution of resistance to the Sdh inhibitor fluxapyroxad in replicate populations of Z. tritici generated from the SDHI sensitive reference isolate IPO323 (Kema & van Silfhout, 1997). The IPO323 derived populations were exposed to increasing inhibitor concentrations in replicate populations at three different starting concentrations, each with or without exposure to UV light to increase mutation rate. After adaptation to ten stepwise increases in fungicide concentration, mutants carrying different Sdh mutations were found in most populations. One population without exposure to UV showed relatively low levels of SDHI insensitivity in the absence of target-site mutations. Studies on archived populations over time showed the mutation rate was not a limiting factor in UVmutagenized lines and that the haplotype replacement of Sdh variants was governed by dose-dependent selection as the "selective window" of the fungicide concentration changed.

| Generation of fluxapyroxad-resistant mutants
The reference Z. tritici isolate IPO323 (Kema & van Silfhout, 1997) was used as the parental strain in all experiments. For this, aliquots of 100 μl of IPO323 spore suspensions (10 7 spores/ml) were plated out onto 60-mm petri dishes of YPD agar amended with 0.4% (v/v) DMSO, and 0.04, 0.06, or 0.08 μg/ml of fluxapyroxad formulated as an emulsifiable concentrate (62.5 g/L EC; BASF, Ludwigshafen, Germany). Four YPD plates per fluxapyroxad concentration were inoculated, of which two were then exposed to 300 J/m 2 of UV light using an UV Crosslinker (model: XLE-1000/FB; Spectroline, New York, NY, USA). The UV light exposure of 300 J/m 2 allowed approximately 45% survival. After exposure, all cultures were incubated at 21°C in the dark for seven days.
After seven days' incubation, spores were harvested into sterile distilled water and quantified by haemocytometer count. Cultures with at least 10 7 spores/ml were exposed to an increased concentration of fungicide. An aliquot of 100 μl of spore suspension (10 7 spores/ml) from each culture was transferred to new YPD plates amended with 0.4% (v/v) DMSO and a twofold increased concentration of fluxapyroxad and left untreated or exposed to 300 J/m 2 of UV light before further incubation. The remaining spore suspension of each culture at each transfer was suspended in 80% (v/v) glycerol and stored at 80°C for further studies. Cultures with less than 10 7 spores/ml were plated on the same fluxapyroxad concentration and left in the dark at 21°C for another generation of seven days until they reached the spore concentration needed to be transferred on to a doubled fungicide concentration. Every series of fungal cultures was kept separately, as an independent population, labelled as "L", "I," or "H" for populations initially exposed to a low (0.04 μg/ml), intermediate ( Populations were diluted to produce single colonies on unamended YPD plates, and twenty colonies from each population were picked and subcultured on unamended YPD for seven days at 21°C in the dark, and stored in 80% glycerol (v/v) at −80°C for further use. Mutants were named in progressive number order per population from which they came (e.g., from L−1.1 to L−1.20).

| Fungicide sensitivity testing
In vitro fungicide sensitivity assays were carried out as described by Pijls, Shaw, and Parker (1994), see Supporting information. After four days of fungicide exposure at 23°C in the dark, fungal growth was measured using absorbance readings at 630 nm (A 630 ) with a FLUOstar OPTIMA microplate reader (BMG Labtech GmbH, Offenberg, Germany). Absorbance was measured in a well-scanning mode with a 2 × 2 matrix of scanning points set at 3 mm diameter.
Fungicide sensitivity was determined as the concentration, which inhibited growth by 50% (EC 50 ). The EC 50 (μg/ml) values were calculated using a dose-response curve (four-parameter fit) using OPTIMA software (v2.20OR2). The fungicide sensitivity test was carried out twice, and the progenitor/reference isolate IPO323 was included as control.
The EC 50 values are the average of the two independent tests.
To determine variation among the populations, the EC 50 values were analysed using ANOVA to compare the populations, taking account of initial fungicide concentration and UV light treatment, in GenStat (2014, 17th edition, VSN International © , Hemel Hempstead, UK). The standard error of the difference (SED) based on the residual degrees of freedom (df) from the ANOVA was used to construct a least significant difference (LSD) at the 5% level of significance with which to compare relevant means.

