Increasing seed set and pollen proofing in Brassica juncea and Brassica napus through novel non-woven synthetic pollination control bags

ABSTRACT Brassica juncea and B. napus species are predominantly self-pollinated. Parchment paper or muslin cloth bags are used for controlled selfed or cross-seed production. We evaluated three non-woven synthetic pollination control bags (PCBs) for two years as replacement for the commonly used PCBs. The new synthetic PCBs were fully pollen proof since the molecular marker analysis of the random seeds set on two cytoplasmic male sterile (CMS) lines isolated with PCBs were all maternally originated. DWB03 bags returned 47% significantly more seeds per bud than the parchment paper bags. Larger DWB03 bags showed 38% higher mean seed yield (g/plant) than the muslin cloth bags. Re-used DWB03 bags were statistically on par with new bags for the seed yield per plant. On average, DWB03 bags produced 57% more hybrid seed set on CMS lines than the muslin cloth bags. The micro-climate in DWB03 bags appeared to be more conducive for good seed set as compared to the outside and muslin cloth bags. Non-woven PCBs offer better options for replacing paper or muslin cloth bags in the breeding of Brassicas.


Introduction
Indian mustard (Brassica juncea L.) and oilseed rape (B. napus L.) are cultivated worldwide as key edible and industrial oilseed crops. Both are predominantly self-pollinated crops with varying amounts of outcrossing in the entomophilous flowers, which attract honeybee foraging (Labana and Banga 1984;Becker, Damgaard, and Karlsson 1992). Breeders maintain genetic integrity of thousands of breeding stocks by isolating floral shoots or complete plants with pollination control bags (PCBs) made of parchment paper or muslin cloth. The number of Brassica germplasm accessions alone exceeds 74,000 globally (Singh and Sharma 2007). PCBs are also used to manage bird damage in the breeding stocks (Ormerod and Watkinson 2000;Gitz et al. 2013Gitz et al. , 2015Schaffert, Virk, and Senior 2016). However, seed set in standard bags is generally low due to poor aeration, elevated temperatures, high humidity, reduced light transmission, and high incidence of pathogens and pests (Foster and Wright 1970;Gitz et al. 2015;Schaffert, Virk, and Senior 2016, 2019Yun et al. 2017). Furthermore, the parchment paper bags are prone to tearing due to high intensity winds, rains, and the pressure of growing inflorescences (Bridgwater et al. 1998;Gitz et al. 2013). Muslin cloth bags made of loosely woven fabric carry the added risk of stray pollen grains infiltrating through its holes. Of late, PCBs made from novel nonwoven fabrics are becoming popular as these avoid alien pollen ingress through randomly laid multiple fabric layers, which create a "tortuous path" and not allow the pollen to infiltrate (Hayes and Virk 2016;Clifton-Brown et al. 2018). Synthetic PCBs with greater air permeability and strength, and re-usability (PBS International 2022a, b) have been found to be more effective than standard PCBs in sorghum (Gitz et al. 2015;Schaffert, Virk, and Senior 2016, 2019Gaddameedi et al. 2017), sugar beet, wheat, and Arabidopsis (Clifton-Brown et al. 2018), Miscanthus (Hayes and Virk 2016), grasses (Vogel, Sarath, and Mitchell 2014;Adhikari et al. 2015), and oil palm (Bonneau et al. 2017). However, no corresponding studies on synthetic PCBs are available in Brassicas. Therefore, the major objectives of the present investigation, conducted on ogura-cytoplasm-based male sterile lines and fully fertile genotypes across two years, were to assess the efficiency of novel non-woven synthetic fabric PCBs for pollen proofing and biological output in comparison to the standard practices in Indian mustard and oilseed rape.

Materials and methods
Pollination control experiments were conducted during winters of 2019-2020 and 2020-2021 at the Oilseeds Research Farms of Punjab Agricultural University (PAU), Ludhiana (30.9010° N,75.8573° E), India (Figure S1, S2). In the first year, three PCBs were tested, followed by more detailed confirmatory studies on the best PCB in the second year. The flowering time of the crop from mid-February to mid-March in both years coincided with the COVID-19 pandemic. Despite work interruptions resulting in forced unbalanced experimental structure, data were amenable to robust statistical analyses.

