Crop Science Journal of Natural Resources and Life Sciences Education
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Published online 1 February 2006
Published in Crop Sci 46:671-680 (2006)
© 2006 Crop Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Ruiz, R. A.
Right arrow Articles by Maddonni, G. A.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Ruiz, R. A.
Right arrow Articles by Maddonni, G. A.
Agricola
Right arrow Articles by Ruiz, R. A.
Right arrow Articles by Maddonni, G. A.
Related Collections
Right arrow Seed Quality
Right arrow Crop Physiology & Metabolism
Right arrow Sunflower

CROP PHYSIOLOGY & METABOLISM

Sunflower Seed Weight and Oil Concentration under Different Post-Flowering Source-Sink Ratios

Ricardo Adolfo Ruiz and Gustavo Angel Maddonni*

Dep. de Producción Vegetal, Fac. de Agronomía, Univ. de Buenos Aires, Av. San Martín 4453, Ciudad de Buenos Aires (C1417DSE), Argentina

* Corresponding author (maddonni{at}agro.uba.ar)


   ABSTRACT
TOP
 NOTES
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Sunflower (Helianthus annuus L.) seed weight and oil concentration are commonly related to post-flowering source of assimilates (e.g., leaf area index duration, LAD). A predictive variable including both the source of assimilates and the sinks (i.e., seed number) would better account for seed weight and seed oil concentration variability of crops with contrasting seed number and canopy size. We established quantitative relationships between oil weight per seed components and post-flowering source–sink ratio. Field experiments were conducted in Argentina from 1998 to 2001. Four hybrids were cultivated under contrasting plant populations and nutrient supplies. A wide range of LAD (913–3130 m2°Cd m–2), seed number (4270–8880 seeds m–2), and seed weight (41–62 mg) was recorded. In contrast, seed oil concentration was not modified (about 530 mg g–1). Post-flowering source–sink ratio (LAD per seed) better accounted (r2 = 0.69) for seed-weight prediction than LAD (r2 = 0.42). Maximum seed weight (60 mg) was attained with source–sink ratios ≥0.33 m2°Cd seed–1. Results from our data set pooled together with others of different agro-ecological regions reveal that sunflower crops are normally growing under limiting post-flowering source–sink ratios and a 47% reduction of seed weight occurs when post-flowering source–sink ratio is dramatically (100%) reduced. Seed weight is only 23% increased at saturated source–sink ratios. In contrast, for the wide range of post-flowering source–sink ratios analyzed, seed oil concentration did not vary.

Abbreviations: fIPAR, fraction of photosynthetically active radiation intercepted • GLAI, green leaf area index • IPAR, intercepted photosynthetically active radiation • LAD, leaf area index duration • RMSE, root mean square error


   INTRODUCTION
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
IN GRAIN CROPS, such as maize (Zea mays L.), wheat (Triticum aestivum L.), soybean [Glycine max (L.) Merr.], and sunflower, grain yield per unit land area is the product of two main components: grain number per unit area and grain weight. Sunflower grain yield components are commonly referred to as seed number per unit area and seed weight. Since this crop is mostly cultivated for its oil, the marketable yield is expressed as oil yield per unit area, multiplying grain yield by the oil concentration of seeds.

In Argentina, current sunflower husbandry is based on the cultivation of hybrids under rainfed conditions with no fertilizer application (Mercau et al., 2001). Recent studies have revealed that Argentine commercial hybrids have more seeds ({cong}8000 vs. 5000 seeds m–2), which are slightly lighter ({cong}55 vs. 65 mg seed–1) and of greater oil concentration ({cong}500 vs. 400 mg g–1) than the open-pollinated cultivars used before 1970 (López Pereira et al., 1999a). Oil yield per square meter, however, exhibited a positive response to seed number and saturated at about 7500 seeds m–2 (López Pereira et al., 1999a). Above this threshold, oil yield per unit area did not vary (about 200 g m–2) in response to seed number. Thus, further increases in oil yield per unit area could be achieved by maintaining seed number per unit area around this threshold but increasing seed weight and/or oil concentration of seeds.

Seed weight and seed oil concentration are determined during the seed-filling period, i.e., from the end of flowering to physiological maturity (Aguirrezábal et al., 2003). Consequently, changes in the source of post-flowering assimilates could be reflected in final seed weight and in its oil concentration.

The source of post-flowering assimilates can be indirectly quantified as post-flowering leaf area index duration [(LAD); López Pereira et al., 1999b; de la Vega and Hall., 2002a] or as the intercepted photosynthetically active radiation (IPAR) per plant during seed filling (Dossio et al., 2000; Aguirrezábal et al., 2003). Both quantifications give a good description of differences in seed weight and seed oil concentration. For example, treatments imposed at flowering aimed to modify IPAR per plant during the seed filling period, decreased (shading treatments) or increased (thinning treatments) seed weight (Andrade and Ferreiro, 1996; Santalla et al., 2002). In these papers, however, the magnitude of seed-weight response to IPAR per plant modifications was not quantified. A quantitative approach was proposed by Dossio et al. (2000). A negative exponential function was used to describe the relationship between final seed weight and the accumulated IPAR per plant during the seed-filling period. The same function significantly described the relationship between seed oil concentration and the IPAR per plant of a high oil concentration potential (about 520 mg g–1) hybrid. In contrast, seed oil concentration of a low oil concentration potential hybrid (about 415 mg g–1) did not respond to IPAR per plant modifications (Dossio et al., 2000).

