Pollination of Rapeseed ( Brassica napus ) by Africanized Honeybees ( Hymenoptera : Apidae ) on Two Sowing Dates

In this study, performed in the western part of the state of Paraná, Brazil, two self-fertile hybrid commercial rapeseed genotypes were evaluated for yield components and physiological quality using three pollination tests and spanning two sowing dates. The treatments consisted of combinations of two rapeseed genotypes (Hyola 61 and Hyola 433), three pollination tests (uncovered area, covered area without insects and covered area containing a single colony of Africanized Apis mellifera honeybees) and two sowing dates (May 25th, 2011 and June 25th, 2011). The presence of Africanized honeybees during flowering time increased the productivity of the rapeseed. Losses in the productivity of the hybrids caused by weather conditions unfavorable for rapeseed development were mitigated through cross-pollination performed by the Africanized honeybees. Weather conditions may limit the foraging activity of Africanized honeybees, causing decreased cross-pollination by potential pollinators, especially the Africanized A. mellifera honeybee. The rapeseed hybrids respond differently depending on the sowing date, and the short-cycle Hyola 433 hybrid is the most suitable hybrid for sowing under less favorable weather conditions.


INTRODUCTION
Rapeseed (Brassica napus L. var.oleifera Metzger) is an important oilseed crop in Brazil; however, the average yield of 1,226 kg ha -1 (CONAB 2012) is much lower than that of the world average of 1,820 kg ha -1 (USDA 2012).The introduction of Apis mellifera L. honeybee colonies to rapeseed crops may increase production rates (Abrol and Shankar 2012).
The rate of cross-pollination in rapeseed varies with the diversity of pollinating insects, cultivar and climatic conditions (CFIA 1999).The efficiency of self-pollination and cross-pollination processes that lead to grain production is dependent on factors related to the reproductive mechanisms of the plant (protogyny or protandry) and on environmental conditions.In Sweden, for example, a study of crosspollination rates in the B. napus cultivar "Topa" at different locations, indicated a variation of 12% to 47% in cross-pollination rates, demonstrated by isoenzyme analysis (Becker et al. 1992).
The most favorable regions for rapeseed cultivation are those that exhibit an average temperature of 20 °C during the plant cycle (Thomas 2003).Temperatures higher than 27 °C may result in floral sterility and the subsequent loss of grain yield (Morrison and Stewart 2002).In addition, high temperatures during pre-anthesis may cause pollen sterility and can retard the growth of the pollen tube during anthesis (Prasad et al. 2001).
Climatic factors can also act indirectly on rapeseed grain production by limiting insect activity during the collection of floral resources.Lower flower foraging activity by pollinators can result in lower pollen removal from the anthers and, therefore, less pollen deposition on the stigma, which limits the effect of cross-pollination by the insects (Thomson andGoodell 2002, Wilcock andNeiland 2002).Africanized honeybees exhibit optimal temperature and relative humidity ranges for the collection of nectar and pollen (Malerbo-Souza and Silva 2011).
An understanding of the possible interactions between ecological and environmental factors that might influence yield and seed quality components in rapeseed are of great importance for obtaining the maximum productive potential in rapeseed crops.In this study, possible interactions between pollination tests, rapeseed hybrids and sowing dates relative to production and seed quality variables were analyzed.The effects of environmental factors on yield and seed quality were investigated in rapeseed genotypes sown on two different dates.

