Pollen production pattern in the capitulum of the cultivated sunflower ( Helianthus annuus L . )

The intra-plant pollen production in cultivated sunflower (Helianthus annuus L.) was quali- and quantitatively studied. Two modern self-pollinated, high oleic sunflower hybrids were grown during two consecutive seasons: 2009 in three planting dates, and 2010 in two planting dates. Two cross-pollinated varieties of sunflower, Hopi and Havasupai, were also studied to determine if an anomalous intracapitulum pollen gradient observed in preliminary studies could be a remnant character inherited from primitive genotypes. Pollen grains per flower (PGF), anther volume (AV), pollen grain volume (PGV) and pollen viability (PV) were measured in three capitulum sectors, external (ES), middle (MS) and internal (IS). At harvest, fully developed fruits (FDF) and incompletely developed fruits (IDF) per capitulum sector were counted. Based on seasonal averages of hybrids and planting dates, the capitulum IS showed the highest PGF number (40864 ± 6519) followed by the MS (35243 ± 5242) and the ES (28273 ± 3102). Anther length (AL) and width (AW) were higher in the IS (5.3 and 0.8 mm, respectively) compared with the ES (6.4 and 0.6 mm, respectively); ES and the MS had intermediate values (5.2 and 0.7 mm, and 4.9 and 0.6 mm, respectively). No significant differences (p=0.18) were observed in PGF per sector in capitula of Havasupai (average= 42679 ± 4107). On the other hand, Hopi flowers of the MS and IS produced a larger number of pollen grains (average = 53412 ± 3553) than those of the ES (37050 ± 5947). PGV showed no significant differences in both hybrids, capitulum sectors and growing seasons (average = 22168 ± 1042 μm3). A positive relationship between AL and AW and the total number of pollen grains was also observed. Pollen viability in both growing seasons was greater than 97% in all cases. At harvest, in all cases, the higher number of FDF was observed in the ES and MS (598 and 456 respectively) and the number of IDF was higher in the IS (132). Given that intra-plant pollen production of both, pollen quantity and viability were not limiting, the origin of IDF in different regions of the capitulum could not be explained by the lack of pollen.The finding of an a typical positive centripetal gradient of pollen grains in the capitulum as compared with other racemose inflorescences is interpreted as an adaptive response to the time sequence of anthesis in this species to ensure a homogeneous pollination in the capitulum.

Astiz V & LF Hernández, FYTON 83 (2014) was greater than 97% in all cases.At harvest, in all cases, the higher number of FDF was observed in the ES and MS (598 and 456 respectively) and the number of IDF was higher in the IS (132).Given that intra-plant pollen production of both, pollen quantity and viability were not limiting, the origin of IDF in different regions of the capitulum could not be explained by the lack of pollen.The finding of an atypical positive centripetal gradient of pollen grains in the capitulum as compared with other racemose inflorescences is interpreted as an adaptive response to the time sequence of anthesis in this species to ensure a homogeneous pollination in the capitulum.

INTRODUCTION
The most common reasons for the origin of seedless or "incompletely developed" fruits or IDF (Lindstrom et al., 2004) in the sunflower capitulum are failures in fertilization, physiological or morphological defects in the ovary and embryo, and vascular deficiencies at the receptacle-ovary interphase (Hernández & Orioli, 1991;Alkio & Grimm, 2003;De Grandi-Hoffman & Chambers, 2006).
Regarding pollination, even though the literature indicates that the sunflower is a self-incompatible insect-pollinated plant, to reduce the dependence on pollinators for seed production under field conditions, breeders have broken the selfincompatibility system and selected self-fertility in cultivated varieties (Roumet et al., 2012 and references therein).The selective pressures on self-compatible or self-incompatible varieties might have been very different, but always expected to have an optimum pollen presentation pattern producing very different "optimal solutions".Whereas in self-incompatible varieties, fruits cannot be produced without visits of pollinators and the patterns of pollen presentation could be crucial for maximum seed set, pollinators are not strictly required for seed set in selfcompatible varieties.Thus, the production of pollen could account by itself for successful pollination and eventually yield.
From a biological point of view, most primitive sunflower varieties and old varieties are self-incompatible and require cross-pollination for fruit setting.The choice of pollinators among those genotypes producing large amounts of efficient pollen could be of strategic importance for fruit production.This choice could be particularly important in modern sunflower hybrids.
In sunflower, IDF causes important reductions in seed and oil production.Knowledge of pollen production, and determination of the reasons and the degree of the observed reduction in fecundity, are essential pre-requisites for sunflower management, improvement and breeding programs.
To date, knowledge about the amount of pollen grains produced by different florets of the sunflower inflorescence (capitulum) has received little attention, and only a few papers are available regarding this topic (Hernández, 2008;Astiz et al., 2011;Astiz & Hernández, 2013).
The aim of the present work was to determine quali-and quantitatively the intraplant pollen production and IDF generation in sunflower.High oleic hybrids were used, since these genotypes have recently been incorporated into the value chain of sunflower production, and have relevant importance for the future sunflower production in Argentina.

