In vitro embryo rescue and plant regeneration following self-pollination with irradiated pollen in cassava ( Manihot esculenta Crantz )

1 National Crops Resources Research Institute (NaCRRI), Namulonge, P. O. Box 7084, Kampala, Uganda. 2 Department of Biological Sciences, College of Natural Sciences, Makerere University, P. O. Box 7062, Kampala, Uganda. 3 Max-F. Perutz Laboratories, University of Vienna, Dr. Bohrgasse 9, 1030 Vienna, Austria. 4 Institute of Biology, Pedagogical University of Kraków, Podchorążych, 30-084, Kraków, Poland. 5 International Center for Tropical Agriculture (CIAT), Apartado, Aéreo, 6713, Cali, Colombia.


INTRODUCTION
Cassava (Manihot esculenta Crantz) is the world's second most important root crop that is grown in over 100 countries in Africa, Asia, the Pacific, and South America (FAOSTAT, 2014).Cassava is becoming popular for three main reasons: (i) Its yields are the highest with respect to the amount of starch per unit area when compared to other staples; (ii) cassava starch is amenable to diverse food and non-food uses; and (iii) it has a flexible harvesting schedule that is suitable for most rural and subsistence farmers.Also, the high resilience of cassava and its adaptation to climatic changes when other major food staples fail is an attribute for sustainable food security (Burns et al., 2010;Jarvis et al., 2012).
Globally cassava yields have increased by ca.1.8% per annum over the last decade to 12.8 tons per hectare, resulting into a global cassava harvest of more than 280 million tons in 2012, reflecting a 60% increase since 2000 (FAO, 2013).This output is set to accelerate further due to enhanced recognition of cassava's potential.Therefore, the current and anticipated demands for cassava starch necessitate concerted and systematic efforts tailored to the development of higher yielding cassava varieties (Aerni, 2006;Bull et al., 2011).
Although conventional breeding may still deliver high yielding varieties (Ceballos et al., 2004;2012), systematic progress in the genetic improvement of cassava is challenged by the heterozygous nature of the crop (Kawano et al., 1978;Meireles da Silva et al., 2003;El-Shakaway 2004;Bull et al., 2011).Similar limitations in maize (Zea mays) were overcome by introducing inbreeding and use of inbreds over a century ago (East, 1908;Shull, 1908;1909;Crow, 1998).Inbred lines (or pure, homozygous lines) have two copies of the same genome, such that after self-pollination (same genotype is both male and female); all the offsprings are identical to the parent.Similarly, double haploids (DHs) (genotypes formed when haploid cells undergo chromosome doubling), eliminate heterozygosity and achieve approximate complete homozygosity in one generation, compared to many generations in conventional inbreeding (Jain et al., 1996;Maluszynski et al., 2003a, b).The success attained in maize hybrid breeding (Crow, 1998) spurred the use of inbred lines or doubled haploids for several other economically important crop species such as cucumber, flax, oil palm, barley, solanaceous crops (Magoon and Khanna, 1963;Maluszynski et al., 2003a, b;Nasertorabi et al., 2012;Nasution et al., 2013;reviewed in Andersen, 2005;Dunwell, 2010;Germanà, 2011).
Based on the fact that cassava is outcrossing and monoecious, utilization of heterozygous parents to generate F 1 's for onward evaluation and selection is the norm since the inception of cassava breeding (reviewed in Ceballos et al., 2004;2012).However, several advantages in using inbred progenitors in cassava have been reported (Ceballos et al., 2012).In maize, genetic homogeneity and inbred lines can be obtained in one generation through use of doubled haploid inducers, which are attained by crossing target lines with a DH inducer line and thereafter selection among progeny individuals that are doubled haploids (Kebede et al., 2011).
