Three B. napus spring cultivars, namely Westar, Topas, and DH12075, were used for this study. Seeds were sterilized with 20% bleach, washed twice in sterile distilled water, vernalized at 4°C for 24 h, germinated on plates containing MS medium for 5 days (21°C, 16 h light/8 h dark period, 150 μmol/m²/s), and transferred to soil in 0.7 L pots. After 2 weeks in phytotron (21°C, 16/8 L/D period, 150 μmol/m²/s), the plants were fertilized with KRISTALON™ Start [N-P-K (19-6-20) + 3% Mg + 7.5% S] and 1 week later transferred to 1.5 L pots. With the first visible flowering buds, the pots were transferred to the greenhouse chambers (Photon Systems Instruments, s.r.o.). Ten plants per each cultivar per replicate in four biological replicates were tested for each temperature regime. Plants from each cultivar were cultivated in blocks of individual pots distant from the plants of other cultivars in each of the greenhouse chambers (120 m³ cultivation space) to avoid cross-pollination. The chambers were maintained with a long-day regime (16 h light/8 h dark), LED lights with a light intensity of 150 μmol/m²/s, 35%–45% humidity, and 18°C during the night. During the day, control (CT), mid- (MT) and high-temperature (HT) chambers were set to 21°C, 28°C, and 34°C, respectively, with ramping of the temperature up and down by 4°C per hour (Supplementary Figure S1). During this study, the temperature growth regime 21/18°C is defined as the optimal growth temperature (control temperature, CT), 28/18°C is a suboptimal growth temperature (mid-temperature, MT), and 34/18°C is a stressful high growth temperature (high-temperature, HT). The temperature was uniformly distributed in each chamber by an air circulation system. Temperatures were controlled with a thermostat in the middle of each chamber, situated on the wall at 150 cm from the ground, and adjusted by water heaters and an adiabatic cooling system. Plants were well watered manually in the trays to avoid any effect associated with drought stress. Plants were once fertilized with KRISTALON™ Fruit and Flower [N-P-K (15-5-30) + 3% Mg + 5% S] at the flowering start and kept in the respective chamber till seed harvest. The experiment was performed between October 2019 and March 2020.

The number of leaves (NL) was counted at the beginning of flowering as a controlling trait. The length of the main flowering stem (LMS, cm), the duration of flowering time (FT, days), the number of primary branches on the main stem (NB), the number of flowers on the main stem (NF), and the number of ovules per pistil were evaluated during each experiment (four replicates, 10 plants per cultivar). FT was measured as the duration between the first and the last opened flowers.

Pollen viability was tested by both acetocarmine and Alexander staining. Alexander staining solution was prepared according to Alexander (1969). Pollen grains were tapped on a microscopic slide with 10 μl of Alexander solution, incubated at 50°C for 1 h, and counted under the light microscope Zeiss Axioscope.A1. For acetocarmine solution, 2% carmine solution was made in 95% glacial acetic acid (Marutani et al., 1993). Pollen grains were stained for 10 min and evaluated using the same microscope.

For silique length (SL, cm), 30 flowers were hand-pollinated for each cultivar (five flowers on six plants, four biological replicates). The siliques were measured every 24 h for 11 days to monitor silique growth over time. For correlation analysis, the final length and number of seeds (NS) per silique were recorded at harvest. Siliques with a length <1 cm were excluded (no seeds). The Relative Growth Rate (RGR) from the exponential growth model was used to examine the temperature effect on silique growth after pollination. RGR is estimated by the coefficient (r) from the exponential function: Y = C ∗ e r t , where Y is the variable modeled and C, a constant term. When r = 0.1, the silique grows by an average of 10% every time point t.

Seeds from five hand-pollinated flowers on 10 plants were evaluated for the number of seeds and different phenotypes of mature seeds after harvest. Moreover, seeds from two plants were collected at 3, 4, 5, 6, 7, and 8 DAP (days after pollination). The ovule number (NO) was calculated by adding the number of developed seeds and unfertilized ovules, counted under the stereomicroscope in all siliques. From siliques at 5–8 DAP, the number of fertilized, aborted, and non-fertilized was similarly evaluated. Seeds were cleared in a chloral hydrate solution (chloral hydrate, Sigma C8383/Glycerol/Water, 8/1/3 w/v/v) for embryo phenotyping at DAP indicated in the text. Observations were made using a light microscope Zeiss Axioscope.A1 equipped with DIC optics and ZEN blue software for picture analysis at CEITEC MU Cell Imaging Core Facility. Premature seed germination (pre-harvest sprouting, PHS) was evaluated at 26 DAP. After harvest, seeds were stored in paper bags at room temperature. The 100-seed weight and germination rate with seedling viability assessment were analyzed 4 months after harvest to allow equal seed desiccation, stored in paper bags in the dark at room temperature.

