Effect of sunflower seeds quality on dehulling process in order to produce protein
content guaranteed meal
DAUGUET Sylvie1, GUILLEMAIN Céline1, CARRE Patrick2, MERRIEN André3, KROUTI
Mohammed3, CHAMPOLIVIER Luc4
1CETIOM, Technical Institute for Oilseeds, 11 rue Monge 33600 Pessac, France, dauguet@cetiom.fr
2CREOL, 11 rue Monge 33600 Pessac, France, carre@cetiom.fr
3CETIOM, Ardon Analysis Laboratory, 270 av de la Pomme de Pin Ardon 45166 Olivet, France,
merrien@cetiom.fr
4CETIOM, Centre INRA, bat. AGIR 31326 Castanet Tolosan, France, champolivier@cetiom.fr
ABSTRACT
 Sunflower seeds dehulling increased the protein content in meal but the high variability of
seed quality was a problem for optimizing industrial process. The aims of this study were: to
observe the parameters influencing the protein content and dehulling ability of sunflower
seeds and to assess strategies in order to improve dehulling process.
 During this study, a first task was to measure variability of sunflower seeds quality delivered
in a crushing factory during three months: protein content expressed on defatted dry matter,
and dehulling ability obtained on a pilot equipment. In a second task, we studied genetic,
climatic, and crop management effects on these characteristics, with seed samples coming
from an investigation in farmers’ fields.
 Variability of seeds quality was important at the entrance of the factory: protein content
from 29.8% on DDM (defatted dry matter) to 35.8% DDM and mean value 33.8% DDM;
dehulling ability from 9.2% to 18.4 % (weight of extracted hulls expressed on initial weight
of seeds) and mean value 12.4%. In the second task, the climatic effect was observed as
sunflower grown in 2008 had higher protein content than sunflower grown in 2009 (same
variety NK Country: 31.8% protein on DDM in 2008 and 30.8 in 2009). Also, there was a
significant difference on dehulling ability (23% in 2008, and 19% in 2009). Hydric stress
was higher in 2009, especially high just before flowering. In 2008, hydric stress occurred
after flowering what could improve dehulling ability, and protein content. No relationship
was observed between protein content expressed on DDM and oil content, but there was a
significant negative linear relationship between dehulling ability and oil content. Other
interactions have not been clearly demonstrated (nitrogen fertilization, potential of the
fields) due to a wide variety of situations and agricultural practices encountered in the
farmers’ fields.
 This work shows that the origin of lots of sunflower will affect the quality of sunflower
seeds. This provides opportunities for crushing plants to optimize the process of dehulling to
produce consistent quality meal, with a guaranteed content of protein, while applying a
dehulling rate as low as possible. Indeed, one can consider making a management of seeds at
the entrance to the factory according to the agricultural region of origin, identifying areas
low in protein and rich in protein every year, to supply with a mixture of seeds more
homogeneous.
 In order to make better use of sunflower meal, dehulling is very interesting. It can be
considered an optimization by search of a better quality of crops, for example by genetic
improvement, without penalizing the seeds oil content, but also by a rational management of
crushing plants supply.
Key words: crushing factories - dehulling ability - sunflower - meal - protein – seed quality
INTRODUCTION
The whole seed sunflower meal (WSSM) is characterized by a high content of poorly digestible fiber
derived from the hulls, which represent about 20-25% of the seed, while protein content is only 30-32%
of dry matter (Borredon et al, 2011; Dauguet et al, 2012). These characteristics are quite interesting for
animals like ruminants, rabbits, or some poultry needing low energy feeding. But, this WSSM cannot
integrate rations of animals requiring higher content of energy and protein, like dairy cows, pigs and
chickens. Dehulling, partial or complete, sunflower seeds before oil extraction provides meals with higher
content in protein, and less in fiber. Obtaining partially dehulled sunflower meal (PDSM) with 36%
protein content expressed on whole matter would give competitiveness in sunflower, as the selling price
of this kind of meal is higher than the selling price of WSSM, from 20 to 40% more (market prices in
France during 2011). So the introduction of the dehulling process in French crushing factories would
help to improve the quality of sunflower meal and improve competitiveness in sunflower. This
improvement would be more beneficial if it is accompanied by an improvement in the quality of seeds:
protein content and ease with which hulls can be removed from seeds (or dehulling ability, called
hullability by some authors). Indeed, extracted hulls can be recovered for energy production on site, but it
is more economically interesting to extract a minimum of hulls and to produce a maximum of meal that
meets the requirements of protein content. In these quality requirements, the regularity of protein content
is an essential component. That is why the high variability of seed quality was a problem for optimizing
industrial process. The aims of this study were: to observe parameters influencing protein content and
dehulling ability of sunflower seeds and to assess strategies in order to improve dehulling process.