The use of DH inducer technology boosted maize breeding as inbred lines are readily tested and used within a shorter time.In contrast, cassava has a long breeding cycle (that is, generation of sexual seeds) of more than a year, followed by five to six years of evaluation and selection to identify best performing genotypes (Kawuki et al., 2011).Therefore, conventional breeding approach has limited genetic precision particularly when responding to emerging threats such as cassava brown streak and cassava mosaic diseases in Africa and thus remains limited to match the emerging demands for cassava resources (Aerni, 2006;Bull et al., 2011;Ceballos et al., 2012).Doubled haploids (DHs) are a preferred method for developing homozygous lines in a timely fashion (Jain et al., 1996;Maluszynski et al., 2003a, b;Wedzony et al., 2009).DHs can be produced by tissue culture methods that use immature pollen (microspores) or anthers (androgenesis), ovules (gynogenesis), immature embryos from distant crosses and after pollination with irradiated mature pollen to produce a fully homozygous plant in a short time, enabling the production of pure lines from a donor plant within one generation (reviewed in Forster et al., 2007;Wedzony et al., 2009).The outstanding contributions of DH technology have been reported in several crops, especially in Brassica spp., maize, barley and wheat (Alison et al., 2011).The possibility to raise DH lines in cassava has just been investigated by Perera et al (2014a, b) focusing on androgenesis, gynogenesis and crosses with castor bean (Ricinus communis L.).While all the three methods gave promising results in terms of ability to generate calli, neither of them has so far made a breakthrough to generate cassava plantlets (Perera et al., 2014a, b).Moreover, there is no optimized method to regenerate microspore or anther-derived embryos and plants.
The use of irradiated pollen in combination with embryo culture is an efficient method for induction of haploids in several plants notably plum (Prunus domestica L; Peixe et al., 2000), oil pumpkin (Curcurbita pepo; Koṧmrlj et al., 2013) and melon (Cucumis melo; Sari et al., 1994;Nasertorabi et al., 2012).In cassava, embryo rescue methods are available (Fregene et al., 1999;Akinbo et al., 2010;Yan et al., 2014), which further motivates the use of the pollen irradiation technology.Recently, considerable progress was made for early embryo rescue (two to three weeks after pollination) in cassava by cassava crosses (Restrepo, 2014).Taken together, the prevailing challenges in cassava breeding arising from cassava reproductive biology can be overcome using DH technology to facilitate attainment of homozygosity in a short time.
Regardless of the method chosen for doubled haploid production, fast and inexpensive identification of candidate plants and the development of accurate methods or tools chromosomal duplication are critical.Flow cytometry is widely used to evaluate DNA content and determination of the efficiency of haploid-induction (Nasution et al., 2013).
Other tools used in the identification of haploids include molecular markers, especially microsatellites (simple sequence repeats, SSR), due to their stability and codominance, allowing the separation of homozygotes from heterozygotes (F 1 ) (Drumeva et al., 2005;Belicuas et al., 2007).
The aim of this study was to regenerate cassava plantlets from embryos after self-pollination with irradiated cassava pollen of cassava genotype Nase 14.To our knowledge, this is the first attempt of using irradiated pollen to generate plantlets for on-ward development of doubled haploid cassava lines.

