Degree of infestation

Mite infestation per plant was statistically significant (α≤ 0.05) and two groups of contrasting cultivars (Figure 1) were identified. In the first group cv. catalina was the most susceptible to the mite; the second group included ambiance as the toughest. This is contradictory because in the current rose cultivated a reduction should be expected in genetic variation due to advanced process of domestication and improvement linked to only 7 of more than 100 species used (^{Martin et al., 2001}), but its many and complex hybridizations are likely to be the main factor of this variability.

Figure 1 Number of mites per plant recorded 15 rose cultivars from flower bud formation to the phenological stage of “large peas”. On the bars the standard deviation of the mean is indicated. Bars with the same letter are the same significantly (Tukey, α= 0.05). 

In Mexico enterprises engaged in production of pink do research that are related to the development of new cultivars from the exploitation of genetic variation; no studies on self-defense of the plant either directly, through synthesis products, or indirectly attraction of natural enemies, against herbivores are known (^{Agrawal et al., 2002}; ^{Boege y Marquis, 2005}). All cultivars are imported, which increases production costs.

Environmental variables associated with the mite population development

Variations in temperature, relative humidity and their interaction affected the population growth of mites per plant per season evaluated (Figure 2). In spring 2012 there was a temperature of 26 °C and a relative humidity of 60%; both they favored the increase in the population of T. urticae. These results coincide with those of other studies where the intervals 25-28 °C and 60-70% represented the optimum development of this pest conditions (^{Kheradpir et al., 2007}). In spring 2010 the average temperature was very similar to 2012, but the relative humidity was 56.5%. In autumn 2010 the temperature and relative humidity were 20.7 °C and 58% so the number of mites per plant was below those recorded in the other two production cycles (Figure 2).

Figure 2 Relationship between number of mites per plant, temperature and relative humidity in the variety catalina recorded in the three experiments. 

The optimal development mite temperatures also promote the increase of females (^{Rasmy et al., 2011}); perhaps the differences observed between spring 2012 with the other two production cycles attributed to an increase in them. Another environmental factor that can affect mite population growth is the fertilization of the plant (^{Chow et al., 2009}; ^{Flores et al., 2011}).

The differential response in the mite population development, also could be affected by defense mechanisms of the plant as physical barriers or metabolic products. The total phenolic content measured in some cultivars (Figure 3), elected by their means and extreme values in incidence of mites (Figure 1) also showed natural variation; could be detected tend to be higher in cultivars with lower incidence of red spider.

Figure 3 Contents of total phenols in photosynthetically active leaves of flowering stems of certain rose in the phenological stages arrocillo, marble and cut point. On the bars the standard deviation of error is indicated. 

Other authors have reported that phenolic compounds can be constitutively in plants or synthesized in response to infection pest (Gardner et al., 1999), where they accumulate more and more rapidly resistant mite species, with the presence of a pathogen (^{Khatun et al., 2009}), in healthy leaves (^{Coruh and Ercisli, 2010}), or even the agronomic management and environmental variables present; their presence also represent an evolutionary gain delay damage in adverse to the plant environmental conditions (Gardner et al., 1999).

Other biochemical factors of primary metabolism (anti-stress molecules) and secondary (terpenes, tannins and essential oils) constitutive and induced character and morphological structures (stomata, waxes, glands, sheet thickness, etc.) can also influence attracting or repelling mite (^{Graham et al., 2008}; ^{Mumm and Dicke, 2010}; ^{Flores et al., 2011}), all within a complex relationship between parasite and host must individually explored (^{Hernandez et al., 2002}; ^{Vivanco et al., 2005}). All these elements can be complementary criteria in integrated crop management and genetic improvement (^{Chow et al., 2009}). Recent studies, in addition to environmental management, report the use of volatile compounds from plant leaves as Cleome gynandra L./ Gynandropsis gynandra (L.) Briq.), which in co-culture with Rosa x hybrida for cut flower production without affecting its performance and quality, it has substantially reduced the populations of T. urticae (^{Nyalala et al., 2013}).

Variables associated with quality floral stem

Highly significant differences among cultivars (α= 0.01) were observed in all variables associated with quality floral stem, which directly reflect the existing phenotypic variability. The characteristics with greater range of variation among cultivars were number of petals, leaf area, stem length, productive stems, blind stems, number of leaflets and stem diameter (NP, AF, LT, TP, TC, NF, DT; Table 2). The environmental effects associated with the three production cycles also affected the development of cultivars to be observed highly significant differences in all variables except AF. The significant interaction C x CU observed in all variables except the chlorophyll content, suggest a differential response of cultivars to environmental factors prevailing in the three cycles of growth and development; in cv. beautiful pearl, the lowest temperatures observed in autumn 2010 in relation to the other two cycles, originated slow growth of the flower stem and had a greater number of short stems without flower (blind stems), probably also affected by the photoperiod. The cutoff point average in all cultivars also extended phenological cycle in 20 days.

Table 2 Mean squares and statistical significance of the F values of three experiments established for evaluating rose cultivars. 

Variations in chlorophyll contents among cultivars were not significant; this fact could be considered the less variable varietal effect influencing the color tone of the leaves, but a greater dispersion in tissue damage was observed with mites, which highlights the decrease of photosynthetic activity in certain cultivars. Similar results were also observed in coffee leaves (coffea sp; ^{Neves et al., 2006}), also reports synthesis of metabolic products.