During this study, a first task was to measure variability of sunflower seeds quality delivered in a
crushing factory during three months: protein content expressed on defatted dry matter, and dehulling
ability. The second task was to obtain these both characteristics for 8 sunflower varieties of market,
samples coming from experiments on sunflower varieties. And, in the third task, we studied genetic,
climatic, and crop management effects on these characteristics, with seed samples coming from an
investigation in farmers’ fields.
MATERIALS AND METHODS
For the first task, 170 samples of sunflower seeds were collected at the entrance to a crushing factory.
Sampling is performed at different scales: truck (10 samples per truck from 5 trucks), day (10 trucks per
day sampled over 10 days) and month (a representative sample of a day made by taking seeds of each
load in this day and this about eight days per month). For samples at the day scale, the loading place of
the different trucks and supplier storage companies were identified, in order to identify differences in
quality of seeds depending on the place of origin.
In the second task, 156 seed samples were collected from a network of farmers’ fields constructed
within a project on the improvement of sunflower oil yield at a production area scale, in the South-West
of France (one sample represented one field). The studied high oleic varieties were: NK Countri and
PR64H32 in 2008, and NK Countri and NK Ferti in 2009.
On all these collected seeds samples, oil content, expressed on Dry Matter (DM), was measured by
Nuclear Magnetic Resonance and protein content, expressed on Defatted Dry Matter (DDM) was
obtained by Dumas method: Protein (%DDM) = Protein (%DM)/ (1 - Oil (%DM)).
Dehulling ability was obtained measuring the weight of extracted hulls expressed on the initial
weight of seeds: Extracted hulls (%) = (extracted hulls in g)/(initial seeds in g)*100
Dehulling was performed on a pilot equipment based on centrifugal process (Figures 1 and 2), seeds
going through equipment 3 times with a rotational speed of 2000 turns per minute.
Figures 1-2. Pilot dehulling equipment Techmachine and laboratory sorting equipment
The seeds were kept in cold storage, and next, to deliver them to ambient conditions prior to the
dehulling, the seeds were placed in Petri dishes open for 48 hours. The lots containing a high level of
impurities were cleaned manually using a sieve. Each sample was divided into 4 conical dividers on lots
of about 15 g (3 replicates for a dehulling and for measuring the water content of seeds). The samples
were weighed before being passed three times dehuller to 2000 turns/min. This method of dehulling was
determined by a previous work: this is a moderate dehulling corresponding to an industrial dehulling
process conducting to 10% of extracted hulls.
After a mechanical sorting, the different fractions (almonds, fines and hulls) were weighed (to nearest
0.01 g). The measured percentage of extracted hulls corresponded to the average of three replicates.
The water content of seeds was determined by the difference in seed weight before and after 15 hours in
the oven at 103 °C.
For seeds coming from the first task (sampled at the entrance of the crushing plant), the seed moisture
at which the dehulling ability was determined was the one of seeds coming from storage companies:
about 6-7% (mean moisture 6.7%, and standard deviation 0.46%).
For seeds coming from the second task, the seed moisture at which the dehulling ability was
determined was lower, as the seeds were previously slightly dried for a safe long-term storage: about 4.56.5% (mean moisture 5.6% and standard deviation 0.74%)
Data were analyzed using ANOVA with F test and differences were evaluated with StudentNewman-Keuls Test (software SAS Enterprise Guide). The coefficients of determination, and probability
associated (Student) were determined using also SAS Enterprise Guide.
In the first task, we calculated the percentage of hulls that should be extracted from the seeds if the
aim was to produce a partially dehulled sunflower meal with a protein content of 36% expressed on raw
matter (95% of batches above this value).
RESULTS
In the first task, the measurement of the oil content, protein content and dehulling ability showed that
there was variability in the quality of seeds delivered in a crushing plant (Table 1). On the truck scale, we
observed significant differences in seed quality according to the sources of seeds on all three studied
criteria: from 29.8% DDM to 35.8% DDM for protein content, and from 9.2% to 18.4% for extracted
hulls. On the day scale and on the month scale, homogenization occurred by mixing the seeds of several
sources and therefore the differences were not significant. There was more variability within a single day
(higher standard deviation) than within a truck, given the different sources of seeds received on a day.
