Evaluating the relationship between leaf chlorophyll concentration and ...

Photosynth Res (2007) 91:37–46 DOI 10.1007/s11120-006-9077-5

REGULAR PAPER

Evaluating the relationship between leaf chlorophyll concentration and SPAD-502 chlorophyll meter readings J. Uddling Æ J. Gelang-Alfredsson Æ K. Piikki Æ H. Pleijel

Received: 14 March 2006 / Accepted: 31 May 2006 / Published online: 7 March 2007 Springer Science+Business Media B.V. 2007

Abstract Relationships between chlorophyll concentration ([chl]) and SPAD values were determined for birch, wheat, and potato. For all three species, the relationships were non-linear with an increasing slope with increasing SPAD. The relationships for birch and wheat were strong (r2 ~ 0.9), while the potato relationship was comparatively weak (r2 ~ 0.5). Birch and wheat had very similar relationships when the chlorophyll concentration was expressed per unit leaf area, but diverged when it was expressed per unit fresh weight. Furthermore, wheat showed similar SPAD–[chl] relationships for two different cultivars and during two different growing seasons. The curvilinear shape of the SPAD–[chl] relationships agreed well with the simulated effects of non-uniform chlorophyll distribution across the leaf surface and multiple scattering, causing deviations from linearity in the high and low SPAD range, respectively. The effect of non-uniformly distributed chlorophyll is likely to be more important in explaining the nonlinearity in the empirical relationships, since the effect of scattering was predicted to be comparatively weak. The simulations were based on the algorithm for the calculation of SPAD-502 output values. We suggest that SPAD calibration curves should generally be parameterised as non-linear equations, and we hope

J. Uddling (&) Æ J. Gelang-Alfredsson Æ K. Piikki Æ H. Pleijel Department of Plant and Environmental Sciences, Go¨teborg University, P. O. Box 461, SE-405 30 Go¨teborg, Sweden e-mail: [email protected]

that the relationships between [chl] and SPAD and the simulations of the present study can facilitate the interpretation of chlorophyll meter calibrations in relation to optical properties of leaves in future studies. Keywords Absorbance Æ Chlorophyll Æ Non-uniform chlorophyll distribution Æ Reflectance Æ Scattering Æ SPAD Abbreviations a Absorbance A Absorptance ac-650 Absorbance of control sample with uniformly distributed chlorophyll and no scattering C Concentration of absorbers [chl]area Chlorophyll concentration per unit leaf area [chl]fwt Chlorophyll concentration per unit fresh weight I Intensity of transmitted light I0 Intensity of incident monochromatic light k Coefficient in SPAD algorithm l Length of light path through absorbers M Output SPAD value p Absorbance ratio between samples with and without scattering; also called apparent scattering pathlength R Reflectance Ri Internal reflectance Rs External leaf surface reflectance T Transmittance Molar absorption coefficient

123

38

Introduction The leaf chlorophyll concentration, [chl], is usually determined by extraction from leaf samples and subsequent spectrophotometric measurements (e.g., Arnon 1949; Porra et al. 1989). Such in vitro determinations are destructive, expensive, and time consuming, and may therefore not be applicable for all purposes. There are also more rapid methods for estimating the leaf [chl] non-destructively, in vivo. These methods exploit the optical properties of leaves and are based on the reflectance and/or absorbance of radiation by chlorophyll. In the 1980s, a hand-held absorbance-based dualwavelength chlorophyll meter was developed by Minolta (SPAD models 501 and, later, 502; Minolta corporation, Ltd., Osaka, Japan). The SPAD-502 meter measures the transmittance of red (650 nm) and infrared (940 nm) radiation through the leaf, and calculates a relative SPAD meter value that should ‘‘correspond to the amount of chlorophyll present in the sample leaf’’ (Minolta 1989). The use of the SPAD chlorophyll meter has increased dramatically within agriculture and research during the last decade, and currently there are more than 200 published studies using the instrument in the scientific literature. Only a small fraction (~10%) of these studies quantify the relationship between the leaf [chl] determined in vitro and the SPAD readings. The studies that do perform such calibrations of the SPAD meter usually parameterise linear relationships (e.g., Marquard and Tipton 1987; Dwyer et al. 1991; Fanizza et al. 1991; Schaper and Chacko 1991; Gratani 1992; Xu et al. 2000; Yamamoto et al. 2002; Esposti et al. 2003; Wang et al. 2004), which is in accordance with the proportional relationship between pigment concentration and absorbance predicted by Beer’s Law (Eisenberg and Crothers 1979). However, a number of other studies report on curvilinear relationships between [chl] and SPAD values (Monje and Bugbee 1992; Markwell et al. 1995; Castelli et al. 1996; Bindi et al. 2002; Richardson et al. 2002; Jifon et al. 2005). Several plausible explanations for this non-linearity have been proposed, such as non-uniform distribution of chlorophyll and/or radiation across the leaf surface and differential scattering and reflection of photons at 650 nm and 940 nm (e.g. Monje and Bugbee 1992; Markwell et al. 1995). The discrepancy between the absorption spectra of a leaf in vivo and an in vitro solution of chlorophyll-protein complexes is well-known and mainly caused by the non-uniform distribution of chlorophyll and multiple scattering in the intact leaf

123

Photosynth Res (2007) 91:37–46

(Richter and Fukshansky 1996). So far, no study has quantitatively investigated how such deviations from the assumptions underlying Beer’s Law can be expected to influence the relationship between [chl] and the SPAD readings. The objectives of this study are (1) to determine empirical relationships between leaf [chl] an SPAD502 meter readings for three different species (silver birch (Betula pendula), wheat (Triticum aestivum L., cv. Dragon and Lantvete) and potato (Solanum tuberosum L., cv. Bintje)) and (2) to interpret these relationships in relation to the optical properties of leaves. The interpretation is based on the simulated effects of non-uniform [chl] across the leaf surface and multiple scattering on the shape of the SPAD–[chl] relationship.

