MATERIALS AND METHODS

Location. The experiments were established in the 2018 spring-summer cycle in three regions that are representative of the rainfed barley producing area of the Mexican highlands: 1) Relinas, Axapusco, State of Mexico, located 19° 50’ 33.32” LN and 98° 53’ 36.73” LW and at an altitude of 2322 masl, where the predominant soil was Feozem; 2) Moxolahuac, Tlahuapan, Puebla, 19° 26’ 31.37” LN, 98° 32’ 31.60” LW and an altitude of 2763 masl, where the predominant soil is Andosol, of volcanic origin, rich in organic matter, a high water retention and cationic exchange; 3) Soltepec, Tlaxco, Tlaxcala, 19° 35’ 40.07” LN, 98° 18’ 36.48” LW at 2532 masl; the predominant soil type is Feozem, which is porous and rich in organic matter.

Experimental design. The experimental design was a randomized complete blocks with a factorial arrangement (6 x 3) resulted in 18 treatments and four repetitions, for a total of 72 experimental units or plots for each location. The experimental unit consisted of four furrows, each one 3 m in length and with a space of 30 cm between them, corresponding to an area of 3.6 m², considered a useful plot.

Germplasm. Six barley varieties were evaluated: Apizaco, Esmeralda and Maravilla, released by the Instituto Nacional de Investigaciones Forestales, Agrícolas y Pecuarias (INIFAP); Doña Josefa, released by the ICAMEX, and AC Metcalfe and ABI Voyager, varieties from Canada and the USA, recently introduced into Mexico. The first four varieties are of six rows, while the latter two are of two rows. These varieties are malting varieties, except for Maravilla and Doña Josefa, which are forage grain varieties. The level of resistance to YR in the varieties released by INIFAP indicate that Esmeralda is resistant (^{Zamora et al., 1997}), Maravilla is cataloged as tolerant (^{Zamora-Díaz et al., 2017}) and Apizaco, as susceptible (^{Cuéllar-Zambrano et al., 2015}), whereas Doña Josefa is resistant (^{Guzmán-Ortíz et al., 2019}). There are no data published for Mexico indicating the level of resistance of varieties AC Metcalfe and ABI Voyager.

Fungicides. The fungicides tested were Azoxistrobin 11.1% + Tebuconazole 18.4% (Azimut^® 320 SC) and Tebuconazole 25% (Folicur^® 25 EW) and the dose used was the one recommended by the manufacturer (0.7 L ha^-1 for Azoxistrobin 11.1% + Tebuconazole 18.4% and 0.5 L ha^-1 for Tebuconazole 25%). The 18 treatments consisted of all the varieties without fungicide, all varieties with Azoxistrobin 11.1% + Tebuconazole 18.4% and all varieties with Tebuconazole 25%. In all treatments we used the coadjuvant polyether polymethylsiloxane copolymer (Break Thru^®) at a dose of 0.25 L ha^-1. The fungicides were applied using a SWISSMEX knapsack compression sprayer (15 L) with a flat TeeJet ASJ 11003 spray tip. Fungicides were applied twice: the first was 50 days after planting, when plants were in development stage 41-49 according to the scale proposed by ^{Zadoks et al. (1974}) and observing a severity of 1 to 5% in the susceptible variety Apizaco, and the second one was 20 days after the first application.

Inoculum. The incidence of yellow rust appeared naturally, from the growth stage of booting (development stage 41-49; ^{Zadoks et al., 1974}), since the three locations display favorable weather conditions for the development of the disease, such as cold to temperate temperatures (5 to 18 °C) and the formation of dew.

Variables. The following variables were recorded: days to heading (DE - days between planting and the moment in which 50% of plants displayed visible spikes), days to physiological maturity (DM - days between harvest and the moment in which the spike’s peduncle turned hay yellow), plant height (AP - height in centimeters from the surface of the soil to the tip of the terminal or higher spike), grain yield (REND - the weight of grain produced by all the spikes of all useful plots, registered in grams, which was later converted into kg ha^-1), test weight (PH - weight of the grain per volume unit (kg hL^-1)) final disease severity (SFE - highest level of damaged foliar area, registered as a percentage and using the scale modified by Cobb) (^{Peterson et al., 1948}). The first reading of yellow rust was carried out just before the products were applied for the first time, and was followed by three more, once every 10 days. With the readings of severity (four) taken in time intervals, the area under the disease progress curve (AUDPC) was calculated using the equation described by ^{Bjarko and Line (1988}).