Plant material
A cassava variety, Nase 14, was used in this study.Isolated plots of ca.1.5 acres (0.60 ha) were established at National Crops Resources Research Institute (NaCRRI), Namulonge, Uganda (~ 200 m away from the nearest cassava plants of another genotype.The variety Nase 14 was selected because it flowers profusely and is highly responsive to tissue culture (unpublished personal observation); it is thus likely that its explants or progeny will respond well in tissue culture during downstream applications.Mature male flowers (unopened buds) were harvested from cultivar Nase 14 and subjected to gamma irradiation.

Pollen irradiation
Harvested mature male flowers were irradiated using a Co 60 gamma ray source (GWGP-80, Cobalt-60, Nuclear Power Unit Institute, China, Installed at Mulago Hospital, Kampala, Uganda).This irradiation was performed in small cardboard boxes (5 × 7 × 2 cm).The irradiation source had a half-life T½ of 5.26 years and the mean energy of 1.2 MV with the maximum source to surface distance of 80 cm.The field size was set using a hand control panel.The treatment couch table was adjusted by moving it vertically, laterally and horizontally to position the samples in the center of the radiation beam.
Based on the fact that this was a pioneering pollen irradiation experiment in cassava, there was no prior knowledge of the most appropriate dosage to be used.Therefore, the most frequently used ranges in other crops were used (Peixe et al., 2000;Nasertorabi et al., 2012), taking keen interest in dosage rates that have been reported to induce haploids in those crops.
For each irradiation treatment, ~150 male buds were detached from the plants in the morning (0630 h local time) including buds for the control treatment and placed in (90 × 30) mm Petri dishes.Each irradiation treatment was represented by five Petri dishes, with each Petri dish having ~40 male buds.Only mature male buds (just before anthesis) were irradiated (treatments running from 0800 to 1500 h) and their extracted pollen used for pollination in the evening (1700 to 1830 h).Thus, pollen detachment from mother plants, irradiation and pollination were all done on the same day (within 13 h).Pollination with non-irradiated pollen was also done in the evening, as stated above.

Pollen germination
Pollen was cultured on modified Brewbaker and Kwack (1963) medium containing 100 mg/l boric acid; 300 mg/l calcium nitrate; 200 mg/l magnesium sulfate; 100 mg/l potassium nitrate; and 5% sucrose adjusted to pH 5.8.Petal-like bracts were removed from the male flowers and pollen was brushed into the medium.Thereafter, the pollen was incubated in an oven maintained at 40°C and 100% relative humidity.Aniline blue was used for staining, prepared by dissolving 1 g of aniline blue in 100 ml of 85% ethyl alcohol.Aniline blue stains callose that is located in the inner layer of the pollen tube, and although it is not commonly used to stain pollen from the genus Manihot; evidence showed that it is an effective method for observing and studying pollen tube development, pollen morphology, longevity, viability and crossability in species of Manihot and other genera (Fang, et al., 2015;Vieira et al., 2015).Pollen germination and viability were assessed after 24 h using a sample of 300 pollen grains per treatment.Three replicates were performed per treatment.Pollen was considered to have germinated when pollen tube length was at least equal to or greater than the grain diameter.Pollen viability and germination ability were observed by direct microscopy through an inverted microscope (Nikon Alphaphot-2 YS2, Japan) equipped with a mercury lamp of 100 W and an excitation UV light.

Pollination with irradiated pollen
Female flowers were bagged for one day (with white interwoven a nylon bag that is routinely used in cassava pollinations) to avoid undesirable pollinations.Only bagged and opened (receptive) female flowers were pollinated.Pollination was done by hand by brushing the stigmas with irradiated pollen, and thereafter the pollinated flowers were re-bagged for three to seven days to eliminate contamination by foreign pollen.This is the same protocol used in controlled pollinations in cassava (Kawano, 1980).Control pollinations with non-irradiated pollen were performed at the same time.All pollinations experiments were conducted in the field at NaCRRI as described above.The cassava field was laid out in plots such that each irradiation dosage rate was assigned a separate set of plants/plot onto which the irradiation treatment was only applied.Field records were taken on the total number of pollinations made per dosage rate.Beginning with three days after pollination (DAP), the numbers of surviving fruits (pollinated female flowers) per treatment were recorded up to 42 days, when embryo rescue was done.Records were also taken on seed-set (seeded or seedless fruits), and number of embryos recovered.Five rounds of pollinations with irradiated pollen were undertaken.