MATERIALS AND METHODS
The experiment was performed on the rapeseed The soil at the experimental area was classified as Eutrudox Red Latosol (Embrapa 2006).Crambe (Crambe abyssinica), quinoa (Chenopodium quinoa), corn (Zea mays) and B. napus crops were growing near the experimental area, and an apiary containing six colonies of Africanized honeybees was located at a distance of approximately 1 km.
The rapeseed hybrids used in this experiment were medium-cycle Hyola 61, characterized by a broad adaptation to the different environmental conditions characteristic to Brazil, and short-cycle Hyola 433, which exhibits a high requirement for favorable environmental factors, especially high soil fertility (Tomm et al. 2009).
Environmental conditions data were collected and were recorded during the experiment at the Experimental Weather Station of the Unioeste and are shown in Fig. 1.
The experimental design was a randomized complete block with a 3 x 2 x 2 factorial scheme and four replications.The treatments entailed a combination of the following: three pollination tests (demarcated area free to insect visitationcontrol; area covered with a pollination cage and containing an A. mellifera colony -area visited only by Africanized honeybees; and area covered with a pollination cage without honeybees -area of total insect exclusion), two sowing dates (May 25th, 2011 andJune 25th, 2011) and two rapeseed hybrids (Hyola 61 and Hyola 433).
The experimental area measured 90 m long x 21 m wide (representing a seeded area of 1.512 m 2 ).
The rapeseed hybrids were sown using a direct (no-till) planting system.Corn was grown in the intercrop period.The 12-15-15 NPK fertilizer (200 kg ha -1 ) was used for the base dressing.
The hybrids were sown on May 25th, 2011 and June 25th, 2011 in 48 plots, each measuring 24 m 2 .
In each plot, the plants were sown in eight rows spaced 0.45 m apart, with a 0.05-m space between each plant.The sowing depth was 0.01 m.At the two-to-four true leaf stage, thinning was performed to obtain a uniform density of 40 plants m -2 in the plots.At the phenological stage (four developed leaves), 30 kg of N ha -1 in the form of ammonium sulfate were applied as a top dressing in accordance with technical guidelines for growing rapeseed (Tomm et al. 2009).Pollination cages were constructed using a 2 x 2-mm nylon mesh supported by ¾-inch PVC tubes, which formed cages measuring 4 m wide, 6 m long and 2 m high at the tallest point, and encompassed a 24 m 2 area (Chiari et al. 2005) (Fig. 2).
The cages were assembled 5 d before flowering and were disassembled at the end of the flowering period to allow complete plant development.At the onset of flowering, the plots caged with honeybees received a colony of Africanized A. mellifera consisting of five frames, three containing the brood and two containing the food.Throughout flowering, the individual colonies were supplied with drinking water and a food supplement consisting of 0.5 liter of syrup containing approximately 50% sugar (Free 1993).The harvest was performed manually after the onset of pod dehiscence.The hybrids Hyola 433 and  Grain yield (GY, kg ha -1 ) was estimated after cleaning, classification and weighing of the grains harvested from the four central meters of the two plant rows located at the center of each plot, representing a total area of 3.6 m 2 .The GY was corrected to 10% moisture, which is the reference percentage moisture used commercially and indicated for short-term storage (Tomm et al. 2009).Seed moisture content was determined in two replicates of each experimental unit using the oven method at 105 °C ± 30 °C for 24 h (Brasil 2009).
To measure the grain mass per plant and to estimate the grain components, 10 plants were harvested at random from each plot, and the mean values of the following variables were calculated: a) grain mass per plant (GMPl, in g plant -1 ), obtained by individually weighing the grains from each plant; b) number of pods per plant (NPP, in units), obtained by counting the pods from 10 plants per treatment; and c) number of grains per pod (NGP, in units), obtained by randomly collecting 20 pods per treatment.
Calculations of the mass of 1,000 grains (MTG, in g), vigor (VG) and germination (GER) were based on the Rules for Seed Analysis (Brasil 2009).The ether extract content was quantified by the "hot" extraction method (two replicates per treatment) using the Soxhlet extractor, and petroleum ether (pa) was used as the solvent.
The data were subjected to analysis of variance in a triple factorial scheme.The Student-Newman-Keuls test was used to compare the means of the variables analyzed among the classes of the pollination test factor.The effects of hybrid and sowing date on the response variables were tested using the F-test.A 5% probability level was adopted in both analyses.Statistical analyses were performed using the Statistical Analysis System (SAS, SAS Institute 2012) software.

