Contribution of Awns to Seed Yield and Seed Shattering in Siberian Wildrye Grown under Irrigated and Rainfed Environments

: The seed yield of grass species is greatly dependent on inﬂorescence morphological traits, starting with spikelets per inﬂorescence and seeds per spikelet, to kernel size, and then to awns. Previous studies have attempted to estimate the contribution of these traits on the harvested yield of major cereal crops, but little information can be accessed on the inﬂuence of awns on seed yield of forage grass species. Siberian wildrye ( Elymus sibiricus L.) is a widely important perennial forage grass used to increase forage production in arid and semi-arid grasslands. The grass has long inﬂorescences with long awns developed at the tip end of the lemmas in the ﬂorets. In order to evaluate the effect of awns on Siberian wildrye seed production, awn excision analyses from 10 accessions were performed at ﬂowering stage under irrigated and rainfed regimes. Overall, awn excision reduced thousand-seed weight and seed size under both irrigated and rainfed regimes, which decreased ﬁnal seed yield per plant. De-awned plants produced signiﬁcantly more seeds per inﬂorescence, but spikelets per inﬂorescence was not inﬂuenced by awn excision in either condition. Moreover, histological analyses showed a high degradation of the abscission layer in the awned plants than de-awned ones, and awn excision evidently improved average seed breaking tensile strength (BTS), and thus decreased the degree of seed shattering. In conclusion, the observed signiﬁcant impact of awn excision on different yield-related traits mirrored the impact of awns on the performance of Siberian wildrye under diverse growing conditions. These results provide useful information for plant breeders, seed producers, and researchers to efﬁciently improve seed production in Siberian wildrye. growth between early seedling emergence and the anthesis (May, June, July), than seed development stages. Moreover, the average temperature of both seasons increased during anthesis and seed development stages, and this increment was more pronounced in the second season than the ﬁrst.


Introduction
Siberian Wildrye (Elymus sibiricus L.) is a perennial, self-pollinating allotetraploid bunchgrass species that is important for arid and semi-arid regions [1]. Siberian wildrye has an erect culm ranging from 40 to 90 cm long, and the leaves are basal and cauline. The inflorescence of Siberian wildrye is a nodding spike containing 15-30 spikelets, with 4-5 fertile florets in each spikelet, and bristle, curling awns developed at the tip of the lemmas in the florets, which can reach 20 mm in length. The grass is native to highaltitude regions of western and northern China, and widely distributed in Northern Europe, Northern Asia, and North America [2]. The species plays an important role in maintaining and developing the alpine pastures because of its seedling vigor and resistance to pests and stress conditions. Owing to its forage yield potential and palatability (for

Plant Materials
The contribution of awns to seed yield, seed shattering, and other seed yield-related traits was analyzed in a total of 10 Siberian wildrye accessions (Table 1). These accessions were selected by evaluating awn length and were screened from 80 Siberian wildrye accessions. In this species, long-awned genotypes usually have awn lengths of about 20 mm, while short-awned genotypes have awns of less than 10 mm in length [4]. To ensure phenotypic variation, both long-awned and short-awned accessions were selected. The seeds of these accessions were obtained from the U.S. Department of Agriculture Germplasm Resources Information Network (GRIN), Sichuan Agricultural University, Sichuan Academy of Grassland Science, and Lanzhou University. Before planting, the seeds were germinated in plastic boxes containing a moistened double layer of blotter paper at 25 • C, 85% relative humidity, and a photoperiod of 12 h light, 12 h dark. Seedlings were grown under greenhouse conditions until they reached 8 weeks.