This task revealed the difference in quality of sunflower seeds at the entrance to the crushing plant,
which could affect the production of the meal. Indeed, in order to produce a meal with a protein content
guaranteed, it would be important to be able to dehull enough, and to have raw material with a protein
content high enough and relatively stable. In these samples, we could not identify a relation between
percentage of extracted hulls and seed moisture (R²=0.018 in a linear regression).
Table 1. Oil content, protein content, and percentage of extracted hulls, from different scales at the
entrance of a crushing plant
Scale
Number of
Oil (%DM)
Protein (%DDM)
Extracted hulls (%)
samples
Mean
Truck1
10
51,82 (B)
Truck 2
9
46,77 (C)
Truck 3
10
53,47 (A)
Truck
Truck 4
10
53,08 (A)
Truck 5
10
53,26 (A)
Mean
51.68
Day 1
10
53,61 (A)
Day 2
8
52,27 (A)
Day 3
10
51,60 (A)
Day 4
10
51,83 (A)
Day 5
10
53,08 (A)
Day
Day 6
9
53,08 (A)
Day 7
10
52,17 (A)
Day 8
10
52,11 (A)
Day 9
10
52,46 (A)
Day 10
10
53,10 (A)
Mean
52.53
Month 1
14
52,84 (A)
Month 2
7
51,99 (B)
Month
Month 3
3
52,40 (AB)
Mean
52.41
Means within a column (oil, protein, extracted hulls), and within
significantly different (P<0.05).
SD
0,90
0,25
0,84
0,22
0,43
0.53
1,25
1,95
1,82
2,13
1,91
0,90
1,28
1,25
1,43
0,90
1.48
0,45
0,38
0,52
0.45
a scale
Mean
SD
Mean
SD
34,36 (B)
1,09
9,54 (CD)
0,84
29,81 (C)
1,85
18,42 (A)
0,80
35,56 (A)
1,07
9,16 (D)
0,56
33,35 (B)
0,69
10,04 (C)
0,59
35,78 (A)
0,84
14,72 (B)
0,64
33.77
1.11
12.38
0.69
33,88 (A)
1,45
10,68 (B)
1,91
33,32 (A)
0,81
11,17 (AB)
1,65
32,72 (A)
1,70
11,36 (AB)
1,14
33,68 (A)
1,68
12,00 (AB)
2,77
33,12 (A)
0,91
12,65 (AB)
2,60
32,66 (A)
1,59
14,42 (A)
4,35
33,62 (A)
1,95
12,96 (AB)
2,77
33,44 (A)
1,38
12,50 (AB)
1,78
31,96 (A)
1,26
12,14 (AB)
2,64
32,90 (A)
1,62
12,58 (AB)
1,90
33.13
1.44
12.25
2.35
33,30 (A)
1,29
11,06 (A)
0,81
32,43 (A)
0,88
11,63 (A)
0,76
33,69 (A)
0,33
11,45 (A)
1,65
33.14
0.83
11.38
1.07
(truck, day, month) followed by the same letter are not
The results of the second task are presented in Table 2. The climatic effect was observed as sunflower
grown in 2008 had higher protein content than sunflower grown in 2009, as the NK Country hybrid had a
protein content of 31.8% DDM in 2008 and 30.8% DDM in 2009. Also, there was a significant difference
on dehulling ability according to year, as the NK Country hybrid had 23% of extracted hulls in 2008, and
19% in 2009. The NK Country hybrid had lower oil content than PR64H32 and than NK Ferti hybrids.
PR64H32 had higher protein content than NK Countri, while there was no significant difference between
NK Countri and NK Ferti for this parameter. Concerning dehulling ability, NK Countri showed a good
dehulling ability, higher than the one of NK Ferti. The comparison could not be achieved for the year
2008 as the Shapiro-Wild test rejected the normality of the data for PR64H32.
Table 2. Mean yearly results sunflower hybrids from the network of farmers’ fields (task 2)
Year
Hybrid
Number of
observations
Oil content (% DM)
Mean
SD
Protein content (%
DDM)
Mean
Extracted hulls (%)
SD
Mean
SD
NK Countri
48
48.4 (A) (b)
1.9
32.1 (A) (b)
2.5
23.8 (A)
2.3
PR64H32
48
51.8 (a)
2
33.3 (a)
2
9.8
4.2
NK Countri
33
46.2 (B) (b)
2.4
30.78 (B) (a)
1.9
19.5 (B) (a)
3.9
2009
NK Ferti
41
49.6 (a)
1.7
30.3 (a)
1.8
14.2 (b)
3.1
Means within a column followed by the same letter are not significantly different (P<0.05). Capital letters (A, B) compare the years
for the NK Countri hybrid. Small letters (a, b) compare the hybrids within a year. SD = Standard Deviation.
