Resistance to the alfalfa weevil in the Iranian collection of Medicago sativa Resistencia al gusano verde en la colección iraní de alfalfa

The Iranian collection of Medicago sativa (alfalfa, 502 accessions) was evaluated under field conditions to identify variation in resistance to the alfalfa weevil (Hypera postica Gyllenhal) between accessions over 4 years. Twenty-one accessions were identified as resistant (score 3) to semi-resistant (score 5) based on K-means cluster analysis. These materials were grouped into 10 groups by means of Duncan's Multiple Range Test. Susceptibility to alfalfa weevil was significantly correlated with some of the agro-morphological traits such as plant height (r = 0.314, P<0.01), regrowth rate (r = 0.435, P<0.01) and fall dormancy score (r = 0.284, P<0.01). In general, the resistant germplasm originated from elevated regions at high latitudes, where a co-evolution between the plant and the pest may have taken place. Multiple regression analysis showed a positive linear relationship between susceptibility to the alfalfa weevil and traits such as growth habit, plant height, length and width of the central leaflet and regrowth rate. The resulting bi-plot based on factor analysis ranked accessions for susceptibility to the pest. The identified germplasm should prove useful in alfalfa breeding programs to develop accessions resistant to the alfalfa weevil.


Introduction
Alfalfa (Medicago sativa L.) is one of the most important forage legumes in the world, and is adapted to a wide range of environments. It is the main forage legume in Iran, being cultivated in about 600,000 ha of mostly irrigated lands (Ahmadi et al. 2015). Alfalfa originated in Vavilov's "Near Eastern Center" Asia Minor, Transcaucasia, Iran and Turkistan (Bolton 1962). In the wild, M. sativa and related perennial species are found throughout Eurasia and as far north as Siberia (Hanson et al. 1988). Some parts of Iran are known as the center of diversity for diploid and tetraploid forms of M. sativa (Lesins 1976). Studies have shown that the probability of accessing traits, such as resistance to alfalfa weevil, Hypera postica (Gyllenhal) (Coleoptera: Curculionidae), is much higher in material collected from the centers of genetic diversity than other parts of the world (Jolivet 1998;Thompson and Cunningham 2002).
The alfalfa weevil is the most destructive pest of alfalfa crops in Iran and originated in the area surrounding the Mediterranean Sea. Essig and Michelbacher (1933) reviewed the history of the pest and concluded that it occurred throughout Europe, western Asia and the northern coast of Africa, plus the Madeira and Canary Islands. In temperate zones, alfalfa weevil can damage up to 100% of the first harvest and 50-100% of the second harvest of alfalfa (Behdad 1996).
In a laboratory study, Campbell and Dudley (1965) reported that a round-stemmed variety of alfalfa, Medicago sativa var. gaetula Urb., was almost immune to oviposition. From a field study, Norwood et al. (1967) reported that oviposition preference accounted for 26% of the variation in larval feeding damage among alfalfa clones. While Busbice et al. (1967) evaluated 456 alfalfa introductions and found all to be susceptible to the weevil, they noted that plants with thin stems were less preferred for oviposition. In a subsequent study, Busbice et al. (1978) reported that laboratory and greenhouse efforts to breed weevil resistance in alfalfa failed to identify field resistance and concluded that the only effective method of selection for field resistance to the weevil was field selection under natural weevil infestations. These authors suggested that heritability of weevil resistance was moderately high.
Karaj is one of the most important hot-spots for infestation of alfalfa weevil in Iran (Habibi 1976;Mazahery-Laghab and Yazdy-Samadi 1994), so is ideally suited to perform field evaluation of resistance to the weevil. A study on 124 alfalfa varieties showed a significant correlation between weevil damage and some agro-morphological traits (Mazahery-Laghab and Yazdy-Samadi 1994). Although no significant correlation was observed between the level of saponin and alfalfa weevil larval damage (Yazdi-Samadi et al. 2004), Bakhtiyarian and Mazaheri-Laghab (2016) indicated that resistant landraces showed higher levels of saponins in the thin layer chromatograms.