Seed weight depends not only on the source of assimilates during the seed-filling period but also on seed number per plant (i.e., the sink of assimilates). Seed number per plant may also be increased (thinning treatment) or decreased (shading treatment) when manipulative treatments are imposed within the critical period for seed set (–30d to +20d after flowering; Cantagallo et al., 1997, 2004). But because the effect of the treatment on seed number per plant was of a lower magnitude (less than 5 or 20% for shading and thinning treatments, respectively) than those applied on the source (more than 50 or 150% for shading and thinning treatments, respectively), seed weight was mainly related to the accumulated IPAR per plant (Dossio et al., 2000). Similarly, testing short periods of severe IPAR reduction (50 and 80%) during the seed filling, the 80% of both seed weight and seed oil concentration variability of a high oil concentration potential cultivar was accounted for by the accumulated IPAR per plant variability (Aguirrezábal et al., 2003). Because seed number per plant was slightly (about 27%) or not affected by treatments, an independent variable including both the source and the sinks (i.e., the source–sink ratio) did not improve seed-weight prediction. The source–sink ratio, however, would be an explanatory variable of seed weight, when comparing treatments which significantly affect both the source size and the sinks. Using this approach, variations in LAD per seed accounted for 72% of the seed-weight variability of the actual hybrids (about 8000 seeds m–2) and old open-pollinated cultivars (about 5000 seeds m–2), while seed oil concentration increased slightly with decreasing source–sink ratio (López Pereira et al., 1999b).

Finally, in field conditions many environmental restrictions such as water and nutrient supply, commonly take place early and/or late in the sunflower growth cycle (Mercau et al., 2001; Chapman and de la Vega, 2002), affecting leaf expansion, seed number per plant, and leaf area duration (Connor and Jones, 1985; Steer et al., 1986; Sadras et al., 1993a; Trápani and Hall, 1996). Moreover, plant population density, a decisive cultural practice aimed to maximize oil yield per unit area (Villalobos et al., 1994; Andrade, 1995; Barros et al., 2004), has an actual impact on seed number per plant, leaf area index, and LAD (Steer et al., 1986; Sadras and Hall, 1988; Ferreira and Abreu, 2001; Barros et al., 2004). Thus, variations of the post-flowering source of assimilates per seed could be recorded by the effect of the mentioned factors on the source, the sinks or both variables. Hence, for these conditions, seed weight and seed oil concentration variability would be best accounted for by the post-flowering source–sink ratio than by the source size. In this work, we have analyzed a data set from field experiments in which four modern Argentine hybrids were cultivated under contrasting environmental conditions (promoted by a combination of sites, years, plant populations, and nutrient supplies) aimed to obtain a wide range of post-flowering source–sink ratios. Our objectives were (i) to establish relationships between oil weight per seed components (i.e., seed weight and seed oil concentration) and the post-flowering source–sink ratio or the post-flowering source size and (ii) to define a quantitative approach for determining the magnitude of seed weight and seed oil concentration changes in response to post-flowering source–sink ratio modifications. For the second objective, we have reanalyzed published and unpublished data sets from several sunflower agro-ecological regions pooled together with our data set.


   MATERIALS AND METHODS
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Experimental Design and Growing Conditions
Field experiments were conducted in Argentina at the experimental unit of the Department of Plant Production of the University of Buenos Aires (34° 25'S, 58° 25'W) on a silty clay loam soil (Vertic Argiudoll, Soil Taxonomy, U.S. Department of Agriculture) under rainfed conditions.

Experiment 1 (E1) was sown on 26 Oct. 1998, Experiment 2 (E2) on 19 Oct. 1999 and Experiment 3 (E3) on 10 November 2000. In E1, hybrids Dekasol 3900 (DK3900), Dekasol 4030 (DK4030) and Nidera Paraiso 20 (Paraiso 20) were sown at a population density of 6 plants m–2. In E2, the same hybrids were sown at two population densities: 3 and 6 plants m–2. In E3, hybrids were sown at a density of 6 plants m–2; hybrid DK4030 was replaced by Nidera Paraiso 30 (Paraiso 30). Paraiso 20, DK4030, and DK3900 have achenes with a phytomelanin layer present, uncolored hypodermis, and black stripes in the epidermis. Achenes of Paraiso 30 lack the phytomelanin layer, have uncolored hypodermis, and have brown stripes in the epidermis. In E1 and E3, hybrids were arranged in a randomized complete block design with three and four replicates, respectively. In E2, treatments were arranged in a split plot design with three replicates, with plant population density as the main factor and hybrids as the subfactor. Each plot or subplot was made up of five (E1 and E2) or eight (E3) rows, 0.70 m apart, and 5 m (E1 and E3) or 10 m (E2) long. All experiments were hand-planted at three seeds per hill and thinned to the desired plant population at V2 to V4 (Schneiter and Miller, 1981). All experiments were kept free of weeds, insects, and diseases.