The results (multiple pairwise comparisons between different conditions-temperatures) were analyzed using generalized linear mixed models (Poisson or negative binomial mixed model for count data and linear mixed model for continuous data), and ANOVA followed by Tukey’s post hoc test in R software (R Core Team, 2014) in RStudio (RStudio Team, 2019). Mixed models were used, so the randomness between the different batches is considered (McCulloch and Neuhaus, 2005). The level of statistical significance was set at p ≤ 0.05 for all tests. Packages “glmmTMB” and “emmeans” were used for the fitting of mixed models and implementing pairwise comparisons, respectively (R Core Team, 2014; Brooks et al., 2017; RStudio Team, 2019; Russell, 2020).

Harvested mature seeds were analyzed for their oil composition and glucosinolates content. Three technical replicates for each of the three biological replicates from CT and MT regimes were analyzed for the three cultivars. Mature seeds were collected for each plant separately. A total of 210 seeds from three plants (3× 70 seeds, more than 0.5 g) was sampled for one technical replicate. Seeds for each technical replicate originated from three different plants from the same set of plants grown for phenotyping. Each biological replicate for seed metabolites content analysis corresponds to plants grown from a biological replicate for the phenotyping analysis. Sample measurements were performed by a spectrophotometer FT-NIR Antaris II (Thermo Fisher Scientific Inc., United States) on the integration sphere in reflectance mode in a spectral range of 10,000–4,000 cm⁻¹ using OMNIC for Antaris software. Whole seeds were measured in rotary circular cuvettes with quartz bottom permeable for NIR radiation. The resulting spectrum of each sample was obtained as an average of 64 scans with a resolution of 2 cm⁻¹. Calibration models for quantitative analysis of the oil, main fatty acids (palmitic, stearic, oleic, linoleic, linolenic), and glucosinolates content were developed using Partial Least Squares algorithm in Thermo Scientific TQ Analyst software. Data measured by routine laboratory reference methods were used to construct FT-NIR calibration models. Determination of oil content was performed by extraction method according to ISO 659 (2009), which defines the weight determination of oil content after extraction with petroleum ether, distillation of the solvent, and drying the extracted fat. The dry matter content was determined by a gravimetric method, according to ISO 665 (2000), after 4 h of drying at 103°C. To determine the GSL content, the HPLC/UV–VIS method, according to ISO 9167-1 (1992), was used, which defines the GSL determination in the form of desulfoglucosinolates. The representation of individual fatty acids in the form of fatty acid methyl esters was detected by GC/FID according to the internal methodology of the Research Institute of Oilseed Crops. All methods used are validated and routinely used. Data of total oil are presented as % content in dry matter. The content of each fatty acid is shown as % of the total oil content. The nitrogen compound content (%) is a measure of protein content in seeds. The number represents the mean value ± SD. Statistical evaluation for significant difference in heat treatments was performed by a paired Student’s t-test (*, **, and *** correspond to value of p 0.05 > p > 0.01, 0.01 > p > 0.001, and p < 0.001, respectively). The fatty acid composition is also expressed by the ODP (Oleic Desaturation Proportion) and LDP (Linoleic Desaturation Proportion), derived from the formulae: O D P = 100 × % Linoleic + % Linolenic % Oleic + % Linoleic + % Linolenic and L D P = 100 × % Linolenic % Linoleic + % Linolenic (Cherif et al., 1975). These parameters are proportional to the action of enzymes for the desaturation of oleic and linoleic acids, respectively. Graphs are done with GraphPad Prism 9.0.

