Quantitative genetic analysis of methylxanthines and phenolic compounds in mate progenies

The objective of this work was to determine the contents of methylxanthines, caffeine and theobromine, and phenolic compounds, chlorogenic and caffeic acids, in 51 mate progenies (half‐sib families) and estimate the heritability of genetic parameters. Mate progenies were from five Brazilian municipalities: Pinhão, Ivaí, Barão de Cotegipe, Quedas do Iguaçu, and Cascavel. The progenies were grown in the Ivaí locality. The contents of the compounds were obtained by high performance liquid chromatography (HPLC). The estimation of genetic parameters by the restricted maximum likelihood (REML) and the prediction of genotypic values via best linear unbiased prediction (BLUP) were obtained by the Selegen ‐ REML/BLUP software. Caffeine (0.248–1.663%) and theobromine (0.106–0.807%) contents were significantly different (p<0.05) depending on the region of origin, with high individual heritability (ĥ2>0.5). The two different progeny groups determined for chlorogenic (1.365–2.281%) and caffeic (0.027–0.037%) acid contents were not significantly different (p<0.05) depending on the locality of origin. Individual heritability values were low to medium for chlorogenic (ĥ<0.4) and caffeic acid (ĥ<0.3). The content of the compounds and the values of genetic parameters could support breeding programs for mate.


Introduction
Mate (Ilex paraguariensis St. Hil.), a species of the family Aquifoliaceae, is native to southern South America, and almost 80% of its natural occurrence is in the five southernmost states of Brazil, where it shows relevant economic and social importance.
The contents of these compounds are determined by plant origin, genetic and environmental variability, harvest time, and processing (Athayde et al., 2000). Contents also vary according to the seasons (Da Croce, 2002), varieties and progenies (Scherer et al., 2002;Cardozo Junior et al., 2007).
Estimation of genetic parameters is important for quantitative information of the genic action, to indicate the most appropriate selection scheme, and to estimate the selection progress (Resende, 2002a). The genetic control of a certain trait may be evaluated by its heritability, which corresponds to the proportion of the total variability of a genetic nature, or the ratio between the genetic variance and the total variance. The narrow sense heritability is important for sexual propagation and allows estimating the possibilities of gain of a character (Sturion, 1993).
Using the test applied in the present research, Sturion et al. (2004) observed that mate from the Barão de Cotegipe, RS, Quedas do Iguaçu, Cascavel, and Ivaí, PR, regions are the most productive. They also determined the narrow sense heritability at family mean level for tannins, phenolic compounds, and caffeine contents, ranging from medium to high, which indicates a good possibility of genetic gain by progeny selection.
There are few studies on the breeding of mate aiming at the improvement of the traits of its commercial products and, possibly, its biological properties.
The objective of this work was to evaluate the methylxanthines and phenolic compounds present in the leaves of 51 mate progenies (half sib-families) and to estimate genetic parameters, such as the variance coefficients and heritability.
The Ivaí climate is Cfb-type, mesothermal humid subtropical. Warm summers and infrequent frosts occur in this region, and there is a tendency to rain concentration in the summer and without a well-defined dry season (Iapar, 2008). The soil is Latossolo Vermelho distrófico (Rhodic Hapludox) comprising 72% clay and 4% slope (Santos et al., 2006).
Experiments were arranged in a randomized block design in ten replications with six plants in a 3x2 m spacing. Progeny selection was randomized without hierarchic arrangement within each geographical origin. The plants were hand harvested in August 2004 at a leaf age of 12 months; in November 2004, leaf age of 18 months, and in June 2005, leaf age of 24 months.
The samples analyzed correspond to a combined half-sib family test of the geographical origins and progenies. Provenances with high leaf mass production were selected. Leaves (100 g) from the medium third of the plant were collected and immediately brought to the laboratory. Each sample was boiled in water for 10 s; the green and clean leaves were separated and dried in a forced draft oven (45ºC, 48 hours). Afterwards, the leaves were ground, placed on paper, wrapped with plastic, and refrigerated until analysis.