Crops were exposed to contrasting weather and soil conditions. Despite the fact that all experiments were conducted at the same experimental site, plots of E3 had impoverished physical soil properties (top soil compaction) for crop growth. In E1, urea was applied 30 d after seedling emergence (50 kg N ha–1) to minimize N restrictions. The other experiments were not fertilized. Owing to differences in sowing date, dates of flowering (7 January 1999, 2 January 2000 and 24 January 2001 , for E1, E2 and E3, respectively) and physiological maturity (22 February 1999 , 12 February 2000 and 14 March 2001, for E1, E2, and E3, respectively) differed among experiments. Total amount of rainfall in E1 (635 mm) and E3 (694.5 mm) was greater than that recorded in E2 (305 mm). Similarly, total rainfall during the pre- and postanthesis periods of E1 (276 and 359 mm, respectively) and E3 (306 mm and 388.5 mm, respectively) was greater than that registered during the same periods of E2 (106 and 199 mm, respectively). Mean air temperature (about 22.5°C) and solar radiation (about 20.3 MJ m–2 d–1) were similar in all the years. The flowering period in E3 was exposed to lower solar radiation (about 20.4 vs. about 23.5 MJ m–2 d–1) and to higher (about 25.6°C) or similar temperatures than those of the previous experiments (about 22.7 and 26.4°C in E1 and E2, respectively). A similar trend was recorded for the post-flowering period: solar radiation in E1 (about 20 MJ m–2 d–1) and E2 (about 21.9 MJ m–2 d–1) was higher than that of E3 (about 18.5 MJ m–2 d–1) and air temperature in E2 (about 25.7°C) and E3 (25.5°C) was higher than that of E1 (about 22.8°C).

Measurements
At V2 to V4, six plants were tagged in the central row of each plot or subplot. Phenology was recorded weekly on tagged plants from first anthesis (R5.0, Schneiter and Miller, 1981) up to physiological maturity (R9).

The fraction of photosynthetically active radiation intercepted by the crops (fIPAR) at flowering (R5.5) was calculated from PAR measured above the canopies and PAR below green leaves but above senesced leaves, at the bottom of the canopies. Five independent measurements were made at each position within each plot, between 1100 and 1400 h on clear days, with 1 m of a line quantum-sensor (LI-191SA, LI-COR, Lincoln, NE). These measurements were made diagonally across the rows to locate the sensor bar between two inter-row spaces.

Green leaf area at R5.5 was measured on tagged plants by a nondestructive method (Pereyra et al., 1982). Green leaf area index (GLAI) was calculated as the product of green leaf area per plant and plant population density. Area of senesced leaves (i.e., when half or more of their areas had yellowed) was discounted from GLAI at R5.5, to obtain weekly values of GLAI until maturity. Leaf area index duration was the integral of the evolution of GLAI values on a thermal time basis (base temperature of 4°C; Villalobos and Ritchie, 1992). Postanthesis source–sink ratio was calculated as the quotient of LAD and seed number per square meter.

Oil yield per square meter and oil yield components were determined at physiological maturity. Tagged plants were harvested to determine seed number per plant, seed weight and seed oil concentration. Samples were dried at 80°C until constant weight. Seed oil concentration was determined on one random sample of seeds of the whole head per replicate, by magnetic resonance analysis (Oxford 4000, Oxford Analytical Instruments). Oil weight per seed was calculated as the product of seed weight and seed oil concentration.

In E2 and E3, seed-weight and seed oil concentration dynamics of DK3900 and Paraiso 20 were followed during the entire seed-filling period. For this purpose, about 10 plants per hybrid and replicate with similar development were tagged at R5.0 for seed sampling. Every 7 d, one tagged plant per replicate was sampled. One hundred seeds per capitulum were taken from the ninth to the eleventh floral circle from the periphery of the floral disc. At least four samples during the effective seed filling period and one sample when maximum seed weight was recorded, were selected to determine seed oil concentration using magnetic resonance analysis.

Statistical Analysis
Results were subjected to analysis of variance to evaluate the effects of treatments and their interaction on GLAI, LAD, oil yield per square meter, and oil yield components. Association among variables was investigated using linear and nonlinear models. A bilinear model with plateau (Eq. [1] and [2]) was fitted by using the nonlinear routine of Table Curve V 3.0 (Jandel TBLCURVE, 1992).

Formula 1[1]

Formula 2[2]
where a and b are the intercept and the slope, respectively, of the linear regression corresponding to the first stage, x is the independent variable and the constant c is the unknown breakpoint of the function indicating the x value above which y is maximized (z).

When the bilinear model with plateau was fitted between seed weight and thermal time after flowering, b and c parameters estimated seed-filling rate and the duration of seed-filling period, respectively (de la Vega and Hall, 2002b).