IAA, its metabolites, and ABA were determined following the methods described by Pěnčík et al. (2018). Samples of Westar pistils, 5 DAP, and 26 DAP seeds (two biological replicates, each with three technical replicates) containing 10 mg of frozen homogenized tissue were extracted with 1 ml of 50 mM phosphate buffer (pH 7.0) containing 0.1% sodium diethyldithiocarbamate and a mixture of stable isotope-labeled internal standards. One portion of the extract (200 μl) was acidified with HCl to pH 2.7 and purified by in-tip micro solid-phase extraction (in-tip μSPE). Another 200 μl of the extract was purified directly without acidification to determine IAOx. The last 200 μl portion was derivatized by cysteamine, acidified with HCl to pH 2.7, and purified using in-tip μSPE to determine IPyA. After evaporation under reduced pressure, samples were analyzed using HPLC system 1,260 Infinity II (Agilent Technologies, United States) equipped with Kinetex C18 column (50 mm x 2.1 mm, 1.7 μm; Phenomenex) and linked to 6,495 Triple Quad detector (Agilent Technologies, United States). Data are presented as pmol for 1 g of fresh tissue. The number represents the mean value ± SD. Statistical evaluation for significant difference in heat treatments was performed by a paired Student’s t-test (*, **, and *** correspond to value of p 0.05 > p > 0.01, 0.01 > p > 0.001, and p < 0.001, respectively).

Tissue (50–100 mg per replicate) collected from CT and HT regimes was directly frozen and stored in a −80°C freezer. For isolation, samples were ground into a fine powder using mortar or ceramic beads added to 2 ml Eppendorf tubes. The total RNA from leaves, pistils, and young seeds at the early globular embryonic developmental stage (7 DAP from CT and 5 DAP from HT) were isolated following a protocol for Trizol (Invitrogen). RNA from 26 DAP seeds was isolated using a NucleoSpin RNA Plant and Fungi kit (Macherey-Nagel) according to the manufacturer’s protocol. All samples were treated with a rDNase, RNase-free (Macherey-Nagel). M-MLV Reverse Transcriptase (Promega) was used for reverse transcription on two μg of DNA-free RNA.

Expression of genes involved in auxin and ABA biosynthesis, degradation and signaling, and genes related to mRNA splicing pathway and regulators of flowering and thermogenesis were quantified. Gene sequences for primer design were available in the Genbank database (LOC number listed in Supplementary Table S1). The selection of the candidate reference genes for normalization was made using the combination of the algorithms provided by Normfinder (Andersen et al., 2004), geNorm (Vandesompele et al., 2002), and BestKeeper (Pfaffl et al., 2004). The geometric mean of the Cq values of the selected genes was used as a normalization factor for calculating the log2-fold change. The selected reference genes were BnaACT7 (leaves, pistils, and 26 DAP seeds), BnaEIF5A (leaves, pistils, young seeds, and 26 DAP seeds), and BnaTMA7 (pistils and young seeds). These genes were chosen as their expression is not affected by the temperature treatment in the given tissues. Temperature-dependent expression of targeted genes (Supplementary Table S1) was evaluated in the presented tissues (leaves, pistils, seeds). All primers sequences used in this study are listed in Supplementary Table S1. Gene transcript abundance was quantified by RT-qPCR using FastStart Essential DNA Green Master (Roche) on a Lightcycler 96 (Roche). The experiment was performed with three biological replicates, each with three technical replicates. Analysis of fold changes was performed using the 2^−ΔΔCt method (Livak and Schmittgen, 2001). The significance of the expression changes was evaluated by a t-test. The qPCR reactions were performed according to MIQE Guidelines (Bustin et al., 2009).

Three B. napus cultivars were evaluated for main growth characteristics in the three temperature regimes. At the beginning of the experiment, NL (number of leaves) was evaluated with no significant difference among the temperatures demonstrating a comparable vegetative growth of all the plants randomly distributed to the different greenhouse chambers (Supplementary Figure S2A). For all the cultivars, significantly higher LMS (length of the main stem) and NF (number of flowers) were detected in HT conditions, while MT only increases LMS to a limited extend and cause no change of NF (Figures 1A,B; Supplementary Figure S2). The flowering stem’s growth speed is more accelerated than the production of flowers by the floral meristem on the main stem, resulting in a decreasing trend in NF/cm ratio (NF/LMS; Figure 1C). A strong correlation was detected in CT (control growth temperatures, 21/18°C) between LMS and NF with a comparable level among all three cultivars (>0.77). In MT (mid growth temperature, 28/18°C), a strong correlation was maintained in Topas (0.90) and Westar (0.74) but dropped to 0.43 in DH12075. In HT (high growth temperature, 34/18°C), on the contrary, this correlation dropped for Topas (0.44) to almost no correlation for Westar (0.22) and increased to 0.77 for DH12075 (Supplementary Table S2).

