Geographic Distribution and Genetic Analysis of Physiological Racesof Setosphaeria turcica in Northern China

Five hundreds and forty-six isolates of Setosphaeria turcica, the causal agent of Northern Corn Leaf Blight, were collected in 61 corn-growing locations throughout six provinces of Northern China during 2000-2002, to determine their pathogenicity on two sets of host differentials: OH43/Huangzao4, OH43Ht1Huangzao4Ht1, OH43Ht2/Huangzao4Ht2, OH43Ht3/Huangzao4Ht3 and OH43HtN/Huangzao4HtN. The isolates were grouped into 13 different physiological races (0, 1, 12, 3, 13, 23, N, 1N, 2N, 12N, 3N, 23N and 123N) based on their infection types on the host differentials. Incidence analyses indicated that races 0 and race 1 were still dominant in Northern China and took 40.66% and 18.32% of total isolates tested respectively, while other races sparsely occurred from 1.28-6.59%. The emergence of race 123N that was toxic to corn lines woth all four existing major resistant genes implies the possibility of present hybrids to loss their resistance in some regions of China. Further analysis of race distribution in Northern China demonstrated that the occurrence and composition of physiological races of S. turcica varied among provinces. Genetic analysis of within and between pathogenic race groups by random amplification polymorphic DNA (RAPD) markers revealed high genetic diversity in S. turcica population. Relatively high genetic similarity (70.46-95.12%) was identified within race groups and the results suggested that race groups were genetically similar within one geographic locations, but genetic migration could possibly happen between some locations which might lead to relative high genetic diversity within one geographic location. The similarity indexes derived from 13 race groups varied from 20.31 to 82.81% indicating great genetic variation between race groups. The UPGMA dendrograms generated by NTSys software grouped the 13 races into two not very robust but relatively distinct clusters: cluster I (0, 1, N, 12, 1N, 2N and 12N) and cluster II (3, 13, 23, 3N, 23N and 123N) with 63.7 and 72.3% P value respectively. The cluster analyses suggested that the pathogen might have a great genetic change while mutating to be virulent Ht3 resistant gene, but the details about the mechanisms were remain unaware. key words: Zea mays, corn, Setosphaeria turcica, differential lines, genetic diversity, geographic distribution


Setosphaeria turcica Leonard and Suggs, anamorph
Exserohilum turcicum Leonard and Suggs, the causal agent of Northern Corn Leaf Blight (NCLB), was first reported in Italy in 1876 [1] . It had been one of the most important foliar diseases in corn worldwide [2] . S. turcica isolates could be grouped into different physiological races based on their infection types on corn genotypes with different dominant resistant genes Ht1, Ht2, Ht3, or HtN [3] . Inbred lines carrying these resistant genes had been used in corn breeding program. Because the pathogen's physiological races were greatly related to its pathogenicity, the identification of physiological race was very important in the disease control and the study of corn and pathogen interaction. During the past two decades, an increasing number of novel races had been identified in China (0 and 1), in Mexico(23N, 23 and 2N), in Zambia (23, 23N and 0) and Uganda (0, 2, N, 23N) [4] .
In China NCLB was first reported in 1899 and was primarily found in North and Northeast regions of corn production, where the environmental conditions were favorable for disease development [5] . The disease did not become serious until the hybrid corns were widely planted in 1960s since the resistant genes of inbred lines had not been integrated into the hybrids yet. Inbred lines with Ht1, Ht2, Ht3 and HtN resistant genes were introduced in corn breeding programs in succession and generated a number of hybrids that could reduce the yield loss due to S. turcica. However, the causal fungus rapidly evolved into novel physiological genotypes, such as race 1, races 23 and 23N that led to rapid resistance loss in many hybrids containing Ht1, Ht2, Ht3, or HtN gene [6,7] . Therefore, the occurrence of novel pathogenic races was continuously a threat to corn with single-locus resistant genes and the work monitoring the occurrence of new pathogenic races of S. turcica was always conducted in some NCLB research groups, including the mycotoxin laboratory of Agricultural University of Hebei, China. Since 1999, we sent over 1000 mails to farmers and local researchers in the main corn-grown regions throughout Northern China every year. We asked them collect the leaves infected by NCLB and put them into envelopes before mailed us. We sent students and professional teachers to collect the samples too. In this way, we had built a big bank of S. turcica isolates and it made us possible to study the race composition and genetic variance of the causal pathogen of NCLB in Northern China.
The use of RAPD (random amplification of polymorphic DNA) markers was widely used in taxonomy and genetic variance of microbial populations and was ever applied to study the genetic polymorphism of S. turcica in Africa and Isalel [8] , in tropical and temperate [9] and in Europe [10] . The studies enhanced the ability to make inferences about phylogeny and provided vital information about possible genetic mechanisms of S. turcica.
In this study, we will focus on (1) the pathogenic race identification of S. turcica in main corn-growing regions of Northern China, (2) the description of occurrence and geographic distribution of the pathogenic races and (3) the analysis of genetic variance within and between the races by RAPD molecular markers.