Statistical analysis. The data of the variables obtained in the three locations were analyzed statistically in a joint manner using the program SAS 9.3 (SAS Institute^®, USA) and comparisons of averages were carried out for the variables under study using the DMS test (p=0.01).

RESULTS AND DISCUSSION

Analysis of variance: The joint analysis of variance of the three evaluated locations found highly significant differences for most of the variables evaluated. The fungicides showed a highly significant difference for days to maturity (DM), test weight (PH), grain yield (REND), final disease severity (SFE) and area under the disease progress curve (AUDPC). Meanwhile, in the variety*fungicide interactions it was the same case for PH, SFE and AUDPC, whereas in location*fungicide, significant differences were found for DM, PH and AUDPC. These results indicate a different behavior between varieties, where the locations play an important part in the expression and behavior of the genetic materials, as mentioned by ^{González-González et al. (2016}); similar results were reported by ^{Rodríguez-García et al. (2020}) for wheat. The application of fungicides shows that there is a positive response of the varieties in the control of the disease, the crop cycle and yield-related variables, based on the product applied, which is conditioned by the environment. For variables DE and AP, no significant effects of the chemicals were observed, therefore the behavior of these variables depend on the variety and the environment.

Comparison of means for fungicides, locations and varieties. The average values (Table 1) show that there are no statistical differences for variables DE and AP, with and without fungicides. For the remaining variables (DM, PH, REND, SFE and AUDPC), there is a response between the application of fungicides and the control without fungicide. However, when analyzing the information of the products applied, no significant statistical differences are observed between them. The application of fungicides for the control of YR affects the value of most of the evaluated variables in a positive way, with the most notorious increases in PH and REND. ^{Devlash et al. (2015}) reported similar behaviors in the application of several fungicides for the control of YR in barley, and point out that applying fungicide, regardless of the product, significantly reduces SFE and increases the yield in comparison with the control without fungicide.

In the average values per location (Table 2), locations played a crucial part in the expression of the values of each one of the variables evaluated. For DE, the locations of Relinas and Soltepec displayed a similar behavior when reaching this stage 60 days after planting (dds). In Moxolahuac, the cycle was later, requiring 10 additional days. A similar behavior was observed in DM, with Moxolahuac being the location with the longest cycle (143 dds). Plant height varied between locations, with lower heights observed in Relinas (48 cm), whereas plants in Soltepec and Moxolahuac surpassed the height of 70 cm. The grain test weight (PH) was greater in Soltepec (62.16 kg hL^-1), followed by Moxolahuac and Relinas. Grain yield was higher than 3 t ha^-1 in the locations of Soltepec and Moxolahuac, whereas in Relinas, the average yield hardly surpassed 1 t ha^-1. This behavior may be due to the availability of water during the crop cycle, since Soltepec and Moxolahuac had 428 and 417 mm of rain respectively during the crop cycle, yet the location of Relinas had 286 mm (data registered by the Barley Genetic Breeding Program of INIFAP- Valle de México Experimental Field), causing the low yields observed. The highest incidence of YR was registered in the locations of Relinas and Soltepec, in comparison to those found in Moxolahuac.

The behavior of the varieties (Table 3) displayed significant differences between them. The earliest materials (DE and DM) were, in general terms, the six-row varieties, whereas the two-row varieties were late. Esmeralda was the early variety, reaching heading and maturity at 54 and 121 dds, respectively. The variety ABI Voyager and AC Metcalfe were the latest, reaching maturity at 140 and 137 dds, respectively. The height of plants was distributed in a range of 64 to 72 cm, with the two-row varieties being taller than the six-row ones. PH was numerically higher in the Esmeralda variety (62.04 kg hL^-1), although similar statistically to the values obtained in AC Metcalfe and Doña Josefa; Apizaco was the variety with the lowest PH, with 58.5 kg hL^-1. In grain yield, Doña Josefa (3234.5 kg ha^-1) statistically surpassed all other varieties. Esmeralda and Maravilla had statistically similar results to AC Metcalfe and ABI Voyager, whereas Apizaco was the lowest yielding (2337 kg ha^-1). A similar behavior was observed in locations and genotypes by ^{González-González et al. (2016}) and ^{Jalata et al. (2011}), who mention that the environmental variation between locations and the genetic variability have an influence on the expression and the potential of the genotypes.