| Sequencing of the Sdh subunit B, C, or D encoding genes
The binding site of SDHI fungicides is formed by Sdh subunits B, C, and D, and mutations conferring resistance in Z. tritici have been previously identified in all three subunits (Fraaije et al., 2012;Scalliet et al., 2012). To determine whether Sdh alterations were associated with SDHI insensitivity, two fluxapyroxad-insensitive mutants with each of the highest, median, or lowest fluxapyroxad EC 50 values were selected from each population and their SdhB, C, and D genes sequenced.
Genomic DNA for sequencing was extracted and quantified according Rudd et al. (2010) from strains grown on YPD plates at 15°C in the dark for seven days. PCRs were carried out on a Biometra T3000 thermocycler (Biotron, Göttingen, Germany) in a final volume of 50 μl containing 50 ng of fungal template DNA. PCRs for amplification of SdhB or D contained 0.5 μM for each primer (Table S1) (Table S1) and 200 μM dNTP and contained 1× of Easy-A reaction buffer and 2.5 units of Easy-A High Fidelity PCR cloning enzyme (Agilent Technologies, Cedar Creek, TX, USA). Amplification conditions were 95°C for 1 min, followed by 30 cycles at 95°C for 15 s, 70°C (SdhB and D) or 65°C (SdhC) for 30 s, and 72°C for 1 min with a final DNA extension at 72°C for 5 min. PCR products were sequenced by MWG Eurofins Genomics GmbH (Ebersberg, Germany) using specific primers for each SdhB, C, or D gene (Table S1). Sequences were assembled and aligned with Geneious v.6.1.4 software (Biomatters Ltd., Auckland, New Zealand), and amino acid substitutions determined after sequence analysis.

| SNP detection pyrosequencing assays
To determine changes in frequency of key DNA mutations linked with SDHI resistance at each round of selection, six populations: three exposed to UV and with a greater diversity of mutations (I+1, H+1, and H+2), and the others without UV exposure and with fewer or no mutations (L−1, I−1, and H+1) were tested using SNP detection pyrosequencing assays using the PyroMark Q96 system (Biotage, Uppsala, Sweden) as described by Carter, Cools, West, Shaw, and Fraaije (2013) (see Supporting information). Aliquots of 10 μl of spore suspension from the last round of selection, and stored glycerol spore suspensions of previous RS, were plated out on YPD agar, and incubated for seven days at 15°C in the dark. DNA was extracted from each population. Mutants carrying different Sdh variants and the reference isolate IPO323 were used as positive and negative controls for the SNP detection pyrosequencing assays. Frequencies of mutations underlying SdhB-H267L, B-H267Y, B-N225T, C-T79I, C-S83G, C-H152R, and D-I50L were estimated with the PyroMark Q96 ID software version 2.5. Frequency values are the mean of two technical replicate pyrosequencing reactions. The detection limit of SNP detection pyrosequencing is approximately 3% for each allele (Gruber, Colligan, & Wolford, 2002;Wasson, Skolnick, Love-Gregory, & Permutt, 2002).

| Effect of fluxapyroxad concentration on the selection of SdhC-T79I and C-H152R variants
To investigate whether lineage replacement is mutation-limited or due to alleles being favoured only at certain stages of the selection regime, population H+1 spore suspensions from RS 3, 4, or 7 were regrown on YPD plates for seven days at 15°C in the dark and subsequently exposed to different concentrations of fluxapyroxad. For this, aliquots of 100 μl of RS3, RS4, or RS7 spore suspensions at 10 7 spores/ml were plated out onto YPD plates amended with 0.4% (v/v) DMSO, and 0, 0.32, 0.64, or 5.12 μg/ml of fluxapyroxad. These concentrations correspond to the most recent round of selection at which these populations were originally selected. Inoculated plates were incubated for seven days at 21°C in the dark, and DNA was extracted. Frequencies of SdhC-T79I or C-H152R variants in these populations were determined using SNP detection pyrosequencing assays.

| Gene expression studies
Although mutants of population L−1 were less sensitive to fluxapyroxad, no target-site mutations were found in the SdhB, C, or D gene.
To investigate whether changes in gene expression of a range of nontarget-site resistance conferring candidate genes were associated with fluxapyroxad insensitivity, we examined changes in their mRNA levels following the introduction of the fungicide. The reference Z. tritici isolate IPO323 and the fluxapyroxad-adapted mutant L-1.7, with no target-site mutations in the SdhB, C, or D gene, were in vitro exposed to fluxapyroxad at concentrations inhibiting growth by 50 or 80% (EC 80 ), and changes in gene expression was measured (see Supporting information).
Relative transcript abundances [relative quantities (RQ)] from genes encoding SdhB, C, or D, alternative oxidase (AOX), seven ABC transporters, seven major facilitator superfamily (MFS) drug efflux transporters, seven glutathione S-transferase (GST), and β-tubulin were determined with quantitative RT-PCR using specific oligonucleotide primers (Table S3). The selected genes encoding transporters or GSTs were the seven most up-regulated genes in the Z. tritici isolate IPO323 after exposure to chlorothalonil or folpet in the lag or log phase of growth (Gutierrez-Alonso, 2015). The Aox gene was included because overexpression can confer lower sensitivity to other fungicides that affect respiration (QoI) through nontarget-site mutations (Miguez, Reeve, Wood, & Hollomon, 2004 (Pfaffl, 2001), using β-tubulin as the endogenous control and samples from the untreated IPO323 isolate as calibrator. NRQ values of selected target genes were transformed to the logarithmic (log 2 ) scale and analysed using ANOVA in GenStat. The log 2 (NRQ) values were calculated assuming an efficiency of two for the target and reference gene (Pfaffl, 2001). Results are presented as ratios to the control treatment, using the log 2 (NRQ) means.

