Non-destructive estimation of chlorophyll and nitrogen content in leaf of Rosa damascena (Mill)

Abstract

Chlorophyll (Chl) and nitrogen (N) status of leaves provide valuable information about the physiological condition of plants. The conventional methods for measuring Chl and N contents in leaves are destructive, costly, time-consuming, and do not allow repetitive measurement of the same sample. The Damask Rose (Rosa damascena Mill) is an important aromatic crop in the western Himalaya region in India. Generally, flower yield and oil yield of the Damask rose are correlated with nitrogen, phosphorus, and potassium (NPK) levels in the leaf at the bud development stage. The dynamics of N within the rose plant have not been reported clearly. Thus, there is a pressing need for non-destructive techniques to estimate Chl and N content in the leaf of the Damask rose. Our objective was to establish an appropriate mathematical relationship between the Chl content index (CCI) value and the total Chl/N contents for non-destructive estimation of total Chl and N in the leaf of the Damask rose. The regression models were developed with destructively measured parameters (total Chl and N) as the dependent variable and a parameter derived from CCM-200 as the independent variable (CCI). We found that polynomial regression models are suitable for non-destructive estimation of total Chl, and the model predicted values were very close to traditionally measured values with a root mean square prediction error (RMSEp) less than 0.20 mg g⁻¹ of Chl. In the case of N estimation, a power regression model was appropriate with lowest Akaike's information criteria (AIC) and root mean square validated error (RMSEv) value. Significant correlations (P ≤ 0.001) were observed between traditionally measured values and our model predicted values in both cases.

Keywords:

• chlorophyll
• Damask rose
• model
• nitrogen
• non-destructive.

Introduction

Chlorophyll (Chl), the essential green pigment of plants, harvests solar energy and converts it into chemical energy through the process of photosynthesis. The absorption of solar radiation by a leaf is a function of the photosynthetic pigment content in the leaf (Foyer et al. 1982; Steele et al. 2008). Chl tends to decline rapidly when plants are under stress or during leaf senescence (Hendry et al. 1987; Penuelas and Filella 1998). On the other hand, nitrogen (N) deficiency decreases Chl status in leaf (Kowalczyk-Jusko and Koscik 2002). So, Chl status furnishes an indirect evaluation of N status because a considerable amount of leaf N is incorporated in Chl (Moran et al. 2000). A low concentration of Chl in the leaf also reduces photosynthetic activity and primary production (Filella et al. 1995; Richardson et al. 2002). Since Chl and N content status provide such valuable information about plants, these parameters may be used to assess the overall photosynthetic capacity of the canopy and productivity of the plant.

The conventional methods of Chl and N analysis require the destruction of the leaf sample, and prevent studies on the same leaf over time. It is also time-consuming and expensive to process the samples. The most important aim of precision agriculture is optimization of field-level nutrient management while reducing environmental risks and the footprint of farming. The conventional techniques for assessing Chl and N status in plants cannot satisfy the requirements of precision farming. Non-destructive optical techniques based on leaf absorbance and reflectance of light by leaves have been proven as alternative time-saving and simple techniques to quantify Chl in a number of agricultural species such as birch (Betula pendula Roth.), wheat (Triticum aestivum L.) and potato (Solanum tuberosum L.) (Uddling et al. 2007), maize (Zea mays L.) and soybeans (Glycine max L. Merrill.) (Gitelson et al. 2005), sorghum (Sorghum bicolor L. Moench.) and pigeonpea (Cajanus cajan L. Millsp.) (Yamamoto et al. 2002). Strong relationship between CCI and nitrogen was also reported in sugar maple (Acer saccharum Morsh.) (Van den Berg and Perkins 2004), and rice (Oryza sativa L.) (Peng et al. 1993).

Out of 200 species of the genus Rosa, Rosa damascena (Mill) is recognized for its high-value essential oil, which is used in the pharmaceutical, flavouring and fragrance industries (Lawrence 1991). Flower yield and oil yield of the rose are correlated with nitrogen, phosphorus, and potassium (NPK) levels in the leaf at the bud development stage (Orlova 1984). The concentration of N, P and K decreases by 35%, 67% and 25% respectively, between bud development and the main flowering stage (Weiss 1997). Rationally, the Damask rose requires 200 kg N ha⁻¹ which is applied in two equal doses, half at the time of pruning and half 20 d later. The dynamics of N within the Damask rose plant have not been reported clearly. Hence, the non-destructive techniques for estimation of Chl and N will help to understand the dynamic of N and physiological status within the Damask rose plant. Furthermore, it will be helpful as a decision-making support tool for precision farming.