Materials and methods Data sets The present study is based on data sets with birch, wheat, and potato. The data sets were collected during different years and with different replication (number of leaves sampled and number of measurements averaged for an individual leaf). Two different absorbancebased chlorophyll meters were used: a SPAD-502 meter for birch and potato and a Hydro N-Tester (Yara International ASA, Oslo, Norway) for wheat and potato. Since both meters were used for measuring the same potato leaves they could be intercalibrated. The SPAD-502 and Hydro N-Tester measurements were conducted in the field between 10:00 and 16:00 h. The adaxial side of the leaves was always placed toward the emitting window of the instrument and major veins were avoided. Leaves at different development stages were selected, including premature, fully mature and senescent leaves. Leaves were detached and collected immediately after the SPAD502 and Hydro N-Tester measurements and were kept cool (~0C) until sub-samples were punched in the laboratory. The punched leaf material was weighted and then immediately frozen and kept at –25C until chlorophyll extraction. The area of the punched wheat leaf material was determined using an area meter (Delta-T Devices Ltd., Cambridge, England), while two circular 1.0 cm diameter leaf discs from one side on the midrib were punched for birch and potato. The spatial distribution of punched leaf material matched the distribution of the SPAD-502 and Hydro N-Tester readings.


The leaf samples were ground in 80% acetone (5 · 5 ml in five repeated steps) with a tissue homogeniser. After sedimentation, the extract solutions were filtered and additional acetone was added to a total volume of 25 ml. The absorbance was determined at 645, 663, and 720 nm using a spectrophotometer (UV2401PC, Shimadzu Corp., Kyoto, Japan), and the [chl] of the solutions were determined using the equations published by Porra et al. (1989). The [chl] was calculated per unit fresh weight [chl]fwt as well as per unit leaf area [chl]area. Birch A total of 60 leaves from ten different juvenile trees (height: 2–4 m) were sampled at the Asa Forest Research Station (5710¢ N, 1447¢ E, altitude 190 m) in southern Sweden on August 21 2003. The SPAD-502 values ranged from 1 to 46. Ten SPAD-502 readings (five on each side of the midrib) evenly distributed over the leaf area but not near the leaf edges were averaged for each leaf. Wheat ¨ stad Flag leaves of wheat growing in a wheat field at O sa¨teri, 50 km north-east of Go¨teborg, Sweden (5754¢ N, 1224¢ E; alt. 62 m), were sampled during the 1997 and 1999 growing seasons. In 1997, 72 leaves of the Dragon cultivar were sampled, while in 1999 45 leaves each of the cultivars Dragon and Lantvete were sampled. The Dragon data set from 1997 has been described in more detail by Gelang et al. (2000). The wheat cultivar Lantvete was grown in southern Sweden between 1900 and 1910 and has since then not been actively subject to plant breeding. Leaf area was measured for 60 of the 1997 Dragon leaves, but not at all in 1999. Hydro N-Tester values ranged from 0 to 681. A total of 30 Hydro N-Tester readings evenly distributed over the leaf were averaged for each leaf. Potato ¨ stad sa¨teri A total of 24 leaflets from a potato field at O were sampled on August 20 1999. Each leaf was measured with both the SPAD-502 and the Hydro N-Tester. The SPAD-502 values ranged from 6 to 40, and the Hydro N-Tester values ranged from 12 to 549. A total of 30 readings (15 on each side of the midrib) evenly distributed over the leaf area were averaged for each leaflet.

39

Theory According to Beer’s Law, the absorbance (a) is proportional to the molar absorption coefficient (), the length of the light path through the absorbers (l), and the (uniform) concentration of absorbers (C) (Eisenberg and Crothers 1979): a ¼ elC

ð1Þ

The law can also be expressed in terms of the intensity of the incident monochromatic light (I0) and the intensity of the transmitted light (I): a ¼ log10

I0 I

ð2Þ

The assumption behind this logarithmic relationship is that each layer of a solution absorbs the same fraction (i.e., not the same amount) of the remaining beam of light. The law assumes an ideal optical system with a uniform concentration of absorbers and no scattering or reflectance of light. The SPAD-502 and Hydro N-Tester chlorophyll meters exploits Beer’s Law and the efficient absorbance of red light (650 nm) by chlorophyll. Since some of the photons at 650 nm are absorbed by molecules other than chlorophyll, the transmittance at a reference wavelength where absorbance by chlorophyll is insignificant (940 and 960 nm for the SPAD-502 and Hydro N-Tester, respectively) is also measured by the instruments. The SPAD-502 meter calculates an output SPAD value (M) according to the following equation: M ¼ k log10