Experimental Conditions and Trial Design
The field trial was carried out at the Yuzhong experimental station of Lanzhou University in Gansu province, China (latitude: 35 •  growth between early seedling emergence and the anthesis (May, June, July), than seed development stages. Moreover, the average temperature of both seasons increased during anthesis and seed development stages, and this increment was more pronounced in the second season than the first.
second season than the first. The plots were arranged in a randomized complete block design with three r cates. Individual treatment plots were 30 m 2 (6 m long × 5 m wide; 50 cm row spacing 20 cm inter-plant spacing), and seedlings were transplanted at an optimal 10-15 cm de After transplanting, a total of 50 mm of water was applied immediately in both irrig and rainfed plots and no fertilizers were applied afterwards. A total of 75 mm was a tionally added to the irrigated plots between heading to maturity. When plants reac anthesis stage, two treatments involving the presence (awned) and absence (de-awned awns were imposed. Within each plot, treatment was replicated three times. Excisio awns was performed for basal, central, and apical spikelets with scissors.

Awn Length and Seed Yield Components
Seed-related traits were measured at early maturity stage (21 days after anthe Seed samples were oven-dried at 80 °C for 2 days, and clean and dry seed samples w stored in paper bags at 20 °C and 15% relative humidity before being subjected to m urements. For awn length (AL), seed length (SL), and seed width (SW), seeds sample the basal, central, and apical spikelets were used; fifty random seeds were chosen in e sample. Seeds were scanned by Epson Perfection V700 Scanner (Seiko Epson Corp., gano, Japan). The images were analyzed by Image J software [24]. Thirty random inflo cences were used to evaluate spikelets per inflorescence (SpI), seeds per inflorescence florets per spikelet (FS), and seeds per spikelets (SSp). All these measurements were m ured from the same spikelets. To obtain thousand-seed weight (TSW), thirty random florescences from each plant were selected. The TSW was determined by averaging weight of three thousand seed samples from basal, central, and apical spikelets. The plots were arranged in a randomized complete block design with three replicates. Individual treatment plots were 30 m 2 (6 m long × 5 m wide; 50 cm row spacing and 20 cm inter-plant spacing), and seedlings were transplanted at an optimal 10-15 cm depth. After transplanting, a total of 50 mm of water was applied immediately in both irrigated and rainfed plots and no fertilizers were applied afterwards. A total of 75 mm was additionally added to the irrigated plots between heading to maturity. When plants reached anthesis stage, two treatments involving the presence (awned) and absence (de-awned) of awns were imposed. Within each plot, treatment was replicated three times. Excision of awns was performed for basal, central, and apical spikelets with scissors.

Awn Length and Seed Yield Components
Seed-related traits were measured at early maturity stage (21 days after anthesis). Seed samples were oven-dried at 80 • C for 2 days, and clean and dry seed samples were stored in paper bags at 20 • C and 15% relative humidity before being subjected to measurements. For awn length (AL), seed length (SL), and seed width (SW), seeds sampled in the basal, central, and apical spikelets were used; fifty random seeds were chosen in each sample. Seeds were scanned by Epson Perfection V700 Scanner (Seiko Epson Corp., Nagano, Japan). The images were analyzed by Image J software [24]. Thirty random inflorescences were used to evaluate spikelets per inflorescence (SpI), seeds per inflorescence (SI), florets per spikelet (FS), and seeds per spikelets (SSp). All these measurements were measured from the same spikelets. To obtain thousand-seed weight (TSW), thirty random inflorescences from each plant were selected. The TSW was determined by averaging the weight of three thousand seed samples from basal, central, and apical spikelets.

Spike Dry Matter
Spike dry matter (SDM) was estimated at early maturity stage (21 days after anthesis). Ten random tillers of each plant in each treatment were selected, and their spikes were careful hand-cut just below the spike and oven-dried at 80 • C for 2 days [5]. Before measurement, awns were also removed for awned spikes to avoid differences that could be caused by the presence of awns. The dried spikes were weighed, and data were expressed as g tiller-1.

Seed Yield
In early August of each year (2018 and 2019) when seeds had fully matured (28 days after anthesis), a total of 10 plants of each treatment were randomly selected, and the heads of each plant were carefully hand-cut. To avoid border effects, seed samples were taken from the center of each plot. Seed samples were harvested when the seed moisture content was approximately 45%. Seed yield per plant (SYP) measurement was carried out after samples were weighed and when the seed moisture content was about 10%.