| Sensitivity of fluxapyroxad-resistant mutants to SDHIs
Isolates from the fungicide-adapted populations were less sensitive to fluxapyroxad than the reference isolate IPO323, which had an average EC 50 value of 0.02 μg/ml in this study (Figure 1a; Table S4). No significant differences (p = .2) were detected between EC 50 values of the nine populations studied. Fluxapyroxad sensitivity values ranged from 0.29 to 1.93 μg/ml (Figure 1a). There was no strong association between EC 50 values and exposure to UV or the initial fluxapyroxad concentration.
Most of the fluxapyroxad-resistant mutants were also less sensitive to fluopyram than isolate IPO323, which has an average EC 50 value of 0.11 μg/ml ( Figure 1b; Table S4). There were significant (p < .001) differences between populations, with a clear decrease in sensitiv- The IPO323-derived mutants were also less sensitive to carboxin, with EC 50 values ranging from 4.73 to 146.5 μg/ml (Figure 1c) compared to 1.36 μg/ml for the parental isolate, regardless of the treatment given to the population from which they came (Figure 1c; Table S4). This lower sensitivity to carboxin was expected, as fluxapyroxad and carboxin are SDHI carboxamide fungicides (FRAC, 2015).

| Detection of SdhB, C, and D variants
Based on the results of the fluxapyroxad sensitivity test, 57 fluxapyroxad-resistant mutants were selected and the genes encoding SdhB, C, and D sequenced (Table 1)

| Dynamics of SDHI resistant alleles in IPO323derived fluxapyroxad-adapted populations
A continuous replacement of mutants carrying different amino acid substitutions was measured in UV-exposed populations during subculturing of successive generations on agar with increasing fluxapyroxad concentrations. The mutation underlying B-H267L was detected in populations I+1 and H+2 at round of selection (RS) 5 or 4, respectively ( Figure 3). After an initial detection of 8% at RS5, the frequency of B-H267L in population I+1 increased gradually to approximately 80% at RS10; whereas in population H+2, B-H267L increased up to approximately 8% at RS-7 then decreased to <3% at RS10 (Figure 3).
The C-H152R allele frequency increased to approximately 95% in population H + 1 after RS9, whereas in population H + 2, the allele frequency increased to 71% at RS9 then decreased to approximately 32% at RS10. Similarly, C-H152R increased to approximately 40% at RS7 then decreased to <16% at RS10 in population I+1 (Figure 3).

| Effect of fluxapyroxad concentration on the selection of SdhC-T79I and C-H152R variants in population H+1
Changes in frequency of C-T79I or C-H152R depended on fungicide concentration during the experiment. When aliquots of the selected populations were regrown without fungicide amendment as for the original DNA extractions, the frequency of C-T79I and C-H152R remained comparable to that in the parent population, with a maximum difference of 13% and 2%, respectively, for RS4. However, following further selection at the intermediate fluxapyroxad concentrations of 0.32 and 0.64 μg/ml, the frequency of C-T79I increased from the low parental level in RS3 (Figure 4a), but was unchanged from its intermediate level in samples from RS4 and RS7 (Figure 4a and b, respectively). At 5.12 μg/ml fluxapyroxad, there was a substantial decrease in frequency of C-T79I in samples taken from RS4 and RS7 (Figure 4b and c, respectively), but an increase in the sample taken from RS3 ( Figure 4a). Frequency of C-H152R increased sharply at this concentration in samples from all three RS, confirming that it was already present at RS3, but undetectable, below the SNP detection pyrosequencing threshold.