I0ð650Þ Ið940Þ Ið650Þ I0ð940Þ

ð3Þ

where k is a confidential proportionality coefficient (technical bulletin on the SPAD-502, Spectrum Technologies Inc., Plainfield, Illinois, USA). The output values given by the SPAD-501 meter are based on the same equation, but with the wavelengths 644 and 790 nm instead of 650 and 940 nm and a k-value of 40. The SPAD-502, which is designed to have data compatibility with the SPAD-501, would also have a k-value of 40 at 644 and 790 nm. Since the wavelengths of this instrument are different, the k-value of Eq. 3 is not exactly 40. The actual k-value for Eq. 3 and the SPAD-502 meter has not been released by the manufacturer (Alf Karlsson, Konica Minolta Photo Imaging Svenska AB, Solna, Sweden, personal communication). The Hydro N-Tester is technically based on the SPAD-502, but the k-values differ between the

123

40


instruments (Frank Brentrup, HYDRO International Research Centre Hanninghof, Duelmen, Germany, personal communication). A photon incident on a leaf is reflected, absorbed, or transmitted. Mathematically, this can be expressed as {1 = R + A + T}, where R is reflectance, A is absorptance, T is transmittance, and 1 is the normalised incident radiation (I0/I0). A more thorough expression of M, explicitly incorporating the R and A at both wavelengths, is therefore: I0ð650Þ I0ð940Þ ðIA þ IR Þ940 ; M ¼ k log10 I0ð940Þ I0ð650Þ ðIA þ IR Þ650

ð4Þ

where IA and IR are the radiant fluxes subject to absorption and reflection, respectively. Simulations In two different simulations, we investigate how a nonuniform distribution of chlorophyll across the leaf surface and multiple scattering affect the relationship between the [chl]area determined in vitro and the SPAD values, M. Both simulations are compared to a control with the same mean [chl]area and the absorbance ac-650. Reflection and scattering are absent (R = 0) in this control simulation, where the chlorophyll is uniformly distributed and is the only absorber of radiation (A940 = 0). The control is hence an ideal optical sample that follows Beer’s law, and serves here to represent the in vitro [chl] determinations. Non-uniform distribution of chlorophyll across the leaf surface The non-uniform distribution of chlorophyll molecules within the leaf is influenced by the structural organization of grana within chloroplasts, chloroplasts within cells and cells within tissue layers; patterns which may differ strongly among species (Fukshansky et al. 1993). In this simulation it is assumed that the [chl]area has a normal distribution around its mean value (l). The standard deviation (r) of l was set to values between 10% and 50% of l in order to simulate the impact of different degrees of heterogeneity in [chl]area. For the small proportion of the [chl]area distribution with [chl]area < 0, the [chl]area was set to zero. Chlorophyll was assumed to be the only absorber of light in this simulation, and R was neglected. Multiple scattering Some of the radiation incident on a leaf is subject to reflectance caused by reflection at the external leaf

123

surface (Rs), and the radiation entering the leaf is scattered in all directions by reflection and refraction at the curved air-cell wall interfaces (McClendon and Fukshansky 1990a). Scattering causes an increase in the mean pathlength traveled by photons within the leaf and, hence, an increase in a and A (Richter and Fukshanky 1996). This increase is stronger at wavelengths with lower absorption as well as at low concentrations of the absorber. Scattering also causes internal reflectance (Ri), and the proportion of the initial photon flux subject to Ri decreases as the A increases, and vice versa (McClendon and Fukshansky 1990b). The adaxial Rs, which is usually around 4–6% (McClendon and Fukshansky 1990a) and approximately the same at red and infrared wavelengths (Vogelmann 1989), is assumed to be 5% at both 650 and 940 nm in the present simulation. The Ri, on the other hand, is highly dependent on wavelength and [chl]. Subtracting a few percent attributable to Rs from the total R, the wavelength-dependent variations in R (i.e., in Ri) and T over the 400–850 nm spectrum are remarkably similar for a broad range of species (Knapp and Carter 1998). The non-absorbed radiation is therefore roughly equally divided between R and T, implying efficient scattering of light in all directions within the leaf. In this simulation we assume that Ri = T. Internal scattering results in an increase in the absorbance of radiation entering the leaf by a factor p (p also called apparent scattering pathlength, since l and a in Eq 1 are proportional), compared to the absorbance of a sample with uniformly distributed chlorophyll and no scattering (i.e., ac-650). To simulate the dependence of p on ac-650, we apply a model fitted to Fig. 7 in McClendon and Fukshansky (1990b), where {p = 3.00 – 2.43 ac-650 + 1.34 a2c-650 – 0.271 a3c-650}. The p predicted by this model is multiplied by 2/3, 1, and 4/3 to investigate the impact of different degrees of scattering. Finally, the absorptance at the reference wavelength 940 nm is set to 0.02. In summary, this simulation assumes Rs = 0.05, Ri = T, a650 = pac-650, A650 = (1 – Rs) – (1 – Rs)/(10pac-650), p as described above, A940 = 0.02, and a uniform [chl]area.