In vitro embryo rescue
Fruits were harvested 42 DAP, washed under running tap water, and surface-sterilized for 20 min with 15% sodium hypochlorite (NaOCl) containing two to three drops of Tween 20.Thereafter, the fruits were rinsed three times with sterile distilled water in a laminar flow hood as previously described (Fregene et al., 1999).Some fruits did not have seeds.Setted seeds were excised from the fruits under sterile conditions in a laminar flow hood and the immature embryos were isolated under a stereomicroscope using sterile forceps.
The immature embryos were each separately cultured in culture tubes containing 15 ml of solid medium (M6) containing modified Murashige and Skoog (MS) medium; 2% sucrose, 1 mg/l GA3, 0.7% agar, adjusted to pH 5.8.This methods was adopted from Catano et al. (1993).Immature embryos were cultured for approximately one month in the growth room with a temperature of 28±2°C for 16 h in the light and 8 h of darkness.Embryos that had experienced reasonable growth (with 1-2-cm long shoots) were transferred to baby jars containing 40 ml of solid medium of MS supplemented with 2% sucrose, vitamins and 0.3% phytagel adjusted to pH 5.8.Subsequently, each plantlet was sub-cultured by putting the single stem node cutting on new medium containing solid MS-medium with 2% sucrose.The generated plantlets were subjected to microsatellite genotyping and ploidy analysis.

Microsatellite genotyping
To ascertain zygosity status of the 62 plantlets generated from the rescued embryos, DNA was extracted using the Qiagen kit (Qiagen, Venlo, The Netherlands), following manufacturer's instructions.The DNA was assayed using 16 genomic simple sequence repeat markers (SSRs; EME425, NS158, NS169, SSRY100, SSRY103, SSRY106, SSRY135, SSRY148, SSRY179, SSRY181, SSRY19, SSRY20, SSRY215, SSRY240, SSRY28 and SSRY59) (Mba et al., 2001).All these markers/loci were heterozygous in the mother genotype (Nase 14), and thus deviations from the original heterozygous state to homozygous state (for the generated plantlets) would indicate attainment of homozygosity.The increased homozygosity attained through inbreeding could be inferred from the plantlets derived from the non-irradiated treatments.Amplifications with SSR primers were performed in 10 µl reactions containing 50 ng of DNA, 1 pmole of each primer, 1x Taq DNA Polymerase buffer, 2 mM MgCl2, 0.2 mM dNTPs and 0.375 U Taq DNA Polymerase (New England Biolabs Inc.).The PCR cycling parameters were 95°C for 2 min, followed by 30 cycles of 95°C for 30 s, 55-57°C for 1 min, and 72°C for 1 min and a final extension at 72°C for 30 min.Differently-sized amplification products at the different loci and variable fluorescent labels [(NED, 6-FAM, PET and VIC (by manufacturers, MWG-Biotech)] on the forward primer allowed multiplexing of amplicons from the same individual.
For each co-loading set, 1 to 2 µl, depending on amplification efficiency, of the different amplicons were mixed and briefly vortexed.Aliquots of 1 µl of the mixture were added to 9 µl of a master-mix containing Hi-Di formamide and GeneScan 500 LIZ size standard (1 ml of Hi-Di and 12 µl of 500-LIZ).Amplicons were denatured at 95°C for 3 min, and subjected to capillary electrophoresis using the ABI 3730 DNA sequencer (Applied Biosystems) and allele calls made using GENEMAPPER ® software version 3.7 (Applied Biosystems).All genotyping was done at the Biosciences Eastern and Central Africa hub in Nairobi, Kenya.The DNA of the heterozygous Nase 14 was used as a control.In addition, genetic relationships among the generated plantlets were calculated using the simple matching Euclidean distance (DE) followed by cluster analysis using the weighted neighbour-joining algorithm.Relationships were visualised as a dendrogram.This analysis was done using PowerMarker version 3.25 (Liu and Muse, 2005).

Ploidy analysis
Eight-week old plantlets (since initiation onto embryo rescue medium) were obtained from rescued embryos and subjected to ploidy analysis using flow cytometry technique according to Otto (1990).For each plantlet, the fresh innermost part of approximately 25 mg of a young leaf tissue was cut with a sharp razor blade in a glass Petri dish containing 0.5 ml of cold OTTO1 buffer (0.1 M citric acid monohydrate and 0.5% Tween-20).The contents were then passed through a nylon filter of 50 μm pore size.The filtrate was centrifuged at 1000 rpm for 5 min, and the pellet was re-suspended in OTTO I buffer.Thereafter, 400 µl of OTTO II buffer [containing 0.4 M anhydrous Na2HPO4, 4 µg/ml of DAPI (4,6-diamidino-2phenylindole), and 1µl/ml mercaptoethanol] was added and the samples run for analysis.Diploid mother plants of cassava (cv.Nase 14) were used as internal controls for ploidy determination.The channels or peak mean of diploid cassava were used to compute a ratio that could be used to discriminate the ploidy levels of the samples.