The CCM-200 leaf chlorophyll meter (Opti- Sciences, USA), one of several hand-held Chl meters available on the market today, measures leaf transmittance at two different wavelengths of light: c. 660 nm (red) and c. 940 nm (near-infrared). But this instrument generates a Chl content index (CCI) value that indicates relative Chl content but not absolute Chl content per unit leaf area, or concentration per unit of leaf mass. Thus, it is necessary to establish a mathematical relationship between meter output and actual Chl concentration, so that a direct estimation of Chl may be possible. It may also be used to determine N content because Chl concentration is correlated with N concentration in the leaf.

The mathematical relationships between CCI value and total Chl or N contents vary with species (Schaper and Chacko 1991; Yamamoto et al. 2002) and growing conditions (Bullock and Anderson 1998). Thus, an independent mathematical model should be established for each species for a particular zone. To the best of our knowledge, there is no non-distractive method to evaluate Chl and N status of the Damask rose. Thus, the objective of this study was to establish the non-destructive estimation of total Chl and N in the leaf of the Damask rose.

Materials and Methods

Plant material

This study was conducted at the experimental field of the Institute of Himalayan Bioresource Technology, in Palampur, India. Rosa damascena Mill (cv. Jawla) leaves were used as plant material. Fertilizers were applied at the rate of 200 kg N, 100 kg phosphorus pentoxide (P₂O₅) and 50 kg potassium oxide (K₂O) ha⁻¹y⁻¹. One hundred leaf samples with as wide a range of Chl contents as possible, from very pale green to very dark green, were collected from fields which were planted on the same day. The sample leaves were taken from different part of the plant. Sixty out of 100 leaves were used for establishing and validating the mathematical relationship between CCI and Chl content. The remaining 40 leaf samples were used to establish the regression equation of N% (dependent) on CCI reading (independent) and to validate the equation. We randomly separated the sample into two equal parts in both cases. Half of the samples were used for developing the mathematical models and half were used to validate the models.

CCM reading, Chl extraction and N analysis

Using the CCM-200 (Opti- Sciences, USA) hand-held Chl meter, five readings were recorded from each of the 100 leaf samples. The CCM-200 then calculates a CCI. The arithmetic mean of these five CCI readings was calculated to establish a single representative CCI value per leaf. Once CCI readings were completed, major veins were discarded, and a 200-mg leaf sample was separated from each leaf for extracting the Chl. Chl was extracted simultaneously from leaves in a solution of 80% acetone (v/v). Finally, absorbance of extracts was recorded with a spectrophotometer (model T 90 + uv/vis PG Instrument Ltd.) at 645 and 663 nm. Total Chl (mg g⁻¹) was calculated from the absorbance values as per standard equations suggested by Arnon (1949).

In the case of N, after recording CCI leaf samples were placed in an oven at 65°C until dry weight was constant. Then the dried leaf materials were ground for determination of N content. The Kjeldahl method was used to measure N content in percentage of dry weight of leaf samples.

Data analysis

In our quest to find the appropriate mathematical models for non-destructive estimation of Chl and N contents in leaves, we established different regression equations (e.g. linear, power, exponential, and polynomial) between laboratory analysis values (dependent variable) and CCI values (independent variable). We used the root mean square calibration error (RMSE _c ) and coefficient of determination (R²) to test the goodness-of-fit of the models. We also calculated the Akaike's information criteria (AIC) values, which are indicators of the actual success of the mathematical models. We further calculated the root mean square error of validation (RMSR _v ) and coefficient of variance (CV) to validate the model. We used the standard formula for calculating the RMSE, AIC and CV:

where Y _i and X _i represent the model predicted and observed value, respectively, and

where n is the number of observations, SS the sum of square of the vertical distances of the points from the curve and K is the number of parameters in regression equations. In the present study, we used bias adjustment, because the ratio of n/K is <40. Then, the adjusted AIC (AIC_c) is:

Generally, a good regression model should possess higher R² and lower AIC and RMSE values. Moreover, coefficient of variation (CV) is also used to validate the models:

Here, RMSE substitutes the standard deviation and is the mean of observed value.

