Pigment levels vary with species, season, and physiological state. Consequently, landscape-level detection of relative pigment levels could provide useful insights into vegetation distribution and function. However, conventional wet chemical methods of pigment quantification involve tedious and destructive sampling that cannot be readily applied at scales larger than the single leaf. We used hyperspectral reflectance – reflectance with many adjacent narrow bands – to explore alternate methods of pigment quantification in the intact leaf as a basis for further work at canopy and landscape scales. By using the full spectral information, good correlations were found between spectral reflectance and estimated pigment levels for three major pigment groups: chlorophylls, carotenoids, and anthocyanins. Additionally, since this procedure involves estimation of pigment absorptances, this approach may be useful for characterization of the fate of absorbed light in photosynthesis and photoprotective mechanisms. Further work is examining the physiological significance of changing pigment levels and exploring the application of hyperspectral reflectance for quantification of pigment levels and absorptance at canopy and landscape scales. By utilizing the fundamental characteristics of pigments common to all plants, our goal is to develop a procedure that can be applied uniformly across ecosystems.

Single leaf methods:

Reflectance (400-1000 nm) of individual leaves was measured with a leaf clip connected to a spectrometer (Unispec, PP Systems) like the one pictured above. This clip allowed measurement of a very small, well defined portion of the leaf, as well as providing constant illumination through one side of a bifurcated fiber optic. Leaf transmittance and absorptance of some of the leaves were also measured in an integrating sphere. Following optical measurements, the leaves were frozen and stored for later wet chemical measurement of pigment contents.
 

Anthocyanins are red pigments, commonly found in flowers, fruits and leaves. They have a single absorption peak around 529 nm in the green region of the spectrum. Anthocyanins function as attractants for flower pollinators, and seed dispersers. In leaves, they may function to protect the photosynthetic system from excess light and/or to deter herbivory.

Anthocyanins tend to increase in leaves with low photosynthetic rates, either because they are young or are under stress. Consequently, measurements of leaf anthocyanin content may be useful in assessing the leaf’s physiological state.
 
 

Wet chemical techniques for pigment measurement:

Effect of anthocycanins on spectrophotometric measurement of chlorophyll and carotenoids:

Traditional equations for the measurement of chlorophyll and carotenoids in leaf extracts have not taken into account the effects of anthocyanins.

The absorption spectrum of anthocyanin in a neutral pH extract is broadened considerably compared to that in acidic methanol. Consequently, there is substantial overlap between anthocyanin absorption and that of both chlorophyll and carotenoids.
 
 

Equations for 4 pigment mixtures:

We developed the following equations for spectrophotometric estimation of chlorophyll a, chlorophyll b, anthocyanin and carotenoids in mixtures. These equations are based on the data published for chlorophyll and carotenoids by Lichtenthaler (1987) and the molar extinction coefficients for anthocyanin shown in the figure at left.
 
 

Anthocyanin = 0.0821*A₅₃₄ - .00687*A₆₄₃ - 0.002423*A₆₆₁

Chlorophyll b = 0.02255*A₆₄₃ - 0.00439*A₅₃₄ - 0.004488*A₆₆₁

Chlorophyll a = 0.01261*A₆₆₁- 0.001023*A₅₃₄ - .00022*A₆₄₃

Carotenoids = (A₄₇₀ – 17.1*(Chl a + Chl b) – 9.479 * Anthocyanin) / 119.26

Failure to correct for the absorption of anthocyanins in leaf extracts containing both chlorophyll and anthocyanin can result in considerable errors, particularly in the estimation of chlorophyll b content, and thus the chlorophyll a/b ratio.
 
 
 

Estimation of pigment contents from reflectance:

These wet chemical methods of pigment quantification involve tedious and destructive sampling that cannot be readily applied at scales larger than the single leaf. We used hyperspectral reflectance – reflectance with many adjacent narrow bands – to explore alternate methods of pigment quantification in the intact leaf as a basis for further work at canopy and landscape scales.
 

These wet chemical methods of pigment quantification involve tedious and destructive sampling that cannot be readily applied at scales larger than the single leaf. We used hyperspectral reflectance – reflectance with many adjacent narrow bands – to explore alternate methods of pigment quantification in the intact leaf as a basis for further work at canopy and landscape scales.

Canopies
 

Canopy level methods:

Reflectance (400-1000 nm) of whole plant canopies was measured by holding a straight optical fiber, connected to a spectrometer (Unispec, PP Systems), vertically over the canopy as pictured above. That section of canopy was then harvested, separated into leaves, stems, fruits, and flowers as appropriate, projected areas of each group measured and representative samples frozen for pigment analysis. Total canopy pigment contents were calculated by multiplying total areas of leaves, stems, fruits and flowers by their measured pigment contents per unit area.

Measurements were made in annual grassland, coastal sage scrub and chaparral in the Santa Monica Mts of California and in mixed mojave scrub on the Nevada Test Site north of Las Vegas, NV. All measurements are single species canopies.
 
 

Canopy Anthocyanin:

The red/green reflectance ratio was not correlated with the anthocyanin/chlorophyll ratio for whole canopies. Soil, bark and dead vegetation had rather high red/green ratios which caused an apparent anthocyanin signature even where there in fact was little or none. Further work is examining techniques to correct for this effect.
 
 

Chlorophyll

Chlorophylls are the primary photosynthetic pigments. They have absorption peaks in the red and blue portions of the spectrum. Light energy absorbed by chlorophylls can be used by photosynthetic reactions, reemitted as fluorescence or dissipated as heat. Leaf chlorophyll concentrations can be used to assay the level of plant stress and estimate the physiological state of plants. In addition, estimates of light absorptance by chlorophyll are a crucial parameter in canopy photosynthesis models.