Differential growth imposed by MT and HT was also reflected in flowering time (FT). In MT, Westar and Topas reacted with shortening their flowering time (Figure 1D; Supplementary Figure S2). Topas kept a reduced flowering time in the HT regime due to problems with the floral meristem, restricting growth, while DH12075 and Westar displayed a significantly longer flowering time in HT (Figure 1D; Supplementary Figure S2). Therefore, these two cultivars developed more flowers per day in MT but not HT (Figure 1E). On the other hand, due to a shorter flowering time, Topas had a significantly higher number of flowers per day in both MT and HT (Figure 1E). In general, a strong correlation was found between the flowering time and the number of flowers in CT and MT, with a decreasing tendency to a low correlation in HT (Supplementary Table S2). Moreover, all the cultivars developed more branches in HT treatment (Supplementary Figure S2).

Longer main stems, prolonged flowering duration, and more branches in DH12075 and Westar cultivars (Figure 1; Supplementary Figure S2) suggest changes in hormonal levels, maintenance of growth, and decreased apical dominance in HT growth conditions. On the other hand, in Topas, the floral meristem arrested its growth earlier at HT and, therefore, developed more flowers per day. However, the other measured parameters showed that Topas appeared less negatively affected by both MT and HT growth regimes on the whole plant level than the other cultivars (Supplementary Figure S2D).

The relative expression of PHYA and ELF4, genes related to thermosensing, was assessed. We found both PHYA and ELF4 to be significantly downregulated in HT in leaves and pistils (Figure 1F). The expression of the FLOWERING LOCUS T gene, indirectly regulated by ELF4 through the Evening Complex in response to high temperature (Kumar et al., 2012; Box et al., 2015; Ezer et al., 2017), was not significantly changed in leaves, but significantly upregulated in pistils (Figure 1F). A lower amount of PHYA and ELF4 may de-repress the thermomorphogenesis pathway (Ezer et al., 2017).

The success of reproduction relies on the regulated development of ovules within the pistil and pollen grains inside the anthers. It is also determined by the proper timing of pollination and fertilization events for subsequent seed development. The number of ovules per pistil (NO) was comparable in all the temperature regimes (Table 1; Supplementary Figure S3A). For Westar, we found a minimal but significant difference between CT and HT, which may be caused by the higher variability of this trait in the Westar cultivar, but its biological relevance is relatively small.

Pollen viability decreased by less than 5% by any temperature treatments (MT and HT), which was not considered a significant effect (Supplementary Figures S3B– D). Fertilization and abortion rates were monitored in siliques from 5 to 8 DAP. In all three cultivars, the percentage of non-fertilized ovules increases with increasing temperatures, resulting in fewer developing seeds. Moreover, Topas and Westar cultivars also exhibited an increasing percentage of aborted seeds at the early stages of development. On the contrary, the abortion rate in DH12075 decreased in HT compared to CT and MT (Table 1).