MATERIALS AND METHODS
Plant material.Two self-fertile high-oleic sunflower hybrids, Dekasol Oilplus (DKO) 3845 and DKO3945 (Monsanto Argentina S.A.I.C), were sown at the Agronomy Department -UNSur (Bahía Blanca,38˚ 45' S;62˚ 11' W ). Experiments were conducted during two consecutive growing seasons: 2009 with three planting dates, and 2010 with two planting dates.Preliminary results obtained in our laboratory showed an anomalous intracapitulum pollen gradient, quite different from other racemose inflorescences.To help determine if these results could correspond to a remnant character inherited from primitive genotypes, two crosspollinated varieties of sunflower (i.e., Hopi and Havasupai: Heiser, 1951) were also grown during the second growing season.
After seedling emergence, plant density was adjusted to 6 plants/m 2 by manual thinning.Each growing season, a completely randomized split-plot experimental design was used, with hybrid assigned to main plots and planting date to subplots.Three replicates were used per treatment.Each subplot had three rows 0.70 m apart and 6.0 m long.
Soil moisture was kept near field capacity by drip irrigation, and weeds were controlled manually.At the developmental stage V4 (Schneiter & Miller, 1981), plants were fertilized with 50.0 kg N/ha using NH 4 NO 3 .During the two-year study, mean air temperature (˚C), solar radiation (mmol/ m 2 /s), and rainfall (mm) were recorded daily at one-hour intervals.Table 1 summarizes the seasonal pattern of environmental variables recorded for the two experimental periods.Both years had similar environmental conditions during Table 1.Mean environmental data for three developmental crop phases for each year.Anthesis (A) and physiological maturity (PM) dates used for defining the mean crop stages represent the average for both hybrids in each experimental year.Crop phase A represents the 15-d interval centered on the mean date of full anthesis.Crop interval S-A is the time elapsed between sowing and 7 days before the mean date of full anthesis.A-PM is the interval between 7 days after full anthesis and the mean date of PM for both hybrids.

Environmental factor
Hybrids developmental phases three study phases of crop development.Maximum and minimum temperatures during the reproductive period, as well as the average daily incident radiation, were within the optimum range for sunflower (Connor & Hall, 1997).
Pollen count and viability determination.Three plants per replicate were used for measurements.Quali-and quantitative pollen analysis were made in three capitulum sectors, external (ES), middle (MS) and internal (IS), each equal to 1/3 of the capitulum radius.Pollen grains per flower (PGF) were counted as follows: five pre-anthesis flowers (i.e., 25 anthers) per capitulum sector, planting date, season, and genotype were removed from the inflorescences, and placed in 2.5 mL Eppendorf™ vials containing FAA fixative solution (Ruzin, 1999).The pollen was removed from the anthers at room temperature within 5-7 days from sampling.To facilitate the removal of pollen grains from the anthers, FAA was drained from each vial and replaced by 2 mL of a 1% aqueous solution Triton™ X-100 (Octyl Phenoxy polyethoxyethanol).Flowers were then crushed with a glass rod into the vial and stirred in a vertical vortex mixer to achieve a homogeneous suspension of pollen.A Neubauer hemocytometer was used to count the pollen grains, following Godini (1981).Two drops of the suspension containing the pollen grains were placed with a micropipette into each of the two main chambers of the hemocytometer.After placing the cover glass, pollen grains were counted in each of the nine small chambers forming the two main chambers.The number of pollen grains per anther was consequently calculated as the product of the observed pollen grains in each small chamber by a dilution factor, divided by the number of anthers in each vial (i.e., 25).Finally, the num- Hernández, FYTON 83 (2014) ber of pollen grains per flower was calculated by multiplying the number of anthers per flower times the number of pollen grains per anthers.
Pollen viability (PV) was determined using Alexander's stain (Alexander, 1969) using Malachite green, which stains the cell wall cellulose, and Acid Fuchsine, which stains the protoplasm.Viable pollen grains stain dark, while non-viable ones remain light green.Pollen grain diameter was measured in samples of 60 grains taken from flowers on each capitulum sector.Pollen grain volume (PGV) was then calculated using the formula of the sphere.Using five flowers per each capitulum sector used for pollen count, anther length (AL) and width (AW) were also measured considering a transverse distance of both pollen sacs at the equatorial zone.Those values were used to estimate the volume of each pollen sac (VPS), calculated as the volume of an ellipsoid of revolution: PSV (mm 3 ) = 4/3 .π .[(AW/2) 2 .AL]. Anther volume was then obtained as: AV (mm 3 ) = PSV (mm 3 ) . 4.
Pollen grain diameter and anther size measurements were made microscopically using an ocular micrometer.
Yield analysis.Six plants per replicate were randomly chosen and labeled at the reproductive stage R5.1 (Schneiter & Miller, 1981).The harvest was made with a fruit moisture content of 13 to 18%.Capitula were taken to the laboratory for analysis where they were split into three sectors, IS, MS and ES.The fruits were counted per sector and classified into two categories: fully developed fruits (FDF) and IDF (Lindström et al., 2004).