In previous studies, only small alfalfa collections were evaluated for susceptibility to the pest in this country. Hence, the current study presents the results of field evaluation of a comprehensive Iranian alfalfa collection for resistance to alfalfa weevil under natural field infestation in an important hot spot (Karaj, Iran) for the pest.

Materials and Methods
The Iranian collection of Medicago sativa, consisting of 502 accessions (Appendix 1) held by National Plant Gene Bank of Iran, was planted in the field at Karaj Experimental Station, Iran (3552´ N, 5059´ E; 1,520 masl) in September 2003. Each accession was planted in 2 rows of 1 m length and 50 cm apart with distance between plants within rows of 10 cm. The field was irrigated every 8 days from spring to fall throughout the study. During the following 4 years, susceptibility to alfalfa weevil and agromorphological traits were evaluated according to IBPGR (1984) descriptors. Growth habit was recorded annually using a scale of: 1 = prostrate, 2 = semi-erect and 3 = erect, at the end of April. Before the second harvest in each year, width and length of the central leaflet at the fourth node of stems were visually scored as: width 1 = very narrow to 9 = very wide, and length as 1 = very short to 9 = very long. Plant height was recorded at 10% flowering for each cutting cycle at which stage each accession was harvested throughout the study. Harvests were performed 4-6 times each year depending on particular accessions. Regrowth rate (cm/day) was determined using plant height at 15 days after each harvest. At the autumn equinox in each year all accessions were harvested, and fall dormancy was recorded as plant height 4 weeks after this harvest, allocating 1 point for every 5 cm of plant height. Plant hairiness was scored as: 1 = hairless, 2 = sparse (base of nearest neighbor further than length of hair) and 3 = dense (base of nearest neighbor closer than length of hair) before the second harvest in the first year. Alfalfa weevil susceptibility (AWS) of accessions was ranked as: 1 = no damage (very resistant), 3 = partial damage to a few terminal leaves (resistant), 5 = partial damage to most leaves (semi-resistant), 7 = severe damage to terminal leaves and partial damage to others (susceptible) and 9 = severe damage to all leaves (very susceptible). The plants were exposed to natural infestation in the field and no chemicals were applied during the 4-year experiment. Since economic damage to crops occurs from mid-March to early May annually in temperate zones like Karaj, evaluations for weevil damage were carried out at 2 and 4 weeks after the spring equinox. Means of the 2 evaluations in each year were recorded for assessing susceptibility to the pest. Passport data of collection sites were used only in the correlation analysis.

Statistical analyses
Spearman's method was applied to compute correlation coefficients between susceptibility to alfalfa weevil and other traits. Accessions were grouped based on alfalfa weevil susceptibility over the 4-year experiment by K-means cluster analysis to identify 3 categories (resistant, semi-susceptible and susceptible). One-way analysis of variance (ANOVA) was performed to evaluate year and accession effects. Transformed data (square root) were used in the ANOVA. Duncan's Multiple Range Test grouped 21 resistant/semi-resistant accessions and 1 susceptible accession to 7 groups. Multiple regression using the stepwise method was used to estimate the role of agro-morphological traits (as independent variables) on susceptibility to alfalfa weevil (as dependent variable). The factor analysis was performed on the mean of each trait with PCA method and varimax rotation. These analyses were performed by SPSS version 15.

Results
Natural infestations of the pest over the 4-year study were high (Table 1, Figure 1A, Appendix 1). Evaluation of the whole collection (WC, Appendix 1) of 502 accessions of M. sativa collected from farmers' fields and wild habitats (Table 1) showed that most accessions were susceptible ( Figure 1B). Twenty-one accessions were identified as resistant (score 3) to semi-resistant (score 5) based on K-means cluster analysis and are suggested for use in breeding studies as sources of genes for resistance to alfalfa weevil [called resistant collection (RC), Table 2, Appendix 2]. The means of susceptibility scores from the first to fourth year differed from 8.1 to 8.6 (1 = very resistant to 9 = very susceptible) in the WC and 3.4 to 5.3 in the RC (Table 1). In each year more than 400 accessions were awarded a score of 9 (very susceptible, Figure 1B) indicating the level of natural infestation of the pest over the 4 years. Resistance of individuals varied over time and semi-resistant to semisusceptible accessions showed fluctuations among years.