Effects of radiation on pollen, fruits, and embryos
Pollen germination done 13 h after irradiation treatments showed that irradiation doses used did not prevent pollen germination.Microscopic observations also indicated that pollen grains germinated across all irradiation doses (Figure 1).However, the frequency of pollen germination was reduced for irradiated pollen as compared to fresh or non-irradiated pollen (Table 1).Indeed, compared to fresh pollen, the 13 h old non-irradiated pollen (0 Gy) had the lowest reduction in pollen germination (24%) while the irradiated pollen registered reductions of up to 67% with 200 Gy irradiation dose (Table 1).
At the time of embryo rescue (42 DAP) (Figure 2A-E), less than 10 fruits per irradiation treatment were still surviving following pollination with irradiated pollen (Table 1).Fruit length and width did not differ significantly (P > 0.05) across the irradiation doses.However, average fruit length (1.4 cm) and fruit width (1.2 cm) were consistently lower than those recovered from non-irradiated pollen treatments, which were 1.6 and 1.5 cm, respectively (Table 1).Significant differences (P < 0.05) were observed in number of fruits, seeds and embryos generated with the non-irradiated pollen treatments having significantly higher numbers (Table 1).Under natural conditions, the cassava pistil is trilocular with a single ovule in each locule, producing a maximum of three seeds from each fruit at maturity (Ogburia et al., 2000).However, for all pollinations undertaken with irradiated pollen, some fruits did not contain seeds while some recovered seeds did not contain embryos.For example, at 42 DAP in 100 Gy treatment, an average of 9.8 (ca.10) fruits were rescued containing 21.2 (ca.21) seeds (instead of the expected maximum 30 seeds in a trilocular pistil) from which only 5.4 (ca.5) embryos were recovered (Table 1).On the other hand, the nonirradiated pollen treatment had on average 26.2 (ca.26) fruits at 42 DAP from which 64.8 (ca.65) seeds were recovered and 45.6 (ca.47) embryos were rescued (Table 1, Figure 2D-F).On average, fruits resulting from pollinations with irradiated pollen had 1.9 seeds per fruit, while those from non-irradiated pollen had 2.5 seeds per fruit (Table 1).For the seeds that had an embryo, one embryo was found for each seed.Two weeks after embryo rescue treatments, the first leaf appeared (Figure 2F) and by eight weeks, a total of 62 cassava plantlets were recovered from embryo rescue procedures (Figure 2H-J).

Ploidy and zygosity analysis
The regenerated 62 plantlets (Figure 2I-J) were subjected to zygosity and ploidy analysis.Each plantlet was considered a different genotype and thus analyzed separately.On calibration using diploid Nase 14 as the internal control, the channel or peak mean was 78.8 (Table 2).Ideally, a haploid plant should have half of this mean (that is, channel mean of 39.4).The ratio of channel means of mother plant, Nase 14, were all above 39.4 (Table 2), indicating that none of the plantlets was haploid.
When analyzed with 16 SSR markers that are heterozygous in the mother plant (Nase 14), none of the 62 plantlets was completely homozygous (Supplementary Table S1).However, varying levels of homozygosity were observed.For instance, the four plantlets resulting from 150 Gy had homozygous loci in a frequency varying between 6.0% (one locus out of 16 loci) and 50.0%(eight loci out of 16 loci), while the four plantlets from 250 Gy had homozygous loci varying between 12.5 and 68.0%(Table 2, Supplementary Table S1).The five plantlets from 0 Gy had percentages of homozygous loci ranging from 43.0 to 68.0% (Table 2).On average the highest homozygosity was observed from plantlets from 0 Gy and the lowest was from plantlets from 200 Gy (Table 2).These results suggest that the plantlets obtained may have originated in self-pollinations rather than through parthenogenesis as it was hoped.

