The field experiment was conducted at the experimental farm of CSIR-Institute of Himalayan Bioresource Technology (32°06′05″N; 76°34′10″E), Palampur, India, during the cropping seasons of 2012–2013 and 2013–2014. According to the USDA soil taxonomy classification system, the soil of experimental area belongs to Alfisols (Sharma and Kumar, 2003). The experimental unit is situated at the altitude of 1393 m from mean sea-level. The amount and distribution of rainfall, maximum and minimum temperature, relative humidity, and sunshine hours during the two growing seasons were also presented (Figure 1). The soil of experimental plot was silty clay in texture, and the reaction of the soil was acidic. The physico-chemical properties of the experimental soil are presented in Table 1.

In this study, 5-year-old plantation of R. damascena (cv. Jwala) field was used, and the planting geometry was 1.5 m between rows and 0.75 m within rows. A basal dose of 100 kg nitrogen (N), 21.85 kg phosphorus (P), and 41.50 kg potassium (K) was applied by urea (46% N), single super phosphate (16% P₂O₅), and muriate of potash (60% K₂O), respectively, during both the cropping seasons. Irrigation was not given during the course of study, since the crop was grown under rain-fed conditions. However, other recommended agronomic practices for R. damascena were adopted as per requirement for better growth and development. The experiment was laid out in randomized block design (RBD) with two-factors factorial arrangement and three replications. Ten treatment combinations consisting two different types of pruning [complete pruning (C) and partial pruning (P)] and five different concentrations of MgSO₄ [water spray control (M₀), MgSO₄ @ 5.0g L^-1 (M₁), MgSO₄ @ 10.0g L^-1 (M₂), MgSO₄ @ 15.0g L^-1 (M₃), and MgSO₄ @ 20.0g L^-1 (M₄)] were tested. The pruning operation was done during 50th meteorological standard week (MSW) at 90 cm height from the ground level in 2012–2013 and 2013–2014. In case of partial pruning system, five new shoots were left without pruning in each bush, and the remaining shoots were pruned at 90 cm height from the ground level. The MgSO₄ solutions for different treatments were diluted with water (about 800 L ha^-1), and sprayed twice; the first foliar spray was applied at axillary bud development stage, and the second spray was done at flower bud appearance stage.

Two plants were randomly selected for each treatment from each replication, and the selected plants were tagged for growth and yield observation. After pruning, number of old shoots (No. bush^-1) was recorded. New shoot initiation rate (No. old shoot^-1) was also recorded. The data on number of flowers (No. shoot^-1), flower weight (g flower^-1), flower yield (g new shoot^-1 and g bush^-1) were recorded day-to-day basis from initial date of harvesting to end of the flowering. The flowers were harvested by manual picking in the morning (6:00–9:00 AM) to prevent the loss of volatile compounds from the flower.

The essential oil was extracted from fresh flowers harvested separately from each plot. The oil was extracted by hydro-distillation for 4 h on a Clevenger-type apparatus using a 5.0 L distillation system. The flower and water ratio was 1:2 (w/v). The essential oil from each plot sample was measured, and the oil content (w/w) in flower was expressed as percentage on a fresh weight basis. The extracted essential oil was dehydrated by anhydrous Na₂ SO₄ (Merck) and collected in a glass vial. The sealed oil samples were stored in a dark place at 4°C until analysis.

The compounds of oil samples were identified by using a Shimadzu QP2010 GC-MS system (Shimadzu, Tokyo, Japan) attached with an AOC-5000 auto injector and a ZB-5 (SGE International, Ringwood, VIC, Australia) fused silica capillary column (30 m × 0.25 mm i.d., and film thickness 0.25 μm). The conditions for analysis were identical to those previously described (Pal et al., 2014). The retention indices (RI) for all volatile compounds were computed by using homologous series of n-alkanes (C8–C24). Then the constituents of essential oil were identified by comparing their RI and mass spectra with those of authentic samples and with those stored in the NIST-MS (National Institute of Standards and Technology-mass spectral) database (Stein, 2005).

All the GC analyses of rose oil samples were carried out by a Shimadzu GC-2010 gas chromatograph (Shimadzu, Tokyo, Japan) equipped with flame ionization detector (FID) and a ZB-5 capillary column (30 m × 0.25 mm, fused silica, and film thickness 0.25 m). The operating conditions for analysis were identical to those previously described (Pal et al., 2014). The nitrogen gas was used as carrier with the velocity of 1.05 mL min^-1. Then the individual compounds were quantified based on peak area percentage of the chromatogram.