The bilinear model with plateau was also fitted between (i) seed oil concentration and thermal time after flowering, (ii) seed number per square meter and GLAI, (iii) LAD and GLAI, (iv) oil weight per seed components and LAD, and (v) oil weight per seed components and LAD per seed.

Statistical comparisons between the bilinear with plateau and simple linear models were performed by calculating the root mean square error (RMSE) described in Potter and Williams (1994).

The fIPAR values at R5.5 were related to the corresponding GLAIs and an exponential function (Trápani et al., 1992) was fitted (Eq. [3]).

Formula 3[3]
where k is the light attenuation coefficient.

Database and Database Analysis
To study quantitatively the response of seed weight and seed oil concentration to the post-flowering source–sink ratio, we pooled together our data set with those published in previous works (Table 1). Works included in the database were those where the post-flowering source of assimilates was quantified (as accumulated IPAR per plant or LAD) and oil yield per square meter components were reported. Thus, post-flowering source–sink ratio was calculated as the quotient of (i) accumulated IPAR and seed number per plant or (ii) LAD and seed number per square meter. For each data set, post-flowering source–sink ratio, seed weight, and seed oil concentration of the same growing season were expressed in relative terms, by calculating the quotient of (i) the difference between the individual value of each variable and the corresponding annual average and (ii) the annual average. The representation of variables in relative units allowed for comparisons of data from different environmental conditions, cultivars and source quantifications. Linear and bilinear models with plateau (Eq. [1] and [2]), were fitted to the response of both relative seed weight and seed oil concentration to the relative post-flowering source–sink ratio. All cultivars and environments of the database (n = 173) were included in the models. We also fitted a response curve to the 10% least responsive and 10% most responsive data to describe the extremes of the database, following Borrás et al. (2004) methodology. Briefly, we sorted all data disregarding their origin by post-flowering source–sink ratio values and for each set of 10 continuous values the highest and lowest seed weight and seed oil concentration were used. Then, from the total of 173 data points, the 17 most and the 17 least responsive seed weights and seed oil concentrations across the magnitude of the post-flowering source–sink ratios were used. We fitted linear models to these two groups of values representing the extreme responses. It is important to note that the most responsive cases are the ones that had the largest seed weight or seed oil concentration increase per unit of post-flowering source sink-ratio increase and the largest seed weight or seed oil concentration decrease per unit of post-flowering source sink-ratio decrease. Conversely, for the least responsive cases, they were the data points that increased or decreased the least per unit of post-flowering source–sink ratio increase or decrease, respectively.


View this table:
[in this window]
[in a new window]
  		Table 1. Description of sowing date, plant population density, number of tested cultivars, growing conditions and manipulative treatments, and the country (including the latitude and longitude of the location) where the experiments were conducted.

 

   RESULTS
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Oil Yield per Unit Land Area and Oil Yield Components
Oil yield per square meter varied from about 90 to 230 g m–2 (Table 2) and significant differences among experiments (P < 0.001) and tested hybrids (P < 0.10–0.13) were detected. At the highest plant density, crops in E1 (about 218.5 g m–2) and E2 (about 199.3 g m–2) had higher oil yield per square meter than in E3 (about 109.2 g m–2). Hybrid DK3900 presented the lowest oil yield per square meter. Among oil yield components, seed number per square meter was the most sensitive to treatments (about 6343 seeds m–2, CV 0.22). Seed weight was less variable (about 52.2 mg, CV 0.17), and seed oil concentration was almost unaffected (about 530 mg g–1, CV 0.03). Consequently, differences (P < 0.01–0.10) among treatments and experiments in oil weight per seed were mainly related to seed weight (oil weight per seed = 0.90 + 0.51 seed weight; r2 = 0.97, n = 41, P < 0.001).


View this table:
[in this window]
[in a new window]
  		Table 2. Oil yield and seed number per unit land area, oil weight per seed and oil weight per seed components (seed weight and seed oil concentration) for different sunflower hybrids.

 
A comparison of seed-weight and seed oil concentration dynamics of achenes located at intermediate positions of the capitulum revealed that plant population only affected the temporal evolution of the former, without affecting that of the latter (Fig. 1 ). At the lowest plant density, seeds of DK3900 and Paraiso 20 exhibited a higher (P < 0.05) seed filling rate (about 0.067 mg °Cd–1) than at highest plant density (about 0.046 mg °Cd–1). At both plant densities, seeds of Paraiso 20 had similar seed filling duration (about 784°Cd), but seeds of DK3900 had longer seed-filling duration at 6 plants m–2 (796°Cd) than at 3 plants m–2 (657°Cd). In contrast, a common seed oil concentration rate (about 1.025 mg g–1 °Cd–1) and duration (about 545°Cd after flowering) was observed. Maximum oil concentration (about 542 mg g–1) was reached earlier (about 240°Cd before for Paraiso 20; and 112°Cd and 251°Cd before for DK3900 at 3 and 6 plants m–2, respectively) than maximum seed weights (about 57 and 40 mg at 3 and 6 plants m–2, respectively).


