Gladiolus (Gladiolus communis L.) is considered a rust-free crop in the northern part of the State of Mexico (^{CESAVEM, 2013}). However, rust has been unofficially reported in the last crop cycles, but there is no information available on damages or about its spatial distribution.

Field tours have confirmed the presence of yellowish-orange pustules which develop across leaf veins. These are symptoms typically caused by the fungus Uromyces transversalis (Thumen) G. Winter. Though this pathogen mainly attacks gladiolus crops, it has also been detected in other members of the Iridiaceae family, including Crocosmia, Freesia, Melasphaerula, Tritonia and Watsonia (^{Schubert et al., 2006}). The fungus mainly attacks the leaves and lowers quality of flowers for marketing, causing strong economic damages to production and reducing exports (^{Valencia-Botin et al., 2013}). When damages exceed 60%, the flowers commercial value is totally lost; if the infection occurs at early stages, the quality of flowers and their corm thickness may be diminished. When the disease emerges at late stages it favors inoculum spreading through infested corms to other plots or regions (^{SENASICA, 2008}). The fungus can be disseminated by the wind, infected leaves and stalks, and by corms, rhizomes and flowers infected with rust spores (^{SENASICA, 2008}).

To develop strategies for managing this emerging disease, it is necessary to carry out research on its time-space dynamics. Through this research the spatial behavior of gladiolus rust in the municipality of Jiquipilco, State of Mexico, was determined, and maps of its spatial distribution by plot were developed.

Sampling. In the 2013 and 2014 winter-summer and summer-fall cycles, two 1-hectare commercial plots were evaluated in the municipality of Jiquipilco, State of Mexico. Plots were sown with the Roja Borrega variety using 0.80 m row spacing and a density of 250,000 plants/ha. In the winter-summer cycle sampling was performed 60 days after sowing (DAS, growth stage) 80 DAS (bract stage) and 113 DAS (flowering stage). In the summer-fall cycle, sampling was performed at 78, 103 y 120 DAS. For this, a 100-m wire mesh was placed on each side of the plots; then, sampling points at 10-m intervals were established for a total of 121 per plot. The severity and recurrence of the disease in a plant was determined at each point. The Pustules of the fungus were identifiable for the development horizontally across the veins, perpendicualr of infected leaf. Samples of the infected tissue were taken for morphological recognition of the structures in the lab. Rust severity was estimated using a 1-6 scale: 1 (healthy with no visible symptoms), 2 (<15 % leaf area damaged), 3 (16 a 30 %), 4 (31 to 45 %), 5 (46 to 60 %) and 6 (>60 %).

Geostatistical analysis. Plants at each sampling point were geographically located using a Differential Global Positioning System (dGPS) (Model SPS351, Trimble). The experimental semivariogram was calculated according to the model of ^{Isaaks and Srivastava (1989)}; the best model with the best fit was selected and the spatial interpolation was performed using the kriging method. The experimental semivariogram was developed using VarioWin 2.2 (Software for Spatial Data Analysis in 2D).

The spatial dependency level was calculated by dividing the nugget effect by the sill value (^{López-Granados et al., 2002}). Estimates of the spatial distribution of the disease at different phenological stages were mapped using Surfer 9.0 (Surface Mapping System, Golden Software Inc. 809, 14^th Street Golden, CO).

Detection. No rust was detected in the 2013 gladiolus production cycles or the 2014 winter-summer cycle. In the 2014 summer-fall cycle the disease was detected in the two evaluated plots. In plot 2, gladiolus rust was detected starting at the bract stage, while in plot 1 it was observed at flowering. Rust occurrence was associated to leaf wetness and an average temperature of 14.6 ºC (INIFAP, 2015, which favor uredospores germination and survival (^{Aloj et al., 1981}; ^{Peterson and Berner, 2009}), as well as a reduction in the frequency with which farmers apply fungicides.

