Evaluation of sainfoin accessions exposed to powdery mildew disease at four locations in Iran

In order to evaluate resistance of sainfoin (Onobrychis viciifolia) to powdery mildew, seeds of 19 accessions were collected from different parts of Iran and sown at 4 locations, i.e. Kheirabad, Khoramabad, Semirom and Tabriz, in 2014. Accessions were evaluated for powdery mildew severity index (DSI), forage dry matter yield (DM), dry matter digestibility (DMD) and crude protein (CP) and water soluble carbohydrate (WSC) concentrations over 4 years. Based on Duncan’s test, accessions 15353 and 3001 showed disease severity index lower than 25% and were nominated as resistant to powdery mildew. Accessions Oshnavieh and Polycross were considered semi-resistant due to their DSI ranging from 25 to 50%. Other accessions were considered susceptible because their DSI was higher than 50%. The resistant accessions (15353 and 3001) with average yields of 3,341 and 3,304 kg DM/ha were ranked as having high DM production, in addition to displaying high DMD plus high CP and WSC concentrations. Severity of powdery mildew infection was linked negatively with all 3 quality traits, i.e. DMD and CP and WSC concentrations. According to Eberhart/Russell regression results, stability of accessions 3001 and 15353 for DSI and DM yield was confirmed across 4 locations. We recommend the use of accessions 3001 and 15353 in future breeding programs to increase resistance to powdery mildew, while at least maintaining yield and quality attributes. Evaluation of other sources of sainfoin germplasm should continue to identify further resistant accessions.


Introduction
Common sainfoin, Onobrychis viciifolia Scop. (syn. O. sativa Lam.), is a most important perennial forage legume, which is highly regarded by farmers due to high levels of palatability and nutrient concentrations (Delgado et al. 2008). Sainfoin was introduced to agriculture as a drought-and salinity-tolerant plant, can produce yields comparable with those of alfalfa and due to its deep roots is well adapted to dry and desert ecosystems (Soares et al. 2000). It contains condensed tannins which lower bloat incidence in grazing animals and improve protein digestion in the intestines (Rumball and Claydon 2005). Sainfoin, also known as holy clover, is often intercropped with forage grasses to improve soil fertility and pasture quality via nitrogen (N) fixation (Lu et al. 2000). The ability of this forage crop to fix N can reduce applications of chemical fertilizers (Greub et al. 1984). The honey obtained from nectar of sainfoin flowers is very bright and sweet, with a distinct flavor, and displays concentrated crystallization at any temperature with yields in the range of 20-51 kg/ha (Pérez-Arquillue et al. 1995). Sainfoin is also tolerant of Hypera postica (Gyllenhal) (Curculionidae), a common pest of alfalfa crops (Allen and Allen 1981).
The Onobrychis genus comprises 69 species (both annual and perennial) in Iran (Mabberley 1997). The species are concentrated in the Zagros mountains of Iran (60 species) and Turkey (52 species) (Çelik et al. 2011), and these areas appear to be the major origins of genetic diversity in sainfoin populations (Mohajer et al. 2013). The plant is commonly grown in both irrigated and dry lands of Iran, including Charmohal Bakhtiary, Lorestan, Fars, Kerman, Kordestan, Kermanshah, Zanjan and Mazandaran provinces (Hidarian and Mollaei 2001) and is capable of acceptable growth in cropping systems, which are inappropriate for clover and alfalfa cultivation.
However, more widespread cultivation of sainfoin in the main growing regions of Iran is limited by the incidence of powdery mildew, which reduces forage yield during the second harvest (Majidi 2010). Severe infections of powdery mildew occur in Azarbayjan, Charmohal Bakhtiary, Esfahan, Fars, Kerman, Kermanshah, Kordestan, Lorestan, Mazandaran and Zanjan provinces (Bamdadian 1991;Behdad 1996). Naseri and Alizadeh (2017) reported a number of climatic indicators of development of powdery mildew infections in Zanjan province. This disease can appear at the end of the growing season and causes noticeable yield losses at the last cutting. The causal agent of this disease is a fungus known as Leveillula taurica, with Oidiopsis sp. as the asexual form. This pathogen also infects other plant species such as alfalfa, sunflower, safflower and hemp. The aerial plant parts severely infected by this pathogen develop spots, become dehydrated and are shed. Depending on the region, the disease appears in August-September in Iran (Sharifnabi and Banihashemi 1990;Ershad 1995;Naseri and Marefat 2008). Alizadeh and Jafari (2014) evaluated the susceptibility of 56 accessions of sainfoin to powdery mildew in Alborz province during 2010-2012. Their results showed that 4 accessions (Polycross, Oshnavieh, 3001 and 15353) were more resistant than the remaining accessions and were considered desirable parents for breeding superior sainfoin cultivars.
The objectives of this study were: (i) to test the resistance to powdery mildew disease of these promising accessions across 4 geographical regions of Iran in comparison with a set of 15 inbred accessions collected in different parts of the country; and (ii) to identify accessions with superior dry matter yield and forage quality.

