A set of 377 KASP markers ( Supplementary Table 1 ) known to be polymorphic between wheat and T. timopheevii (King et al., 2022) were used to characterize the 24 wheat-T. timopheevii lines. DNA was extracted from 3–5 cm sections of 10-day old leaf material collected from each genotype into a 96-well collection plate. After being freeze-dried and ground using a TissueLyser (Qiagen), DNA extraction was performed as described in Thomson and Henry (1995). Genotyping was performed as described in Grewal et al. (2020). In summary, for each sample a 5 µl reaction containing: 1 ng genomic DNA, 2.5 µl KASP reaction mix (ROX), 0.068 µl primer mix and 2.43 µl nuclease free water was made using an automated PIPETMAX 268 (Gilson, United Kingdom). PCR was performed in a ProFlex PCR system (Applied Biosystems by Life Technology) and a QuantStudio 5 (Applied Biosystems) was used for fluorescence detection of the reactions. PCR conditions consisted of 15 min at 94°C; 10 touchdown cycles of 10 s at 94°C, 1 min at 65-57°C (dropping 0.8°C per cycle); then 35 cycles of 15 s at 94°C followed by 1 min at 57°C. The QuantStudio™ Design and Analysis Software V1.5.0 (Applied Biosystems) was used for data analysis. The raw genotyping data is available form the corresponding author on request.

DNA extraction for sequencing was carried out using the extraction buffer (0.1 m Tris–HCl (pH 7.5), 0.05 m EDTA (pH 8.0), 1.25% SDS) of 10 days old leaf materials. Library preparation and DNA sequencing was performed by Novogene (UK) Company Limited. Each library was skim-sequenced to 0.05x coverage on a NovaSeq 6000 S4 flow cell with PE150 strategy. Sequence data analysis was carried out using two bioinformatics pipelines, to assess both the wheat background genome and the introgressed T. timopheevii genomic regions, as outlined by Coombes et al. (2023) and Adhikari et al. (2022), respectively. The former used IWGSC RefSeq v1.0 as the wheat reference genome assembly (IWGSC et al., 2018) to align the sequence reads whereas the latter pipeline used the recently published, chromosome-scale genome assembly of T. timopheevii (Grewal et al., 2024). This analysis allowed for the precise identification of the location and the size of the introgressed wild relative segments in the introgression lines. Raw skim-sequencing reads for T. timopheevii introgression lines have been deposited at the European Nucleotide Archive (ENA) under project accession PRJEB6515.

In Sub-population 1, Tim23 and Tim18 exhibited a significantly longer filament compared to the parental wheat line Paragon ( Figure 5a ), showing an increase in length of 22%. In Sub-population 3, which included only four lines, Tim6 demonstrated a significantly longer filament than both parental genotypes, Paragon and Highbury, as well as the positive control Piko, with an improvement of up to 52% ( Figure 5b ).

In Sub-population 1, Tim24 had significantly smaller pollen grains compared to the parental line Paragon with a 14% reduction in pollen size ( Figure 5c ). In addition, phenotyping of Sub-population 2 showed that Tim28 had significantly smaller pollen grains than both Paragon and Chinese Spring, as well as the two positive controls Apache and Piko ( Figure 5d ). A reduction of 9% and 6% in pollen diameter was observed when compared to the Chinese Spring and Apache controls, respectively. Within the same sub-population, Tim8, while not significantly smaller, still showed smaller pollen grains than all the wheat controls included in the screening ( Figure 5d ). Both Tim8 and Tim28 contained introgressions from T. timopheevii chr5G. Sequencing and KASP genotyping showed the presence of an overlapping region from the long arm of T. timopheevii chromosome (Chr) 5G between these two lines ( Figure 1 ; 5G.B3 and 5G.B8).

Two additional lines screened in Sub-population 1, Tim10 and Tim18, and one line in Sub-population 2, Tim9, also contained segments from the long arm of T. timopheevii Chr5G ( Figure 1 , 5G/5B). Although Tim9 carried the same 5G introgression (5G.B3) as Tim8, its pollen size was larger than Tim8 but still smaller than the wheat parental lines Paragon and Chinese Spring ( Figure 5d ). Neither Tim10 nor Tim18 from Sub-population 1 showed a small pollen grain phenotype, as their introgressions did not include the overlapping region found in Tim8 and Tim28. Thus, this indicates that the overlapping region within T. timopheevii Chr5G introgression potentially carries a gene(s) responsible for the smaller pollen grain phenotype.

Three plants exhibited longer filament length; Tim18 and Tim23 from Sub-population 1 and Tim6 from Sub-population 3 ( Table 4 ; Figures 5a, b ). Tim6 contains two introgressions (from Chr2G and Chr3G). Only the Chr3G introgression ( Figure 1 ; 3G.B5) is partially unique to Tim6 suggesting that the gene(s) conferring a 52% increase in filament length may be located within this introgression.