At the time of peak flowering stage, the leaves were collected from each experimental unit for estimation of Chl content. Chl was extracted from 200 mg fresh leaf tissue sample for each treatment in the solution of 80% acetone (v/v). The absorbance of the extracts at 645 and 663 nm was recorded with a spectrophotometer (model T 90 + UV/vis, PG Instrument Ltd.). Finally, the total Chl content (mg g^-1 tissue) was calculated based on the absorbance values as per standard equations (Arnon, 1949).

On the other hand, the leaves were collected from each experimental unit at the end of the both cropping seasons for the estimation of N, P, and K content in the leaves. After drying, the leaf samples were prepared with a laboratory grinder having a sieve spacing of 0.7 mm. For N estimation, the samples were digested with concentrated H₂SO₄ and a catalyst mixture of potassium sulfate and copper sulfate (10:1). Then, Kel Plus nitrogen analyzer unit was used for estimation of total N content in the leaves. In case of P and K estimation, a mixture of concentrated H₂SO₄ and perchloric acid (5:1) was used for digestion. Then, a spectrophotometer (model T 90 + UV/vis, PG Instrument Ltd.) and a flame photometer (model BWB XP, BWB technologies UK Ltd., UK) were used for the estimation of total P and K, respectively, as per standard procedure (Prasad et al., 2006).

The growth and yield data obtained from R. damascena for consecutive 2 years were subjected to analysis of variance (ANOVA) to test the sole effect of pruning system and foliar application of MgSO₄, and pruning system × MgSO₄ interaction by using Statistica 7 software (Stat Soft Inc., Tulsa, OK, USA). In this experiment, a two-factor factorial RBD was used with three replications. Differences among the treatment means were assessed by the least significant difference (LSD) value at P = 0.05. Correlation matrix was conducted by using Statistica 7 software to investigate relationships between the yield and yield attributes. The regression equation between yield and MgSO₄ doses was also developed using same software. However, the heat maps of chemical profiling of essential oil were prepared with the help of R data analysis software (version 3.1.3). Principal component analysis (PCA) was performed to evaluate the influences of treatment combinations on chemical profiling of essential oil as a bi-plot, and nature of variations among the treatment combinations was also projected. The factor loading values represent the correlations of each variable with the principal components (PCs).

The analyzed data revealed that two main yield attributes of R. damascena, new shoot initiation rate (No. old shoot^-1) and number of flowers (No. bush^-1), were significantly (P ≤ 0.05) influenced by the system of pruning during both the years (Table 2). In this study, partial pruning system registered significantly (P ≤ 0.05) higher new shoot initiation rate (4.63 and 10.80 No. old shoot^-1) and number of flowers (350.47 and 266.80 No. bush^-1) compared with the complete system, irrespective of foliar application of MgSO₄. However, number of petals (No. flower^-1) and flower weight (g flower^-1) were not influenced by the system of pruning during both the years, and these two parameters remained inconsistent over the years.

The effects of foliar application of MgSO₄ on new shoot initiation rate (No. old shoot^-1), number of petals (No. flower^-1), and flower weight (g flower^-1) were not significant (P ≥ 0.05). However, number of flowers (no. bush^-1), the important yield component of R. damascena, was significantly (P ≤ 0.05) influenced by the foliar application of MgSO₄; and maximum numbers of flowers (357.50 and 250.67 no. bush^-1) were recorded with MgSO₄ @ 15.0g L^-1 (Table 2). Irrespective of system of pruning, the application of MgSO₄ @ 15.0g L^-1 registered about 30 and 36% higher number of flowers compared with water spray control during 2012–2013 and 2013–2014, respectively. The percentage of blind shoots was also significantly (P ≤ 0.05) affected by the system of pruning during 2013–2014 cropping season; however, the effect of foliar application of MgSO₄ on blind shoot (%) was insignificant, least percentages of blind shoot (14.15 and 20.71%) were recorded with the foliar application of MgSO₄ @ 10.0g L^-1 during both the years. The interaction effects of pruning system and foliar application of MgSO₄ on blind shoot were significant (P ≤ 0.05) during 2012–2013 (Table 2).