Symptoms observed included yellowish-orange pustules on leaves, located horizontally across leaf veins. The increased level of incidence from the sphate stage to the flowering stage in plot 1 (Table 1) suggests that the disease was able to spread from plant to plant within the same plot due to production of new inoculum. Microscopic examinations showed ovoid, ellipsoidal or oblong uredospores with a hyaline cell wall and isolated dark germinal pores, which are similar to those reported by ^{Rodríguez-Alvarado et al. (2006)}, ^{Blomquist and Thomas (2007)}. This is the first time the presence of gladiolus rust is reported in the northern region of the State of Mexico.

Final disease incidence was 30.6 % for plot 1 and 69.4 % for plot 2 at 120 DAS (Table 1); this was lower than those reported by ^{Blomquist and Thomas (2007)} in California, and by ^{Rodríguez-Alvarado et al. (2006)} in Michoacán. Intermediate severity in plot 1 was 1.4 and 2.5 in plot 2.5 (Table 1).

Table 1 Incidence, severity and statistical values and parameters of the models adjusted to semiovariograms of gladiolus rust sampling in Jiquipilco, Mexico, in the 2014 summer-fall cycle. 

^aDAS: Days after sowing.

Rust emergence in the studied region may have been caused by spores disseminated in corms carried by farmers to sow new fields, yet spores can also be carried long distances by the wind (^{Agrios, 2004}; ^{SENASICA, 2008}). Rust is being detected at the same time that Hemileia vastatrix is being reported in coffee crops in Mexico (^{Mora-Aguilera et al., 2014}) and Central America. It might be that changes in weather patterns contribute to quick germination of uredospores and therefore a greater production of gladiolus rust inoculum (^{Peterson and Berner, 2009}).

Semivariograms resulting from samplings were adjusted to the spherical theoretical model (Cuadro 1), which shows the level of rust severity, and were expressed as specific points within the plot compared to the rest of the sampled points. Results suggest that the disease could have developed from contaminated materials (^{Roberto et al., 2002}) such as corms (^{SENASICA, 2008}), that are the primary source of inoculum. The high spatial dependence (Table 1) observed suggests that the severity of the disease shows an aggregated distribution among the different sampling points in the plot, which is explained by the estimated semivariograms.

The maps we developed allowed us to visualize the spatial behavior of gladiolus rust severity in aggregation points and disease gradients in the two plots in 2014 (Figure 1). The map of plot 1 at 120 DAS showed the presence of aggregation points distributed across the eastern part of the plot, from the northern part to the southern part; there was a slight slope in that part. On the other hand, maps of plot 2 showed a greater number of aggregation points compared to map developed during the previous bract stage (Figure 1). Aggregation points were located in the northern part of the map, from the eastern to the western area. The aggregation points in this area of the map were observed to have greater intensity.

Figure 1 Semivariograms and maps the spatial behavior of gladiolus rust in different sampling dates in Jiquipilco, State of Mexico. DAS: days after sowing; S2: variance of estimated values 

The aggregation observed suggests that the management strategy used by farmers in the northern region of the State of Mexico, which includes spraying the entire gladiolus plot, can be replaced by a specific site or point disease management (^{Carvalho et al., 2009}). This means carrying out an analysis of the spatial patterns of the disease. Also, spatial aggregation suggests that different control measures should be taken and sampling activities should be conducted in specific areas or sites (^{Navas-Cortés et al., 2008}) where the disease is present.

The use of aggregation maps, such as those developed for gladiolus rust, allows targeting control strategies precisely to specific areas of infestation, as well as sources of inoculum (infected corm, among others) which cause the disease to spread. This represents the basis of site specific management, which would result in cost savings and less environmental impact before the disease spreads. The application of contact fungicides up to the bract stage, removal of damaged plants or leaves in the specific infestation points, especially in the early stages of the disease and targeting the sampling activities justify the use of precision agriculture techniques to control damages caused by gladiolus rust in the northern region of the State of Mexico.