Seed Shattering Breaking Tensile Strength (BTS)
Seed shattering was determined by measuring the minimum breaking tensile strength (BTS) required to detach seeds from pedicels. The spikes were carefully harvested at early maturity (21 days after anthesis). Each spike was attached vertically upside down to a digital force gauge (HANDIPI, China), and forceps were used to pull down each seed. The BTS measured was recorded in gram-force (gf) units. A total of 50 seeds from five spikes representing each plant in each treatment were measured.

Scanning Electron Microscope (SEM)
For histological analysis, abscission zone tissues of the two accessions (TZ02 and ZN03) with contrasting awn length and degree of seed shattering were used. Samples were collected from three central spikelets from each inflorescence at early maturity stage (28 days after heading). The samples were fixed in FAA solution [ethylalcohol (70%):glacial acetic acid:formaldehyde (40%) = 90:5:5)] at 4 • C. Transverse sections were achieved by hand-cutting through the seed and pedicel junction. The sections were dehydrated in a series of ethanol solutions (50, 70, 90, and 100%), substituted in 3-methylbutyl acetate, and vacuum dried. Sections were then gold plated and observed using a Hitachi S-3000N variable pressure scanning electron microscope.

Statistical Analysis
The statistical analyses were performed using R statistical software (version 4.0.1; R Foundation for Statistical Computing, Vienna, Austria). The data were first checked for normality and error variance heterogeneity. The normality tests showed non-significant error variances (data not shown), and the data were left untransformed. The data of seed yield per plant, yield components, and seed shattering were subjected to analysis of variance (ANOVA). The presence (awned) and absence (de-awned) of awns and growing conditions (irrigated and rainfed) were treated as fixed factors, while block and season were treated as random factors. Means were compared by using the Student's t-test and Turkey's HSD post hoc test with a level of significance of p < 0.05. To ensure reproducibility, all the experiments were replicated at least three times.

Seed Yield
The ANOVA showed that SYP was not significantly impacted by awn excision but was significantly influenced by the environment and the interaction between awn excision   (11) under control and awn excision treatments across rainfed and irrigated conditions. Error bars represent mean ± standard deviation. ns: not significant, * p < 0.05, ** p < 0.01, *** p < 0.001.

Seed Yield
The ANOVA showed that SYP was not significantly impacted by awn excision but was significantly influenced by the environment and the interaction between awn excision  (11) under control and awn excision treatments across rainfed and irrigated conditions. Error bars represent mean ± standard deviation. ns: not significant, * p < 0.05, ** p < 0.01, *** p < 0.001.

Seed Yield
The ANOVA showed that SYP was not significantly impacted by awn excision but was significantly influenced by the environment and the interaction between awn excision and the environment ( Table 2). The influence of awn excision on the average SYP was more pronounced in the rainfed regime (awned, 7.32; de-awned, 7.02; p < 0.05) than the irrigated one (awned, 6.98 g; de-awned, 7.40 g; p > 0.05) (Figure 7). Assessments across both environments revealed awn excision influenced the SYP of 10 accessions, although this was non-significant ( Figure 8). The highest average SYP was found for PI499456 (awned, 6.98 g; de-awned, 7.40 g; p > 0.05), while the lowest value was found for ZN03 (awned, 6.20 g; de-awned, 6.92 g; p > 0.05).
The structures of the abscission layers of pedicel junctions that were manually excised are shown in Figure 9. Cross-sections showed that the shape and arrangement of abscission layers were similar in the short-awned genotype (TZ02) for awned and de-awned seeds. For the long-awned genotype (ZNO3), the smooth fracture on the rachilla was observed in awned seeds, while a well-defined boundary between the pedicel and the spikelet was observed in de-awned seeds.
The structures of the abscission layers of pedicel junctions that were manually excised are shown in Figure 9. Cross-sections showed that the shape and arrangement of abscission layers were similar in the short-awned genotype (TZ02) for awned and deawned seeds. For the long-awned genotype (ZNO3), the smooth fracture on the rachilla was observed in awned seeds, while a well-defined boundary between the pedicel and the spikelet was observed in de-awned seeds.