Descriptive statistics of agro-morphological traits for RC and WC are shown in Table 1. While most accessions in the WC displayed erect growth habit, 15 of the RC accessions were semi-erect in growth habit, while 2 were erect and 4 were prostrate. RC accessions also showed short and narrow leaflets. Plant height in the RC varied from 20.5 to 49 cm with a mean of 38.1 cm, compared with 20.5 to 97 cm (mean 65.5 cm) in the WC (Table 1).  (3), semi-resistant (5), susceptible (7) and very susceptible (9).   RC accessions were generally collected from northern parts of Iran, so had low fall dormancy scores (range 0.1-3.8, mean 1.2). However, the WC showed a larger range (0.1-9.9, mean 3.7) for this trait. In the same manner, regrowth rate and central leaflet width manifested greater ranges in WC than RC. In general, a sparse to dense hairiness was observed in the RC (Table 1).
ANOVA results showed significant differences in RC and WC for susceptibility to the pest over the 4 years (Table 3). Duncan's Multiple Range Test grouped the RC accessions into 10 groups (Table 2). Accessions 51TN0669 and 51TN0672, which originated from northwest Iran, were the most resistant. The resistant to semi-resistant accessions (scores 2 to 5, respectively) were mostly collected from provinces in northern, southwest and central Iran at higher geographical latitudes, which experience a cold-temperate climate ( Figure 2).
Susceptibility to the alfalfa weevil showed significant correlation with some agro-morphological traits (Table  4). In WC, susceptibility was significantly and positively correlated with growth habit (r = 0.532, P<0.01), plant height (r = 0.314, P<0.01), regrowth rate (r = 0.435, P<0.01), fall dormancy score (r = 0.284, P<0.01) and width of central leaflet (r = 0.450, P<0.01) ( Table 4). There was no significant correlation (P>0.05) between susceptibility to the pest and elevation or longitude, from which the accessions had been collected, but susceptibility was significantly and negatively correlated with latitude of origin (r = -0.296, P<0.01) ( Table 4).   Multiple stepwise regression analyses revealed a positive linear relationship between alfalfa weevil susceptibility (Y) and traits such as: growth habit (X1), plant height at flowering (X2), length of central leaflet (X3), regrowth rate (X4) and width of central leaflet (X5) with R 2 = 0.32 as shown in the following Equation 1: Factor analysis in the WC showed that 65.2% of the variance was explained by 4 primary factors (Table 5). The first factor, which accounted for 28.1% of the variance, is affected mainly by 3 traits (regrowth rate, fall dormancy score and days between successive harvests).
The second factor explained 21.1% of the variation and was affected by traits such as alfalfa weevil susceptibility, width and length of central leaflet, growth habit and plant height. Distribution of accessions in a bi-plot based on Factors 1 and 2 is shown in Figure 3. 1 (very short) to 9 (very long). 5 1 (very narrow) to 9 (very wide).   (1), sparse (2) and dense (3). 4 1 (very short) to 9 (very long). 5 1 (very narrow) to 9 (very wide).