Use of NDVI to estimate chlorophyll content:

The widely used normalized difference vegetation index is in essence a chlorophyll index since it measures the spectral feature produced by chlorophyll absorbance around 660 nm. However, the commonly used form of this index, based on reflectance at 680 nm saturates very quickly with increasing chlorophyll concentration (see fig at right).

A modified index based on reflectance at the red edge (705 nm) was developed by Gitelson (1994) and provides much better sensitivity to high chlorophyll concentrations. Our data set provides a test of this index over a much wider range of chlorophyll contents and leaf structures.

We made a further modification to the Gitelson index to account for high leaf surface reflectance of pubescent or waxy leaves (see representative reflectance plots below). This surface reflectance tends to shift the whole reflectance curve up and results in low NDVI. Leaves of annuals, herbaceous perennials and drought deciduous shrubs deviated most from the overall relationship. A correction based on reflectance at 445 nm (site of maximum chlorophyll and carotenoid absorbance) substantially improved the correlation with chlorophyll content.

Use of NDVI to estimate leaf PAR absorptance:

Absorptance of photosynthetically active radiation (PAR) is a crucial parameter in models of canopy photosynthetic rate. Using data for leaf absorptance measured in an integrating sphere, we tested the ability of NDVI to predict absorptance. In contrast to estimates of chlorophyll content, the traditional NDVI based on reflectance at 680 nm is most linearly related to PAR absorptance.

Carotenoids

Carotenoids are yellow accessory pigments associated with chlorophyll in the photosynthetic system. They have broad absorption peaks in the blue region of the spectrum. Carotenoids modulate the flow of energy into and out of the photosynthetic system. The carotenoid violaxanthin absorbs blue light and transfers the energy to chlorophyll a for use in photosynthesis. In contrast, the related pigment zeaxanthin removes excess energy from the photosynthetic system. As light intensities increase, violaxanthin is reversibly converted into zeaxanthin, thereby modulating energy flow into the photosynthetic system.

Leaf carotenoid to chlorophyll ratios have been suggested as measures of plant stress. It is also possible to measure the interconversion of the xanthophyll cycle pigments based on changes in reflectance and this has been used to estimate photosynthetic rates.
 

Carotenoid to chlorophyll ratios:

In leaves, the contents of chlorophyll and carotenoids were highly correlated. This probably reflects the tight coordination between these pigments within the photosynthetic system. Substantial variation in this ratio was found only in senescent leaves with very low contents of both carotenoids and chlorophyll. However, this variation was still small in comparison to yellow flowers.

Given this low variation in chlorophyll to carotenoid ratio, the reflectance indexes we tried for prediction of this ratio provided very poor results. The best predictor of carotenoid content was the chlorophyll content.

NDVI and chlorophyll at the canopy level:

The relationship between canopy chlorophyll content and NDVI was much less sensitive to the form of the NDVI equation than was the case at the leaf level (see below). This was apparently a result of the greater importance of total leaf area, as opposed to chlorophyll content per unit leaf area, in determining canopy NDVI. NDVI was strongly correlated with total green tissue area at the canopy level (see fig at right).

Pleuraphis rigida (a desert perennial grass) had a large quantity of dead leaf material which obscured its green tissues, resulting in very low NDVI. For this reason it was excluded from the regressions in these plots.

The effect of canopy structure (leaf angle and leaf overlap) can be seen in the fig below where we compare the relationship between NDVI and chlorophyll at the leaf and canopy scales.
 
 

Carotenoid and chlorophyll were even more highly correlated at the canopy level than they were at the leaf scale. This is particularly surprising given that some of the plants had yellow flowers at the time of measurement. Partly, this good correlation is a result of the expanded range of values resulting from differences in leaf area per unit ground area.

Conclusions:

Pigment assays:

Anthocyanin in leaf extracts can substantially affect spectrophotometric determination of chlorophyll. We developed equations to correct for this problem. As an additional benefit, these equations provide a new method for measurement of anthocyanins in leaf extracts and this measure appears to be better correlated with actual pigment color in leaves than is the traditional method.

Single leaves:

1. Leaf anthocyanin content of young leaves was well correlated with the red/green reflectance ratio, aver(600-699nm)/aver(500-599nm)

2. We developed a new chlorophyll index, based on a red edge NDVI corrected for baseline reflectance at the leaf surface, which substantially improves the estimation of leaf chlorophyll from reflectance.

3. In contrast to chlorophyll content, leaf PAR absorptance was best correlated with a traditional NDVI based on reflectance near the chlorophyll absorbance maximum.

4. Leaf carotenoid content was highly correlated with leaf chlorophyll, which proved to be the best predictor of carotenoid content.
 

Canopy level:

1. The red/green reflectance ratio was not correlated with anthocyanin/chlorophyll ratio at the canopy scale. Soil and dead vegetation confused this signal. Further work is exploring methods to correct for this effect.

2. In contrast to the leaf level results, the form of the NDVI equation had little effect on the linearity of the correlation with chlorophyll content. Canopy NDVI appeared to be determined more by total green leaf area as opposed to chlorophyll content per unit leaf area.
Carotenoid and chlorophyll contents were even more highly correlated at the canopy scale than they were at the leaf scale.
 

Acknowledgements:

We wish to thank David Fuentes, Misha McKinney, Miriam Schmidts, John Surfus, and Brian Zuta, for help with field work and pigment analyses. This work conducted with the support of the CEA-CREST program at California State University at LA.