Results The relationships between leaf [chl] and SPAD values were curvilinear for all three species (Fig. 1). The slope of the relationship between [chl] and SPAD increased with increasing SPAD, and the data fitted well to exponential functions with two parameters (Table 1). For birch and wheat the relationships were slightly stronger for [chl]area than for [chl]fwt while the opposite

-2

Chlorophyll concentration (g m )


41

0.8

(a) birch wheat potato birch wheat potato

0.6

0.4

0.2

0.0

0

10

20

30

40

50

40

50

-1

Chlorophyll concentration (mg g )

SPAD value 6

(b) 5 4 3 2

potato leaflets had a strong linear relationship with a significant SPAD-intercept at zero Hydro N-Tester (Fig. 3). A non-uniform distribution of chlorophyll across the leaf surface results in lower M/k (Fig. 4a) and higher T (Fig. 4b) as compared to a leaf sample with the same amount of chlorophyll uniformly distributed. The difference in M/k between samples with uniform and nonuniform [chl]area increases with increasing M/k and is very small at M/k lower than 0.5. Normal distributions of [chl]area with r of 40 and 50% of the mean were predicted to transmit 2.4 and 5.1% of the radiation, respectively, at a [chl] that would transmit 0.1% if the chlorophyll was uniformly distributed (Fig. 4c). Multiple scattering results in higher M/k compared to a sample with no scattering (Fig. 5), and this effect increases with increasing apparent scattering pathlength (p-value). Simulating the dependence of p on ac650 resulted in a curvilinear relationship at low M/k. In the lower M/k range ( < 0.5), the M/k for a sample without scattering increases less per unit M/k for a sample with scattering (i.e., lower slope in Fig. 5), as compared to higher in the M/k range.

1 0 0

10

20

30

SP AD value Fig. 1 Leaf chlorophyll concentration, expressed per unit (a) leaf area or (b) leaf fresh weight, in relation to the SPAD-502 values for birch, wheat (Dragon data set from 1997), and potato. The equations of the exponential regression lines and their coefficients of determination (r2-values) are given in Table 1. The SPAD-502 values for wheat was estimated from Hydro N-tester readings using the calibration curve given in Fig. 3

was true for potato (Table 1). The relationships for birch and wheat were almost identical when based on [chl]area (Fig. 1a) while they differed in the high SPAD range when based on [chl]fwt (Fig. 1b). For potato, the relationships were weaker, and the [chl]area and [chl]fwt at a given SPAD value (only at SPAD >20 for [chl]fwt) were higher and lower, respectively, as compared to birch and wheat. The birch leaves with very low SPAD ( < 5) still contained considerable amounts of chlorophyll, while wheat leaves had very low [chl] in the SPAD range 5–8. It should be noted though, that the SPAD values for wheat were calculated from Hydro NTester data and not measured directly. There was little variation between the SPAD–[chl] relationships determined for the two different wheat cultivars (Fig. 2a) or for the cultivar Dragon during the two different growing seasons (Fig. 2b). The SPAD-502 and Hydro N-Tester values measured for the same

Discussion Empirical SPAD–[chl] relationships All three species investigated in this study had nonlinear SPAD–[chl] relationships with increasing slope with increasing SPAD (Fig. 1). Most studies in the literature that quantify the relationship between [chl] and SPAD values employ linear regression (Yadava 1986; Marquard and Tipton 1987; Dwyer et al. 1991; Fanizza et al. 1991; Schaper and Chacko 1991; Gratani 1992; Xu et al. 2000; Yamamoto et al. 2002; Esposti et al. 2003; Kapotis et al. 2003; Murillo-Amador et al. 2004; Wang et al. 2004). Linear regressions between [chl] and SPAD determined for the present data sets resulted in lower r2-values and a systematic pattern of the residuals, with underpredictions of [chl] in the low and high SPAD ranges and overpredictions in the intermediate range (not shown). Furthermore, the [chl] at SPAD = 0 was predicted to be substantially below zero by linear regression models. These consequences of employing linear regression in the calibration of the SPAD meter are obvious in most of the SPAD calibration studies in the literature (Marquard and Tipton 1987; Dwyer et al. 1991, Shaper and Chacko 1991 (some species); Gratani 1992; Xu et al. 2000; Yamamoto et al. 2002; Wang et al. 2004), but some studies report on relationships that are truly linear over

123

42


Table 1 Equations and coefficients of determination (r2-values) for the exponential regression lines between chlorophyll concentration and SPAD values depicted in Figs. 1 and 2 Data set (Figure)

Unit for chlorophyll concentration

Equation x = SPAD-502 value y = chlorophyll concentration

Birch (1a) Wheat (1a) Potato (1a) All species (1a)a Birch (1b) Wheat (1b) Potato (1b) All species (1b)a Wheat, Dragon 1999 (2a) Wheat, Lantvete 1999 (2a) Dragon + Lantvete 1999 (2a)a Wheat, Dragon 1997 (2b) Wheat, Dragon 1999 (2b) Dragon 1997 + 1999 (2b)a

g m2 projected leaf area

y y y y y y y y y y y y y y

a

mg g leaf fresh weight

= = = = = = = = = = = = = =

0.0641 e0.0467x 0.0599 e0.0493x 0.913 e0.0415x 0.0691 e0.0459x 0.363 e0.0452x 0.305 e0.0545x 0.409 e0.0342x 0.236 e0.0588x 0.467 e0.0416x 0.435 e0.0455x 0.481 e0.0418x 0.305 e0.0545x 0.467 e0.0416x 0.360 e0.0495x