DISCUSSION
This study reports for the first time in the history of cassava breeding the use of irradiated pollen and in vitro embryo rescue procedures to regenerate cassava plantlets as candidate materials for the development of doubled haploids.Irradiation dosages 50 to 250 Gy were studied across five pollination events, while dosages 300 and 500 Gy were used in one pollination event.Although irradiation reduced pollen germination (between 30-67%), none of the irradiation dosages was able to completely inhibit pollen germination completely.Non-irradiated 13 hour-old pollen showed a 24% reduction in pollen germination, a finding that has strong implications towards ongoing efforts to address flowering and fruit-set bottlenecks in cassava crossing nurseries.Pollination with irradiated pollen and subsequent embryo rescue resulted in the recovery of 62 diploid cassava plantlets a Numbers in parentheses indicate percent germination based on counts of 300 pollen grains across replicates.Assessments of nu mber of fruits, fruit length, fruit width, number of seeds and embryos rescued, all done at 42 DAP.Data for dosages 300 and 500 Gy are not presented.that are currently undergoing in vitro growth to allow future field evaluations and characterization.Successful in vitro pollen germination in the family Euphorbiacea is rare with limited success (for example, Orrego and Hershey, 1984;Mbahe et al., 1994;Chavarriaga-Aguirre and Halsey, 2005;Viera et al., 2012); however, efficient pollen germination was achieved in our study using modified Brewbaker and Kwack (1963) medium at pH 5.8.These results annul an earlier hypothesis by Viera et al. (2012) claiming the presence of a specific substance found in the stigma but not in the medium that induces pollen germination, which opens up opportunities to study the impact of irradiation in cassava.The effects of ionizing radiation on pollen germination have been studied in a number of plants (Visser and Oost, 1981;Pooler and Scorza, 1997;Musial and Pryzwara, 1998;Yue and Zou, 2012;Koṧmrlj et al., 2013).These studies generally concluded that irradiation results in a reduction of pollen germination.Similarly, our study show that pollen germination in cassava is possible up to 500 Gy, although there was a general reduction in the frequency of pollen germination with irradiated pollen.Studies in peach (Prunus persica) have shown that pollen germination is still possible at 1500 Gy (Pooler and Scorza 1997).Furthermore, huge differences in pollen germination frequency between non-irradiated pollen (that is, freshly collected pollen and 13 h old pollen) indicate that viability of cassava pollen after shedding drops rapidly, something that is indirectly implied by poor fruit set observed in cassava crossing nurseries (Nassar and Ortiz, 2006).Future experiments will reduce the time between pollen collection and pollination; develop protocols for preserving pollen viability, and to determine a clear dose response.Nonetheless, irradiation may have had an effect on the pollen as revealed by the zygosity patterns of the generated plantlets.In addition, source activity (for example, Gy/min) along with total radiation is an important factor to take into account.Ideally, future studies should use high-activity sources that will require a short time of exposure to achieve the desired and probably higher levels of radiation.
Pollinations with irradiated pollen ranged between 494 (for 250 Gy) to 595 (for 50 Gy), potentially translating to 494 and 595 fruits, respectively.However, by 15 DAP over 80.0% of the fruits had aborted (highest abortion in the 250 Gy treatments).By embryo rescue time at 42 DAP, fewer fruits and seeds were obtained from irradiated treatments as compared to the non-irradiated treatments (P < 0.05).Due to the fact that this was the first time to use irradiated pollen in cassava, no direct comparisons can be made.Studies on other plants, notably peach (Prunus persica) have reported fruit set of less than 15 % for pollinations with pollen irradiated with 530, 820 and 5000 Gy (Pooler and Scorza, 1997).This suggests that more than 85% fruits or seed loss is observed following pollination with irradiated pollen.In kiwifruit (Actinidia deliciosa), the number of abnormal seeds drastically increased in fruits when pollinations with irradiated pollen were performed, that is by 68% for 700 Gy and 74% for 900 Gy (Musial and Pryzwara, 1998).Related studies show that the number of seeds generated by pollination with irradiated pollen is a function of genotype (Nasertorabi et al., 2012), while the frequency of haploids generated using irradiated pollen is a function of the recipient female parent and the pollen donor (Koṧmrlj et al., 2013).Also, haploid production from the same genotype is not consistent across different pollination events with irradiated pollen (Koṧmrlj et al., 2013).Our study consisted of only one cassava genotype, Nase 14.This decision was made in order to get an initial insight based on one genotype to guide follow-up studies on the use of irradiated pollen on many other cassava genotypes.
High fruit abortion was observed in all cases resulting from pollination with irradiated pollen, irrespective of low (50 Gy) or five-fold higher doses (250 Gy).Each irradiation dose was associated with a significantly lower number of seeds that had no embryos.In kiwifruit, seeds resulting from pollination with irradiated pollen at 700 and 900 Gy had 90% of the seeds containing only endosperm and 9.6% seeds containing both endosperm and embryos (Musial and Pryzwara, 1998).On the other hand, in plum (Prunus domestica), there was a continuous fruit drop in number from 30 to 70 DAP due to absence or occurrence of fertilization, followed by a rapid rejection of the male genome (Peixe et al., 2000).