Tim18 contains two introgressions from Chr5G and Chr6G, neither of which can be completely ruled out. Equivalent Chr5G introgressions are present in two other lines, Tim10 and Tim28, both of which exhibited shorter filament length. Thus, it is unlikely that the Chr5G introgression in Tim18 is carrying a gene(s) for longer filament length. However, Tim18 also carries a large and unique introgression from Chr6G which could potentially harbor gene(s) contributing to this phenotype. Further work is needed to reduce the size of this large Chr6G introgression and create lines with smaller introgressions covering different regions of Chr6G, ideally with overlapping segments. Phenotyping these lines would help identify which genomic regions are important for the increase in filament length.

Tim23 possesses an Chr5A^t introgression that is also present in several other lines screened for filament length, including Tim4 and Tim25. KASP genotyping indicates that the Chr5A^t introgressions in Tim4 and Tim25 completely match the introgression in Tim23 ( Figure 1 ). Although neither of these two lines showed a significantly longer filament, both exhibited longer filaments than those of the wheat lines in their background (Paragon, Chinese Spring and Highbury). Thus, the Chr5A^t introgression requires further investigation to determine if it carries genetic diversity for filament length. Crossing all three introgression lines (Tim23, Tim4 and Tim25) several times to Paragon to achieve a more uniform genetic background, would be an essential first step in this investigation.

Unlike anther extrusion, which is known to be a polygenic trait (Lu et al., 2013; Skinnes et al., 2010), the genetic basis of filament length in wheat has not been extensively studied. In Arabidopsis thaliana, genetic control of filament length is mainly regulated by hormones that influence gene expression and cell elongation within the filament (Ge et al., 2010). Previous studies have consistently highlighted the significant positive correlation between anther extrusion and filament length (Beri and Anand, 1971; Langer et al., 2014; Boeven et al., 2018; Denisow et al., 2022), a finding also supported by this study. This correlation is not surprising, as anther extrusion is a combined effect of flower opening and filament elongation, which pushes the anthers out of the floret during anthesis (Joppa et al., 1968; Beri and Anand, 1971; De Vries, 1974; Sage and Isturiz, 1974; Boeven et al., 2016). In wheat, flower opening during anthesis is driven by lodicule swelling (De Vries, 1971). While both lodicule enlargement and filament elongation are triggered by water influx associated with potassium accumulation, this osmotic influx could be independently regulated in these organs (Heslop-Harrison and Heslop-Harrison, 1996). Interestingly, despite the increased filament length in Tim6, Tim18 and Tim23, this did not translate to better anther extrusion in these introgression lines, unlike the wild relative species T. timopheevii. This suggests that flower opening, filament length and anther extrusion could be under separate genetic control mechanisms.

Hexaploid wheat typically produces relatively large pollen grains, with a diameter of approximately 58 µm (Goss, 1968; De Vries, 1971). In this study, the wheat controls had an average pollen diameter of 55 µm, ranging from 51 µm in Highbury to 59 µm in diameter for Piko, with Paragon, the common wheat parent of all the introgression lines, having a pollen diameter of 57 µm. While the weather conditions significantly influence pollen dispersal (Faegri and Van Der Pijl, 1979), pollen size also plays a crucial role, directly affecting on the aerodynamic properties of the pollen grain and consequently, pollination efficiency (D’Souza, 1970; De Vries, 1971; Waines and Hegde, 2003). Size differences impact the fall rate of pollen and therefore, the distance it can travel from the source. The relatively heavy pollen grains of hexaploid wheat (De Vries, 1971) have a sink rate of 55–60 cm/sec over 1 meter in still air (Lelley, 1966). As a result, more than 90% of pollen grains typically settle within 3 meters of the source (Hegde and Waines, 2004). Thus, in a hybrid wheat system, smaller pollen grains are desirable as they can travel further from the source, promoting optimum dispersal to the female plants and enhancing the potential of cross-pollination.