All the equations and graphs were developed through Microsoft Excel. Subsequent analysis was also done with Microsoft Excel.

Results

We used four regressions (linear, polynomial, exponential and power) to find the appropriate mathematical models for predicting Chl and N contents in leaves of the Damask rose from CCM-200 readings (Fig. 1). The root mean square error of validation (RMSR) values indicated that the prediction efficiency of the developed mathematical models to estimate Chl and N contents in leaves of the Damask rose is quite high (Table 1). The extractable total Chl concentration of 60 leaves of the Damask rose varied from 3.484 to 6.492 mg g⁻¹ fresh leaf tissue with a mean of 4.689 (±0.682). Among the developed regression models, the polynomial model displayed the lowest RMSEc (0.2383, 0.1875) and maximum R² values (0.886, 0.847) for estimating total Chl and N (Table 1). However, the exponential model and the power model registered the lowest AIC values (−78.33, −25.33 for total Chl and N, respectively). In the case of N estimation, the maximum RMSEc (0.2147) and lowest R² (0.815) were recorded with the power regression model. So it may be biased to assess the models based on their calibrated parameters without validation. To ascertain the best model, we validated all models through RMSEp and CV values which were obtained from the relationship between conventionally analyzed (actual) values and model predicted values (Table 1). Validated data (Table 1) indicated that the polynomial model was the best model for predicting the total Chl with the lowest RMSEp (0.1909) and CV (4.03 %) values. Figure 2a indicated that the model predicted values of total Chl were very close to conventionally analyzed (actual) values with RMSEp less than 20 mg g⁻¹.

Figure 1. The linear and non-linear relationships between chlorophyll content index (CCI) and total chlorophyll (mg g⁻¹) (a), and CCI and nitrogen (N) content (%) (b) in leaves of the Damask rose (Rosa damascena Mill). R² is the coefficient of determination. Root mean square error for calibration (RMSEc) of respective regression lines are given in Table 1.

Figure 2. Validation of the developed models for estimating total chlorophyll (Chl) (a) and nitrogen (b) content in leaf. Solid lines are Chl _pred  = Chl _obj (a) and nitrogen _pred  = nitrogen _obj (b); dotted lines are best-fit function for predicted values versus observed values. *** indicates the level of significance (P < 0.001).

Table 1. Regression models and estimated parameters for non-destructive estimation of chlorophyll and nitrogen content

		Estimated parameters and statistics of developed models						Statistics for model validation
Dependent variable (Y)	Regression Models	α	β	γ	AIC	R^2	RMSEc	RMSEp	CV(%)
Total Chl	Linear (Y = α + βX)	1.1315	0.1696		−77.00	0.878	0.2469	0.2073	4.38
Nitrogen		1.4264	0.0768		−23.86	0.840	0.1917	0.2085	6.65
Total Chl	Exponential (Y = αe^βX)	2.1645	0.0363		−78.33	0.883	0.2415	0.1911	4.03
Nitrogen		1.7382	0.026		−22.42	0.846	0.1886	0.2239	7.14
Total Chl	Power (Y = αX^β)	0.5092	0.73		−75.54	0.870	0.2530	0.2147	4.53
Nitrogen		0.7057	0.4824		−25.34	0.815	0.2011	0.1936	6.17
Total Chl	(Y = α + βX + γX^2)	2.7257	0.0116	0.0038	−76.45	0.886	0.2383	0.1909	4.03
Nitrogen		1.9807	0.0164	0.0015	−18.37	0.847	0.1875	0.2341	7.45
Akaike Information Criterion (AIC), coefficient of determination (R^2) and root mean square error of calibration (RMSEc) were estimated for selection of models. Prediction accuracy of model was estimated using root mean square error of prediction (RMSEp) and coefficient of variance (CV) throughout cross-validation procedure. Chl, chlorophyll.

In the case of N estimation, the prediction efficacy of the power model was marginally better with the lowest RMSEp (0.1936) and CV values (6.18%). The power model registered 7.67%, 15.66% and 20.91% lower RMSEp over linear, exponential and polynomial regression models, respectively. Thus, in the present study, the power model is the best model for N estimation. The trend line between observed values and model predicted values was close to a 1:1 line (Fig. 2b). Thus, a single criterion for the selection of a model is perhaps ambiguous. Moreover, the correlation coefficient between observed values and our model predicted values was highly significant (p ≤ 0.001) in both cases. Clearly, one cannot say with certainty that the Chl or N content in the leaf must have a particular amount with a particular value of CCI. We also built up a 95% confidence interval (Fig. 3) for two suitable models (polynomial for Chl and power for N) which indicated that there was a 95% chance that a new observation will fall within the lower and upper bounds.