Discussion
While most of the evaluated accessions were landraces and local varieties, resistant accessions were collected mainly as wild ecotypes from near farmers' fields, on roadsides or within rangelands of mountainous areas in the country. Only 2 landraces collected from farmers' fields showed resistance to the pest (Table 2). Table 1 shows that wild accessions do not rate highly for some evaluated traits such as plant height, re-growth rate and width and length of central leaflets, so they have low potential for forage production. The resistant accessions, including wild accessions 51TN0669 and 51TN0672, were generally collected from northern parts of Iran (Table 2), where cold conditions prevail. ANOVA results (Table 3) showed that pest damage was different in different years, which could result from increases in pest density in the field from the first year onward that is usually expected. The significant negative correlation between susceptibility to the pest and latitude (r = -0.296, P<0.01) in the WC (Table 4) suggests that resistant material was distributed at higher latitudes, where cold conditions prevail (Figure 2). Some resistant to semi-resistant accessions were collected from Hamedan and Azerbaijan, the origins of the 2 important cold alfalfa landraces, Hamedani and Gharah-Younjeh (Abbasi et al. 2007), whereas the germplasm collected from warm and subtropical regions lacked resistance to the pest (Figure 2). Earlier studies showed that alfalfa weevil tended to be most prevalent in cold areas, mainly in the cold zones of the mountainous areas (Essig and Michelbacher 1933;Behdad 1996). In such areas, a coevolution may take place between the pest and the crop, leading to the development of resistant germplasm. The current study indicated that such a phenomenon probably existed between the pest and alfalfa in cold climates (higher latitudes) of the plant's origin. Co-evolution of species is one of the main functions contributing to the Earth's biodiversity. Jolivet (1998) reviewed the interrelationship between insects and plants and explained the co-evolution phenomenon between an insect and its host plant. According to Thompson and Cunningham (2002), local structure of co-evolutionary selection remains consistent through successive generations. Alfaro et al. (1999), after reviewing literature on genetic resistance in the weevil/spruce system, concluded that there was coevolution between the weevil and its hosts. Likewise, Mendel (1998) reported that resistance to the Israeli pine bast scale, Matsucoccus josephi (Homoptera, Matsucoccidae), which developed in pines of the subsection Halepenses, viz. Pinus brutia, is believed to be the result of co-evolution.
Another factor in the susceptibility to the weevil of tropical accessions may be the change in environment (from tropical to temperate conditions). Tropical alfalfa germplasm was collected from southern parts of the country and evaluated in Karaj, where the climate is temperate. Therefore, inherent resistance within these materials may not have been expressed optimally, if the environment was not conducive to natural growth of the germplasm. Abiotic influences like temperature, humidity, solar irradiance, photoperiod, water availability and elevation all could have an impact on the effectiveness of host plant resistance and biological growth of the weevil. The pest does not cause economic damage to the crop in the warm conditions prevailing in southern parts of the country.
The results of correlation analysis were in accordance with some previous reports (Mazahery-Laghab and Yazdy-Samadi 1994; Kakaei et al. 2016). The significant correlation between susceptibility to the pest and plant hairiness (r = 0.114, P<0.05) (Table 4) agrees with reports from earlier studies of a significant negative relationship between hairiness and resistance to the alfalfa weevil (Shade et al. 1975;. However, since the correlation coefficient was so low, this trait was not inserted into the final regression equation. Equation 1 particularly implies that tall erect plants are most susceptible to the pest. Although Busbice et al. (1967) reported that fast growth is an important characteristic contributing to alfalfa weevil resistance, Norwood et al. (1967) showed that laying of pest eggs and larval feeding tended to be least on plants with prostrate growth habit, poor growth and small thin stems. Our results implicitly confirmed these findings.
According to factor analysis and the produced bi-plot, susceptible accessions came together closely towards the center of the bi-plot and semi-resistant materials appeared mainly apart from them in Section III of the bi-plot, while resistant material appeared mostly in Section III, but clearly separated from the semi-resistant accessions (Figure 3). There was a similarity between the results of K-means cluster analysis and the produced bi-plot in identifying the susceptibility of accessions to the alfalfa weevil (Table 2, Figure 3).
This classification of the accessions in the Iranian alfalfa collection under severe natural infestation of alfalfa weevil satisfies the requirements of Busbice et al. (1978) that the only effective method of selection for resistance to the weevil is field selection under natural weevil infestations. Accessions identified as resistant to weevil attack should provide genetic material for use throughout the world in alfalfa breeding studies to develop cultivars resistant to H. postica through production of hybrids. The high degree of diversity for most desirable agronomic traits in the resistant collection should ensure that other attributes should not necessarily suffer by using these lines.