Types of pollination bags used
Pollen-size variations in Brassica have been reported by various authors, e.g., Arora and Modi (2011), and Saha and Begum (2020) reported B. juncea pollen = 13 μm polar (P) axis and 11 μm equatorial (E) axis); and B. napus pollen = 16 P × 16 μm E. Perveen, Qaiser, and Khan (2004) estimated pollen sizes of B. napus = 20 P × 20 E μm; and B. juncea = 33 P × 30 E μm. We selected three non-woven fabrics with pore sizes little larger than the Brassica pollens. Physical properties of these materials (named DWB01, DWB03 and DWB23) are available in Table 1. PCBs were made in two sizes: (a). Two-dimensional (2-D) small bags (40 × 15 cm) to cover an inflorescence with counted number of floral buds (25 to 50 buds) and (b). Three-dimensional (3-D) bags (130 × 24.5 × 24.5 cm) to cover a whole plant for selfing or two plants (one plant of cytoplasmic male sterile (CMS) line plus one plant of fertility-restorer line) together for hybrid seed production. The standard controls were parchment paper bags for the smaller PCBs and muslin cloth bags for the larger PCBs. An open-pollinated control was also included to estimate reduction in selfed-seed output following covering of the whole plants. The best performing PCB (DWB03) from 2019-20 trials was studied more extensively during 2020-21 by increasing the sample size to 35 plants. Used DBW03 bags from the first year were saved and tested for re-usability (Hayes and Virk 2016). Data on per cent relative humidity (RH %) and temperature in Centigrade ( 0 C) were recorded, by placing electronic data loggers inside the PCBs from 15 February 2020 to 15 March 2020 ( Figure S3A-F) and from 10 February 2021 to 15 March 2021, at intervals of 30 min ( Figure S4A-F). This generated 1435 and 1615 data points for 2020 and 2021, respectively. Estimates of outside temperature and RH for the corresponding outside environment were obtained from the Metrological Station, PAU Ludhiana. The estimates of daily average, maximum and minimum temperature and RH were plotted separately for the two years ( Figures S3,  S4). The variance estimates of the daily temperature and RH recordings for new and re-used DWB03 bags were plotted for 2021 only ( Figure S5). A two-tailed paired t-test was used to test the significance of difference of means of average, maximum and minimum daily temperatures ( 0 C) and RH (%) within DWB03 new and Muslin cloth bags from the outside estimates over the two years (31 days in 2020 and 34 days in 2021). However, this comparison for the re-used DWB03 bags was only possible for 2021.  (Gudi et al. 2020).

Description of experiments
Experiment 2: This experiment tested the selfed-seed production in fully fertile plants of B. napus cv. GSC22 and B. juncea cv. AJR102B. Small 2-D bags were used to isolate a counted number of floral buds on inflorescences (25 to 50 floral buds), and the large 3-D bags were used to isolate single plants to study the impact of reduced light transmission due to relative opacity of the PCB fabric on total seed yield per plant (Figures S1, S2). Parchment paper or muslin cloth bags were used as the control treatments for small bags and large bags, respectively. Open-pollinated seed set per plant was also included as a treatment during 2019-20 to investigate plant productivity under normal conditions. Total number of seeds from each bag was counted and divided by the number of buds isolated to arrive at the number of seeds per bud for statistical analysis. Number of pods per bag was also recorded. This allowed the computing of number of seeds per pod for analysis during 2020-21. Two biological traits, plant height (PH in cm) and seed yield (SY) per plant in grams (g) were recorded at maturity for plants under large PCBs. Used bags from 2019-20 were included as an extra treatment with small and large bags on B. napus cv. GSC22. A combined statistical analysis was performed using Minitab-17 statistical package (ANOVA in Tables S1, S2). Experiment 3: We enclosed one male sterile plant and one fertility restorer plant in large synthetic PCBs to study hybrid seed set on the male sterile plants. Muslin cloth bags were used as control. CMS plants produced hybrid seeds due to pollination by the fertility restorer. In contrast, seeds set on fertility restorer plants were due to self-pollination. Data were recorded at maturity on plant height (cm) and seed yield (g per plant) from the male sterile as well as fertility restorer plants. During 2019-20, an analysis of variance was performed (ANOVA in Table S3) for number of hybrid seeds collected from CMS plants and selfed-seeds from the fertility restorer plants as factors. However, during 2020-21, analysis of variance (ANOVA in Table  S4) was performed only on hybrid seed set on CMS plants since no data could be recorded for selfed-seed on restorer-line plants.
Mean values for different bag treatments from experiments 1 and 2 were tested for significant differences using Fisher's pairwise least significant difference (LSD) comparison at 95% probability.