r2

0.93 0.89 0.46 0.84 0.92 0.86 0.58 0.85 0.81 0.85 0.82 0.86 0.81 0.83

Regression lines not shown in figure

a wide SPAD range and passes through the origin (Yadava 1986; Shaper and Chacko 1991 (some species); Kapotis et al. 2003). Non-linear relationships between [chl] and SPAD have been parameterised previously, employing second-order polynomials (Monje and Bugbee 1992; Castelli et al. 1996; Azia and Stewart 2001; Bindi et al. 2002; Richardson et al. 2002; Jifon et al. 2005) or exponential equations (Markwell et al. 1995; Castelli et al. 1996). The choice of an exponential equation in the present study was based on its fewer and more readily interpretable parameters rather than on a better fit to data, compared to the quadratic equation. The exponential function suggested by Markwell et al. (1995) (y = 10x^k) was not used here, since it forces [chl] to 1 at SPAD = 0. The species-specific relationships between [chl] and SPAD for birch and wheat converged when [chl] was expressed per unit leaf area instead of per unit fresh weight (Fig. 1a, b). This was not the case for the thicker potato leaflets, but it should be noted that the potato SPAD–[chl] relationships were comparatively weak (Table 1). Inter-species variations in the relationship between [chl]area and SPAD have been found previously (Marquard and Tipton 1987; Shaper and Chacko 1991; Castelli et al. 1996; Yamamoto et al. 2002), although several other studies report on similar SPAD calibrations for different species (Gratani 1992; Monje and Bugbee 1992; Markwell et al. 1995). Thus far, no study has investigated the causes of this speciesspecific variation. The similar SPAD calibrations parameterised for different wheat cultivars or for the same cultivar during different growing seasons (Fig. 2) suggest that the

123

calibration is robust within the wheat species and over time. A study with corn reported on similar SPAD calibrations for six different hybrids (Dwyer et al. 1991), while different calibrations were determined in a study with different citrus cultivars (Jifon et al. 2005). The growing seasons 1997 and 1999 were comparatively similar in terms of meteorology, and therefore variations in growing conditions that may affect the SPAD–[chl] relationship (Campbell et al. 1990; Martinez and Guiamet 2004) were probably of limited importance in the present comparison. The strong linear relationship between the SPAD502 and the Hydro N-Tester (Fig. 3) was expected, since the Hydro N-Tester is technically based on the SPAD-502 (Frank Brentrup, HYDRO International Research Centre Hanninghof, Duelmen, Germany, personal communication). The reason for the significant SPAD-intercept at Hydro N-Tester = 0 is not clear, however. The SPAD-502 meter may be more reliable at very low values, since the two lowest Hydro N-Tester values (0 and 12) were recorded for leaves that contained considerable amounts of chlorophyll (0.09 and 0.13 g m–2, respectively). Interpretation The quantitative interpretation of the SPAD–[chl] relationship has been complicated by the fact that the algorithm for the meter output value is not given in the SPAD-502 manual (Minolta, 1989). Markwell et al. (1995) proposed an equation for the SPAD-502 which was identical with Eq 3 except that the k-value was left out, causing that equation to be physiologically unre-


43

5

(a) 4

Dragon 1999 Lantvete 1999 Dragon 1999 Lantvete 1999

3

2

1

0

0

10

20

30

40

50

Chlorophyll concentration (mg g-1)

SPAD value 6

(b) 5

Dragon 1997 Dragon 1999 Dragon 1997 Dragon 1999

4

effects can be expected to influence the general shape of the SPAD–[chl]area relationship. Both simulations were compared to a control that represents an ideal optical system with uniformly distributed chlorophyll and no scattering. In Figs. 4a and 5 the M/k on the y-axes therefore represent the [chl]area determined for chlorophyll solutions in vitro, while the x-axes represent the simulations of leaves in vivo with either nonuniform [chl]area (Fig. 4a) or multiple scattering (Fig. 5). Both simulations predicted curvilinear SPAD– [chl]area relationships with increasing slope with increasing SPAD (Figs. 4a, 5). The deviation from linearity mainly occurs at high [chl]area in a sample with non-uniform [chl]area (M/k > 0.5 ~ SPAD > 20), while scattering cause deviation in the low range (M/k < 0.5 ~ SPAD < 20). All empirical SPAD–[chl]area relationships agree with this prediction (Figs. 1, 2) since they clearly show an increasing slope with increasing SPAD. The effect of non-uniformly distributed chlorophyll is likely to be more important in explaining the non-linearity in the empirical relationships, since the effect of scattering was predicted to be rather weak (Fig. 5). The positive [chl] intercepts at zero SPAD could not be explained by the simulations, and their cause therefore remains unclear. Similar intercepts were observed by Markwell et al. (1995), who speculated that they could be caused by the stronger absorbance by chlorophyll b than by chlorophyll a at 650 nm in combination with a more pronounced deficiency in chlorophyll b at low [chl] in higher plants. The latter suggestion was not supported by our data (not shown). A non-uniform distribution of chlorophyll across the leaf surface results in lower A (and M/k) and higher T compared to a sample with the same amount of