| Gene expression
Quantitative RT-PCR detected a low but significant (p < 0.05) upregulation of genes encoding SdhB, C, or D in response to fluxapyroxad treatment in the reference isolate IPO323 (Fig. S3A, Table   S5). In the mutant L-1.7, only genes encoding SdhB or D were significantly (p < 0.001) slightly up-regulated (between 1.1-and 2.1-fold change; Fig. S3B and Table S6). Reference isolate IPO323 and the L−1.7 showed similar AOX gene expression patterns. Overexpression of AOX was only significant different in IPO323 when exposed  F I G U R E 3 Frequencies of key SdhB, C, and D target-site mutations in IPO323-derived fluxapyroxad-resistant populations during ten rounds of selection (RS) in vitro on YPD amended with increasing concentrations of fluxapyroxad. Fluxapyroxad sensitivity values of IPO323-derived mutant populations (A). Populations I−1 and I + 1 were exposed to 0.06 μg/ml as starting fungicide concentration. Populations H−1, H + 1, and H + 2 were exposed to 0.08 μg/ml as starting fungicide concentration. Fungicide concentration was increased twofold at every round of selection. Additionally, populations I + 1, H + 1, and H + 2 were exposed to 300 J/m 2 of UV light at each cycle. Relative allele frequency of positive controls: B-H267L=1, B-H267Y=1, B-N225T=1; C-T79I=1, C-S83G=1, C-H152R=1, and D-I50L=1. Frequency values are the mean of two technical replicates between replicate pyrosequencing reactions to fluxapyroxad at its approximate EC 50 and EC 80 concentrations ( Fig. S3, Table S5).
GST-4 was also significantly up-regulated (4.0-fold change, p < 0.05 or 10.5-fold change, p < 0.01), along with GST-1 (2.5-fold change, F I G U R E 4 Frequency of SdhC-T79I and C-H152R alleles in H + 1 populations grown for seven days at various fluxapyroxad concentrations. Samples were taken from the population after three (a), four (b), and seven (c) rounds of selection (RS). Cultures grew on YPD plates with and without addition of fluxapyroxad for seven days at 21°C in the dark. The arrows indicate the frequency in the parent population from which the samples were taken. Frequency values are the mean of two technical replicates between replicate pyrosequencing reactions. The detection threshold of both SNP detection pyrosequencing assay was 3%. Error bars denote two SEM F I G U R E 5 Expression levels of genes encoding putative ATP-binding cassette transporters (ABCt) in the reference Zymoseptoria tritici isolate IPO323 (a) and the laboratory mutant L−1.7 (b) after 24-h exposure to their respective fluxapyroxad EC 50 (light grey) or EC 80 concentration (dark grey), measured relative to the untreated control using quantitative RT-PCR. Mean of three biological replicates. Statistical significance for each gene is marked by asterisks (*p < .05 or **p < .01). See Table S5 and S6 for the means of log 2 fold-change data for the treatments with standard error of the difference for the comparisons made p < 0.01 at EC 80 ), in the reference isolate IPO323 in the presence of fluxapyroxad (Fig. S4A, Table S5).

| DISCUSSION
As the SDHI sensitive isolate Z. tritici IPO323 was used as progenitor of all populations, no external genotypes entered the mutant populations and no genetic exchange among populations occurred, so mutation was the only source of genetic variability. Therefore, selectively advantageous mutations conferred steadily reducing sensitivity to fluxapyroxad over time in all nine surviving populations of Z. tritici.
Seven of the nine populations were also less sensitive to fluopyram or carboxin; two were more sensitive to fluopyram. This negative cross-resistance of B-H267Y was reported before for Z. tritici by Fraaije et al. (2012) and Scalliet et al. (2012) and has also been reported for other plant pathogens (see Sierotzki & Scalliet, 2013). In one mutant, two mutations leading to B-N225T and D-I50L

| Resistance mechanisms in IPO323-derived fluxapyroxad-resistant mutants
were found simultaneously. This haplotype was associated with a high level of insensitivity to fluxapyroxad and lower levels of insensitivity to carboxin and fluopyram (Table 1). Scalliet et al. (2012) reported B-N225H/I in laboratory mutants, conferring lower sensitivity to carboxin, isopyrazam, fluopyram, and boscalid. An equivalent alteration in Botrytis cinerea, B-N230I, has also been linked with SDHI resistance (Leroux, Gredt, Leroch, & Walker, 2010). Equivalent mutations to D-I50L are not known and the precise role of D-I50L, alone or in combination with B-N225T, in SDHI binding needs further studies. Mutants carrying C-T79I, C-S83G, or C-H152R are insensitive to fluxapyroxad, fluopyram, and carboxin (Table 1). Scalliet et al. (2012) reported that C-T79I, C-H152R, or C-S83G variants can confer resistance to a range of SDHIs (i.e., carboxin, isopyrazam, fluopyram, and boscalid). Interestingly, they reported that C-S83G variants have higher levels of resistance to fluopyram than variants C-T79I or C-H152R. Similar results were found in our study.
Fluxapyroxad-insensitive isolates with no target-site (SdhB, C, or D) alterations were also less sensitive to fluopyram and carboxin. The gene expression study indicated that overexpression of an ABC transporter may contribute to the phenotype. The ABCt-2 gene, encoding a putative Z. tritici ABC transporter (Goodwin et al., 2011), was constitutively overexpressed up to 9.5-fold in the fluxapyroxad-resistant IPO323 mutant L−1.7 (Figure 5b). Similarly, Cowen et al. (2002)  including SDHIs and azoles (Omrane et al., 2015).