Results
Restricting to overall best performing DWB03 bag, its daily average temperature in both years was lower than outside but either higher than or closer to muslin cloth ( Figure S3). The re-used bags showed a slightly lower average temperature than the new bags. The maximum temperature was higher in DWB03 in both years but equal or higher than muslin cloth, with re-used bags showing lower temperature than the new bags ( Figure S3). The minimum temperature inside DWB03 was lower than the outside but comparable to re-used bags and higher than muslin cloth in both years ( Figure S3). Average and maximum RH in DWB03 was higher than outside but close to muslin cloth, with re-used bags showing highest estimates. The minimum RH inside DWB03 was close to muslin cloth but re-used bags showed higher estimates compared to the outside, which showed large fluctuations ( Figure  S4). The variances of daily temperature and RH for DWB03 new and muslin cloth bags were very similar, but re-used bags showed the lowest estimates ( Figure S5).
New and re-used DWB03 bags showed significantly lower average temperature by 1 to 2°C than the outside temperature, in line with muslin cloth bags (Table 2). However, the mean maximum temperature of DWB03 bags was about 5 degrees higher, which was slightly higher than muslin cloth but about 2 degrees higher than DWB03 re-used bags. All PCBs showed about 2 degrees lower minimum temperature than the outside. RH estimates in PCBs were generally higher than the outside ( Table 2). The mean for average RH was 14%, 16% and 22% higher for muslin cloth, DWB03 new and DWB03 reused bags compared with the outside. The mean maximum RH was 12%, 15%, 15% higher than the outside for DWB03 new, muslin cloth and DWB03 re-used bags. The mean minimum RH for DWB03 new and muslin cloth bags was on par with the outside but the DWB03 reused bags showed significantly (13%) more RH than the outside (Table 2).

Pollen-proofing ability of pollination control bags
No seed set was recorded on the isolated male sterile plants of B. napus CMS line AG24 (both during 2020 and 2021) and B. juncea CMS line DTM81 (B. juncea) during 2021. However, sporadic seed set was recorded on a few plants of CMS DTM81, isolated with different bag sizes during 2020. The lowest number of 14 seeds was observed following isolation with DWB23 and DWB01 bags versus 39 from DWB03 and 36 from the muslin cloth bags. The latter two bags also showed pre-harvest sprouting of seeds. Only 28 of the 103 seeds germinated and 27 of these were found to be of maternal origin as per KASPar marker assays ( Figure 1). However, the plant germinated from one seed obtained under the muslin cloth bag was heterozygous (Rfo rfo) for fertility restoration gene (Rfo) (Figure 1).

Comparing relative biological outputs in plants enclosed with different PCBs
A combined analysis of variance (across two years) showed that the main effects of years, genotypes and bag types were significant for the numbers of seeds per bud and seeds per pod for the plant inflorescences covered by small bags (Table S1). However, the seed output for new vs. re-used PCBs did not differ significantly (Table S1). Interactions years × genotypes and genotypes × bags were significant for both seed traits, but they had little consequence because of their low (2 to 7%) contributions toward total sum of squares (Table S1) compared with large contributions of the main effects.
A comparison of pairwise combination of least squares-fitted mean values revealed that DWB03 was significantly superior to all bag types Table 2. Comparative mean of daily average, maximum and minimum temperature ( 0 C) and mean of daily average, maximum and minimum relative humidity (RH %) within DWB03 bags (new and re-used) and outside over two years (2020 and 2021). evaluated in the overall analysis for seed number per bud and seed number per pod (Table 3). Compared with the parchment paper control, it yielded 58% more seeds per bud in both years separately. Overall, its superiority across two years was 47%, less than the average of two years because of the unequal number of bag samples for DWB03 compared with parchment paper bags, especially in 2021 (Figure 2, Table 3). In contrast, the commonly used parchment paper bags performed better than DWB01 and DWB23 synthetic bags, which showed 40% and 52% lower mean seeds per bud (Table 3).
A combined analysis of variance for plant height and seed yield per plant across two years from the fully covered plants with large bags revealed significant differences between the years and bag types (Table S2). Genotypes differed significantly for plant height but not for seed yield. Interestingly, there were no significant differences between the new and reused DWB03 bags for plant height and seed yield (Table S2). The interactions genotype × bag type and genotype × year were significant for both plant height and seed yield. However, interactions contributed little (0.2 to 7.5%) to the total SS for both the traits (Table S2).
The fitted mean values across two years showed that the mean plant height was significantly higher for open-pollinated plants than the most bagged plants, except for those covered by DWB03 bags (Table 4). The mean seed yield per plant (g) and plant height (cm) were highest in the plants enclosed with DWB03 bag. These values were comparable with those in openpollinated control (Table 4). DWB03 was as good as open-pollinated condition, with 6% overall increase in plant height over muslin cloth bags (Figure 3a). DWB03 returned 8% more seed yield than the corresponding open-pollinated plants and 38% more seed yield than the muslin cloth bags (Table 4, Figure 3b). The standard practice of using muslin cloth bags led to   (Table 4).