The average response of SFE and ABCPE indicate that Apizaco was the variety with the greatest YR infection levels, which is a susceptibility response of this genotype, since it is one of the old varieties released by the INIFAP and has no effective resistance genes for the current races of this pathogen or due to its release before the pathogen was established in Mexico. The three remaining six-row varieties (Esmeralda, Maravilla and Doña Josefa), were statistically similar, and the ones with the highest levels of resistance to the disease. The two-row varieties, AC Metcalfe and ABI Voyager, were, along with Apizaco, the most susceptible to YR. These results show a clear superiority in resistance of the varieties released by INIFAP (Esmeralda and Maravilla), in comparison with the two-row varieties recently introduced into Mexico for their use in the brewing industry. When comparing the yields of AC Metcalfe, ABI Voyager and Esmeralda, these are observed to have statistically similar values. However, Esmeralda tolerated best the presence of the pathogen, since it is a variety that, when released, it was described by ^{Zamora et al. (1997}) as the first barley variety developed in Mexico for rainfed conditions with a certain level of resistance to yellow rust. This variety keeps its resistance until now.

Effect of the fungicides and agronomic behavior of the varieties. The originated treatments of the combination of varieties with the fungicides applied (Table 4) show the behavior of each one of the varieties under the effect of the product. The fungicides affected the behavior of the varieties for the variables evaluated, except for DE and AP. The Esmeralda variety was the earliest in all its treatments for DM, whereas the two-row varieties were the latest; a tendency to increase cycle duration was observed when applying a product, as opposed to when it is not applied. The results obtained show that Esmeralda was the earliest genotype when no fungicide was applied (Esmeralda-Sf) and the combination of ABI-Voyager-Azoxistrobin 11.1% + Tebuconazole 18.4% was the latest. In general terms, the treatments without fungicide were the earliest in each variety, and when Azoxistrobin 11.1% + Tebuconazole 18.4% was applied, the behavior of the varieties Esmeralda, Maravilla and ABI-Voyager was to delay leaf senescence, whereas with Tebuconazole 25%, it was the varieties Apizaco, Doña Josefa and AC-Metcalfe. ^{Bertelsen et al. (2001}) reported that the fungicide Azoxistrobin causes a delay in the senescence of leaves, although this does not necessarily mean that there is an increase in biomass or in grain yield.

For the variable PH, a greater test weight can be found within each variety when a fungicide is applied than when it is not applied (Table 4), except in the combination Maravilla-Az+Teb. Also, a numerical superiority is observed when applying Azoxistrobin 11.1% + Tebuconazole 18.4% over Tebuconazole 25% in all varieties except in Maravilla, in which Tebuconazole 25% was superior to Azoxistrobin 11.1% + Tebuconazole 25%. The average PH values obtained in each treatment, except for Apizaco without fungicide, comply with the values specified in the Mexican norm NMX-FF-043-SCFI-2003 for the marketing of barley, as indicated by ^{González-González et al. (2016}). The highest PH values were obtained in the combinations of AC Metcalfe-Azoxistrobin 11.1% + Tebuconazole 18.4% (62.56 kg hL^-1) and Doña Josefa-Azoxistrobin 11.1% + Tebuconazole 18.4% (62.53 kg hL^-1).