Statistical analysis.
All data were statistically analyzed by a three-way analysis of variance (ANOVA) using the Infostat statistical software package v. 2010 (Di Rienzo et al., 2008).The least significant difference (LSD) test was used to separate differences between planting dates, hybrids, capitulum sectors and their interactions.In the case of cross-pollinated varieties one-way ANOVA was used to assess differences between capitulum sectors.Probabilities equal to or less than 0.05 were considered significant for main effects and interactions.

RESULTS
Pollen analysis.Table 2 shows pollen grain number per flower (PGF) for each capitulum sector in the different planting dates and for both hybrids and experimental years.In the first year, the internal sector showed the highest number of pollen grains per flower (30979 to 40448 PGF), while lowest values were shown in the external sector (21924 to 31625 PGF); the middle sector (26063 to 35396 PGF) showed intermediate values.The differences between the external and internal sectors were significant (p<0.05) in both hybrids for the three planting dates, except for the case of the third planting date when DKO3945 did not show significant differences (p=0.25), with values between 27167 and 30979 PGF (Table 2).In the second year, pollen grain number per flower showed highly significant differences (p<0.01)among capitulum sectors, showing again the highest pollen production in the internal sector (43542 to 50250 PGF), and the lowest one in the external sector (27007 to 32000 PGF) (Table 2).
In both experimental years and in all planting dates, both hybrids showed a similar variation in the number of pollen grains per flower according to the sampled capitulum sector.Thus, the highest production of pollen was observed in flowers at the internal sector, followed by those at the middle sector, and finally the lowest production was observed in flowers at the external sector.

Sunflower pollen production
Table 3. Anther length (AL, mm) and width (AW, mm); anther volume (AV, mm 3 ) and number of pollen grains per flower (PGF) for each capitulum sector in both cross-pollinated varieties: Havasupai and Hopi.

Capitulum
Tabla 4. Longitud de la antera (AL, mm) y su ancho (AW, mm); volumen de la antera (AV, mm 3 ), volumen del grano de polen (PGV, μm 3 ) y viabilidad del polen (PV, %) para cada sector del capítulo.Los datos presentados son la combinación de resultados para todas las fechas de siembra para los dos híbridos.In Fig. 1A arrows denote points that were considered outliers for the linear fit.No significant differences (p=0.18) were observed in PGF per capitulum sector in capitula of Havasupai (Table 3).On the other hand, in Hopi, flowers of the MS and IS produced a larger number of pollen grains than those of the ES (Table 3).