ABA and auxin metabolites content were quantified in pistils from CT and HT in Westar to understand why the fertilization rate was reduced. Increased levels of tryptophan (TRP), indole-3-acetamide (IAM), and indole-3-acetonitrile (IAN) were found in pistils without any changes in indole-3-acetic acid (IAA) and degradation metabolites levels except for decreased levels of 2-oxoindole-3-acetyl-1-O-ß-d-glucose (oxIAA-Glc; Figure 2; Supplementary Table S3). Expression of MYB34, encoding an MYB transcription factor regulating the expression of the CYP79B2/3 enzymes converting Trp into indo-3-acetaldoxime (IAOx; Celenza et al., 2005), was significantly increased in HT pistils and leaves (Figure 3A). However, no significant changes in the expression of CYP79B2/3 in HT pistils and leaves were observed (Figure 3A). Regarding auxin degradation, only GRETCHEN HAGEN3.5 (GH3.5) is significantly upregulated. GH3 enzymes catalyze the conjugation of IAA with amino acids (Wojtaczka et al., 2022). Despite GH3.5 upregulation in pistils, the Asp-conjugated IAA levels remain unchanged, and GLU-conjugated IAA levels are below detection limits (Figure 2). Also, the DAO genes expression was down-regulated in pistils at HT (Figure 3B). DAOs are auxin oxidases involved in auxin degradation by the formation of oxIAA from IAA (Mellor et al., 2016). oxIAA is further metabolized by conjugation with glucose into oxIAA-Glc, which levels were decreased in pistils at HT (Figure 2). We did not observe any phenotypic changes of non-pollinated pistils and the auxin homeostasis seems to be maintained in HT conditions. Therefore, the increased TRP, IAM, and IAN levels may be related to secondary metabolites protection rather than auxin biosynthesis. Also, we found a reduced expression of ABA biosynthetic genes NCED9 and significantly lower ABA content (Figures 3C,D; Supplementary Table S4). The non-accumulation of ABA in pistils developed at HT may suggest that they are less protected against temperature stress, which may eventually contribute to the reduced fertilization rate.

Seeds were investigated to monitor embryo development from 3 DAP to 8 DAP (Table 2). The early embryonic development in B. napus follows the same dynamics described in Arabidopsis (Verma et al., 2021). Following fertilization, the zygote elongates within the 3 days after pollination (DAP). The zygote divides asymmetrically to give rise to a small apical cell and an elongated basal cell (Figure 4A). At 3 DAP, all the cultivars exhibited embryos in the zygote or one-cell stage. The basal cell will divide anticlinally to form a cell file connecting the apical cell and its progeny to the seed coat cells. The apical cell undergoes three rounds of symmetrical divisions to form an 8-celled embryo (Figures 4B,C). At CT, the three studied cultivars have 8-cell embryos at 5-to-6 DAP in Topas and Westar and 6-to-7 DAP in DH12075 (Table 2). The follow-up cell divisions are asymmetrical with the formation of the embryonic epidermal layer (protoderm) and the specification of the root pole at the lower tier of the embryo to form an early globular embryo between 6 and 8 DAP (Figure 4D). Then, the ground and vascular tissues are specified in the mid-globular stage at 8 DAP (Figure 4E). After 8 DAP, the shoot apex is specified in the upper tier at the late-globular stage, followed by the emergence of two cotyledon primordia from the transition stage (Figure 4F) to form a heart embryo (Figure 4G). The embryo development is accelerated with increasing temperatures, leading to two developmental stages difference at 6 DAP (Table 2; Supplementary Figure S4). Faster embryonic development may affect the synchronous development of the seed coat and the endosperm and eventually lead to a higher abortion rate in later stages (Table 1). Less than 1% of defective embryos were observed in CT for all cultivars. This number significantly increased for DH12075 (6%) and Westar (10%) in MT, and for all the cultivars in HT (DH12075, 25.45%; Topas, 16.2%, and Westar, 39.58%; Figure 4L; Supplementary Figure S4).

Defective embryonic patterning was observed at 6, 7, and 8 DAP. The defective embryos exhibited several altered cell division patterns in the embryonic lower tier and cell proliferation in the suspensor altering the apical-basal embryonic axis (Figures 4H–K; Supplementary Figure S5). We observed elongated embryos with a proliferating lower-tier domain (Figure 4J). The aberrant divisions in the lower tier domain and the upper suspensor cells ill-defined the root pole and the shoot-root junction (Figures 4J,K). The root pole appears not to be correctly specified (Figures 4J,K). Aberrant or misregulated cell divisions in the apical embryo domain cause incomplete cotyledon development (Figure 4K). The suspensor cells proliferated, resulting in a long cell file (Figure 4H), never observed in CT. In extreme cases, some suspensor cells start dividing periclinally and, after several rounds of division, appears what looks like a secondary embryo (globular shape; Figure 4I).