Study area:
The research region was located in Northern China where corn was cultivated as one of the main crops (Fig. 1). All corn cultivars were Hybrids since they had been introduced in 1992-1994. The corn was grown as monoculture or in mixed stand that included wheat, soybean or peanuts. Farmers in this region usually bought traditional hybrid seeds for sowing the following year, but more and more farmers bought new hybrid varieties because they could carry them a better harvest at the present. The new hybrids usually carried Ht1, Ht2 and/or Ht3 resistant genes, while traditional hybrids only carried Ht1 resistant gene or carried no gene resistant to S. turcica. Therefore, new hybrids were usually more resistant to NCLB than traditional hybrids. Usually a farmer in the region had discontinuous holdings of less than 1 hectare, so traditional hybrids and new hybrids were often mixed grown in one plot.
Isolates collection: In the fall of each year during 2000-2002, corn leaves with typical NCLB symptom were collected from 61 sites in Northern China (Fig. 1). The region included six provinces, i.e. Heilongjiang, Jilin, Liaoning, Hebei (including Beijing and Tianjin city), Shandong and Henan provinces, where NCLB seriously occurred in China. To obtain pure culture from each sample collected in the field, the infected leaves were washed in the distilled water and then cut the lesions including some healthy tissues into small pieces (about 1 cm 2 ). The leaf pieces were surface sterilized in a solution of 0.5% sodium hypochlorite for 2 minutes, followed by washing three times with sterile distilled water. Sterile leaf pieces were then placed in a plate of Potato Dextrose Agar (PDA) and incubated at 22°C for 6 to 10 days. S. turcica spores in the cultures were harvested, suspended in sterile distilled water and further diluted up to 10 -6 to obtain single conidium culture. Monoconidial isolates were prepared by picking single germinating conidia on 4% water agar, multiplying on PDA plates and storing in test tubes containing 5 ml of sterile PDA medium. Usually one isolate per lesion was obtained.
Growth characteristics of each isolate were recorded by measuring colony size at a 2-day interval for 10 days or longer. No remarkable variation was observed in incubation period between isolates, but there was variation in colony diameters between isolates from different locations. Colony radial growth was influenced by temperature. Some isolates had faster radial growth than others at 20 or 28ºC. The isolates did not exhibit appreciable variation in appearance or in morphology of the colony.
Pathogenic race identification: Two sets of corn differentials, OH43/Huangzao4, OH43Ht1Huangzao4Ht1, OH43Ht2/Huangzao4Ht2, OH43Ht3/Huangzao4Ht3 and OH43HtN/Huangzao4HtN, were used to distinguish pathogenic races of S. turcica isolates. All differentials were corn-inbred lines generously donated by Professor  The isolates were cultured on PDA medium in a Petri dish about 7 to 9 days for obtaining their inoculants. The plates were first moistened with a few drops of sterile water and then the conidia were dislodged from the surface of the colonies with a microscope slide. The conidia were washed out with 10 ml of sterile distilled water and then drained into a clean beaker. The suspension was filtered through double-layered gauze and spore concentration was determined using a hemocytometer. Concentration of spores was adjusted to 10 5 -10 6 per ml for inoculation of differential hosts. A few of drops of Tween-20 and sucrose (3.0% w/v) were added to the solution just prior to inoculation to improve the spore adhesion on the corn leaves [11] .
The sterilized infield loam with about 1 percent composite fertilizer was used to grow the differential host plants in greenhouse at the Agricultural University of Hebei, Baoding, China. Five seeds per line were planted and later three seedlings were kept per pot for inoculation. When the seedlings at a stage with 4-6 leaf (about 4-5 weeks after planting), they were inoculated by spraying spore suspension with a hand-held sprayer. Inoculated plants were immediately covered with polythene sheets up to 48 h to maintain the moisture.
Three replicates were designed for each treatment. In order to avoid cross contamination, inoculation area and sprayer were treated with 5% carbolic acid between the inoculations.
Infection types were classified R-type and S-type and evaluated at 15 days after inoculation on the differentials [12] . The sporulating lesions without obvious chlorosis were defined as susceptible reaction (S-type) while non-sporulating lesions with clear chlorosis as resistant reaction (R-type). Physiological races of S. turcica were named according to the formula described by Leonard [13] . For example, an isolate which is avirulent to differential lines with Ht1 and Ht2 resistant genes but is virulent to Ht3 and HtN genes will be given a virulent formula Ht1, Ht2/Ht3, HtN and subsequently named as race 3N.