Capitulum sector AL (mm) AW (mm) AV (mm
There was a significant interaction (p>0.05) between hybrids and seasons.Anther length, width and volume in both hybrids showed significant differences (p<0.05)among sectors, with the lowest values of the three variables in the ES and the highest ones in the IS (Table 4).In cross-pollinated varieties, only significant differences (p<0.05) were observed in the anther length of Hopi, with the highest values (9.0 mm and 6.1 mm) for ES and MS, respectively, and the lowest ones for IS (5.0 mm).Havasupai did not show significant differences (p>0.05) in any of the anther size parameters (Table 3).Pollen viability in the hybrids was over 97%, and significant differences were found between the external and the middle sectors (Table 4).
Anther length and width showed a direct relationship with pollen grain count (Figs. 1 and 2).The IS showed the largest anther length and width, followed by the MS and ES (Figs. 1 and 2).This was observed in both hybrids (Fig. 1) and in open pollinated varieties (Fig. 2), indicating a positive relationship between the anther dimensions and the number of pollen grains produced.No significant differences were observed (p=0.38) in pollen grain volume among the three capitulum sectors (Table 4).However, anther size parameters (AL and AW) showed significant differences, with higher  values for the internal sector (25%) than those for the external one (Table 4).This was positively related to the number of pollen grains per flower produced in each sector, which was 27% higher in the flowers of the IS than in those of the ES.
The open pollinated varieties showed more variability for AL and AW (Fig. 2).The magnitude of the average values of these variables was higher than that for the hybrids (Fig. 1).In both hybrids, data were distributed in a such a way that two clusters can be distinguished (Fig. 1).The first one corresponds to the highest AL, AW and PGF values, from the capitulum internal sector (black symbols in Fig. 1); the second one corresponds to the lowest AL, AW and PGF values from the external sector (white and grey symbols in Fig. 1).In the open pollinated varieties this pattern was not so clear, with a higher variation with respect to the hybrids (Fig. 2).

Yield analysis.
In both hybrids and planting dates, the number of FDF showed significant differences (p<0.05)among capitulum sectors, with the highest value of FDF in the ES (598), followed by the MS (456) and the IS (358) (Table 5).Significant differences for IDF (p<0.05) were also found among the capitulum sectors, with a higher number of IDF in IS (132), than in MS (83) and ES (62); no significant differences were found between MS and ES (Table 5).Seed set was significantly different between the IS (74.8%) and the ES and MS (92.4% and 85.6%, respectively; Table 5).

Sunflower pollen production
Table 5. Number of fully developed fruits (FDF) and incompletely developed fruits (IDF), and seed set (%) for each capitulum sector, on average for hybrids and planting dates.Values followed by different letters are significantly different (p<0.05).