Figure 1
View larger version (31K):
[in this window]
[in a new window]
  		Fig. 1. Evolution of seed weight and seed oil concentration (oil concentration) of intermediate achenes, of hybrids DK3900 (A) and Paraiso 20 (B) cultivated at two plant population densities. Lines indicate the models fitted to the data. For DK3900: Seed weight = 9.8 + 0.067 thermal time (r2 = 0.97, n = 8, P < 0.001), for thermal time < 657°Cd; Seed weight = 53.6 mg, for thermal time ≥657°Cd (at 3 plants m–2); Seed weight = 6 + 0.042 thermal time (r2 = 0.76, n = 20, P < 0.001), for thermal time < 796°Cd; Seed weight = 39.8 mg, for thermal time ≥ 796°Cd (at 6 plants m–2); Oil concentration = –40 + 1.017 thermal time (r2 = 0.96, n = 18, P < 0.001), for thermal time < 569°Cd; Oil concentration = 539 mg g–1, for thermal time ≥ 569°Cd (at 3, 6 plants m–2). For Paraiso 20: Seed weight = 9.8 + 0.068 thermal time (r2 = 0.98, n = 8, P < 0.001), for thermal time < 737°Cd; Seed weight = 60.1 mg, for thermal time ≥737°Cd (at 3 plants m–2); Seed weight = 1.2 + 0.054 thermal time (r2 = 0.93, n = 20, P < 0.001), for thermal time < 755°Cd; Seed weight = 41.9 mg, for thermal time ≥ 755°Cd (at 6 plants m–2); Oil concentration = 8 + 1.03 thermal time (r2 = 0.95, n = 17, P < 0.001), for thermal time < 524°Cd; Oil concentration = 548 mg g–1, for thermal time ≥524°Cd (at 3, 6 plants m–2).

 
Oil yield per square meter was positively associated with seed number per square meter (oil yield m–2 = –0.42 + 0.027 seeds m–2; r2 = 0.72, n = 41, P < 0.001, RMSE = 27 g m–2. Fig. 2 ). A bilinear model with plateau fitted to oil yield per square meter and seed number per square meter did not improve oil yield prediction (r2 = 0.74, n = 41 P < 0.001, RMSE = 26 g m–2).


Figure 2
View larger version (23K):
[in this window]
[in a new window]
  		Fig. 2. Oil yield per unit land area as a function of seed number per unit land area. The three dashed lines show oil yield of seeds weighing 65, 50, and 35 mg assuming a seed oil concentration of 530 mg g–1. The solid line indicates the model fitted to the data.

 
Seed weight only accounted for 34% of the variation in oil yield per square meter (P < 0.001), but it was evident that for any given seed number per square meter there was a wide range in oil yield because of variations in seed weight (Fig. 2). Slight changes in seed oil concentration were not reflected in oil yield per unit area.

Source Size and Post-Flowering Source–Sink Ratio
The combined effect of years, hybrids, and plant population densities determined a wide range of GLAIs at R5.5 (1.17–4.48) and LADs (913–3130 m2 °Cd m–2) during seed filling (Table 3). At 6 plants m–2, GLAI and LAD significantly (P < 0.001) differed among experiments, with the higher values in E1 and the lower ones in E3. Differences among hybrids in GLAI (P < 0.05) were detected in E2 and E3. In contrast, hybrids did not differ in LAD. In E2, a significant (P < 0.01) hybrid x plant population density interaction in GLAI and LAD was recorded. The mentioned variables in DK4030 had a positive response to the increase of plant population density. In contrast, GLAI and LAD of the other hybrids were not modified by plant population.


View this table:
[in this window]
[in a new window]
  		Table 3. Green leaf area index at flowering (GLAI), leaf area index duration (LAD), and post-flowering source–sink ratio for different sunflower hybrids.

 
Green LAI at R5.5 accounted for 57% of the variation in seed number per square meter (Fig. 3A ). The bilinear model with plateau fitted to the data set indicates that seed number per square meter had a positive response to GLAI up to a threshold value (2.3 ± 0.19, P < 0.001) above which seed number was maximized (7326 ± 310 seeds m–2). This threshold value was close to the critical GLAI value (2.89), i.e., GLAI above which fIPAR ≥ 0.95, estimated from the exponential function fitted to fIPARs and GLAIs at R5.5 (fIPAR = 1– e(–1.034 GLAI); r2 = 0.84, n = 41, P < 0.001).


Figure 3
View larger version (27K):
[in this window]
[in a new window]
  		Fig. 3. Seed number per square meter (A) and leaf area index duration (B) as a function of green leaf area index at flowering. Data points represent values obtained in each experimental plot. Solid lines indicate the models fitted to the data.

 
A bilinear model with plateau also significantly (P < 0.001) described the response of LAD to GLAI for the whole data set (Fig. 3B). For GLAIs < 4.04 ± 0.24 (P < 0.001), LAD linearly increased with GLAI. Crops with GLAI greater than 4.04, had a similar LAD (3143 ± 121 m2 °Cd m–2).