2008
No relationship was observed between protein content expressed on DDM and oil content, but there
was a significant negative linear relationship between dehulling ability and oil content (Table 3). Moisture
content did not influence the percentage of extracted hulls in this task, within a year and within a hybrid.
We studied other relations with protein content and dehulling ability: nitrogen fertilization, yield,
potential of the soils, number of seeds per m², weight per seed. None of these influenced significantly
both parameters.
Table 3. Regression equations for extracted hulls (%) as a function of oil content (% DM)
Year
2008
2009
OC: oil content (% DM)
Hybrid
Regression equation
R²
p
NK Countri
PR64H32
NK Countri
NK Ferti
-0.68*OC + 56.30
-1.61*OC + 93.52
-1.44*OC + 86.26
-1.04*OC + 65.54
0.324
0.594
0.782
0.339
<0.0001
<0.0001
<0.0001
<0.0001
DISCUSSION
Concerning the results of the first task: Using the 97 samples coming from the day scale, representing
97 separate trucks coming from miscellaneous places, we calculated what would be the protein content in
the meal applying different dehulling rates. We calculated that the dehulling rate should be 15.6% of
extracted hulls, in order to obtain a meal with a protein content of 36% expressed on Raw Matter (95% of
batches above this value): mean protein content in meals would be 39.4% of RM with a Standard
Deviation of 2% (Value in 95% of batches = Mean – 1.64*SD). In order to optimize the dehulling
process, we could consider different management strategies of the seeds arriving at the factory. Indeed, if
we could lower the variability of the quality of seeds (protein content), it would be possible to lower the
dehulling rate while achieving the quality objective, what is economically interesting. It could be possible
to create within the factory two seed short-term storage cells: one with high protein content seeds, and
another with low protein content seeds. So the supply of the crushing circuit would be with a blend of
these two storage cells. Different management strategies could be considered: measuring protein content
in each truck arriving at the plant (difficulty: there is no instantaneous analytical method for protein
content in sunflower seeds like NIRS), assessing early in the season the protein content of sunflower
according to the agricultural area of origin or according to the supplier companies. We evaluated that
these management strategies would allow reducing the dehulling rate from 1 to 2 points (13.6 to 14.5 of
extracted hulls instead of 15.6% without any strategy). These findings must be put into perspective
because we considered the highest variability, with the protein content of meal obtained calculated for
each truck at the day scale, while a first homogenization occurs in the crushing circuit. Other strategy of
seed management could be considered like seed sifting, into uniform size classes in order to optimize the
dehulling rate according to dehulling ability of each class, but this strategy had limited success (Nel,
2001). The differences of dehulling ability according to seed origins were quite important: in order to
obtain a meal with a protein content guaranteed, the industry should apply a dehulling process intense
enough (seed velocity, number of times the seeds go through dehuller). Previous works showed that seed
drying before dehulling can improve the percentage of extracted hulls, using the same dehulling
conditions (de Figueiredo et al, 2011; Nel, 2001). In our study, we couldn’t see any relation between seed
moisture and percentage of extracted hulls, what could be explained by samples coming from different
places (different cropping conditions, and different sunflower hybrids blends), but that did not question
the fact that drying could be helpful in order to improve dehulling ability if seed quality was not
satisfactory.
Concerning the results of the second task: We observed that the sunflower seeds characteristics of
interest to obtain a protein content guaranteed meal (protein content, dehulling ability) were influenced by
climatic conditions and genotypes. Protein content and % of extracted hulls were higher during 2008 than
during 2009. Water stress was higher in 2009, throughout the cycle of sunflower, especially just before
flowering. In 2008, there were satisfactory water supply for sunflower growth from March to beginning
of July, and then water stress occurred after flowering what could improve dehulling ability, and protein
content. Indeed, it was shown that water stress, in particular during seed maturation, resulted in a
considerable increase in dehulling ability (Beauguillaume et al, 1992; Merrien et al, 1992; Denis and
Vear, 1994); so, environmental effect affected the dehulling ability (Nel, 2001). And protein content
could also be influenced by water stress and variety (Oraki et al, 2011). There was no significant relation
between seed moisture content and the dehulling ability in this second task, within a year and within a
hybrid, which could be explained by the fact that the studied seeds were stored during 19 to 31 months
and were slightly dried. As has been shown in previous work, differences between varieties existed for
the dehulling ability, which could be explained by different tissue structures of the hulls (Beauguillaume,
1992; Evrard, 1996). There was a significant negative linear relationship between dehulling ability and oil
content as shown by other authors (Denis et al, 1994b, de Figuereido et al, 2011). While NK Countri
hybrid had lowest protein content and highest dehulling ability, within this cultivar, the higher was oil
content, the less could be dehulled sunflower seeds. Other interactions have not been clearly
demonstrated (nitrogen fertilization, yield potential of the fields, yield, number of seeds per m², weight
per seed) what could be explained by a wide range of situations and agricultural practices encountered in
the farmers’ fields, although this has been shown in experiments under controlled conditions (Nel, 2001).