Fruit abortion can be a result of rapid collapse of the zygote due to problems related to mitosis in irradiated pollen (Peixe et al., 2000), a phenomenon that could also explain the high abortion rates observed in our study.It is also possible that the aborted cassava fruits had fragile haploid embryos, but this was not determined under this study.However, without systematic efforts to rescue and analyze ploidy and zygosity levels of the embryos, it is rather difficult to make accurate judgments on what is happening in vivo.In cassava elite lines HMC-1, SM1219-9 and MCOL1505, natural open pollination by insects showed that the ideal stage for embryo rescue was 32 to 36 DAP (Yan et al., 2014) compared to 42 DAP in our study.Thus, our current efforts are tailored towards optimizing embryo rescue protocols for cassava embryos that are 10-14 DAP, a period associated with high fruit abortion as observed in our study.This similar strategy could be helpful in rescuing cassava embryos from wild crosses, which are also associated with significantly high abortion rates (Restrepo, 2014).These efforts are critical as the lack of normal fertilization process (due to the irradiation of pollen) would prevent the normal development of endosperm which would result in the abortion of the haploid embryos that would otherwise develop.
Ploidy and zygosity analysis of the 62 plantlets rescued from embryos at 42 DAP indicated that none was haploid, suggesting that the irradiated pollen from each of the dosages was able to germinate on stigma, grow within the style, reach the embryo sac, and thereafter fertilize the egg cell.Alternatively, the development of normal diploid zygotes may have originated in pollinations happening after the bags covering the flowers have been removed.It is because of the interest to develop doubled haploids through different approaches that recent studies have demonstrated that stigmas remain viable for up to three days after anthesis (unpublished data).This information was not available at the time this current study was conducted.Thus, no parthogenesis was initiated in the recovered plantlets.It is also likely that the irradiation doses partially affected the pollen sperm nuclei but did not affect its capacity to fertilize the egg cell.In kiwifruit, the frequency of abnormal seeds was shown to increase following pollination with irradiated pollen (Musial and Pryzwara, 1998).On the other hand, it is also possible that parthenogenesis may have been induced but the resulting embryos would have failed to develop fully as reflected by the high fruit abortion.In kiwifruit, parthenocarpic fruits were obtained at varied doses, that is, at 200 Gy (hexaploids), 500 to 900 Gy (hexaploid and triploids), and at 1500 Gy (only triploids) with embryo rescue done between 50 and 72 DAP (Musial and Pryzwara, 1998).In cassava, the deployed irradiation doses (50 to 500 Gy) were probably not sufficient.Thus, follow-up studies could consider exploring higher irradiation doses that should be accompanied by optimal protocols for embryo rescue within the first 14 days after pollination.Equally important is the need to have short time requirements for attaining irradiation doses and undertaking subsequent pollinations.
The possibility that some mutations or aberrations occurred in the pollen grains and were transferred to the progeny is implied by the genetic diversity observed among the S 1 -inbreds as revealed by heterozygosity data.For example, the S 1 -inbreds resulting from pollination with non-irradiated pollen had 56.2% homozygous loci (implying that some 43.8% could potentially be heterozygous).On the other hand, S 1progeny derived from pollinations with irradiated pollen had an average frequency of homozygous loci between 28.1% (100 and 200 Gy) to 55.0% (50 Gy).This translates to about 71.9 to 45.0% loci being in a heterozygous state.One generation of inbreeding of a completely heterozygous plant is expected to generate 50.0%loci in a homozygous state and the other 50.0%loci in heterozygous state.Indeed, the results from the non-irradiated control approximate this expectation with 56.2% loci being in homozygous state, but not the results of the irradiation treatments (Table 2, Supplementary Table S1).This suggests that new genetic variability may have resulted from radiation-induced mutations; however, we plan to fully characterize these plantlets and to establish whether or not novel traits have been added to the already existing native traits in the mother parent.
Taken together, this study provides evidence that pollen germination in cassava though reduced is not completely inhibited by irradiation dosages of up to 500 Gy, which opens up opportunities to explore the effects of higher irradiation dosages.Despite the high incidences of fruit abortions and seeds without embryos after selfpollination with irradiated pollen, the successful regeneration of 62 diploid cassava plantlets via embryo rescue demonstrates a major advancement towards the development of doubled haploid plants via irradiated pollen for pure line breeding schemes in cassava.As such, the placement of cassava in the current and future agriculture merits adoption of this technology to generate genotypes that match emerging needs.Current follow-up studies are aimed at weaning and hardening of the plantlets to allow field evaluations, self-pollination experiments, implementing these procedures to other genotypes, exploring higher irradiation doses by use of high activity sources, and rescuing embryos at earlier stages of development.