Two introgression lines, Tim24 (Sub-population 1) and Tim28 (Sub-population 2) showed a significantly smaller pollen grain phenotype compared to the wheat parental lines, showing a phenotype more similar to the T. timopheevii accession ( Figures 5c, d ). Comparing the genomic segments present in these lines with those not exhibiting the smaller pollen phenotype suggests that several regions of the T. timopheevii genome could be of interest for this trait. Tim24 carries two introgressions from Chr3G and Chr7G. The Chr3G introgression (3G.B1) is also present in Tim5, Tim6 and Tim35, none of which showed the small pollen grain phenotype. Hence, this introgression can be excluded as a candidate for this trait. The second introgression in Tim24 consists of a large Chr7G segment (7G.B1) that is partially unique to this line ( Figure 1 ). Although Tim7, Tim12 and Tim13 also contain a large Chr7G introgression (7G.D2), their 7G segments recombined with the D genome, whereas the segment in Tim24 recombined with the B genome ( Figure 1 ). There is a notable overlap between the 7G.B1 and 7G.D2 segments. Since the Chr7G introgression in Tim24 is larger than those in Tim7, Tim12 and Tim13 introgression lines, it was possible to narrow down the region of interest on the long arm of 7G chromosome of T. timopheevii. Sequencing data estimates the size of this region at approximately 165 Mbp. While the two Chr7G segments (7G.B1 and 7G.D2) recombined with different wheat genomes (B and D, respectively), the large size of these introgressions provides an opportunity to narrow the 7G region in Tim24 through targeted crossing. This strategy may help identify the specific gene(s) responsible for the smaller pollen grain phenotype.

Tim28 carries two introgressions from Chr1A^t and Chr5G, both of which are potential candidates for contributing to the smaller pollen grain phenotype. The Chr1A^t segment (1A^t.A1) is a small introgression of approximately 30 Mbp that has recombined with the short arm of wheat chromosome 1A and is unique to Tim28. Additionally, four introgression lines (Tim8, Tim9, Tim10 and Tim18), contain introgressions from Chr5G, but only introgression 5G.B3 overlaps with the introgression in Tim28 (5G.B8). Although Tim8 did not exhibit significantly smaller pollen grains compared to the wheat controls, it had the next smallest pollen size in Sub-population 2 after Tim28. Through an inter-crossing program between Tim28 and Tim8, a recombinant introgression line (BC₁F₁-R3) was developed, carrying the genomic region of interest ( Figure 6 ). When homozygous, phenotyping of BC₁F₁-R3 showed that it retained the small pollen grain phenotype, providing strong evidence that the reduced Chr5G region contains a gene(s) associated with smaller pollen size. Interestingly, the recombinant line BC₁F₁-R3 showed a small pollen grain size more similar to Tim28 than Tim8 ( Figure 7 ). Since Tim28 produced smaller pollen grains than Tim8, it is possible that Tim8 carries a second gene conferring a larger pollen size outside the region of overlap with Tim28. In the recombinant lines with the smaller introgressions, this gene will no longer be present. The viability of pollen in a hybrid wheat system is also of significant importance as a reduction in viability will lead to a decrease in pollination efficiency (Whitford et al., 2013). Tim8 and Tim28 both showed reduced viability compared to the wheat parents and controls. However, pollen viability of the lines carrying the reduced introgression was higher than in either Tim8 or Tim28 and was equivalent to the viability seen in the wheat controls. Thus, the larger Chr5G introgressions in Tim8 and Tim28 appear to have a detrimental effect on pollen viability either by introducing a gene(s) contributing to this negative effect or by replacing a gene(s) from wheat itself. Except for Tim33, this negative effect was limited to the lines carrying the larger Chr5G introgressions ( Supplementary Figure 2 ).

It is generally assumed that once a wild relative chromosome segment has been introgressed into wheat, reducing the size of the segment is not feasible due to the lack of recombination between the wheat and the wild relative genome. However, a strategy of inter-crossing two lines with overlapping introgressions was used by Sears (1956) to reduce the size of an Aegilops umbellulata segment carrying rust resistance. This approach has also been used to reduce the size of introgressions from rye (Koebner et al., 1986; Koebner and Shepherd, 1988; Anugrahwati et al., 2008), Haynaldia villosa (Lucaszewski and Cowger, 2017) and Aegilops sharonensis (Khazan et al., 2020). In this study, the presence of an overlapping region between the two wild relative segments in Tim8 and Tim28 enabled recombination to occur between T. timopheevii introgressions, i.e., like with like. Interestingly, while shortened introgressions were obtained through recombination between the two overlapping T. timopheevii Chr5G segments, recombination also occurred between T. timopheevii Chr5G and wheat chromosome 5B. As expected, recombination frequency was higher between the two T. timopheevii segments (24 out of 87, 26%) compared to recombination between T. timopheevii and wheat (3 out of 87, 5%). Both the B genome of hexaploid wheat and the G genome of T. timopheevii are believed to have originated from an Aegilops speltoides-like progenitor (Ogihara and Tsunewaki, 1988; Dvořák et al., 1989). However, chromosome pairing studies suggested the G genome of T. timopheevii and the S genome of Ae. speltoides are more closely related to each other than to the wheat B genome (Rodriguez et al., 2000).