The methylxanthines caffeine and theobromine as well as the phenolic acids chlorogenic and caffeic were extracted from the leaves (1 g) by maceration in 50 mL methanol:water (70:30, v/v) for analysis . The extracts were filtered using 0.45 µm nylon filters and chromatographed on a Shimadzu Mod. SCL-10A, (Shimadzu, Kyoto, Japan) high performance liquid chromatography (HPLC) system. A 5-μm Supelcosil LC-18 (Sigma-Aldrich Inc., USA), 4.6x250 mm analytical column was used. The column was maintained at 30°C using the CTO-10AS (Shimadzu, Kyoto, Japan) HPLC integrated oven. The solvent system consisted of acidulated water with 0.3% acetic acid, and methanol. The solvents were run at a flow rate of 1.0 mL min -1 using the following linear gradient: 15% to 20% methanol for 20 min, 20% to 85% methanol for 5 min, and 85% methanol for 5 min. Detection was monitored at 265 nm for caffeine and theobromine, and at 325 nm for caffeic and chlorogenic acids using an SPD-10A UV-vis detector (Shimadzu, Kyoto, Japan) coupled to the chromatograph. All samples were run in duplicate. Chromatographic peaks were identified by comparing retention times with those of caffeine, theobromine, and chlorogenic and caffeic acid standards (Sigma Chemical Co., St. Louis, USA) recorded in the same conditions. All reagents used were of the purest grade available or chromatographic grade. Calibration curves were obtained through the mentioned standards after dilution in the mobile phase. Linearity was determined by regression and precision as well as accuracy were determined using the variation coefficient (CV<3%). The correlation coefficients obtained were r 2 = 0.9999 for caffeine, r 2 = 0.9976 for theobromine, r 2 = 0.9997 for chlorogenic acid, and r 2 = 0.9995 for caffeic acid.
The genetic and phenotypic parameters were estimated by the restricted maximum likelihood (REML) method for variance component estimation and the best linear unbiased prediction (BLUP) procedure for breeding value prediction and genetic selection. The analyses were performed using the Selegen REML/BLUP software (Resende, 2002b;Missio et al., 2004;Chiorato et al., 2008). This Model includes random block experiments, half-sib progenies, several plants per plot, and several populations. In the individual analysis by trial, the linear model applied to the progeny trial data was: y = Xb + Za + Wc + e, where: y, b, a, c and e are data vector, block effects (fixed), additive genetic effects (random), plot effects (random), and random error effects, respectively; X, Z and W: incidence matrices for b, a, and c, respectively.
Distributions and structures of means and variances: The covariances among all random effects of the model were considered zero. Thus: where: Mixed model equations: where: is the narrow-sense individual heritability in the block; is the coefficient of determination of plot effects; is the coefficient of determination of provenance effect; ŝ2 a is the additive genetic variance; ŝ2 c is the variance among plots; ŝ2 e is the residual variance; A is the additive genetic correlation matrix among individuals under evaluation.
The data follow a normal distribution. The significance of the observed differences was evaluated by SISVAR package (Ferreira, 2008), using random block analysis and the Scott-Knott grouping test at 5% probability. All analyses were performed in duplicate and the results are given as dry matter percentage (w/w).

Results and Discussion
The mean contents of caffeine and theobromine varied from 0.248% (progeny 106) to 1.663% (progeny 11) and from 0.106% (progeny 84) to 0.807% (progeny 121), respectively (Table 1). Based on statistical significance, eight groups stand out for caffeine content and seven for theobromine. Comparison of the caffeine and theobromine contents reveals a significant and inverse (r 2 = -0.841) correlation. The interesting fact verified in the present study is that mate progenies comprising low caffeine contents showed high theobromine contents, which was confirmed by the inverse correlation between these methylxanthines.
The chlorogenic acid content results corroborate those of Filip et al. (2001), while caffeic acid contents are higher. Both chlorogenic and caffeic acids showed low concentrations in mate, and di-caffeoylquinic derivatives such as 3,4-DCQ, 3,5-DCQ, and 4,5-DCQ are predominant (Clifford & Ramirez Martinez, 1990;Filip et al., 2001). The variability observed for the contents of these two phenolic acids is smaller than those of methylxanthines. Table 2 shows the caffeine and theobromine contents and ratios in progenies from five Brazilian geographical origins that showed significant differences. Progenies from Pinhão, PR, had low caffeine and high theobromine contents relative to the other origins. Whereas Table 1. Mean (%, w/w) caffeine, theobromine, chlorogenic and caffeic acid contents in 51 mate (Ilex paraguariensis) progenies grown in Ivaí, PR, Brazil (1) .