Thus, this study showed that sunflower seeds protein content (% DDM) and their dehulling ability
were influenced by the environment and by cultivars, and oil content influenced also dehulling ability.
But they could not show the influence of agricultural practices like nitrogen fertilization, which could
help in improving protein content. In the first task, we identified that there were differences in protein
content of sunflower seeds according to areas of origin. Therefore it might be possible to make a selection
at the entrance to the crushing plant. This selection could be made by determining the regions producing
seeds rich in protein or by a more accurate selection by storage companies. Such a strategy would allow
greater monitoring and short-term improving of partially dehulled sunflower meal production quality and
regularity. Currently, the breeders’ efforts do not address these criteria (protein content and dehulling
ability), given that crushers can adapt their dehulling equipment. However, an increase in protein content
in hybrids could be provided without adversely affecting the oil content, as for example the hybrid
PR64H32 which was richer in protein and oil from NK Countri. And improving the protein content of
sunflower seeds through breeding would help to improve on the long range meal quality and profitability
of the crushing process (Nel, 2001). It would also be useful to monitor that dehulling ability is not
degraded through the genetic improvement of sunflower seeds oil content.
ACKNOWLEDGMENTS
The authors acknowledge SAIPOL industry for the first task of this study; they also acknowledge for the
second task of this study IN VIVO AgroSolutions, ARTERRIS, Terre de Gascogne and the funding by
the CASDAR of the French Ministry of Agriculture, Food and Fisheries.
REFERENCES
Beauguillaume, A., J. Evrard and A. Merrien. 1992. Variété ou milieu, quelles influences sur le
décorticage ? CETIOM – Oléoscope 12: 16-17.
Borredon, M.E., M. Berger, S. Dauguet, F. Labalette, A. Merrien, Z. Mouloungui, Y. Raoul. 2011.
Débouchés actuels et futurs du tournesol produit en France Critères de qualité. Innovations
Agronomiques 14:19-38.
Dauguet, S., C. Peyronnet, A. Merrien, M. Krouti, J.P. Loison and A. Quinsac. 2012. A multi-year
survey on sunflower meal quality produced in France. In: Proc. 18th Int. Sunfl. Conf., Mar del Plata,
Argentina. Int. Sunfl. Assoc., Paris, France.
Denis, L., and F. Vear. 1994. Environmental effects on hullability of sunflower hybrids. Agronomie
14:589-597.
Denis, L., J. Dominguez, M. Baldini and F. Vear. 1994b. Genetical studies of hullability in
comparison with other sunflower seed characteristics. Euphytica 79: 29-38.
Evrard, J., P. Burghart, P. Carre, J. Lemarie, A. Messean, L. Champolivier, A. Merrien and F. Vear.
1996. Improvement of sunflower dehulling ability an interdisciplinary approach. In: Proc. 14th Int. Sunfl.
Conf., Beijing/Shenyang, China. Int. Sunfl. Assoc., Paris, France.
de Figueiredo, A.K., E. Baümler, I.C. Riccobene and S.M. Nolasco. 2011. Moisture‐dependent
engineering properties of sunflower seeds with different structural characteristics. Journal of Food
Engineering 102:58-65.
Merrien, A., J. Dominguez, G.P. Vannozzi, M. Baldini, L. Champolivier and P. Carre. 1992. Factors
affecting the dehulling ability in Sunflower. In: Proc. 13th Int. Sunfl. Conf., Pisa, Italy. Int. Sunfl. Assoc.,
Paris, France.
Nel, A.A. 2001. Determinants of sunflower seed quality for processing. Ph. D. diss. Univ. of Pretoria,
Pretoria, Republic of South Africa. Available at http://upetd.up.ac.za/thesis/available/etd-09012001132144/ (verified 15 Dec. 2011).
Oraki, H., I. Alahdadi and F. Parhizkar khajani. 2011. African Journal of Agricultural Research
6(5)/1246-1250.