DISCUSSION
The existence of a positive centripetal gradient of pollen grain number per flower in the sunflower capitula constitutes a remarkable finding.This positive centripetal gradient of intra-inflorescence pollen, unusual in many plant species, might be considered an adaptive response to the time sequence of anthesis in this species to ensure a homogeneous pollination of the capitulum.Thus the flowers of the SE, which open first, are more likely to have an allogamous pollination, while those that open later (SI) would be forced to autogamy because of temporal, positional and morphological disadvantages.
This intra-inflorescence fluctuation might be caused by its architecture (Buide, 2008).In species with racemose inflorescences the flowers that open first are more likely to receive external pollen, while those that open later (internal sector of the capitulum) will mostly receive pollen from the same plant (Ashman & Hitchens, 2000;Buide, 2004).Also, there is a clear tendency from the basal (flowers that open first) to the distal positions (flowers that open later): distal flowers are typically smaller and produce less and smaller ovules, fruits and seeds.However, they may have larger stigmata, and produce larger amount of pollen (Ashman & Hitchens, 2000).
These results have not been previously described for the Asteraceae in general or for the study genus in particular, since it is always assumed that the amount of pollen in inflorescences declines centripetally (Buide, 2008).However, currently available evidence indicates that this gradient can be generated as the result of gibberellin concentrations.It has been reported that at flowering GA 3 concentration in the sunflower capitulum declines towards its center (Duca, 2006).Since GA 3 arrests microsporogenesis (Duca et al., 2008;Kumar & Srivastava, 2009), it could act as a positional control for pollen grain production.
Gradients may also be due to an architectural effect, and to an anthesis temporal dynamic rather than to resource availability.Peripheral flowers (external 1/3 to 1/2 of the capitulum radius) are more likely to receive external pollen; flowers that open later (internal 1/3 of the capitulum radius) will mostly receive pollen of the same plant (Buide, 2004(Buide, , 2008)).
The primitive-domesticated Helianthus annuus L. lines, Hopi and Havasupai, showed the same quantitative intra-capitulum pattern of pollen grains.Could it be a genus character?In this sense, a positive centripetal gradient of the intra-inflorescence pollen production could be considered as an adaptive response to the time-sequence of the anthesis that will ensure a homogeneous pollination of the capitulum in the study species.Thus, flowers of the ES, that open first, are more likely to be allogamous  Astiz V & LF Hernández, FYTON 83 (2014) while those that open later (IS) would be forced to autogamy.
In some sunflower lines there is evidence of a negative centripetal gradient of pollen viability.It has been observed that in some short statured sunflowers, pollen viability ranges from 60-80% in the peripheral flowers to less than 5% in the internal ones (Ramos et al., 2013).
According to the results shown here, we can infer that when viability is low, there is a need for pollen grains to compensate for the lack of capacity to provide male gametes during fertilization.In the sunflower capitulum the time to complete microsporogenesis is shorter in the anthers of central flowers than in peripheral ones (Dr.C. Sala, Nidera Arg., pers.comm.).Our results would to the scenario presented in the sunflower capitulum upon completion of anthesis, in agreement with the pollen compensation mechanism previously mentioned Figure 3 summarizes a global hypothesis of the pollen reduction pattern in the sunflower capitulum according to the four variables described above: pollen viability, gibberellin concentration, rate of microsporogenesis and pollen grain number per flower.Only the latter one has been proved in the present work.
Finally, higher pollen offer of the internal flowers, as a reward factor to pollinators, deserves a special attention.It is clear that when anthesis is taking place in the central region of the capitulum, the external flowers have already been fertilized and their ovaries are in active development.Therefore, due to a temporal and positional advantage, these flowers have already "won" the competition for resources when the central ones are opening.In this sense a higher pollen production in central flowers of the inflorescence could be an attractive mechanism for pollinators.
Other rewards of the flowers (nectar) have been found of high importance in determining the behavior of pollinators when visiting sunflowers (Neff & Simpson, 1990).However, nectar production related with floret position was not studied here.Nevertheless, it is known that central (distal) florets produce less nectar than flowers of the external (proximal) sector in racemose inflorescences (Devlin et al., 1987;Kudo et al., 2001).This is the result of competition between flowers and fruits for carbohydrates (Pleasants, 1983).Perhaps the change in pollen amount produced by the inner florets could possibly alter the foraging behavior of pollinators because of the reduced amount of nectar offered.
This intra-inflorescence variation in the amount of pollen per flower observed in the present study has been apparent in plants with specialized reproductive systems, mainly monoecious species, where male and female functions are in different flowers.
However, in inflorescences with hermaphroditic florets, considerable variation can occur in the energy investment, particularly focused to build a better display or attractive structures (petal size, nectar volume), biomass distribution and reproductive potential (Stephenson, 1981;Digle, 1995).
Although the average energy investment for the production of pollen is higher than that required to produce ovules (190 Cal vs. 133 Cal, respectively;Smith & Evenson, 1978), in the case of sunflower, with uniovulate florets, this should not be an issue.
Regarding anther size, we found a good relationship between the length and the width of the anthers and the total number of pollen grains.Since the volume of each grain (PGV) does not fluctuate among capitulum sectors (Table 4), the gradient of pollen quantity observed may in part be attributed to the anther size.The fluctuation of these anther parameters in the primitive varieties of sunflower studied in the present work was higher than that observed in the hybrids (Fig. 2).This may be due to the fact that the processes of genetic improvement and selection in the sunflower have led to characteristics that determine that the anther dimension fluctuate less in the current hybrids than in wild species (Poverene et al., 2004).Nevertheless, the anther size/pollen grains relationship per capitulum sector in the primitive varieties was consistent with the response observed in the hybrids.Thereafter, anther size would be an inherited character of the genus that was not modified during genetic improvement in sunflower.Evidence in other species has revealed that a QTL for total anther length explained approximately 35% of the phenotypic variation for this trait (Georgiady et al., 2002;Conner, 2006;Brock et al., 2010).

CONCLUSIONS
It is not known whether failure of pollen germination on the stigma could occur or not.However, we can conclude that the formation of IDF in different regions of the capitulum would not depend on intraplant pollen offer, as both pollen offer and viability were not limiting.These results suggest that neither the amount nor the quality of the pollen would limit sunflower yield under the conditions of this study.The intra inflorescence positive centripetal gradient of pollen quantity observed in our investigation may respond to a legacy of the species as a strategy to invest resources in the sporophytic generation during the late stages of anthesis.Thus, the flowers of the ES, that open first, should have a greater chance of allogamous pollination, while those that open later (IS), would be forced to self-pollination due to temporal, positional and morphological disadvantages.This response could avoid photoassimilate supply competition between proximal and distal regions of the capitulum.
The identification of a positive centripetal gradient of pollen grain number per flower within the capitulum of both hybrids and the two cross-pollinated varieties disagrees with the quantitative description of pollen production in other species with racemose inflorescences.Results from the present study, regarding the fact that early-domesticated genotypes, Hopi and Havasapui, showed the same quantitative

Fig. 3 .
Fig. 3. Schematic representation of a sunflower capitulum showing the proposed model for the response of the four variables described in this work, along the capitulum radius.These variables would explain the positive centripetal gradient of pollen grains per flower observed in the study.