Figure 3. The best fitted regression line for estimating total chlorophyll (a) and nitrogen (b) content in leaf with 95% confidence intervals. The lower and upper dotted curved lines demarcate the range of the 95% confidence interval whereas the solid regression line itself represents the best-fit function for estimating total chlorophyll (a) and nitrogen (b). There is a 95% probability that the actual value is between the upper and lower confidence intervals. CCI, chlorophyll content index; CI, confidence interval.

Discussion

Chlorophylls, which are the key pigments of plants, play a direct role in the conversion of solar energy to biochemical energy in the processes of photosynthesis. In the present study, the second-order polynomial regression is the best model for describing the relationship between extractable total Chl content and CCI values. However, the data points are more scattered at higher CCI values (Fig. 1). The non-linear relationships have been confirmed in previous studies for paper birch (Betula papyrifera Marsh.) (Richardson et al. 2002), birch, wheat and potato (Uddling et al. 2007). Nevertheless, many researchers have also found a linear relationship between Chl content and CCI value (Cate and Perkins 2003; Jifon et al. 2005) in different species. In the present data sets, the linear model registered lower R² and higher RMSEc, RMSEp and CV compared with the polynomial and exponential regression models. So, it is clear that the relationship between Chl content and Chl meter-reading does not exhibit a similar pattern for all species. These differences among species may be related to the non-uniform distribution of Chl in leaves as an effect of the clustered structural organization of Chl molecules in chloroplasts, chloroplasts in cells, and cells in leaves (Fukshansky et al. 1993). The curvilinear relationship suggested that CCM-200 underestimated total Chl at high concentrations (Fig. 1a). Thus, the effectiveness of CCM-200 is lower at high concentrations of Chl. This may be due to a non-uniform distribution of Chl in high-Chl leaves as compared to that in low-Chl leaves (Uddling et al. 2007). The curvilinear response suggested a reduction of light absorption by the leaves at high Chl contents, probably due to an increase in Chl density in chloroplasts rather than an increase in the number of chloroplasts (the sieve effect) and rearrangement of chloroplasts in response to the radiation environment (Terashima and Saeki 1983; Vogelmann 1989).

In the case of N estimation, the prediction efficiency of the power model was marginally higher. However, Van den Berg and Perkins (2004) have reported a linear relationship between Chl meter-reading and N content in leaves of the sugar maple, and 64% of the variation in N was predicted by CCI. In our case, more than 80% of the variation in N was predicted by the CCM-200 reading, probably due to the fact that more N is bound up in Chl and other photosynthetic compounds than in the soluble protein pool in the Damask rose leaf. The uneven distribution of N between soluble proteins and the light-harvesting complex can be intensified by leaf age, growth environment, or cultural practices (Bondada and Syvertsen 2003).

Our results confirmed that foliar Chl and N of the Damask rose could be reliably estimated with the help of a CCM-200 meter. Thus, there is an enormous scope for using the CCM-200 as a device for assessment of changes in Chl and N contents in leaves of the Damask rose over time. In this way, this method will be helpful for the assessment of physiological changes over time, and delineating the effects of N management on photosynthetic activity. The main limitation of this method is that a single prediction model cannot be applied to a wide range of cultivars and environmental conditions. In spite of these limitations, CCM-200 is a valuable instrument if it is properly calibrated.

In this paper, the developed mathematical models using the CCI value as the independent variable have proven to be a useful tool for non-destructive estimation of total Chl and N contents in leaves of the Damask rose. The validation study suggested that the polynomial and power regression models were the most appropriate for non-destructive estimation of total Chl and N contents, respectively, in leaves of the Damask rose. The accuracy of the prediction, however, decreases towards the lower and upper ends of the range of CCI. The models relating CCI value to total Chl or N content can be reliably used within the agro-ecological zone where these models have been calibrated.

Acknowledgments

The authors are thankful to Dr. P.S. Ahuja, Director, IHBT, Palampur for his constant encouragement for the work. The authors acknowledge the Council of Scientific and Industrial Research for financial and infrastructural support.