RESULTS
A summary of the analysis of variance of probability values (P) related to the sources of variation included in the statistical model is shown in Table I.
The GMPl and the grain mass per pod (GMPd) were affected significantly by the triple interaction among the hybrid, sowing date and pollination tests (Table 1).The GMPl value for the uncovered area did not differ from the area covered by cages with honeybees, and both were greater than the GMPl values for the area covered by cages without insects for both hybrids sown on May 25th, 2011 (pollination/sowing date-1: Hyola 61: F = 24.07,df = 2, P = 0.0002; pollination/sowing date-1: Hyola 433: F = 14.54, df = 2, P = 0.0015) (Table II).
The GMPd value from the uncovered area and from the covered area with honeybees did not differ and was greater than the GMPd from the covered area without insects for both genotypes on both sowing dates, with the exception of the Hyola 433 hybrid sown on May 25th, 2011 (pollination/ sowing date-1: Hyola 61: F = 19.81,df = 2, P = 0.0005; pollination/sowing date-1: Hyola 433: F = 1.74, df = 2, P = 0.2297; pollination/sowing date-2: Hyola 61: F = 10.41,df = 2, P = 0.0045; pollination/sowing date-2: Hyola 433; F = 20.36,df = 2, P = 0.0004) (Table II  Means in the same column followed by different lowercase letters differ between pollination tests per each hybrid and sowing date combination according to the Student-Newman-Keuls test (P < 0.05).Means in the same row followed by different uppercase letters differ between sowing dates per each hybrid and pollination combination according to the F-test at the 5% significance level.The means were calculated from four observations.
In addition, the GMPl and GMPd values were greater for the rapeseed hybrids sown on May 25th, 2011 than on June 25th, 2011, with the exception of the Hyola 61 hybrid in the covered area without insects, which exhibited equivalent mean GMPd values between sowing dates (sowing date/cages without honeybees: Hyola 61: F = 4.28, df = 1, P = 0.08) (Table II).
There was no difference in GMPl between hybrids in each pollination test and on both sowing dates, with the exception of the Hyola 433 hybrid in the uncovered area sown on June 25th, 2011, which exhibited a higher GMPl compared with the Hyola 61 hybrid (hybrid/uncovered area: sowing date-1: F = 0.53, df = 2, P = 0.61; hybrid/cage without insects: sowing date-1: F = 0.42, df = 2, P =

TABLE I Summary of the significance probability of the initial
analysis of variance for each dependent variable.
For sowing date June 25th, 2011, the NPG was greater in the covered area with honeybees compared to the uncovered area and to the covered area without insects, and the NPG was smaller in the covered area without insects compared to the uncovered area (NPG-pollination: sowing date-2: F = 59.79, df = 2, P < 0.0001).GY did not differ between the uncovered area and the covered area with honeybees and was higher than in the covered area without insects for sowing date June 25th, 2011 (GY-pollination: sowing date-2: F = 8.26, df = 2, P = 0.0012).There was no difference among pollination tests for the MTG between the genotypes sown on June 25th, 2011 (MTG-pollination: sowing date-2; F = 0.68; df = 2; P = 0.5112) (Table III).
There was no difference in the NGP between sowing dates in the uncovered area and the covered area without insects.However, a difference was observed in the NGP between sowing dates in the covered area with honeybees; an increased NGP was observed in the June 25th, 2011 plants compared with the May 25th, 2011 plants, regardless of the hybrid tested (NGP-sowing date: uncovered area, F = 2.86, df = 1, P = 0.1005; cage without honeybees, F = 3.58, df = 1, P = 0.0671; cage with honeybees, F = 7.33, df = 1, P = 0.0107) (Table III).
From the GY data, the value for the dependency of the rapeseed crop on Africanized honeybees (D rapeseed ) was estimated for the two sowing dates, regardless of the hybrid tested, using the equation proposed by Robinson et al. (1989) as follows: where Y o is the GY (kg ha -1 ) in the covered area with Africanized honeybees and Y c is the GY (kg ha -1 ) in the covered area without insects.Therefore, D rapeseed = (1,816.61-1,216.76)/1,816.61= 0.33 in response to sowing on May 25th, 2011, and D rapeseed = (436.14-213.51)/436.14= 0.51 in response to sowing on June 25th, 2011.Therefore, there was greater dependence on the presence of Africanized honeybees for the plants sown on June 25th, 2011.
For the Hyola 61 hybrid, the NPP did not differ between the uncovered area and the covered area with honeybees, and the NPP values were greater than in the covered area without insects, regardless of the sowing date (pollination: Hyola 61, F = 8.44, df = 2, P = 0.0011).There was no difference in the NPP among the pollination tests for the Hyola 433 genotype, regardless of the sowing date (pollination: Hyola 433: F = 0.003, df = 2, P = 0.9968) (Fig. 4).The Hyola 433 genotype exhibited a higher NGP and higher oil content (OC) than the Hyola 61 hybrid when sown on May 25th, 2011.For the June 25th, 2011 sowing date, the Hyola 433 hybrid exhibited a higher NPP and increased VG and GER compared with the Hyola 61 genotype.In contrast, the Hyola 61 hybrid exhibited a higher MTG than the Hyola 433 hybrid when sown on June 25th, 2011.There was no difference between the hybrids regarding the NPP, VG, GER and MTG when sown on May 25th, 2011 or the NGP and OC when sown on June 25th, 2011 (Table IV).