Treatment effects both on the sources and the sinks were reflected in a wide range of post-flowering source–sink ratios (0.173–0.456 m2°Cd seed–1, Table 3). Crops in E1 exhibited higher (P < 0.001) source–sink ratios than those recorded in E2 and E3. In E2, increased plant population reduced (P < 0.01) source–sink ratio from about 0.307 to about 0.228 m2 °Cd seed–1, and differences among hybrids (P < 0.01–0.05) in this variable were detected in E2 and E3.

Oil Weight per Seed Components, Post-Flowering Source Size, and Post-Flowering Source–Sink Ratio
Variations in LAD accounted for changes in both seed weight (r2 = 0.38) and oil weight per seed (r2 = 0.34). The RMSE associated with the linear regression model fitted to the data was 6.9 and 3.7 mg for seed weight and oil weight per seed, respectively. Seed oil concentration variability was not related to LAD. A bilinear model with plateau improved the prediction of seed weight (r2 = 0.42, RMSE = 6.6 mg) and oil weight per seed (r2 = 0.41, RMSE = 3.6 mg) response to LAD variations (Fig. 4 ). Both variables showed a positive response to LAD and saturated at about 2500 m2°Cd m–2 (P < 0.001). Above this threshold value, seed weight (60 ± 1.9 mg), and oil weight per seed (31.2 ± 1.2 mg) were maximized. Post-flowering source–sink ratio, as the independent variable of the same bilinear model, improved both seed weight (r2 = 0.69) and oil weight per seed (r2 = 0.59) predictions (Fig. 5 ). The RMSE associated with these models (4.88 and 2.93 mg for seed weight and oil weight per seed, respectively) were lower than those where LAD was the predictive variable. Crops with a source–sink ratio greater than 0.33 m2 °Cd seed–1 (P < 0.001) yielded the maximum seed weight (60.1 ± 1.7 mg) and oil weight per seed (31.6 ± 0.92 mg). Linear regression model fitted to these data sets had lower determination coefficients (r2 = 0.59 and 0.50 for seed weight and oil weight per seed, respectively) and higher RMSE (5.7 and 3.2 mg for seed weight and oil weight per seed, respectively) than the bilinear model. Seed oil concentration values were not associated to post-flowering source–sink ratio variations.


Figure 4
View larger version (28K):
[in this window]
[in a new window]
  		Fig. 4. Seed weight (A) and oil weight per seed (B) as a function of leaf area index duration. Data points represent values obtained in each experimental plot. Solid lines indicate the models fitted to the data. RMSE = root mean square error of the models.

 

Figure 5
View larger version (28K):
[in this window]
[in a new window]
  		Fig. 5. Seed weight (A) and oil weight per seed (B) as a function of post-flowering source–sink ratio (leaf area index duration per seed). Data points represent values obtained in each experimental plot. Solid lines indicate the models fitted to the data. RMSE = root mean square error of the models.

 
Quantitative Response of Oil Weight per Seed Components to Post-Flowering Source–Sink Ratio
Crop variables of the database exhibited differences among locations and years in each location, in the mean post-flowering source–sink ratio, seed weight and seed oil concentration (Table 4). Coefficients of variation of post-flowering source–sink ratio (CV = 0.15–0.65) were higher than those of seed weight (CV = 0.08–0.29) and seed oil concentration (CV = 0.02–0.10).


View this table:
[in this window]
[in a new window]
  		Table 4. Annual average values, coefficient of variation (CV), and number of data points (n) of post-flowering source–sink ratio, seed weight, and seed oil concentration of the different data sets included in the analysis of Fig. 6.

 

Figure 6
View larger version (32K):
[in this window]
[in a new window]
  		Fig. 6. Relative seed weight (A) and relative seed oil concentration (B) as functions of relative post-flowering source–sink ratio in a number of experiments. The dashed lines show the theoretical slopes of 1 (full source limitation), and the horizontal dotted lines the slope of 0 (full sink limitation). The solid line in panel A indicates the model fitted to the data. RMSE = root mean square error of the model. Descriptions of experimental conditions of each data set are summarized in Table 1. Previously unpublished data sets Perez Alisedo and Ruiz, 1993; Dedominici and Ruiz, 2000.