(1) Means followed by different letters in the same column are significantly different by Scott-Knott test, at 5% probability.  et al. (2007) in 16 mate progenies from four Brazilian geographical origins grown in three different locations. The analyses of caffeine and theobromine contents of all progenies here reported reveals a variability that may be exploited in breeding programs. The presence of caffeine and theobromine, which are responsible for the stimulating and tonic effects of mate (Cardozo Junior et al., 2007), represents a feasible indicator for the selection of progenies and the development of alternative mate products. Furthermore, progenies comprising very high caffeine and theobromine contents might also be used in improvement programs to obtain raw materials with predefined chemical compositions.
The chlorogenic acid content in Pinhão progenies was significantly higher than the others. The caffeic acid contents ranged from 0.030 to 0.0319% among origins (Table 2).
In general, the coefficient of determination of the plot effects (c 2 plot ), which indicates the environmental variations between plots, was low for caffeine and theobromine (Table 3). The coefficient values of additive genetic variation for individuals (CVgi%), which express the genetic variation of the evaluated compound contents, ranged from 25.82 to 38.91% for caffeine, and from 37.33 to 68.05% for theobromine, indicating a potential for character improvement. The coefficient values of the additive genetic progeny variation (CVgp%), which quantifies the genetic variation between progenies, ranged from 19.44 to 29.03%, for caffeine, and from 18.66 to 46.77%, for theobromine.
There was a change in c 2 plot from intermediate to high values for both chlorogenic and caffeic acids, which may be due to environmental variations between plots.
Moreover, the CVgi%, CVgp%, and heritability values were lower than those of the methylxanthines. Montagnon et al. (1998) obtained similar heritability estimates for chlorogenic acid in Coffea canephora, which correlated with the phenotypic characteristics of the species The estimated coefficient of narrow sense heritability (ĥ 2 ) of caffeine and theobromine in progenies grown in the Ivaí, PR, region were high and associated with low standard deviation, which is related to the estimated accuracy ( Table 4). The ĥ 2 of chlorogenic and caffeic acids varied from low to intermediate values.
Additive genetic variance (ŝ 2 a ) is one of the most important components of variance to be determined. It demonstrates the potential of the population for improvement of a character by sexual reproduction. These findings are of particular interest because the lower c 2 plot values for caffeine (Table 3) indicate that possible environmental differences between plots, mainly in soil characters, did not significantly influence the total variability composition. As noted in the present work (Table 4), the heritability estimates for caffeine and theobromine were high and associated with low standard deviation, which is related to the estimated accuracy. In the case of high heritability, the genetic variation represents most of the phenotypic variation in response to environmental variation (Resende, 2002b). The enhanced high heritability values for caffeine and theobromine indicate high genetic control in the conditions evaluated. A high heritability value indicates a good possibility of genetic gain with selection (Missio et al., 2004). High heritability values for caffeine were also observed in mate progenies from the Missiones region, Argentina (Scherer et al., 2002). Schneider et al. (2006) have reported on the quantification of methylxanthines, phenolic compounds and saponins in mate grown in Argentina. Scherer et al. (2002) verified high saponin index heritability values (narrow sense heritability = 0.59 and broad sense heritability = 0.86) for the progenies evaluated, which were compatible to those of caffeine. Their findings indicate the possibility of applying selection strategies to modify the mate saponin content in a manner similar to that used for caffeine and theobromine. Previous studies in native mate populations revealed high genetic variability.
This high intrapopulation variability may be related to the reproductive trait (allogamy), which permits intense gene flow between populations (Wendt et al., 2007). In addition, Cansian et al. (2008) also found that mate intrapopulation variances were larger than the interpopulation ones. In studies on the provenance and progeny test in the Ivaí region, Sturion et al. (2004) concluded that high genetic gains might be obtained with the selection of individuals based on genetic values of leaf biomass weight.

Conclusions
1. Mate progenies show significant differences in caffeine, theobromine, chlorogenic and caffeic acid contents.
2. Caffeine and theobromine contents in the progenies varied significantly from different geographical origins.
3. The narrow-sense individual heritability in the block (ĥ 2 ), the coefficient of individual additive genetic variance (CVgi%) as well as the coefficient of additive genetic variance between progenies (CVgp%) values of both caffeine and theobromine indicate a good possibility of genetic gain. Table 4. Estimates of narrow sense heritability (ĥ 2 ) of caffeine, theobromine, chlorogenic and caffeic acids in mate (Ilex paraguariensis) progenies grown in Ivaí, PR, Brazil.