40

y = 0.0639x + 5.84

3

r2 = 0 .97

30

2 1 0

0

10

20

30

40

50

SPAD value Fig. 2 Leaf chlorophyll concentration expressed per unit leaf fresh weight in relation to the SPAD-502 values for (a) the wheat cultivars Dragon and Lantvete in 1999 and for (b) Dragon in 1997 and 1999. The equations of the exponential regression lines and their coefficients of determination (r2-values) are given in Table 1. The SPAD-502 values for wheat was estimated from Hydro Ntester readings using the calibration curve given in Fig. 3

SPAD value

Chlorophyll concentration (mg g-1)

alistic. In Eq 3 we give the correct equation for how the SPAD-502 meter output values are calculated, obtained from Spectrum Technologies Inc. (Plainfield, Illinois, USA). The k-value in Eqs 3 and 4 is still unknown, but it can be expected to be very near 40 (the k-value for the SPAD-501 meter), since the A and T are similar at the wavelengths used by the SPAD models 501 (644 and 790 nm) and 502 (650 and 940 nm) (Gausman 1985). The well-known discrepancy between the absorption spectra of a leaf in vivo and an in vitro solution of chlorophyll–protein complexes is attributed to two main effects: (1) the non-uniform distribution of chlorophyll that generally decreases A in the intact leaf and, (2) the multiple scattering of radiation that causes internal reflectance and enhancement of A, both being wavelength dependent (Richter and Fukshansky 1996). In two different simulations, we investigated how these

20

10

0

0

100

200

300

400

500

600

Hydro N-tester value Fig. 3 SPAD-502 values versus Hydro N-tester values in potato leaflets in 1999. The equation of the linear regression line and its coefficient of determination (r2-value) are given in the figure

123

44 2

1.0

(a)

(b) = 10% = 20% = 30% = 40% = 50%

T, uniform [chl]

M/k, uniform [chl]

0.8

1 σ = 10% σ = 20% σ = 30% σ = 40% σ = 50%

0

1

0.6

0.4

0.2

0

2

0.0 0.0

0.2

0.4

0.6

0.8

1.0

T, non-uniform [chl]

M/k, non-uniform [chl] 0.20

(c) 0.15

T, uniform [chl]

Fig. 4 The (a) M/k (defined in Eqs 3 and 4) and (b) transmittance (T) of samples with uniform (y-axis) and non-uniform (x-axis) area based chlorophyll concentration. Graph (c) shows the lower left part of graph (b). The values for the non-uniform sample are calculated for the same mean concentration as the uniform solution, but with the probability density function of the chlorophyll concentrations normally distributed around the mean with different standard deviations (r). Chlorophyll is assumed to be the only absorber of light and scattering and reflectance is zero. The bold gray solid lines represent the 1:1 relationship (i.e., r = 0)


0.10

0.05

0.00 0.00

0.05

0.10

0.15

0.20

T, non-uniform [chl]

chlorophyll uniformly distributed (Fig. 4). This effect increases with increasing [chl]area (and M/k), since the radiation transmitted through areas with low [chl]area constitute a larger proportion of the total T as the absorption increase (Fig. 4b, c); this is the so called ‘‘sieve effect’’. Different degrees of heterogeneity in [chl]area can be expected to cause different degrees of non-linearity in the SPAD–[chl]area relationship as well as diverging relationships at increasing SPAD (Fig. 4a). This could not be observed for the three species investigated in the present study, but different degrees of heterogeneity in [chl]area probably explains some of the inter-species variation in SPAD calibrations reported in the literature, since the degree of heterogeneity is known to differ strongly among species (Fukshansky et al. 1993). The two highest simulated degrees of heterogeneity (r of 40 and 50% of l) predicted that 2–5% of the radiation would be transmitted at a [chl]area that would transmit only 0.1% if the chlorophyll was uniformly distributed. These simulated degrees of heterogeneity are probably physiologically realistic, since similar values of T were reported for four different species by McClendon and Fukshansky (1990b). As a result of the clustered structural organization of chlorophyll molecules in grana and chloroplasts, the distribution of chlorophyll within the cell can be highly

123

non-uniform (e.g. Fukshansky et al. 1993). Furthermore, more or less translucent veins that may comprise 25% of the leaf area are likely to be another important source of heterogeneity in [chl]area (McClendon and Fukshansky 1990b). A normal distribution of chlorophyll was chosen for simplicity in the present simulation, although a binomial distribution may be more physiologically realistic (Fukshansky et al. 1993) as a consequence of the clustered structural organization of chlorophyll. A binomial distribution would cause a steeper slope of the SPAD–[chl] relationship over the whole SPAD range, but does not affect the general observation that the slope increases with increasing SPAD. In the present simulation we did not consider the effect of refraction and focussing of light by epidermal cells near the surface of the leaf (Myers et al. 1994), which may interact with the non-uniform distribution of chlorophyll to influence the SPAD–[chl]area relationship. Although epidermal focussing of radiation may act to directing photons to chlorophyll dense positions in the leaf (Vogelmann 1989) the net effect of chlorophyll heterogeneity is generally to decrease A (Richter and Fukshansky 1996), and the general shape of the simulated SPAD–[chl]area relationship should be the same also if the effect of epidermal focussing was considered. Diurnal chloroplast movements in