TABLE IV F probability values and means ± standard deviation values for number of grains per pod (NGP), number of pods per plant (NPP), oil content (OC), vigor (VG), germination (GER) and mass of 1,000 grains (MTG) per rapeseed hybrid and sowing date combination in
Marechal Cândido Rondon (state of Paraná, Brazil).
Means in the same column followed by different lower case letters differ between hybrids according to the F-test (P < 0.05) per sowing date.Means in the same column followed by different upper case letters differ between sowing dates according to the F-test (P < 0.05) for each hybrid.a Means calculated from 12 observations.

DISCUSSION
Rapeseed is a predominantly self-pollinating species that produces a satisfactory GY in the absence of pollinating insects (Eisikowitch 1981, Mesquida et al. 1988, Mussury and Fernandez 2000, Hoyle et al. 2007).
Therefore, in this study, the hypothesis was that sufficient amounts of pollen grains deposited on the stigma of the same flower in the covered area without insects would result in GMPl, GMPd, NGP and GY values equivalent to the uncovered area and to the covered area with honeybees; however, this was not the case.
The data indicated that for the Hyola 61 hybrid sown on May 25th, 2011 in the uncovered area and in the covered area with honeybees, the GMPl was 56.31% and 60.35% higher (P < 0.05) and the GMPd was 47.76% and 44.47% higher (P < 0.05), respectively, than in the covered area without insects.The Hyola 433 hybrid sown on May 25th, 2011 demonstrated a GMPl that was 40.38% and 44.73% higher (P < 0.05) in the uncovered area and covered area with honeybees, respectively, than in the covered area without insects.
Furthermore, for the Hyola 61 genotype sown on June 25th, 2011, the GMPl was 25.64% and 46.72% higher (P < 0.05) in the uncovered area and in the covered area with honeybees, respectively, than in the covered area without insects.For the Hyola 433 hybrid sown on June 25th, 2011, the GMPd was 83.67% and 77.96% higher (P < 0.05) in the uncovered area and covered area with honeybees, respectively, than in the covered area without insects.EMERSON D. CHAMBÓ et al.In the uncovered area and covered area with honeybees, the NGP was 21.73% and 19.86% higher (P < 0.05) and the GY was 46.16% and 49.30% higher (P < 0.05), respectively, than in the covered area without insects for the sowing date May 25th, 2011.The NGP was 10.76% and 37.09% higher (P < 0.05) in the covered area with honeybees compared with the uncovered area and the covered area without insects, respectively, for the sowing date June 25th, 2011.The GY was 222.90% and 104.27% higher (P < 0.05) in the uncovered area and covered area with honeybees, respectively, compared with the covered area without insects for sowing date June 25th, 2011.
It is likely that only self-pollination occurred in the covered area without insects.However, it is possible that the deposition of pollen on the stigma of the same flower was insufficient for the fertilization of all of the ovules, or the pollen did not come into contact with the stigma during the period when the stigma was receptive.
The lack of synchronization between stigma receptivity and viable pollen availability (DeGrandi-Hoffman and Chambers 2006), as well as the absence of fully-developed embryo sacs at the time of flower opening (Wang et al. 2009), are considered the main contributing factors for the lower production of seeds through self-pollination (DeGrandi-Hoffman and Chambers 2006).
Therefore, the data demonstrated that crosspollination by Africanized honeybees is efficient at increasing production rates in rapeseed crop.
The lack of a significant difference between the uncovered area and the covered area with honeybees for the yield components evaluated likely occurred because of the apiary containing six colonies of Africanized honeybees located approximately 1 km from the experimental area, as well as other potential pollinators non-Apis.The density of Africanized honeybees in the uncovered area because of the proximity of the apiary and the presence of other pollinators nonApis were sufficient to generate a production rate similar to that of the covered area with honeybees.In comparing two sowing dates May 25th, 2011 and June 25th, 2011, a larger proportional increase in GY was observed in the uncovered area and in the covered area with honeybees compared to the covered area without insects.Besides, the GY in the uncovered area was 157.98%, was 469.88% in the covered area without insects, and was 316.52% greater in the covered area with honeybees for the sowing on May 25th, 2011 (P < 0.05) than to the sowing on June 25th, 2011 (Table III).This may have occurred because the climatic conditions for the June 25th, 2011 sowing date were more favorable for pollinators.
Pollination by honeybees depends on climatic conditions because these conditions affect both the crop and the honeybees (Abrol 2007).Africanized honeybees actively gather nectar at temperatures ranging between 15.48 and 31.09°C and at a relative humidity between 43.13 and 92.70%.For pollen, the optimal temperatures are between 15.48 and 27.61 °C, and the optimal relative humidity is between 45.00 and 89.50% (Malerbo-Souza and Silva 2011).