Impacts of Awns on Seed Yield and Components
The observed differences in seed yield and yield components between awned and de-awned plants were associated to the amount of assimilates partitioned to growing kernels, and were in agreement with several reports on cereal crops [18,25,26]. The removal of Siberian wildrye awns resulted in a reduced kernel weight and harvested seed yield, and this reduction was more obvious under rainfed regimes (Figure 7). Awns improved seed length, seed width, and thousand-seed weight, and this improvement compensated for the effect of reduced seeds per inflorescence. The significant reduction in seed weight for de-awned plants resulted in small and shriveled seeds, and this impact was more obvious in apical and basal spikelets (Figure 4). The variation in seed size between apical, central, and basal spikelets in this species could also be associated with flowering order, which starts in central spikelets, and then the top and bottom spikelets, respectively. The observed inverse association between harvested yield and shorter or non-existent awns under rainfed conditions has also been reported in wheat [27,28], barley [29,30], and rye [31]. Under irrigated conditions, long awns required a particular portion of assimilates for development, which decreased the assimilates accumulated in the kernels and thus led to a yield loss, a trade-off previously documented in several cereal crops such as wheat [8,32], barley [33], and rice [20]. Under irrigated conditions, the present study revealed that awns improved spikelet number, although this was non-significant, compensating for the reduced fertile florets per spikelet ( Figure 4). Besides, awn excision treatments significantly reduced seed length and seed width, but this reduction was compensated for by increased fertile spikelets, which directly resulted in a higher frequency of kernels per inflorescence. Thus, small seed yield differences were observed for awned and de-awned plants grown under irrigated conditions.
The contribution of awns on seed yield components varied across seasons. The availability of soil moisture during the growing season promoted plant growth and affected seed yield components. In 2018, rainfall was lower during the vegetative growth stage but increased during anthesis and kernel development stages, resulting in a higher seed yield under the rainfed regime in this season compared to that of 2019. Similarly, Wang et al. [34] indicated that increased rainfall during the anthesis stage contributed to increased seed yield in Russian wildrye. Soil moisture is the key variable for the effective flowering and seed development in Siberian wildrye [35], but the excessive moisture could reduce pollination, which consequently reduces the number of seed sets per spikelet. This illustrates the fluctuations in seed yield per plant observed in 2019, which supports a previous study of Siberian wildrye that observed that excessive precipitation reduced seed setting rate and thus final seed yield [5]. The estimated association of awn length with seed width and seed length was inconsistent across growing conditions, with differences in seed width being greatest in irrigated conditions and differences in seed length being greatest in the rainfed regime ( Figure 3). This indicated that, as postulated by Teich [32], moisture could not always decide the impact of awns as the environments also vary in temperature, radiation, and soil fertility [32]. This is in agreement with our previous study that observed that the contribution of awns to seed yield of Siberian wildrye varied across latitudes, longitudes, and altitudes [4]. Thus, opportunity remains in the selection of awn lengths with comparable yields improvements under diverse environment conditions.

Role of Awns on Partitioning of Spike Dry Matter
Dry matter accumulation in the inflorescence, as in leaves and stems, was thought to be associated with nitrogen supply and soil moisture content [5,36]. Although previous studies have indicated that maximum plant dry weights could be obtained by improving the photosynthetic capacity of leaves [37], the present study mirrored the capacity of awns to support dry matter partitioning into the growing kernels, especially under stress-free regimes (Figure 4). The comparisons of Siberian wildrye accessions revealed that spike dry matter was higher in the awned treatments than de-awned ones (Figure 8). Given this capacity of awns to increase spike assimilates, one could expect that improved spike dry weight by awns is associated with yield increase under irrigated conditions. However, awns develop rapidly under favorable conditions and impose a competing sink with growing kernels, reducing the amount of assimilates allocated to kernel filling and thus causing yield loss. This competition for assimilates between different plant organs and growing florets in the reproductive stage was previously documented in wheat. For instance, potential yield increases by spike dry weights were likely limited by the competition between growing florets and elongating stems, as reported by González et al. [38].
The capacity of awns to improve spike dry matter was non-significant under rainfed conditions. According to Martinez et al. [39], this might be linked to limited moisture and high temperatures. Similarly, in wheat, Wang et al. [5] observed that leaf, stem, and spike dry matter contents were greatly affected by air temperature and soil moisture content. However, in the present work, the decrease in spike dry weights did not compromise the capacity of awns to improve seed yield under rainfed conditions. This was probably related to the reduction in awn lengths observed for several plants grown under rainfed conditions, reflecting a reduction in photosynthates allocated to those awns. The increase in assimilates allocated to growing kernels was associated with a greater increase in kernel size, especially in apical and basal spikelets ( Figure 4). Overall, given the growing pressure to improve harvested yields of grass species, future yield gains should be targeted by improving spike dry matter accumulation [8]. Developing cultivars with optimum awn length will be an attractive option to improve the inflorescence dry weights, and thus the harvest index, in grass species.