45

M/k, zero scattering and reflectance

2.0

2/3 p 1p 4/3 p

1.5

1.0

0.5

0.0 0.0

0.5

1.0

1.5

2.0

M/k, scattering and reflectance Fig. 5 The M/k (defined in Eq 4) of a sample with zero scattering and reflectance in relation to the M/k of a sample with reflectance and scattering (see the Simulations section for details). The increase of absorbance due to scattering (p) was assumed to be a function of the absorbance of uniformly distributed chlorophyll without scattering and reflectance according to a model developed by McClendon and Fukshansky (1990b). The modeled p is multiplied by 1 (solid line), 2/3 (dashed), and 4/3 (dash-dot-dot) to illustrate the impact of different degrees of scattering. The chlorophyll is assumed to be uniformly distributed across the surface in both samples. The bold gray solid lines represent the 1:1 relationship

response to light, affecting the degree of heterogeneity in the chlorophyll distribution and, hence, the SPAD values (Hoel and Solhaug 1998), should be of minor importance in this study, where all SPAD-502 and Hydro N-Tester measurements were conducted during 10:00–16:00 h. Averaging SPAD data that have a non-linear relationship with [chl]area acts to make the non-linearity in the SPAD–[chl] relationship more pronounced, but since this effect is a consequence of non-linearity it cannot cause non-linearity per se. Since, in practice, investigators average several SPAD readings across the leaf area in order to get more representative values, this is also how the calibration should be made. Multiple scattering results in higher A (and M/k) compared to a sample with no scattering (Fig. 5). The more scattering (the higher p), the stronger the increase in a and M/k and the lower the slope of the SPAD–[chl]area relationship. Scattering increases a more strongly at lower [chl]area, and simulating the dependence of the apparent scattering wavelength (p) on the a of a sample with uniformly distributed chlorophyll and without scattering and reflectance (i.e., on ac-650) results in a curvilinear relationship at low M/k.

The deviation from linearity in the SPAD–[chl]area relationship occurs in the low range, since the decrease in p with increasing ac-650 is most pronounced at low ac650 (and low [chl]area; McClendon and Fukshansky 1990b). The impacts of scattering on the general shape of the SPAD–[chl]area relationship are therefore (i) a decreasing slope for samples with an increasing degree of scattering, and (ii) an increasing slope with increasing SPAD in the low SPAD range (Fig. 5). The first effect cannot be supported by our data, since we did not estimate the extent to which scattering increased the apparent scattering pathlengths in the different species. Inter-species differences in the SPAD– [chl]area relationship found in other studies (Marquard and Tipton 1987; Shaper and Chacko 1991; Castelli et al. 1996; Yamamoto et al. 2002) may be caused by this effect, but they could also be caused by different degrees of [chl]area heterogeneity. The second effect is difficult to separate from the effect of non-uniform [chl]area, which causes the same curvature but in the higher SPAD range. In conclusion, the curvilinear shape of the SPAD– [chl] relationships parameterised for birch, wheat and potato agreed well with the simulated effects of nonuniform chlorophyll distribution across the leaf surface and multiple scattering. The simulations were based on the algorithm for the calculation of SPAD-502 output values, and we hope that this study will facilitate the interpretation of SPAD calibrations in relation to optical properties of leaves in future studies. Acknowledgements We are very grateful to Kamilla Fredin for laboratory assistance as well as for a useful suggestion on the interpretation of the data, and to Professor John Markwell for sharing with us the technical bulletin describing the SPAD-502 that Spectrum Technologies Inc. (Plainfield, Illinois, USA) sent him in 1994.

References Arnon DI (1949) Copper enzymes in isolated chloroplasts; Polyphenoloxidases in Beta vulgaris. Plant Physiol 24:1–15 Azia F, Stewart KA (2001) Relationships between extractable chlorophyll and SPAD values in muskmelon leaves. J Plant Nutr 24:961–966 Bindi M, Hacour A, Vandermeiren K, Craigon J, Ojanpera K, Sellden G, Hogy P, Finnan J, Fibbi L (2002) Chlorophyll concentration of potatoes grown under elevated carbon dioxide and/or ozone concentrations. Eur J Agron 17:319–335 Campbell RJ, Mobley KN, Marini RP, Pfeiffer DG (1990) Growing conditions alter the relationship between SPAD-501 values and apple leaf chlorophyll. Hortscience 25:330–331 Castelli F, Contillo R, Miceli F (1996). Non-destructive determination of leaf chlorophyll content in four crop species. J Agron Crop Sci 177:275–283