| Emergence and evolution of in vitro resistance to fluxapyroxad
Emergence of amino acid substitutions in the Sdh target protein occurred in parallel at similar rates in the UV light exposed populations succession of genotypes, each one fitter than its immediate predecessor, until the best adapted remains under specific conditions or threats (Muller, 1932 was driven by changes in fungicide concentration. When the population from an early round of selection (RS3) was exposed to a much higher fungicide concentration (5.12 μg of fluxapyroxad/ml), variant C-H152R was selected (Figure 4). This shows that C-H152R was already present after three RS, but only reaches detectable frequencies under selection at a high fungicide concentration (Figures 3 and 4).  (Scalliet et al., 2012).
Therefore, if lower doses are initially used, strains with lower levels of resistance (e.g., C-T79I) may initially be selected, after which fungicide dose could be increased to regain some degree of control, until other more-resistant strains (e.g., C-H152R, C-S83G, or B-H267L) are selected at higher doses. The patterns of genotype replacement in our data are consistent with the hypothesis that some alleles are only selectively advantageous in restricted fungicide concentration ranges.
If not mitigated by changes elsewhere in the genome, this fitness penalty could play a key role in the development of resistance management strategies. Although higher doses of SDHIs can reduce SLB disease symptoms slightly more effectively , using lower doses of SDHIs may reduce selection for strains with higher resistance levels.
In contrast, in two populations, I−1 and H−1, both exposed to initially greater concentrations of fluxapyroxad (0.06 or 0.08 μg/ml) but not to UV light, only a single mutation resulting in amino acid substitution SdhB-H267Y arose after four or five RS and reached a high frequency, close to fixation (Figure 3). Additionally, in population L−1, not exposed to UV, fluxapyroxad resistance was associated with overexpression of abct 2 and no further mutations were detected over the course of the experiment. This suggests that lineage replacement is less common in a mutation-limited system and/or takes much longer to develop.

| Comparison of SDHI resistance development in the field and in vitro
Adaptation to fungicides by target-site or nontarget-site alterations has already occurred in field populations of Z. tritici (see Clark, 2006;Fraaije et al., 2005;Omrane et al., 2015;Torriani et al., 2009). With regard to SDHI resistance, strains carrying amino acid substitutions SdhB-N225T, B-T268I, C-T79N, C-W80S, and C-N86S, conferring low levels of insensitivity, have been reported in field populations (FRAC, 2015(FRAC, , 2016. In our evolutionary study, resistance to fluxapyroxad emerged first by substitution of tyrosine by isoleucine at codon 79 of SdhC, which is the same position where T79N was found in field isolates. The reported C-T79N field strains have low resistance factors to different SDHIs. In our study, in vitro sensitivity assays indicated that the C-T79I variants had resistance factors of 37, 8, and 12 to fluxapyroxad, fluopyram, and carboxin, respectively (Table 1). Interestingly, one field strain carrying C-T79I was found in Hampshire (UK) during late summer in 2016 and is likely to spread further in the absence of fitness costs (B. A. Fraaije, unpublished).
Interestingly, a single field strain carrying B-N225T has been reported (FRAC, 2015). Mutants carrying B-N225T and D-I50L showed higher resistance factors for fluxapyroxad than C-T79I mutants, but the resistance factors for fluopyram and carboxin were still low. Our studies and Scalliet et al. (2012) showed that several amino acid substitutions (e.g., B-H267L, C-S83G, and C-H152R) with high resistance factors to all SDHIs might emerge in Z. tritici field populations. Indeed, C-H152R strains conferring high levels of resistance (RF >100) to bixafen, penthiopyrad, and fluxapyroxad were recently found in Z. trititici field populations sampled in Ireland (Dooley, Shaw, Mehenni-Ciz et al., 2016) and the UK (B. A. Fraaije, unpublished