Materials and Methods
The study was conducted at 4 locations (Khoramabad, Kheirabad, Semirom and Tabriz) in Iran as described in Table 1. These 4 locations ( Figure 1) were included in the current research to cover a range of agro-ecological conditions for powdery mildew development and sainfoin productivity. In this experiment, reactions of tolerant and semi-tolerant accessions in comparison with susceptible accessions were evaluated during 4 growing seasons under irrigated conditions using a randomized complete block design with 3 replications. The seeds of 19 accessions (Table 2) were collected from plants grown in the earlier research project. For each accession, seeds were sown in 4 drilled rows (2 m long and 0.25 m apart) in sward conditions in April 2014. Irrigation was applied according to the plant requirements. Weeds were controlled mechanically and fertilizers (75 kg N/ha as ammonium nitrate and 150 kg P/ha as superphosphate) were applied based on scientific advice and recommendations. In the establishment year (2014), plots were cut once and no data were collected.   In the second year, assessments of powdery mildew were made under natural infections in which all accessions (tolerant and susceptible) were cultivated adjacent to each other. Thus, disease severity was evaluated when plants were exposed to airborne contamination from natural infections under environmental conditions specific to the 4 study regions. A disease severity index (DSI) was applied to each accession according to the percentage of aerial parts covered with fungal mycelium. This assessment was performed a few days before the second and third harvests, when powdery mildew infections on plants were rated according to a 0-4 scale (Horsfall and Cowling 1978), with a score of 0-2 (resistant) allocated when fungal mycelium covered 0-25% of aerial parts, a score of >2-3 (semi-resistant) when mycelium covered 25-50% of aerial parts and >3-4 (susceptible) when mycelium covered 51-100% of aerial parts as an indicator of plant susceptibility to powdery mildew.
At the 50% flowering stage plants were harvested by hand and forage produced by each plot was weighed immediately. Three harvests were performed at each site, i.e. early May (first harvest), early June (second harvest) and early September (third harvest). A 300 g representative subsample of fresh forage from each plot was collected, placed in a bag and transferred to the laboratory, where it was oven-dried at 75 °C for 48 hours and then weighed to determine DM percentage. DM yields were calculated for each experimental plot. In addition, subsamples were milled and dry matter digestibility (DMD) plus crude protein (CP) and watersoluble carbohydrate (WSC) concentrations were determined in the laboratory of Research Institute of Forests and Rangeland, Tehran, Iran, based on the method of Jafari et al. (2003).

Statistical analysis
To simplify interpretation of statistical analysis of quality-severity-yield data obtained from 19 sainfoin accessions examined over 4 growing seasons at 4 different geographical locations, low variable data over years were pooled and analyzed to estimate the extent of variability among genotypes and locations. Thus, due to the lack of significant effects of year, annual DM yields, disease severity ratings and quality traits (DMD, WSC and CP) were averaged over the study years to be used for combined analysis over 4 locations. Mean comparisons were conducted based on Duncan's method. Although several stability parameters have been proposed, Eberhart and Russell (1966) considered a stable genotype should have a slope (b value) equal to unity and deviation from regression (S 2 d) equal to zero. Stable genotypes would be those having mean yield higher than the average yield of all genotypes under test. This method has been used widely for evaluating yield stability in both annual and perennial plants. Based on the Eberhart/Russell stability regression model, regression coefficient values (bi) and deviation from regression (S 2 di) were calculated for each of the 19 genotypes. Stable genotypes with high mean yields were identified if the regression coefficient equated to one (bi = 1) and deviation from regression equated to zero (S 2 di = 0; Eberhart and Russell 1966). The stability tests were performed for both DM yield and disease severity index (DSI) using AGROBASE (Agronomix Software Inc., Winnipeg, Canada). Minitab 16 was used to illustrate relationships among genotypes and environments.

Results
Considering the lack of significant effects of study year, the combined analysis of variance across 4 environments showed significant differences between locations (except for DMD; P>0.01), accessions and G × E interactions for all traits (P<0.01; Table 3). Mean values for DM yield, CP and WSC concentrations, DMD and DSI at each location and averages over 4 locations for sainfoin accessions are presented in Tables 4-8. While some accessions performed consistently in terms of DM yield across all locations, others performed     Overall mean disease severity index (DSI) values for the 4 locations were much higher at Zanjan (56.1) and Tabriz (48.0) than at Esfahan (31.6) and Koramabad (28.6), indicating a greater incidence of powdery mildew at the former 2 locations (Table 8). However, at all locations accession 3001 consistently displayed the greatest resistance to powdery mildew infection . Over all locations this accession displayed a mean DSI of 12.9, compared with an overall mean for all accessions of 41.1. While accessions 15353, 2759, 9263 and Polycross performed well at Esfahan and Koramabad, they were severely infected at Tabriz and Zanjan.