 
When individual values of the mentioned variables were expressed in relative units, a bilinear model better described seed-weight response to post-flowering source–sink ratio (r2 = 0.44, n = 173, P < 0.001, RMSE = 0.13) than the linear regression model (r2 = 0.38, RMSE = 0.18). Seed weight had a positive response to source–sink ratio (Fig. 6A ) up to a relative source–sink ratio value {approx}0.47 ± 0.13 (P < 0.001) (i.e., a source–sink ratio 47% greater than the mean annual source–sink ratio of each experiment). The slope value (0.47 ± 0.04, P < 0.001) of the responsive section of the model, indicates that a 100% reduction of the mean source–sink ratio would determine a 47% reduction of seed weight. Conversely, crops with relative source–sink ratios > 0.47 (only 7 out of 173 data points) yielded seeds only 23 ± 0.5% heavier than the mean seed weight. For the most responsive cases (n = 17), relative source–sink ratio varied from –0.65 to 0.55 and relative seed weight varied from –0.28 to 0.67. More than 70% of these data corresponded to sunflower hybrids cultivated in late sowing dates in each location. A simple linear regression adequately describe the relationship between variables (relative seed weight = –0.0008 + 0.96 relative source–sink ratio; r2 = 0.67, n = 17, P < 0.001). The slope value (0.96 ± 0.17) was not different from 1, and the origin (–0.0008 ± 0.0047) did not differ from zero, indicating an equal seed weight change per each change in the source–sink ratio (i.e., a complete source limitation for seed growth). In contrast, considering the 10% least responsive data points (n = 17), no change in seed weight (mean relative seed weight = 0.001 CV 0.29) was found in response to either increase or decrease of post-flowering source–sink ratio.

In contrast to seed weight, seed oil concentrations of the database were not related to post-flowering source–sink ratios (Fig. 6B). For the wide range of relative post-flowering source–sink ratios (from –0.66 to 1.27), relative seed oil concentrations were close to zero. Hence, no model was fitted to these variables.


   DISCUSSION
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
CONCLUSIONS
 REFERENCES
 
Our experimental conditions allowed us to explore seed weight and seed oil concentration responses to post-flowering source–sink ratio of sunflower crops with an oil yield varying from the lowest to the potential one of modern hybrids cultivated under temperate climates (López Pereira et al., 1999a). In Argentina, regional and seasonal variation in sunflower oil yield per unit area is mainly accounted for by seed number per unit area, which is related to canopy ground cover at flowering (Mercau et al., 2001). In our experiments, contrasting environmental conditions during the pre-flowering period were reflected in maximum GLAIs determining seed number per square meter variability. Hence, sink demand of post-flowering assimilates was profoundly conditioned by vegetative growth during the previous period.

Post-flowering source size of assimilates, was also related to maximum GLAIs. Despite differences among years, plant populations and hybrids on GLAI at flowering, a single bilinear function with plateau significantly described the response of LAD to GLAI for the whole data set. In a previous work (López Pereira et al., 1999b), LAD response to GLAI was simply described by a linear regression, but almost all tested genotypes exhibited GLAIs < 4. In our work, hybrids fertilized and with adequate water availability had GLAIs greater than 4, but above this value, LAD response to GLAI was saturated. These results suggest a limit to increase LAD on the basis of pre-flowering vegetative growth. Crops with canopy sizes greater than the critical GLAI value (2.89) maximized light interception. Consequently, the lowermost leaf stratum of these canopies was subject to an impoverished light environment, which probably accelerated leaf senescence (Rousseaux et al., 1999), counterbalancing the expected positive effect of GLAI size on post-flowering LAD. Differences in the way that sunflower hybrids keep plant tissue green after flowering, i.e., stay green (Thomas and Howarth, 2000), should be taken into account to break this strong association between LAD and GLAI at flowering. Hybrids tested in our experiments did not differ in the post-flowering LAD. But in a previous study (de la Vega and Hall, 2002a), one hybrid from a wide set of modern genotypes with similar duration of the post-flowering period had the greatest stay green resulting from a slower leaf senescence rate, despite its maximum GLAI > 4.

Seed weight had a saturated response to LAD, similar to that previously observed when the source was quantified by accumulated IPAR during the seed-filling period (Dossio et al., 2000; Aguirrezábal et al., 2003). Source size quantified by LAD may overestimate post-flowering light capture (i.e., accumulated IPAR per square meter) if canopy size is above the critical GLAI value. Once {approx}95% of light is intercepted, GLAIs > 2.89 would continue to increase LAD without affecting accumulated IPAR. Hence, these crops attain a similar seed weight close to maximum values. In our experiments, this situation occurred, especially under favorable environments.

As was hypothesized for crops with contrasting LAD and seed number per square meter, the source–sink ratio improved seed-weight prediction. Crops with adequate resource availability per seed, such as those cultivated at 3 plants m–2 or at 6 plants m–2 but fertilized and irrigated, attained the maximum seed weights. Thus, the bilinear model fitted to data set had a single plateau value. Potential seed size did not vary with plant density, in contrast to that reported by Aguirrezábal et al. (2003). Consequently, in our data set the post-flowering assimilate availability per seed conditioned seed weight. Because seed oil concentration dynamic was not affected by treatments, maximum seed weight and oil weight per seed were attained at a similar source–sink ratio (≥0.33 m2 °Cd seed–1). Hence, differences among treatments in oil weight per seed were mainly related to seed weight (Villalobos et al., 1994).