Comparative hybrid seed set on male sterile plants, enclosed with corresponding fertility restorers
Analysis of variance showed significant differences between bag types for the overall number of seeds set on both male sterile and fertile lines in 2019-20 (Table S3). The number of seeds set on male fertile Vs male sterile plants was also significant (Table S3). The mean number of selfedseeds on restorer plants was significantly higher (by 165%) than the hybrid seed mean on male sterile plants (Table 5). DWB03 retuned significantly higher (48%) seed set over the next best, muslin cloth bags ( Figure 4, Table 5). The other bags (DWB01 and DWB23) were far inferior to the control.  The analysis of variance for 2020-21 also revealed significant differences between the two male sterile lines for the quantity of hybrid seeds produced (Table S4). DWB03 and muslin cloth bags differed significantly for plant height and seed yield produced (Table S4). CMS lines × bag type interaction was significant for both the traits, though having small contribution to the total sum of squares, i.e., 6% for plant height and 4% for seed yield (Table S4). DWB03 bags produced 57% higher seed yield, with plants taller by 27% in 2020-21 over the the muslin cloth bags (Figure 4, Table 6).

Discussion
Parchment paper or muslin cloth bags are commonly used for controlled seed set in oilseeds Brassicas, despite their inherent limitations. Recently, PCBs made of non-woven synthetic fibers have been found to be very effective in many crops (Gitz et al. 2015;Hayes and Virk 2016;Bonneau   The non-woven materials offer more properties by flexing polymer, fiber size, length, shape and manufacturing technique and are known to have high strength to withstand plant over-growth, damage from bad weather (heavy rains and high winds), field pests and bird attack; good air permeability to balance moisture and temperature inside the bag. Sufficient light transmissibility and appropriate porosity to prevent external pollen ingress are other important factors (PBS Intl 2022a, b; Figure S6). Many types of non-woven synthetic fabrics have been found to meet these requirements, which have been shown to increase plant breeding efficiency by producing higher and better-quality seed outputs than the controls (Townson, Virk, and Senior 2020;Trammell et al. 2020). We identified three non-woven fabrics (DWB01, DWB03, and DWB23) based on their physical properties that make them more suitable as PCBs for Brassica crops. Because of different properties of fabrics, all these bags did not perform equally well in the two years of study suggesting a need to identify crop specific fabrics for PCBs (Clifton-Brown et al. 2018). However, at least one bag type (DWB03) performed outstandingly better than the control practices, and in some cases better than the open pollinated conditions, for different biological parameters. Apparently, DWB03 provided a micro-environment that was more conducive to higher seed set even when compared with the outside conditions. This might be attributed to breathability of DWB03 fabrics that led to inside temperature and humidity to levels more suitable for seed production (Table 1). Since a PCB covers reproductive structures of the plant, elevated inside temperature (Ball, Campbell, and Konzak 1992;Gitz et al. 2015) is bound to impact the seed set because of reduced pollen viability (Chowdhury and Wardlaw 1978;Harsant et al. 2013) and or seed set (Prasad, Boote, and Allen 2006). Along with rise in temperature, high humidity can also directly reduce the seed set (Foster and Wright 1970) or by creating micro-climate conditions that are more favorable for diseases and pests to develop (Yun et al. 2017). Further, Brassica specificity of DWB03 fabric is strengthened from the similar microclimate and seed output performance of the re-used and new bags. Bagging completely male sterile (MS) plants should normally result in no seed set. However, sporadic seed set may occur on bagged MS plants by pollen ingress or due to the reversion of male sterility resulting in maternally produced selfed-seeds. Our studies showed that DWB03 was completely pollen proof and selfed-seed set, if any, on CMS plants was of parthenogenetic maternal origin as KASPar marker analysis for fertility gene (Rfo) showed the rfo rfo genotype. Only one seed (out of 28 germinated) produced on CMS plant covered with muslin cloth bag presented a heterozygous (Rfo rfo) genotype that was expected from outcrossing. DWB03 was previously proved to be superior for seed output in sorghum Senior 2016, 2018;Gaddameedi et al. 2017) and grasses (Hayes andVirk 2016: Clifton-Brown et al. 2018). DWB03 are made of the