Grain yield (REND) increased in every variety when fungicides were applied. The varieties where Azoxistrobin 11.1% + Tebuconazole 18.4% were applied gave higher yields than those with Tebuconazole 25%, except Maravilla and AC-Metcalfe, where Tebuconazole 25% was numerically superior to Azoxistrobin 11.1% + Tebuconazole 18.4% (Table 4). The combination of Doña Josefa-Azoxistrobin 11.1% + Tebuconazole 18.4% (3567.4 kg ha^-1) displayed the best yield, although it was similar statistically to that obtained in Doña Josefa-Tebuconazole 25% (3258.3 kg ha^-1) (Table 4). The lowest grain yield was obtained in the Apizaco variety when fungicide was not applied (Apizaco-Sf), a statistically similar value to most varieties when no fungicide was applied or when Tebuconazole 25% was applied on Apizaco and Esmeralda and Azoxistrobin 11.1% +Tebuconazole 18.4% was applied on AC-Metcalfe. ^{Devlash et al. (2015}) proved that YR severely reduces grain yield in susceptible genotypes, and in these cases, the use of fungicides must be promoted to control the pathogen.

In terms of percentages, the increases in yield obtained when applying Azoxistrobin 11.1% + Tebuconazole 18.4% were 35, 31, 24, 16, 15 and 3% in the varieties Apizaco, ABI Voyager, Doña Josefa, Emeralda, Maravilla and AC Metcalfe, respectively, in comparison with not applying fungicide. These results show that the varieties respond differently to the application of these chemicals. Similar results were reported by ^{Kanwar et al. (2018}), who point out that when applying fungicides of the group of triazoles such as Trifloxystrobin 25% +Tebuconazole 50% and Tebuconazole 25.9%, they found an increase in grain yield of 56.5 and 53.27%, respectively. On the other hand, ^{Selvakumar et al. (2014}) reported an average increase in yield of 42.97% when applying Tebuconazole 25.9%.

For SFE, statistically similar values were obtained in most treatments (Table 4); the most notorious statistical differences were observed in the treatments of susceptible varieties such as Apizaco, AC Metcalfe and ABI Voyager, where SFE values reached 57.50, 27.08 and 44.58% in these varieties, respectively, when the disease was not controlled, whereas with the application of fungicides, SFE is notoriously reduced. In varieties Esmeralda, Doña Josefa and Maravilla, SFE was very low in all treatments and were statistically similar amongst them, even when Esmeralda-without fungicide recorded a SFE of 6.33. These results indicate a greater resistance to YR of these three varieties in comparison with the varieties Apizaco, AC Metcalfe and ABI Voyager. The AUDPC shows a similar tendency to that observed in SFE, where the varieties Apizaco, AC Metcalfe and ABI Voyager have the highest values for this variable, whereas Esmeralda, Doña Josefa and Maravilla are the varieties with the lowest AUDPC values, and therefore the most resistant to the disease.

Effect of the fungicides on the severity of the disease. The results obtained indicate a clear effect of the fungicides on the varieties (Tables 1 and 4) and that, in addition, in each location, there was a different level of severity (Tables 2, 3 and Figure 1). This effect on the application of fungicides for the control of YR has a greater relevance when observing the superiority of the yield, PH and a greater health of plants in comparison to when the disease is not controlled. Out of the six varieties, only Esmeralda presented low SFE values (Table 4), similar to those obtained by the two forage varieties (Maravilla and Doña Josefa). In addition, except for Apizaco, it is possible to claim that the six-row varieties display lower SFE than the two-row varieties (ABI Voyager and AC Metcalfe). The analysis of the joint information does not display a statistical superiority of Azoxistrobin 11.1% + Tebuconazole 18.4% on Tebuconazole 25% in most varieties for SFE (Table 4), therefore the application of either one of the two fungicides is effective in controlling the disease and ensures a higher grain yield, particularly in susceptible varieties. This coincides with the report by ^{Devlash et al. (2015}), who point out that applying fungicide, regardless of the product, significantly reduces the severity of the disease and increases yield in comparison with the control without fungicide, and they also point out that Tebuconazole reduced the disease severity by up to 91.8%. On the other hand, ^{Walters et al. (2012}) mention that protecting the plant from disease is better than trying to eradicate the pathogen, indicating that triazoles and strobilurins are effective in the reduction of the incidence of YR. ^{Gangwar et al. (2018}) mention that the fungicides belonging to the group of the triazoles, such as Tebuconazol 25%, and to the strobilurins, such as Azoxystrobin 25%, have proven to be effective in the control of rust in barley.