The analyzed data revealed that the effects of pruning system and foliar application of MgSO₄ on the flower yield (g bush^-1) of R. damascena were significant (P ≤ 0.05) in both the years (Table 2). Regardless of foliar application of MgSO₄, the partial pruning system increased flower yield by about 23 and 75% compared with complete pruning system during 2012–2013 and 2013–2014, respectively. Among the foliar treatments of MgSO₄, the application of MgSO₄ @ 15.0g L^-1 recorded significantly (P ≤ 0.05) higher flower yield (1114.47 and 830.69 g bush^-1) compared with water spray control during both the years (Table 2). Though the effects of the foliar application of MgSO₄ @ 5.0g L^-1, MgSO₄ @ 10.0g L^-1, and MgSO₄ @ 15.0g L^-1 on flower yield were statistically at par (P ≤ 0.05) in 2012–2013; significantly (P ≤ 0.05) higher flower yield was recorded with MgSO₄ @ 15.0g L^-1 compared with MgSO₄ @ 5.0g L^-1 and MgSO₄ @ 10.0g L^-1 in 2013–2014. Regardless pruning system, the foliar application of MgSO₄ @ 15.0g L^-1 registered about 26 and 38% higher flower yield compared with water spray control in 2012–2013 and 2013–2014, respectively. The interaction effect between pruning system and foliar application of MgSO₄ on the flower yield (g bush^-1) was significant (P ≤ 0.05) in 2012–2013 cropping season.

The regression equations were established considering MgSO₄ doses as independent variable and flower yield and number of flowers as dependent variable (Figure 2). In this experiment, the flower yield was increased with the corresponding increasing concentration of MgSO₄ doses up to 15.0 g L^-1, thereafter the yield was declined. Thus, the second degree polynomial relationship (y = 721.31 + 25.707 x - 0.894 x²; R² = 0.752) was found between flower yield and level of MgSO₄ doses (Figure 2). The similar relation was found between number of flower and concentration of MgSO₄ doses with an equation of y = 223.41 + 8.123 x - 0.275 x², and R² = 0.774. The correlation matrix among the flower yield, yield attributes, and growth parameter was also established, and the data revealed that flower yield (g bush^-1) was significantly and positively correlated with the number of flower (r = 1.00; P ≤ 0.01) and new shoot initiation rate (r = 0.74; P ≤ 0.05) (Figure 3). However, a negative correlation was observed with the percentage of blind shoot (r = -0.76; P ≤ 0.05). The negative correlation (r = -0.77; P ≤ 0.01) was also found between the percentage of blind shoot and new shoot initiation rate. A significant (P ≤ 0.01) and a positive correlation (r = 0.78) was found between the flower weight (g flower^-1) and number of petal (Figure 3).

In our study there was no significant effect (P ≥ 0.05) of pruning system and foliar application of MgSO₄ on the essential oil content (%) in the flowers of R. damascena during both the cropping seasons (Figures 4A,B). However, regardless of foliar spray of MgSO₄, complete pruning system registered slightly higher oil content (0.046 and 0.042%) compared with the partial pruning system. Among the foliar treatments, the applications of MgSO₄@ 5.0g L^-1, @ 10.0g L^-1, and @ 15.0g L^-1 registered highest oil content (0.046%) in 2012–2013 cropping seasons irrespective of pruning system (Figure 4B). The lowest oil content (0.040 and 0.044%) was recorded with water spray control and higher doses of MgSO₄ (@ 20.0g L^-1) in both the years.

In this experiment, we have identified a total of 33 compounds in 2012–2013, which contributed about 93–98% of the total volume; whereas 30 compounds were identified in 2013–2014 (Figures 5A,B). The lowest numbers (26 and 23) of compounds were identified with the interaction effects of partial pruning system and water spray control treatment during both the years. However, the maximum contribution (97.25 and 97.54%) by the identified compounds in total volume of essential oil was observed with interactive effects of complete pruning system and foliar application of MgSO₄ @ 20.0g L^-1 and water spray.

The chemical profiles of rose essential oil under different treatment combinations are presented by heat maps (Figures 5C,D). The heat maps containing 33 and 30 compounds which depict the changes of chemical profiling of rose essential oil were observed due to pruning system and foliar application of MgSO₄. The data in the heat map showed that the accumulation patterns of two major monoterpenoids, citronellol, and E-geraniol, were inconsistent over the years (Figures 5C,D). However, the maximum shearing of citronellol (39.82%) and E-geraniol (29.68%) was observed with PM₀ (partial pruning with water spray control) and PM₃ (partial pruning with foliar application of MgSO₄ @ 15.0g L^-1), respectively, in 2012–2013 cropping seasons (Figure 5C). In 2013–2014, CM₄ (complete pruning with foliar application of MgSO₄ @ 20.0g L^-1) and PM₃ (partial pruning with foliar application of MgSO₄ @ 15.0g L^-1) registered maximum quantity of citronellol (48.88%) and E-geraniol (24.84%), respectively (Figure 5D). Flowers obtained from the PM₂ (partial pruning with MgSO₄ @ 10.0g L^-1) treatment registered highest concentration (1.27 and 1.45%) of linalool during both the years. In the present investigation, the two major hydrocarbons, nonadecane, and heneicosane, also show diverse accumulation patterns under different treatment combinations in both the years (Figures 5C,D). The minimal level of nonadecane accumulation was recorded with CM₁ (complete pruning with foliar application of MgSO₄ @ 5.0g L^-1) followed by PM₂ (partial pruning with foliar application of MgSO₄ @ 10.0g L^-1) in 2012–2013 season, whereas in second cropping season, the minimum shearing of nonadecane was recorded with higher concentration of MgSO₄ @ 20.0g L^-1) under both types of pruning system (Figures 5C,D).