The Impact of Awns on Seed Shattering
Seed shattering is an essential adaptive trait for the efficient propagation of offspring in wild plants [40]. Breeding initiatives in grasses have primarily focused on improving grain yield and biomass production rather than reducing their shattering trait, even though seed shattering has greatly hindered harvested yield in many cultivated grass species [22]. Seed shattering is a result of the progressive deterioration of the abscission layers [41], causing mature seeds to detach from the pedicel at these layers [42]. In the present study, analyses of pedicel junctions of awned plants confirmed the presence of an abscission with two small cell layers (Figure 9), controlling cell separation and breakage. Moreover, the evaluation of seed shattering of awned and de-awned plants showed that awned seeds displayed a shattering trait with a complete abscission zone between the seed and the pedicel. However, the removal of awns induced a deficiency in the formation of the abscission zone near the vascular bundle, which caused the seeds to stay on the plant at maturity. These results indicate that long-awned species are more susceptible to shattering before harvest since awns increase seed weight and increase the axis of where the weight is concentrated, exerting additional stress on the abscission layer [22,43].
Our analyses indicated the discrepancies in the breaking tensile strength of several awned and de-awned plants (Figure 8). This illustrated how seed shattering is a highlycoordinated trait not only associated with morphological traits but also with changes in cell structure, metabolism, and gene expression [44]. Previously, a study revealed that abscission is associated with a rapid increase in cellulose and polygalacturonase activity [22]. However, how awns interact with these hormones is not clear and remains a subject of intensive investigation. The study of Magwa et al. [45] on rice indicated a possible interaction between the awn development gene (An1) and the grain-shatteringrelated QTL (sh4-1) on chromosome 1. Furthermore, a number of studies also showed that the expression of seed shattering may be controlled by environmental factors such as moisture and temperature [22,44]. Similarly, in this work, we observed significant differences in seed breaking tensile strength among awned and de-awned plants across irrigated and droughted conditions (Figure 7), indicating the role of the deficit and surplus of water in the expression of seed shattering in Siberian wildrye. Overall, a wide range of germplasm collections, assessments with multiple location and year testing, and both outdoor and greenhouse trials would be vital to better understand the influence of awns on seed shattering.

Conclusions
This study attempted to identify the influence of awns on harvested seed yield of Siberian wildrye. The presence of awns slightly increased the seed yield per plant through increased seed size and kernel weight. Awn excision improved seeds per inflorescence and was associated with a reduction in seed shattering. These results revealed the advantages of growing awned species in conditions with water shortages and high temperatures, and show the existing opportunities of growing awnless grass species under stress-free conditions. Plant breeders should focus on increasing spikelet per inflorescence and the number of fertile florets through awn length manipulation. Moreover, this study revealed the possibility of improving seed production by increasing seed retention through awn length optimization. In the future, a greater number of genotypes, large field experiments, and a divergent selection of multi-location trials and multi-year testing will facilitate the identification of the further impacts of awn traits on the performance of awned grass species.
Author Contributions: F.N. and Y.W. performed the field experiment and collected the data. F.N. and W.L. analyzed the data. F.N. and W.X. wrote the manuscript. W.X. conceived and designed the research. All authors have read and agreed to the published version of the manuscript.