DNA extraction and RAPD amplification:
Genomic DNA was extracted from lyophilized ground mycelium of monoconidial isolates following CTAB (cetyltriethyl ammonium bromide) method [14] . DNA relative purity and concentration were determined by electrophoresis with a known concentration marker and fluorometry at the ratio of OD 260 /OD 280 prior to RAPD amplification [15] .
RAPD analysis was employed to determine genetic variation among physiological races of S. turcica. PCR amplification was carried out in a 25 l reaction mixture, containing 10 ng of template DNA, 0.2 mM 10-mer random primer (Sangon, Shanghai, China), 100 mM of each dNTP, 2 mM MgCl 2 , 0.1% Triton X-100, 0.2 unit of Taq DNA polymerase (TaKaRa, Japan). Twenty-six 10-mer primers (Sangon, Shanghai, China) were used in this study. The amplification reaction was performed in a Biometra 4800 (Whatman Corp., Germany) with a program consisting of a predenature at 94°C for 3 min, 40 cycles (94°C for 1 min, at 37°C for 2 min, at 72°C for 2 min) and a final extension of 6 min at 72°C. Reaction products were electrophoresed on 1.6% agarose gels at 3 V/cm and stained with ethidium bromide before visualization under UV light.

Data analysis:
The banding patterns produced by RAPD markers were scored manually using a binary system (1 for presence and 0 for absence at each band position). The proportion of common bands was used to analyze the genetic polymorphism within races and calculate similarity index to study genetic variation between races, using the following formula: Fij = 2Bij/Aij, where Aij is the numbers of total bands observed for the ith and jth races and Bij is the common bands observed between the given pair of races [16] .
Phylogenetic relationships between races were examined by cluster analysis and phylogenetic dendrograms construction. The dendrograms were generated by the UPGMA method [17] based on the average similarity indices between races using the SAHN program of the NTSYS-pc package [18] . The confidence limits of clusters in the UPGMA-based phenograms were determined by performing bootstrap of the binary data using the program WinBoot [19] . Each phenogram was reconstructed 2000 times by repeated sampling with replacement and the frequency with which a particular grouping was formed was considered to reflect the robustness of the group.

Pathogenic races identification:
A total of 546 isolates of S. turcica were collected from 61 locations across Northern China during 2000-2002 (Table 1). There were 284 isolates collected from Hebei province, 99 isolates from Liaoning, 57 isolates from Heilongjiang, 50 isolates from Shandong, 28 isolates from Henan and 28 isolates from Jilin. During the three years, we attempted to sample as broad geographic areas as possible to obtain an estimate of races distribution of S. turcica population throughout Northern China, especially in Hebei province, whose government gave part financial support of the project. However, because the discontinuous distribution of corn-grown areas made it impossible to sample in every corn-grown area, isolates in one study location were collected randomly. Moreover, because of the difficulties of pathogen isolation and fungal conservation for some samples, only 5-8 isolates were collected from some locations ( Table 1).
The 546 isolates were inoculated onto the two sets of corn differentials and their reactions to the infection were recorded to identify the races. All differential plants exhibited small pinhead size spots 48-72 h after inoculation, however, the spots on resistant plants exhibited obvious chlorotic response. Lesions became easy to see at 7-10 days after inoculation. Resistant reaction showed yellowish chlorotic lesions while susceptible lesions were elliptical gray necrotic and sporulating. The lesions on susceptible plants were either water soaked or had dark margins with profuse sporulation. A summary of race identification and sample size for each race were listed in Table 1. Shangyi 9 (7,2,0) 0(9) Total sample size 546 a : Site number correspond to those in Fig. 1. b : Sample size for the province. Isolates included in the study were collected during 2000-2002. c : The first, the second and the third numbers in brackets were isolates collected in 2000, 2001 and 2002 respectively. d : Two sets of corn differentials, OH43/Huangzao4, OH43Ht1Huangzao4Ht1, OH43Ht2/Huangzao4Ht2, OH43Ht3/Huangzao4Ht3 and OH43HtN/Huangzao4HtN, were used to make races identification. Physiological races were rated by the infection types in the differentials described by Hooker (Hooker, 1963) and named according to the formula described by Leonard (Leonard et al., 1989) e : Numbers in the brackets were the sample size of the corresponding races Thirteen physiological races (0, 1, 12, 3, 13, 23, N ,  1N, 2N, 12N, 3N, 23N and 123N) were distinguished according to their infection types in the differentials. Among them, race 2N was firstly identified in China and races 13, 3, 1N, 3N and 123N were firstly reported in the worldwide corn planting area. The isolates in the same race population had the identical infection types in the two sets of host differentials, although the disease severity might be different between the two sets of differentials.