The results presented in the Table 3 revealed that the effects of pruning system on the total Chl content in leaves were insignificant (P ≥ 0.05) during 2012–2013; however, partial pruning system registered significantly (P ≤ 0.05) higher Chl content (3.58 mg g^-1) in 2013–2014. Although the effect of foliar application of MgSO₄ on total Chl content in leaves was insignificant (P ≥ 0.05) in both the years, the Chl concentration was gradually increased with corresponding increasing concentration of MgSO₄ and the utmost value (3.07 and 3.45 mg g^-1) was attained with MgSO₄ @ 15.0g L^-1 during both the years. We also observed that the accumulation of N, P, and K in the leaf was significantly (P ≤ 0.05) influenced by pruning system, and the maximum values were registered with the partial pruning system during 2013–2014 cropping season. On the other hand, irrespective of pruning system, the effects of foliar application of MgSO₄ on the accumulation of N and P were insignificant (P ≥ 0.05) in both the years. However, K concentration in leaf was significantly (P ≤ 0.05) influenced by the foliar application of MgSO₄ in 2012–2013 cropping season, and the maximum value was recorded with water spray control. The Mg concentration in leaves was not influenced by the system of pruning during both the years. However, the effect of foliar application of MgSO₄ on Mg accumulation in leaves was significant (P ≤ 0.05) during 2012–2013 cropping season and the maximum value was observed with MgSO4 @ 15.0g L^-1. In case of interaction effect between pruning system and foliar application of MgSO₄, the insignificant (P ≥ 0.05) results were found in both the cropping seasons (Table 3).

The principal component analyses (PCA) were performed by using the sets of 20 and 17 compounds of essential oil for 2012–2013 and 2013–2014, respectively. The components, which are quantified in all the treatments, are used for PCA. The results from PCA revealed that the first component (PC₁) and second component (PC₂) jointly explained 83.25 and 68.24% of the total variations for 2012–2013 and 2013–2014 cropping seasons, respectively (Figure 6). The relationships among the variables in the space of the first two components (PC₁ and PC₂) with factor loadings are presented in Figures 6A,C, and indicate how each variable contributes to the PCs. In this experiment, PC₁ has positive coefficients with α-pinene, β-pinene, myrcene, linalool, phenyl ethyl acetate, citronellol, E-geraniol, eugenol, geranyl, acetate, methyl eugenol, and α-guaiene in first cropping season (Figure 6A). However, only four compounds (linalool, citronellol, trans-geraniol, citronellyl acetate) have positive coefficients with PC₁ in 2013–2014 (Figure 6C). On the other hand, highly negative loading values were found with heptadecane (-0.97 and -0.82), octadecane (-0.98 and -0.84), nonadecene (-0.92 and -0.89), nonadecane (-0.97 and -0.97), heneicosane (-0.95 and (-0.94), and tricosane (-0.92 and -0.71) in both the cropping seasons (Figures 6A,C).

The PCA bi-plots also indicated that there were three and four distinct groups among the treatment combinations during 2012–2013 and 2013–2014, respectively. The treatment PM₀ (partial pruning with water spray control) and PM₁ (partial pruning with foliar application of MgSO₄ @ 5.0g L^-1) were separated by the PC₁ from the rest of the treatments and placed in the negative end of the PC₁ in 2012–2013 (Figure 6B). These two treatments were separated from each other by PC₂. In 2013–2014, the treatments PM₂ (partial pruning with foliar application of MgSO₄ @ 10.0g L^-1), PM₃ (partial pruning with foliar application of MgSO₄ @ 15.0g L^-1), and CM₄ (complete pruning with foliar application of MgSO₄ @ 20.0g L^-1) were separated from rest of the treatments, and they formed individual group (Figure 6B).