Correlation analysis
The results of correlation analysis demonstrated that the only associations between DSI and other parameters were a positive correlation (P<0.01) between DM yield and DSI at Esfahan (Table 9) and negative correlations between DSI and both CP% and WSC% at Tabriz and overall, as well as between DSI and DMD overall. There was a significant negative correlation (P<0.05) between DM yield and WSC concentration at Esfahan and Zanjan locations. The strongest relationships from the data were strong positive correlations between DMD and WSC concentration for 3 of the 4 locations and overall (Table 9).

Stability analysis
Results of Eberhart/Russell regression response indices (b), deviation from regression (S 2 d) and mean DM yield and DSI for 19 sainfoin accessions over 4 locations are presented in Table 10. Plots of the relationships between regression coefficients (bi) and mean DM yield for 19 accessions (Figure 2) showed that accessions 8199, 9263, 15353, 15364, 9147 and Oshnavieh had DM yields higher than the average value and b values were near unity, indicating their DM yield was stable across environments. Accessions 19402 and 3001 with bi >1 coupled with high DM yield had above-average stability for high performing environments, Khoramabad and Tabriz. A higher deviation from regression indicated sensitivity to environmental changes for DM yield (Table 10).
Similarly, the regression coefficients plotted against the test accessions for DSI are presented in Figure 3. For DSI, accessions 15353, 3001 and 9262 with b values near unity were considered stable over the 4 locations. Oshnavieh in Khoramabad and accessions 2759, 9263 and Plc (Polycross) in Esfahan with a greater resistance to powdery mildew were stable only for those locations. In general, accessions 15353, 9262 and 3001, located close to the regression line, appeared to be more stable in terms of higher production and lower disease severity index across locations.

Discussion
In order to improve sainfoin as bioactive forage, improvements in resistance to powdery mildew, yield and quality traits are required. As an initial step, quantifying the variability in existing sainfoin germplasm would allow the identification of superior genotypes with disease resistance and yield stability across various environments. According to earlier evidence, Iran is recognized as one of the most important sources of genetic diversity in populations of sainfoin worldwide (Mohajer et al. 2013). From Iranian germplasm the present research has identified 2 promising sainfoin accessions not only resistant to powdery mildew, but also with good forage quality and high DM yield, which was stable across a range of different environments and growing seasons. Since the arid and warm climate in Iran provides a high genetic diversity in populations of Leveillula powdery mildew (Khodaparast et al. 2012), testing under natural conditions in Iran provided an ideal environment for comparing accessions. It would seem reasonable to conclude that differences in climatic conditions across the diverse geographical areas would result in genetically diversified populations of the pathogen providing an ideal challenge to accessions. Moreover, this variability in environmental conditions across different Iranian regions resulted in significant differences in powdery mildew severity, DM yield and quality traits among the same accessions grown at 4 study locations. Although Naseri and Alizadeh (2017) reported the association of climate with sainfoin powdery mildew resistance and yield in Zanjan, the interaction of environmental parameters from different regions with the disease, resistance, quality and yield of this valuable forage crop merits further investigation in future. Our findings demonstrated low disease severity index (<25%) in accessions 15353 and 3001 across 4 different environments, identifying them as powdery mildewresistant sainfoin genotypes. Furthermore, accessions 9262 and Oshnavieh were considered as semi-resistant or semi-susceptible to powdery mildew. While various levels of sainfoin resistance to powdery mildew based on data from a single location or year have been reported previously (Alizadeh et al. 2014;Alizadeh and Jafari 2014), to the best of our knowledge this is the first report of stability of disease resistance in 2 sainfoin genotypes over a 4-year study conducted across different geographical areas. This information is a valuable first step in global breeding programs to increase powdery mildew resistance in sainfoin populations.
It has been reported that forage WSC provides efficient energy needed for plant growth following reduction in photosynthesis, rejuvenation after leaf loss and recovery after drought and freezing stresses (Humphreys and Eagles 1988;Humphreys 1989). According to Wilkins and Lovatt (1989), it is desirable to maintain CP concentration of forage above 12%, to ensure rumen microflora have adequate nitrogen for both milk and meat production by livestock. According to our findings, there were significant negative correlations between DSI and forage quality traits tested, indicating that lower powdery mildew levels on accessions corresponded with higher DMD plus CP and WSC concentrations. There are a number of earlier Iranian reports on the linkage of sainfoin powdery mildew severity with forage quality (Mohajer et al. 2013;Alizadeh et al. 2013); however, none of them explored the stability of these relationships over diverse environments and seasons. Moreover, the present research identified a positive correlation between DMD and WSC and CP concentrations in the accessions of sainfoin tested. This result was in agreement with an earlier report of Alizadeh et al. (2013). Therefore, one of the important benefits of reducing powdery mildew infections via breeding of resistant lines may be maintenance of sufficient CP, DMD and WSC to improve sainfoin revival and produce high quality forage. It seems that accessions 3001 and 15353 are potential candidates for both commercial production and use in breeding programs. Further testing of genetic material to identify other accessions for use in breeding programs seems warranted.