In this study, the temperature was closer to the optimal ranges for Africanized honeybees during the second flowering period.Although the air temperature was higher during the flowering period for the hybrids sown on June 25th, 2011, the temperature was less ideal for the cultivation of rapeseed compared to the temperature recorded during the flowering period for the hybrids sown on May 25th, 2011; the entomophilous pollination mitigated, at least partially, the limitations in the GY, possibly because of the higher temperatures and lower rainfall observed in September of that year (Fig. 1).POLLINATION OF RAPESEED BY HONEYBEES Despite the dependency values of 0.33 and 0.51 of the crop on Africanized A. mellifera on May 25th, 2011 and June 25th, 2011, respectively, under conditions more favorable to rapeseed development, such as on sowing date May 25th, 2011, rapeseed can exhibit a higher self-pollination rate, although the presence of Africanized honeybees significantly increased the crop production rates.
Almost all of the pollen produced in the anthers of rapeseed flowers is released shortly after the completion of flower opening (Mussury andFernandez 2000, Abrol 2007), and high temperatures at anthesis can limit the growth of the pollen tube (Prasad et al. 2001, Kakani et al. 2005) and cause pollen degeneration (Nava et al. 2009), limiting the self-pollination process.
It is possible that the pollen produced and deposited on the stigma of the same flower was not intact because of the higher temperatures, limiting self-pollination during the second flowering period.This limitation in the fertilization of all of the ovules should have favored the cross-pollination process, therefore causing a greater dependency on honeybees for the crop, especially for sowing date June 25th, 2011.
However, although the hybrids sown on June 25th, 2011 demonstrated a greater dependency on honeybees, the GY for this period was significantly lower than for the hybrids sown on May 25th, 2011; therefore, planting at the end of June is not justified.
The analysis demonstrated that the interaction of hybrid x sowing date was significant (P < 0.05) for all of the response variables, with the exception of GY (Table I), indicating that the hybrids responded differently to the time when sowing occurred.Both hybrids performed better when sown on May 25th, 2011 rather than on June 25th, 2011.
Yield and seed quality parameters tend to decrease when sowing is delayed after mid-April (Tomm et al. 2004).After anthesis, each 1 °C increase in temperature can trigger losses in GY of up to 289 kg ha -1 (Si and Walton 2004).
Sowing from late April to early June favors the performance of short-cycle genotypes, such as the Hyola 433 hybrid, compromising the yield in hybrids exhibiting medium and long cycles, such as Hyola 61 (Tomm et al. 2003), even though this latter hybrid may exhibit greater productivity under certain conditions (Tomm et al. 2010).
It is possible that the short-cycle Hyola 433 genotype suffered less because of the climatic conditions, especially the low rainfall in September, compared to the medium-cycle Hyola 61 hybrid (Fig. 1).Moreover, the Hyola 433 hybrid is suitable for soils with a high fertility (Tomm et al. 2009), a characteristic observed in the soils of the study area.
In addition, climatic conditions during the flowering period can influence nectar secretion during the day, which can vary depending on the cultivar (McGregor 1976, Free 1993, Delaplane and Mayer 2000).Cultivars less tolerant to heat can reduce the amount or concentration of secreted nectar as well as the pollen availability throughout the day, causing the flowers to become less attractive to pollinators in spite of an adequate insect density, limiting the cross-pollination process and leading to reduced production (Chambó et al. 2011).
There was no effect (P > 0.05) of pollination on the OC in seeds, VG, GER and MTG, suggesting that these variables were not affected by insect pollination (Table I).
No significant difference was observed among the areas available for visitation by insects and the covered areas with and without honeybees for the MTG values and seed vigor in Adesmia latifolia (Camacho EMERSON D. CHAMBÓ et al. and Franke 2008) and in the seed germination rate in Glycine max (Chiari et al. 2008).In contrast, in rapeseed (cv.OAC Triton), the presence of honeybees increased the seed germination from 83% to 96% (Kevan and Eisikovitch 1990).
Insects do not directly influence features related to the physiological quality of the seed because these features are related to the physiological factors of the plant itself and to the climatic conditions.However, in a number of situations, a direct influence may occur, causing an increase in these features (Camacho and Franke 2008, Chiari et al. 2011, Toledo et al. 2011).
The data demonstrated that the crosspollination process performed by Africanized A. mellifera improved the crop production rates in rapeseed.Productivity in the hybrids was increased in terms of the yield components for the May 25th, 2011 sowing date compared with the June 25th, 2011 sowing date.The hybrid yield loss may have been compensated, at least partially, by honeybee pollination when the temperature was higher and consequently less favorable to crop development, as observed for the June 25th, 2011 sowing date.