These results were sustained when our data set was pooled together with those from different locations, hybrids, and years. Moreover, the analysis in relative units proved useful (i) to quantify the magnitude of seed weight and seed oil concentration responses to post-flowering source–sink ratio, independently of the technique used to define the source, and (ii) to determine to what degree seed-weight differences are due to competition among seeds for insufficient source quantity. For the data analyzed, seed oil concentration was almost unaffected when post-flowering source–sink ratio was drastically modified. Thus, as was observed in maize (Borrás et al., 2002), oil concentration is a very conservative seed component. Conversely, seed-weight variability was explained by changes of post-flowering source–sink ratio. Mean seed-weight change per source–sink ratio change in the responsive part of the curve (0.47 ± 0.04) was close to that (0.51 ± 0.07) described for other sunflower cultivars (López Pereira et al., 1999b). These values are less than 1, suggesting that additional carbon from increased photosynthetical activity during seed filling (Evans, 1993) and/or greater contribution from reserves (Hall et al., 1989; 1990; Sadras et al., 1993b) may have attenuated the effects of a lower source–sink ratio. Under limiting source–sink ratios, sunflower (Sadras et al., 1993b) and soybean (Borrás et al., 2004) are more efficient in the use of assimilates stored for seed growth than maize (Kiniry et al., 1992). Thus, seed-weight sensitivity to reductions in post-flowering source–sink ratio differs among these summer crops [e.g., 0.41 and 0.75 for soybean and maize, respectively (Borrás et al., 2004) and 0.47–0.51 for sunflower (this paper; López Pereira et al., 1999b)]. In addition, considering the saturation source–sink ratios quoted by Borrás et al. (2004), sunflower (0.47, this paper) commonly grows under less source limitations for seed growth than soybean (0.99) but under greater source limitation than maize (0.18).

Under more restrictive environments, such as late sowing dates and high latitudes (Andrade, 1995; Bange et al., 1997; de la Vega and Hall, 2002b), some genotypes perceived a full post-flowering source limitation to seed filling (i.e., the slope was not different from 1). Probably, low temperature and solar radiation values commonly registered under these environments affected assimilate production and translocation to seeds (Bange et al., 1998).

Seed-weight response to high post-flowering source–sink ratios may be conditioned by pre-flowering growing conditions. A reduction of the assimilate availability per plant during floret growth before anthesis, such as that expected at high stand densities, reduces ovary size, conditioning final seed weight (Cantagallo et al., 2004). Thus, any increase of the post-flowering source size, like that produced by thinning treatments (e.g., from 7.2–4.5 plants m–2) during seed filling, could not be reflected in seed weight. These mechanisms could account for the lack of response observed in few cases of the database at high source–sink ratios.


   CONCLUSIONS
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
REFERENCES
 
The study of sunflower crops cultivated at contrasting environmental conditions revealed that oil weight per seed was mainly correlated to seed weight because seed oil concentration was almost constant (about 530 mg g–1, CV 0.03). Variations in post-flowering source–sink ratio better account for seed-weight variability than post-flowering source size. The analyses of sunflower crops from different agro-ecological regions revealed for most of the data, that seeds were growing under limiting source–sink ratios, and a saturation source–sink ratio value could not be accurately determined. Breeders should consider genotypic variability in the efficiency of reserve mobilization to seeds as a trait to improve current potential oil yield.


   ACKNOWLEDGMENTS
 
Authors wish to thank A. Fernández, A. Bertero de Romano and S. Katz for providing the seeds and seed oil analyses, A. Perez Alisedo, S. Dedominici, M. P. López Montes, and M. Tinaro for their valuable help in the experimental studies, A. J. de la Vega, M. López Pereira and J.F.C. Barros for their data sets, and E. Whitechurch for the revision of English style. G.A. Maddonni is a member of Conicet, the Scientific Research Council of Argentina.


   NOTES
TOP
 NOTES
ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Financial support, in part from the Univ. de Buenos Aires (UBACyT JG20). G. A. Maddonni is a member of CONICET.


   REFERENCES
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 




This article has been cited by other articles:


Home page
 Agron. J.Home page
V. D. Zheljazkov, B. A. Vick, M. W. Ebelhar, N. Buehring, B. S. Baldwin, T. Astatkie, and J. F. Miller
Yield, Oil Content, and Composition of Sunflower Grown at Multiple Locations in Mississippi
Agron. J., May 7, 2008; 100(3): 635 - 642.
[Abstract] [Full Text] [PDF]

Home page
 Crop Sci.Home page
N. G. Izquierdo, G. A. A. Dosio, M. Cantarero, J. Lujan, and L. A. N. Aguirrezabal
Weight per Grain, Oil Concentration, and Solar Radiation Intercepted during Grain Filling in Black Hull and Striped Hull Sunflower Hybrids
Crop Sci., March 19, 2008; 48(2): 688 - 699.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Ruiz, R. A.
Right arrow Articles by Maddonni, G. A.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Ruiz, R. A.
Right arrow Articles by Maddonni, G. A.
Agricola
Right arrow Articles by Ruiz, R. A.
Right arrow Articles by Maddonni, G. A.
Related Collections
Right arrow Seed Quality
Right arrow Crop Physiology & Metabolism
Right arrow Sunflower

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
The SCI Journals Agronomy Journal Vadose Zone Journal