Because the observed embryonic phenotypes are similar to known mutants with altered auxin homeostasis, we quantified auxin levels in Westar seeds, containing embryos ranging from 8-cell to early globular stages. A decrease in TRP, IAA, and some degradation metabolites (IAAsp and IAA-Glc) and an increase in oxIAA were detected, indicating altered auxin homeostasis in HT (Figure 2). However, the expression of GH3s, CYP79Bs, and DAOs was not significantly changed in seeds bearing early globular embryos in HT (Figures 3A,B). In addition, temperature-induced misregulation of the spliceosome activity may affect embryo development. Therefore, we analyzed whether HT would alter the expression of SLU7, RNPII30C, and SCL30A genes, involved in the alternative splicing pathway. All three were significantly upregulated in the young seeds developed at HT (Figure 4M). Both MT and HT and altered ABA levels in seeds may regulate the levels of the spliceosome components (Cruz et al., 2014; Jiang et al., 2017). But no changes in ABA levels were detected in 5 DAP seeds (Figure 3D; Supplementary Table S4). As one would expect an increase in ABA linked to the applied temperature stress, the steady ABA levels may relate to the abortion of seeds due to lower stress protection (Table 1). Together all these data suggest that the decrease in rapeseed seed yield due to warm growth temperatures is mainly caused by a decreased fertilization rate and defective embryo development. The perturbance of IAA homeostasis and spliceosome activities by warm ambient temperatures may partly be responsible for the embryonic patterning defects.

Silique length (SL) is believed to correlate with the number of seeds they contain (Bac-Molenaar et al., 2015). In HT, in all three cultivars, we have noticed that about 50% of the ovules are not fertilized and that only 32%–47% of the ovules will develop into seeds (Table 1). Therefore, siliques were assessed to their growth rate for 11 days from hand pollination. Siliques from all three cultivars grew on average by 22%–23% daily at CT during the 11 days (Supplementary Figure S6). This growth rate significantly decreased when growth temperature increased to reach a daily growth rate of 18%–20% and 10%–15% at MT and HT, respectively, for all cultivars. The growth rate at HT was, therefore, half the rate than at CT. There is a strong correlation (0.92–0.96 and 0.92–0.99) between the reduced seed number per silique and the reduced growth rate of the siliques at MT and HT, respectively, compared to CT (0.63–0.83; Supplementary Table S6).

The number of viable seeds per silique and seed phenotypes were examined after harvest (Figure 5). Three categories of seeds: fully filled (Figure 5A), partially filled/shrunken (Figures 5B–D), and germinated/sprouted (Figures 5E,F), were considered viable seeds. Both regimes with elevated temperatures severely affected the viable seed yield, with a significant decrease at MT, and a critical one at HT, resulting in almost no viable harvested seeds (Figure 5G). The Topas cultivar has the highest number of germinated seeds before harvest (PHS). It was evident that part of those seeds germinated even before maturation (Figures 5E,F, 6A,B; Table 3).

Fully and partially filled seeds were investigated for seed weight characteristics. In DH12075 and Westar, the 100-seed weight decreased significantly in MT conditions (Figure 5H). HT was not chosen for this experiment due to a poor yield of fully developed seeds under these temperature conditions. A decrease in weight was not identified in Topas. With the highest number of seeds per silique and stable seed weight, Topas could be considered the best yield provider under MT and HT growth conditions.

Based on the mature seed phenotyping analysis, many Topas seeds displayed a sprouting phenotype, which appeared before seed maturation (Figures 6A,B; Table 3). In these seeds, either the embryo emerged from the seed coat, or the seed coat ruptured due to the embryo pressure from 20 to 35 DAP. PHS rarely occurred in CT (less than 1% for all cultivars), but its rate is significantly higher in HT (30.4%–73.4% in Topas). To a lesser extent, it was also observed in Westar (3.8%–7.9%) and DH12075 (2.8%–13.2%) in HT (Table 3). The occurrence of PHS corresponds with the percentage of germinated dry seeds after harvest. Sprouting seeds are not viable because they did not progress into the seed maturation phase.