Mites vs physiological variables

Population growth mite (NA) in all cultivars rose was correlated negatively and significantly with some quality variables f lower stem as stem length, stem diameter and leaf area (LT, DT and AF; Table 3). The NA with the contents of chlorophyll in the leaves also showed a similar trend.

Table 3 Correlation coefficients between the variables evaluated. 

^{Golawska et al. (2010)}), ^{Calatayud et al. (2008)}, ^{Neves et al. (2006)} and ^{Stone et al. (2001)} observed a significant decrease in the values of physiological variables of the plant to increase the plague poblacionalmente, with direct effects on the commercial quality of the flower stalk (^{Tjosvold and Chaney, 2001}). The interruption in the development of floral structure increases with population growth of T. urticae, but not significantly; probably the soil and climate factors and their interaction with the type of farming have a greater effect on these responses.

Cluster analysis

The data for the 15 cultivars rose and the evaluated variables were subjected to a multivariate analysis method unweighted arithmetic average (UPGMA Method). In this, considering an average distance of 4.17 ligation five groups (Table 4) were identified in group 1 it was only observed in cv. Friendship (FS), improved in France, bicolor flower in red and cream, and tolerance level red spider statistically equal to that shown by the best cultivars, catalina, voodoo and freedom (Figures 1 and 3). The second group contains cv. catalina and voodoo which were characterized by hosting the largest number of mites (NA, Figure 1). The cv. catalina was obtained by breeders in New Zealand and in terms of geographical distance represents the place of isolation from the places where the other cultivars studied, which was in European countries (Table 1) were developed. The cv. voodoo, developed in the Netherlands, is reported in California, United States of America as resistant to diseases (^{Anonymous 3, 1978}). The third group contains cv. poison, freedom and grand gala originating in the Netherlands, Germany and France, respectively; in the last two, red flower, no information about their parents was found, so it is unknown whether both and poison, purple flower, are related.

Table 4 Mean variables evaluated 15 cultivars pink, grouped by cluster analysis. 

G= grupos; Pr= promedio; NA= número de ácaros por planta; NP= número de pétalos; LT= longitud del tallo floral; TP= tallos productivos; TC= tallos ciegos; NF= número de foliolos; DT= diámetro de tallo floral; AF= área foliar; CSD= clorofila sin daño; CCD= clorofila con daño.

The fourth group was integrated with samourai and beautiful pearl, developed in France by different companies and contrasting progenitors and flower color. Both cultivars formed the group with the least damage tolerance caused by mites (Figure 1). The fifth group represents almost 50% of the cultivars evaluated; vega, sweet vita, vision, polar star, polo, and baccara royal ambiance are five tones contrasting flower color. The cv. and polar pole star, white flower, formed a subgroup but their parents are unknown, it is not possible to infer whether there is relationship between them. Based on the perception of farmers in the study area cultivar pole it is one of the most susceptible to pests and diseases and this polar star and were developed by the same company in Germany.

The 15 cultivars analyzed in the dendrogram of Figure 4 belong to the group of hybrid tea, considered the most representative in the category of modern roses, obtained by hybridizing rose’s tea groups and hybrid perpetual category of roses old. Roses present study, strictly classified as hybrid tea, represent a complex mixture reductively located in a narrow range of a wide genetic variation of more than 100 cultivars distributed in 13 groups of categories of modern roses and old (^{Martin et al., 2001}). The principal components (CP) 1, 2 and 3 accounted for 72.5% of the original total variation; almost 50% was represented in the biplot using the CP 1 and 2 (Figure 5).

Figure 4 Dendogram to 15 rose cultivars obtained by the average linkage method (UPGMA method). FS= friendship; VO= voodoo; CA= catalina; PN= poison; GG= grand gala; FR= freedom; SA= samourai; BP= beautiful pearl; VE= vega; DV= sweet vita; VI= vision; PS= pole star; PO= pole; RB= royal baccara; AB= ambiance. 

Figure 5 Biplot 15 rose cultivars (capital letters) and 11 variables (lowercase letters). 

According to Sanchez (1995) and González et al. (2010) suggest these percentages reliability in the interpretation of the approximate correlations that can be detected among cultivars and variables evaluated in this study. In the three groups BIPLOT cultivars were observed for similar values identified by their proximity within a single quadrant. The first group included freedom cultivars, poison and grand gala (FR, PN and GG); the second group cultivars were the polar star, beautiful pearl samourai (PS, SA and BP); group three included vega, vision, sweet vita, ambiance and royal baccara (VE. VI, DV, AB and RB). The voodoo cultivars, chainring and friendship (VO, CA and FS) dissimilar in their scores on both CP 1 and 2, were not included in the above groups.

Principal component analysis

The CP1 is mainly explained by stem length (lt) and the number of leaflets (fol); CP2 was related essentially with number of mites (na) and productive stems (tp); CP3 and the highest scores were recorded in number of petals (np) and blind stems (TC). The CA, AM, SA, FR and GG cultivars had the highest scores in the CP1; CP2 was observed in CA and AB had the highest dispersion in the biplot graph and in the same situation CP3 detected with BO and PO (Figure 5).

In this biplot you can also note that CA, BO, FS and VE had the largest number of mites (NA) and productive stems. SA, GG, FR, BP, PN, PO and PS were the most outstanding in stem length (lt), number of leaflets (fo), leaf area (af), stem diameter (dt), unlike chlorophyll (ddc), number of petals (np) and blind stems (tc). RB, AB, DV and VI are mainly grouped by their content of chlorophyll damage (ccd) without damage (csd).