Laboratory analysis of pigments in tree leaves has traditionally been conducted by extraction methods using several water-soluble organic solvents followed by spectrometric determination based on the maxima absorption spectra of Chlorophylls (Chls) and carotenoids in the red and blue spectral regions ( Minocha et al ., 2009 ). Quantification is routinely made by applying various equations developed for different solvents and spectrophotometer resolutions ( Barnes et al ., 1992 ; Wellburn, 1994 ). The solvent DMSO was introduced for pigment extractions as it avoids the need for grinding and centrifuging foliar samples required by other solvents, allowing a large amount of samples to be analysed and a higher stability of the extracted Chls with time ( Hiscox & Israelstam, 1979 ). The Chl _b peak in the red region is much less intense than the Chl _a and is superposed by the later in fresh leaves extracts. For this reason, Chl _b quantification is very sensitive to extract turbidity or high pigments concentration ( Lichtenthaler & Buschmann, 2001 ).

We investigated the drawbacks in the use of DMSO as solvent for pigment quantification in Quercus ilex L. leaves used for SPAD calibrations. An alternative extraction method with aqueous-acetone was studied.

Pigment concentrations analysed in the current year leaves using DMSO solvent were in the range of 70-190 B5g/cm ² and exhibited low Chl _a /Chl _b ratios (0.8 to 2.3). The SPAD regression performed with the DMSO method produced a significant correlation with the Chl _a concentrations analysed ( R ² =0.64, RMSE=0.20, p< 0.0001) ( Fig. 1A ). However the same regression with Chl _b concentrations was much weaker ( R ² =0.33; RMSE=0.23; p< 0.0001) ( Fig. 1B ). Gonsiorkiewicz Rigon et al . (2012) found a similar dispersion only for Chl _b calibration by DMSO extractions using a ClorofiLOGB.-1030. We measured high and variable absorptions readings at 750 nm that should be close to zero in all the DMSO extracts. A strong significant relation between the absorption at 750 nm and Chl _b was found ( R ² =0.83; RMSE=0.007; p< 0.0001) ( Fig. 1D ). The ratio between Chl _a /Chl _b decreased significantly as absorption at 750 nm increased ( R ² = 0.57; RMSE=0.01; p< 0.0001) ( Fig. 1E ). Likewise, a relationship between absorption at 750 nm and extract weight was observed ( R ² =0.38; RMSE=0.02; p< 0.0001) ( Fig. 1F ). DMSO seemed to co-extract compounds that absorb especially in the 649.1 nm peak used for Chl _b quantification, increasing 2-3 fold its value and narrowing the Chl _a /Chl _b ratio.

Extraction with DMSO increased significantly (20%) the Chl _a concentration in the younger leaves with lower pigment concentration and in 5% in the previous year leaves compared with the aqueous-acetone ( Table 1 ). The differences were much greater and significant, in the case of Chl _b concentrations with an increase of 115% and 110% in the DMSO extracts in current and previous year leaves respectively compared with the aqueous-acetone.

Minocha et al . (2009) detected difficulties in using DMSO for pigment quantification in tree leaves when compared with other solvents due to brown colorations. They attributed the colour interference to the tannins and phenolic compounds co-extracted by this solvent. We detected the same brown colorations only in the DMSO extractions (Figs. S1 and S2 [suppl]). However phenolics and tannins should be present in both extracts, since DMSO and aqueous-acetone are able to dissolve them ( Dai & Mumper, 2010 ; Alves et al ., 2013 ) and their absorption maxima in the UV was distant from the orange-red spectral region ( Fig. 2A ).

Phenolic compounds are easily oxidized and degraded at high temperatures and during long extraction times ( Dai & Mumper, 2010 ) forming stable polymerized brown coloured anthocyanin-tannin complexes ( Giusti & Wrolstad, 2001 ). The main spectral differences between the fresh pigment extracts obtained with DMSO and acetone from the same Q. ilex leaf are located in the 500-650 nm region ( Fig. 2A ). Brownish DMSO extracts showed higher absorbance in this region overlapping the absorption at 649.1 nm and leading to a Chl _b overestimation. Subtraction of absorbance at 750 nm ( Table 1 ) did not achieve reliable Chl _a /Chl _b ratios because the absorbance addition effect in the red region was not constant and higher as in the IR region ( Fig. 2A ). The oxidation process increased notably three days after extraction only in the DMSO extracts ( Fig. 2A ). Under the stronger oxidizing effect of DMSO ( Clark et al ., 2008 ), enhanced by the higher temperature and longer extraction time of this procedure, the co-extracted phenolic compounds are easily oxidized producing the brown colorations that interfere during the photometric measurements.

In the alternative extraction with aqueous-acetone, absorption at 750 nm was negligible. Chl _a /Chl _b ratios obtained for current and previous year leaves were in the range of 2.5-4.4 characteristics of sun-exposed leaves ( Lichtenthaler, 1987 ). Chls concentration during the vegetative period varied in the range 18-119 B5g/cm ² (Table S3 [suppl]). High SPAD-pigment calibration correlations (B5g/cm ² ) were obtained for all pigments: R ² =0.90, RMSE=0.16, p< 0.0001 for Chl _a ; R ² =0.91, RMSE=0.16, p< 0.0001 for Chl _b and R ² =0.90, RMSE=0.02, p< 0.0001 for carotenoids (N=167). The calibration equations obtained for the sum of Chl _a+b and carotenoids in B5g/cm ² were: Chl _a+b =exp(1.8244+0.0408*Spad) and C _x+c =(0.2907-0.0039*Spad) ^-1 . Filtered tissue powder after extraction was completely decoloured (Fig. S1 [suppl]). The weaker oxidizing properties of aqueous-acetone, the low temperature and the shorter extraction time prevented the oxidation of co-extracted polyphenolic compounds when this extraction method was used.

DMSO extracts of several perennial forest and shrub species were additionally analysed to confirm our results ( Fig. 2B ). All species had absorbance at 750 nm under 0.010 with the exception of Q. ilex that exhibited the maximum values (0.050). Furthermore, the ratio Chl _a /Chl _b was under 2 only in the Q. ilex samples. The differences in foliar phenolic concentration between species ( Hermes & Hedges, 2004 ) could explain the differences in the rapid development of colored interferences under the same oxidizing conditions. Analysis of total polyphenols using the Folin-Ciocalteu procedure in forest and shrub species extracted by DMSO were performed (mg gallic acid (GA) per gram dry leaf tissue). The Q. ilex leaves had the highest total polyphenol contents of all samples analyzed: 18.8 mg/g GA while all other leaf species were in the range 2.4 mg/g GA ( Laurus nobilis ) to 7.2 mg/g GA ( Arbutus unedo ). The development of colored interferences in photometric pigment determinations by DMSO could be also possible in leaves of other forest species with high polyphenol contents.

As conclusions, the use of the DMSO solvent should be avoided for photometric pigment analysis in Quercus ilex leaves. Cold extraction of finely homogenized foliar material with aqueous-acetone, free of interferences is recommended as we have proved that it provides more reliable pigment concentrations.

We thank the laboratory of Forest Chemistry (CIFOR-INIA) for the analysis of total polyphenols.