ACKNOWLEDGMENTS
To Bee Research Group -Universidade Estadual de Maringá for providing all cages pollination, equipment and materials.To Universidade Estadual do Oeste do Paraná for giving the experimental area.To Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) for financial support.

Fig. 3 -
Fig. 3 -Grain mass per plant (g) of rapeseed hybrids in response to sowing dates and pollination tests.Different lowercase letters in each pollination test indicate significant differences according to the F-test (P < 0.05).

Fig. 4 -
Fig. 4 -Number of pods per plant for each pollination test and hybrid regardless of the sowing date.Columns followed by different lowercase letters differ between hybrids for each pollination test according to the F-test (P < 0.05).Columns followed by different uppercase letters differ between pollination tests for each hybrid according to the Student-Newman-Keuls test (P < 0.05).
(Brassica napus L. var.oleifera) crop during the 2011 harvest season at the Experimental Station at the State University of West Paraná (Universidade Estadual do Oeste do Paraná, Unioeste), which is located in the municipality of Marechal Cândido Rondon (state of Paraná, Brazil; 24°32'09"S and 54°01'11"W; 392 meters above sea level).

TABLE III Mean and standard deviation values for the number of grains per pod, grain yield and mass of 1,000 grains per pollination test and sowing date combination in Marechal Cândido Rondon (state of Paraná, Brazil).
Means in the same column followed by different lowercase letters differ between pollination tests on each sowing date according to the Student-Newman-Keuls test (P < 0.05).Means in the same column followed by different uppercase letters differ between sowing dates for each pollination test according to the F-test at the 5% significance level.a Means calculated from eight observations.