fiber, known commercially as duraweb® SG1. This material could significantly reduce bird damage and mold infection in sorghum by being strong but highly breathable. It is proven to halt unwanted pollen, reduce contamination and increase seed yield in sorghum (PBS International 2022c). Completely pollen proof characteristic of DWB03 was primarily due to the physical complexity caused by weave architecture of the non-woven fabric used, as pore sizes of PCBs evaluated in our studies were larger than the pollen size of Brassica species investigated. Thermally bonded nonwoven fabric filter samples are characterized by multiple filtration layers of interconnected pores and tortuous pore paths through the fabric thickness (Wang and Gong 2006). This torturous but purposefully effective filtration of pollen through larger pore size may not assure totally impermeable conditions, yet it provides a trade-off in pollination performance while allowing exchange of air and moisture. All the fabrics of PCBs clearly provide an acceptable filtering level of co-optimization of pollen exclusion in the present experiment. DWB03 bags also showed no adverse effects with respect to reduced seed production in fully covered Brassica plants. The better biological performance of DWB03 may be attributed to its higher light transmissibility, especially in 400 to 1000 nm wavelength range, compared with kraft paper (Hannah Senior, PBS International, Pers. Comm.; Figure S6).
Another useful outcome of our study is the possibility of re-using the DWB03 PCBs more than once (Hayes and Virk 2016;Schaffert, Virk, and Senior 2018) which would reduce the actual cost and make them more economical than might first appear from the initial higher investment. However, optimization of the number of times of re-use of nonwoven PCBs needs to be established. Although PCBs enclose reproductive plant parts and are critical in artificial hybridization, plant breeders tend to allocate minimal resources to this in relation to the total cost per cross. This tendency exists even though cheap PCBs made of paper and cellulose are prone to damage by birds (Gitz et al. 2013), insects (Demirel and Cranshaw 2006), wind (Bridgwater et al. 1998); require deliberate daily bag shaking for pollen dispersion, and are affected by water, and diseases (Windham and Williams 2007) or slugs. The parchment paper PCBs are also prone to stress tearing as the plant's growth causes the seams of the parchment bags to give way. Muslin cloth bags collapse on the plant and get wet due to dew in the night, creating a favorable environment for diseases and pests. It is estimated that a medium-sized rapeseed mustard breeding program uses ~30,000 standard PCBs per season, of which around 40-50% may get damaged due to bad weather. Although economic analysis was not envisaged in this research, some conclusions can be derived (Table 7) following Schaffert, Virk, and Senior (2016) and Gaddameedi et al. (2017). Parchment papers are prone to various losses compared with nonwoven PCBs, which gave 47% higher seed yield (Table 7). Extrapolations with these assumptions provide a ratio 2.43 paper bags: 1 non-woven PCB per season. Since the latter are re-usable, the cumulative ratio for the second season = 4.86:1 and for the third season = 7.29:1. Clearly, across three seasons, 100 non-woven PCBs can replace 729 paper bags. In addition, using fewer bags also requires less labor, which may reduce direct and indirect (supervisory) cost. The muslin cloth bags are re-usable but collapse on plants, and attract moisture, increase diseases, and reduce seed purity. These assumptions provide a ratio of 1.86:1 per season, i.e., 100 large non-woven bags can replace 186 muslin cloth bags. Plant breeding programs mandate the production of large quantities of selfed or cross seeds with high genetic integrity and germinability. Failures in these may result in losing a season for testing of new hybrids/ entries in multi-location trials.

Conclusions
DWB03 pollination bag made of non-woven synthetic fiber was found superior for selfed or hybrid seed production under controlled conditions as micro-climate conditions within DWB03 bags were found to be more conducive for higher biological output in comparison to standard practices and outside conditions. Further, this bag can be re-used at least once, as no significant differences were detected between used and new PCBs in the present studies. Despite the need for careful economic analysis of new and standard pollination control practices, the benefits of synthetic PCBs are apparent and Brassica breeders may well adopt them at least on a test scale while scientific and economic impact is assessed.