Losses in grain yield. Table 5 shows yield losses caused by YR in the three locations evaluated. The highest loss percentages were recorded in Relinas, with Apizaco being the variety with the greatest loss (53%), followed by AC Metcalfe (36%) and ABI Voyager (25%); Maravilla was the variety with the lowest percentage of losses, with only 5%. The location of Moxolahuac was the location with the second greatest loss, and where Apizaco and ABI Voyager had percentages of 26 and 25%, respectively. Esmeralda and Maravilla were the varieties with the least losses, with 13% each. For the location of Soltepec, the greatest losses were recorded in the varieties ABI Voyager (23%) and Doña Josefa (19%), while Esmeralda was the best variety with only 6% of yield loss. These results indicate that the varieties released by INIFAP (Esmeralda and Maravilla) are less affected in their yields by the presence of YR, while the two-row varieties, AC Metcalfe and ABI Voyager, along with Apizaco, are the varieties most affected by yellow rust due to their susceptibility.

Figure 1 Average behavior by location of the final severity of the disease in six barley varieties with and without the use of fungicides. A=Apizaco; DJ=Doña Josefa; E=Esmeralda; MA=Maravilla; AC ME=Metcalfe; ABI V=Voyager; Sf=Sin fungicida; Az+Teb=Azoxistrobin+Tebuconazole; Teb=Tebuconazole. 

The yields obtained (average percentage) when Azoxistrobin 11.1% + Tebuconazole 18.4% was applied were of 31, 24, 19, 18, 13 and 12 % in the varieties Apizaco, ABI Voyager, AC Metcalfe, Doña Josefa, Esmeralda and Maravilla, respectively, in comparison with not applying fungicide (Table 5). These results indicate that the varieties respond differently to these products and that they have different levels of resistance, although these differences do not statistically imply a higher yield.

YR has caused significant losses in barley around the world. ^{Marshall and Sutton (1995}) reported losses of up to 72% in susceptible varieties, whereas ^{Dubin and Stubbs (1986}) indicate that losses in yield were estimated in 30 to 70% in Andean countries. On the other hand, ^{Safavi et al. (2012}) reported average losses caused by this disease in yield components such as the weight of one thousand grains of 31% in susceptible genotypes, 3% in genotypes with resistance of a specific race, and 12% in lines with partial resistance, and for grain per spike, they reported losses of 19% (susceptible), 0.2% (specific race) and 8% (partial resistance). ^{Kumar et al. (2020}) reported losses in yield in Alberta, Canada, which range between 17.6 and 91% in susceptible varieties, and they point out that the application of fungicide is necessary when moderately resistant varieties are used and the disease prevail, which will reduce yield losses.

Yellow rust is currently becoming more frequent in Mexico and the varieties planted do not have satisfactory levels of resistance, therefore one of the strategies to avoid losses in yield and the subsequent low-quality grain is to use chemical control methods as a complement to genetic control, using the control rationally and responsibly, to avoid contributing to the deterioration of the natural resources and to the resistance of the pathogen. ^{Murray and Brennan (2010}) point out that the use of chemical products must be in the correct dose in order to avoid producing lower amounts of residues in the environment, but particularly to stop the pathogen from becoming resistant to the products applied. Likewise, ^{Walters et al. (2012}) mention that the control of diseases in barley depends, to a large extent, on the combined use of fungicides and resistant varieties, along with adequate agronomic practices. Therefore, to carry out an efficient control of YR in barley, it is necessary to consider diverse aspects, including the production environment, the variety to be planted and the fungicide product to be applied, which will contribute towards having a lower impact on the environment and a higher income for the farmer.

CONCLUSIONS

The fungicide Azoxistrobin 11.1 % + Tebuconazole 18.4% presented the best control of yellow rust in barley and caused an increase of 35% in grain yield. The use of fungicides gave the greatest grain test weight. The losses in grain yield due to the disease reached 53%. Chemical control is a feasible alternative for the control of yellow rust in the barley crop in rainfed environments.