Growth at HT altered hormonal levels and metabolites content in 26 DAP seeds. Higher TRP content with lower levels of direct auxin precursors (indole-3-pyruvic acid (IPyA), indole-3-acetaldoxime (IAOx), and IAN) was observed in HT seeds (Figure 7A). IAA levels are also decreased in 26 DAP seeds. Furthermore, a shift from the oxidative auxin degradation by the DAO enzymes toward the production of amino acid-auxin conjugates was observed (less 2-oxoindole-3-acetic acid, oxIAA, and oxIAA-Glc, more indole-3-acetic acid-aspartate, IAAsp, and indole-3-acetic acid-glutamate, IAGlu). It was confirmed by an expression analysis showing downregulation of DAOs genes and upregulation of some GH3 genes (Figure 7B). Moreover, 26 DAP seeds are in the seed maturation stage, preparing for dormancy. ABA is known to promote dormancy in many crops (Groot and Karssen, 1992; Lefebvre et al., 2006; Huang et al., 2016). The significant decrease in ABA levels in 26 DAP seeds from HT (Figure 6C; Supplementary Table S4) indicates that HT limits seed dormancy progression by reducing the NCED9 expression (Figure 6D). Lower ABA levels may explain the PHS phenotype when the embryo continues to grow, breaks the seed coat, and finally desiccates during fruit ripening. Accordingly, BnaFBA6 and BnaDRM2, genes involved in sugar metabolism and dormancy release, were significantly upregulated at HT in 26 DAP seeds (Figure 6D). In addition, the genes involved in alternative splicing, SLU7 and SCL30A, were significantly upregulated, while RNPII30C is downregulated in 26 DAP seeds (Figure 4I).

Fully and shrunken-filled seeds were investigated for germination. Seedling viability and appearance were monitored for seeds collected in CT and MT. In general, most of the seeds (>80%) germinated 1 day after vernalization, regardless of the temperature, the seeds developed, reaching >94% of germination rate after the 4th day. Within this range, a reduction in germination was notable for DH12075 (2%) and Westar (4%) seeds (Figure 8A).

In CT, the rate of defective seedlings was 2.1% for DH12075, 5.3% for Topas, and 4.3% for Westar (Figure 8B). In MT, the percentage of defective seedlings increased to 12% for DH12075, 15.4% for Topas, and 16.8% for Westar (Figure 8B). Several types of defects were detected, from which some may partially be the consequence of the defects observed during the early embryo phenotyping analysis (Figures 4, 8C– I). We observed viable seedlings with unequal shape of cotyledons (0.67%–1.1% in CT, 1.5%–2.5% in MT) or only one cotyledon (0.44%–2.7% in CT, 3.2%–10.7% in MT, Figures 8D,E). This defect was in some cases combined with either the absence of a root or shoot apical meristem or both (Figures 8G,I). Those seedlings completely stopped growing or just delayed their development until a new root or shoot was developed, thanks to the high plasticity of plant regeneration. Other detected phenotypes were, for example, seedlings with three cotyledons and secondary seedlings occurring in less than 1% of germinated seeds (Figure 8F).

As Brassica napus is an oilseed crop, the quality and composition of oil and the content of glucosinolates and nitrogen compounds were tested in seeds. In general, the oil content was reduced by growth at MT temperature regime by 2%–6% in all the cultivars in MT compared to CT (Figure 9; Supplementary Table S6). The highest decrease was found in DH12075 cultivar from 42.27% (CT) to 36.24% (MT) in dry matter (Figure 9C; Supplementary Table S6). Changes were also detected in oil quality and quantity of unsaturated fatty acids (Figures 9C,D; Supplementary Table S6). The main compound in rapeseed oil is oleic acid (C18:1), which was increased by ~3% in MT at the expense of desaturated linoleic acid (C18:2) and linolenic acid (C18:3), which decreased in average by 0.86% and 1.49%, respectively. These changes correlate with the changes in the desaturation activity of oleic acid measured by ODP: decreased in average by 2.71% in all cultivars at MT (Supplementary Table S6). Moreover, the desaturation activity of linoleic acid measured by LDP decreased in average by 3.75% with temperature in all cultivars, indicative of the accumulation of linoleic acid over linolenic acid at MT (Supplementary Table S6). Contents of palmitic and stearic, in rapeseed oil are low (<5%), and their changes are not biologically relevant (Figure 9D).

Seeds also contain more proteins measured as levels of nitrogen compounds. Protein levels are increased by more than 4.42% in MT (Figure 9B; Supplementary Table S6). Growth at MT increased the glucosinolate levels by 2.17–5.29 μmol/g in MT. In total numbers, Topas exhibits the lowest increase in glucosinolates (+2.17 μmol/g with 12.77 μmol/g in MT). On the other hand, in DH12075, the level of glucosinolates increased to 16.80 μmol/g (Figure 9A). However, the amount of GSL did not reach harmful concentrations for consumption.