RAPD analysis of S. turcica isolates:
Sixteen reproducible polymorphic RAPD primers were selected from 26 RAPD primers to amplify random sequences from the genomic DNA of isolates belonging to 13 races of S. turcica ( Table 3). The other 10 primers had Total RAPD bands 116 S2035 AAGTGCCCTG Average bands for each primer 7.25 a : Primers purchased from Sangon Technologies Inc. Shanghai, China. b : Primers exhibited reproducible polymorphism in RAPD analyses of races of S. turcica no RAPD amplicons for some samples, so they were not selected to analyze the genetic diversity within and between races.
The sample size of each race population was listed in Table 4. Each race population comprised about 2-5 isolates in every location where the race was identified. However, if the race was only found in one location, for example, race 2N was only found in Zhuolu county of Hebei province, race 23N was only found in Antu city of Jilin province, 5-7 isolates were selected to analyze their genetic diversity within the races.
All of the 16 polymorphic primers gave clear DNA bands on the agarose gels as shown in Fig. 2. A total of 116 bands and an average of 7.25 bands per primer were obtained by RAPD amplification (Table 3).
Variation within races: Among all races, the average percentages of common bands were 66.12% for RAPD analysis. The overall similarity values indicated the detectability of genetic polymorphism within races. The average percentage of common bands within races 0, 1 ,  12N, 1N, 12, N, 3N, 13, 123N, 3, 23, 23N and 2N     Comparing the above results with the geographic distribution of race groups in Table 1 (race 0, 28 sites; race 1, 11 sites; race 12N, 3 sites; race 1N, 6 sites; race 12, 5 sites; race N, 4 sites; race 13, 3 sites; race 3, 3 sites; race 23, 3 sites; race 3N, 2 sites; race 123N, 2 sites; race 23N, 1 site; race 2N, 1 site), it could be derived that the geographic distribution of the races might contribute to the genetic variation within race groups, i.e. the broader geographic distribution of the race groups, the more genetic variation within them. The only possible exception was race 12N group, suggesting that genetic migration could happen between different locations and lead to relative high genetic diversity within one geographic location.

Variation between races:
In order to clarify the variation between races, the isolates belonging to the same pathogenic race were treated as one group. Among all race groups, the band patterns were scored and subjected to calculate the similarity indexes between race groups. The indexes derived from 13 race groups varied from 20.31 to 82.81% (Table 5). These results revealed that great genetic differentiation generated between different race groups, which agreed with the conclusions of previous studies [8] .
In order to elucidate the phylogenetic relationships between race groups, a phylogenetic dendrogram was generated by the UPGMA method [17] based on the average similarity coefficient between race groups using the SAHN program of the NTSYS-pc package [18] . The phenogram was reconstructed 2000 times using the program WinBoot [19] and the frequency was considered to reflect the robustness of the clusters (Fig. 3).
Four clusters, A (including race groups 0, 1, N, 12, 1N and 12N), B (race group 2N), C (including race groups 3, 13, 23, 3N and 23N) and D (race group 123N), were resolved at the 0.16 similarity level. Clusters B and D consisted of only one race group each, so bootstrap values for these clusters were not applicable. In order to determine the bootstrap values for clusters B and D as it appears in the UPGMA dendrogram, two bigger clusters, cluster I (including clusters A and B) and