123

46 Dwyer LM, Tollenaar M, Houwing L (1991) A nondestructive method to monitor leaf greenness in corn. Can J Plant Sci 71:505–509 Eisenberg D, Crothers D (1979) Physical chemistry with applications to the life sciences. Benjamin/Cummings, Menlo Park, California, USA Esposti MDD, de Sequeira DL, Pereira PRG, Venegas VHA, Salomao LCC, Filho JAM (2003) Assessment of nitrogenized nutrition of citrus rootstocks using chlorophyll concentrations in the leaf. J Plant Nutr 26:1287–1299 Fanizza G, Dellagatta C, Bagnulo C (1991) A nondestructive determination of leaf chlorophyll in vitis-vinifera. Ann Appl Biol 119:203–205 Fukshansky L, Martinez A, Remisowsky V, McClendon J, Ritterbusch A, Richter T, Mohr H (1993) Absorption spectra of leaves corrected for scattering and distributional error: a radiative transfer and absorption statistics treatment. Photochem Photobiol 57:538–555 Gausman HW (1985) Plant leaf optical properties in visible and near-infrared light. Texas Tech Press, Lubbock, Texas, USA Gelang J, Pleijel H, Sild E, Danielsson H, Younis S, Sellden G (2000) Rate and duration of grain filling in relation to flag leaf senescence and grain yield in spring wheat (Triticum aestivum) exposed to different concentrations of ozone. Physiol Plantarum 110:366–375 Gratani L (1992) A nondestructive method to determine chlorophyll content of leaves. Photosynthetica, 26:469–473 Hoel BO, Solhaug KA (1998) Effect of irradiance on chlorophyll estimation with the Minolta SPAD-502 leaf chlorophyll meter. Ann Bot-London 82:389–392 Jifon JL, Syvertsen JP, Whaley E (2005) Growth environment and leaf anatomy affect nondestructive estimates of chlorophyll and nitrogen in Citrus sp. leaves. J Am Soc Hortic Sci 130:152–158 Kapotis G, Zervoudakis G, Veltsistas T, Salahas G (2003) Comparison of chlorophyll meter readings with leaf chlorophyll concentration in Amaranthus vlitus: correlation with physiological processes. Russ J Plant Physl+ 50:395–397 Knapp AK, Carter GA (1998) Variability in leaf optical properties among 26 species from a broad range of habitats. Am J Bot 85:940–946 Markwell J, Osterman JC, Mitchell JL (1995) Calibration of the Minolta SPAD-502 leaf chlorophyll meter. Photosynth Res 46:467–472 Marquard RD, Tipton JL (1987) Relationship between extractable chlorophyll and an in situ method to estimate leaf greenness. Hortscience 22:1327 Martinez DE, Guiamet JJ (2004) Distortion of the SPAD 502 chlorophyll meter readings by changes in irradiance and leaf water status. Agronomie 24:41–46 McClendon JH, Fukshansky L (1990a) On the interpretation of absorption spectra of leaves – I. Introduction and the correction of leaf spectra for surface reflection. Photochem Photobiol 51:203–210

123

Photosynth Res (2007) 91:37–46 McClendon JH, Fukshansky L (1990b) On the interpretation of absorption spectra of leaves – II. The non-absorbed ray of the sieve effect and the mean optical pathlength in the remainder of the leaf. Photochem and Photobiol 51:211–216 Minolta (1989) Chlorophyll meter SPAD-502. Instruction manual. Minolta Co., Ltd., Radiometric Instruments Operations, Osaka, Japan Monje OA, Bugbee B (1992) Inherent limitations of nondestructive chlorophyll meters – a comparison of 2 types of meters. Hortscience 27:69–71 Murillo-Amador B, Avila-Serrano NY Garcia-Hernandes JL, Lopez-Aguilar R, Troyo-Dieguez E, Kaya C (2004) Relationship between a nondestructive and an extraction method for measuring chlorophyll contents in cowpea leaves. J Plant Nutr Soil Sc 167:363–364 Myers DA, Vogelmann TC, Bornman JF (1994) Epidermal focussing and effects on light utilazation in Oxalis acetosella. Physiol Plantarum 91:651–656 Porra RJ, Thompson WA, Kreidemann PE (1989) Determination of accurate extinctions coefficients and simultaneous equations for assaying chlorophylls a and b extracted with four different solvents: verification of the concentration of chlorophyll standards by atomic absorption spectroscopy. Biochim Biophys Acta 975:384–394 Richardson AD, Duigan SP, Berlyn GP (2002) An evaluation of nonivasive methods to estimate foliar chlorophyll content. New Phytol 153:185–194 Richter T, Fukshansky L (1996) Optics of a bifacial leaf: 2. Light regime as affected by the leaf structure and the light source. Photochem Photobiol 63:517–527 Schaper H, Chacko E (1991) Relation between extractable chlorophyll and portable chlorophyll meter readings in leaves of 8 tropical and subtropical fruit-tree species. J Plant Physiol 138:674–677 Vogelmann TC (1989) Penetration of light into plants. Photochem Photobiol 50:895–902 Wang QB, Chen MJ, Li YC (2004) Nondestructive and rapid estimation of leaf chlorophyll and nitrogen status of peace lily using a chlorophyll meter. J Plant Nutr 27:557–569 Xu W, Rosenow DT, Nguyen HT (2000). Stay green trait in grain sorghum: relationship between visual rating and leaf chlorophyll concentration. Plant Breeding, 119:365–367 Yadava UL (1986) A rapid and non-destructive method to determine chlorophyll in intact leaves. Hortscience 21:1449– 1450 Yamamoto A, Nakamura T, Adu-Gyamfi JJ, Saigusa M (2002). Relationship between chlorophyll content in leaves of sorghum and pigeonpea determined by extraction method and by chlorophyll meter (SPAD-502). J Plant Nutr 25:2295– 2301