Figure 1 .
Figure 1.In vitro germination of pollen grains of cassava (M.esculenta Crantz) cv.Nase 14 after irradiation with various dosages.Pollen tubes (thick thread-like structures) emerging out of the germinating pollen grains (darkened circular structures) are shown.Upper panel (A-F) shows germination without aniline blue staining, while the lower panel (G-L) shows germination with aniline blue staining.The dosage used was 0 Gy (no irradiation, A and G), 50 Gy (B and H), 100 Gy (C and I), 150 Gy (D and J), 200 Gy (E and K), and 250 Gy (F and L).Images for dosage of 300 and 500 Gy are not shown.

Figure 2 .
Figure 2. Embryo rescue at 42 days after pollination (DAP) and regeneration of cassava plantlets.A and B: Opened and unopened cassava locules containing one embryo each for extraction; C: Freshly extracted embryo; D, E and F: Embryo germination after three days, 1 week, and 2 weeks of culture, respectively;G and H: cassava plantlets after 4 and 16 weeks of culture, respectively, from the time embryo rescue was done; I and J: cassava plantlets from embryos rescued from irradiated pollinations for three and eight weeks, respectively, of being hardened in the screen house.A total of 62 cassava plantlets were successfully regenerated from the rescue embryos.

Table 1 .
Influence of irradiation dosage on cassava pollen germination, mean number of fruits, seeds, and embryos generated following pollination with irradiated pollen.

Table 2 .
Zygosity and ploidy analysis of 62 cassava plantlets recovered from embryos that were generated from crosses with irradiated pollen.Cassava mother genotype that was considered under this study.Number of homozygous loci out of the 16 loci that were ass ayed in each of the samples. a