DISCUSSION
Temperature and humidity were two important factors affecting the corn growth and the pathogenicity of the isolates. Even in environment-controlled greenhouses, late spring and early autumn was appropriate time for the laboratory inoculation. Growth stage of corn seedlings also affectws plant response to S. turcica infection. Seedlings with 4-6 leaves were consistently used in this study.
Leath and Pedersen in 1986 [20] found it impossible to accurately assess lesion number, type and size due to some lines exhibited multitudes of small chlorotic lesions. Only necrotic lesions were assessed for size. Therefore only two reactions were recorded in this study, i.e. the resistant lesions and the susceptible lesions.
Distribution of 13 physiological races (0, 1, 12, 3, 13, 23, N, 1N, 2N, 12N, 3N, 23N and 123N) of S. turcica varied among the provinces covered in this study. Except for races 3N and 23N, all other 11 races were detected in Hebei province, in which the most physiological races of S. turcica were found. Eight races were found in Liaoning province, 4 races in Shandong province, 3 races in Jilin and Heilongjiang provinces and 2 races in Henan province. However, only physiological race 0 was found at 28 locations of all 6 provinces covered in this study. Race 1 was also very popular and could be found at 11 locations in Hebei, Liaoning, Jilin and Shandong provinces. Race 1N were found at 6 locations in which 5 locations were in Hebei province, the other one location were in Shandong province. Race 12 occurred at 5 locations in Hebei, Liaoning and Heilongjiang provinces. Race 3, N and 12N were only found in Hebei and Liaoning provinces. Race 13 occurred at 2 locations in Hebei province and 1 location in Heilongjiang province. Race 23 was found at 3 locations in Hebei, Liaoning and Henan provinces. Race 3N was found at Liaoning province and Shandong province. Race 23N and 2N were only found at Antu city in Jilin province and at Zhuolu county in Hebei province respectively. Race 123N, virulent to corns with all four existed Ht resistant genes, was found at Weichang county and Xianghe county in Hebei province (Table 1 and Table 2). The existence of race 123N had given us a signal that the pathogen had been differentiating at all times and could be mutated into more virulent and more dangerous. A B C D I II Introduction of Ht gene into China in 1970s made a tremendous impact on maize production, but the situation had soon changed because of new pathogenic races of S. turcica evolved. Wu et al [21] first reported the existence of S. turcica race 1 in Dandong district of Liaoning province. Race 23 was first found in Yunnan province in 1989 [6] and subsequently identified in Guizhou province in 1993 [22] . Although new races had been reported since then, race 0 was dominant one during the last two decades. In 1993, Chen [5] reported that race 1 became dominant genotype of S. turcica in Liaoning province and a large volume of research were conducted to understand the occurrence and distribution of those races [2,11,20,[22][23][24][25] . In this study, the race 0 was still dominant in northern China with the average occurrence of 40.66%, followed by race 1 (18.32%). The other races occurred only in certain locations with no more than 7% occurrence frequency.
Races 0 and 1 had been dominant ones in China for more than 20 years with little changes in their occurrence, indicating that the pathogen was a population with high speed of genetic variance but at a stable pace. Many factors might contribute such variation, such as the frequently change of hybrid cultivars and cultivation system, the rapid changes in hybrids composition, the change of environmental temperature and so on. It could be predicated that the dominance of races 0 and 1 might continue for several years in Northern China.
In the past two decades, NCLB was one of the most important diseases of maize (Zea mays L) in warm and humid part of the world [26,27] . In Africa, NCLB was reported to be widespread and destructive disease of maize in Ethiopia, Tanzania and Uganda [28][29][30] . In the United States, northern leaf blight was the major disease of maize [31] . However, after the seven races 0, 1, 2, N, 23, 2N and 23N were reported [4,13] , even the great variation in genetic diversity of the pathogen was found within or between states or regions [9,32] , only Oliari et al [33] reported three new races including two new virulent factors in S. turcica. One of the major reasons for the little discovery of new physiological races could be predicated as the even distribution of corn lines with different resistant genes leading to balanced genetic differentiation in Setosphaeria turcica, which could be suggested by the genetic variation studies within and between race groups in this project.
The studies on the genetic variation within and between race groups in Northern China by RAPD analysis revealed that the pathogen's population had high genotypic diversity, which was identical to the studies of genetic structure of Setosphaeria turcica populations in tropical and temperate climates [9] . Genetic variation within race groups suggested that the race groups were genetic similar within one location, but it could not be excluded that genetic migration could happen between locations. Genetic variation between race groups was revealed by the phylogenetic relationship study using a UPGMA dendrogram generated by NTSys software and confirmed by WinBoot software. Although having not very robust confidence, 63.7 and 72.3% respectively, two distinct clusters were resolved at 0.1 similarity level: cluster I (0, 1, N, 12, 1N, 2N and 12N) and cluster II (3, 13, 23, 3N, 23N and 123N). Race group 2N was relatively distinct from other race groups with 77.6% p-value in cluster I and race group 123N was relatively distinct from other race groups with 62.7% p-value in cluster II. The cluster analyses suggested that the pathogen might have a great genetic change while mutating to be virulent Ht3 resistant gene, but the details about the mechanisms were remain unaware.