Phenology and Floret Development as Affected by the Interaction between Eps-7D and Ppd-D1

Earliness per se (Eps) genes may play a critical role in further improving wheat adaptation and fine-tuning wheat development to cope with climate change. There are only few studies on the detailed effect of Eps on wheat development and fewer on the interaction of Eps with the environment and other genes determining time to anthesis. Furthermore, it seems relevant to study every newly discovered Eps gene and its probable interactions as the mechanisms and detailed effects of each Eps may be quite different. In the present study, we evaluated NILs differing in the recently identified Eps-7D as well as in Ppd-D1 at three temperature regimes (9, 15 and 18 °C) under short day. The effect of Eps-7D on time to anthesis as well as on its component phases varied both qualitatively and quantitatively depending on the allelic status of Ppd-D1 and temperature, being larger in a photoperiod-sensitive background. A more noticeable effect of Eps-7D (when combined with Ppd-D1b) was realised during the late reproductive phase. Consequently, the final leaf number was not clearly altered by Eps-7D, while floret development of the labile florets (florets 2 and 3 in this case, depending on the particular spikelet) was favoured by the action of the Eps-7D-late allele, increasing the likelihood of particular florets to become fertile, and consequently, improving spike fertility when combined with Ppd-D1b.


Introduction
As mentioned in the introduction of the companion paper (Basavaraddi et al., submitted), and elsewhere [1,2], the main genetic factors controlling wheat developmental traits are photoperiod-and vernalisation-sensitivity genes (Ppd and Vrn), both critical for coarse tuning phenology, and the earliness per se (Eps) gene, responsible for fine-tuning development [3]. Furthermore, these genetic factors controlling the duration of the developmental phases may have pleiotropic effects on yield traits [4,5], explaining a selective advantage of certain genes over others. Yield advantages of these genetic factors may depend on the strength of their effect on phenology and whether they also influence the rate and development of organs/primordia along the way. A number of studies showed that genes lengthening the duration of either time until terminal spikelet (TS) or the late reproductive phase (LRP, from then to anthesis) may affect the number of organs initiated during these phases [6][7][8][9][10].
The importance and application of genetic factors controlling developmental traits may depend on the epistatic interaction with one another and, in turn, with environmental conditions. In spring wheat, only the Ppd and Eps genes can be manipulated to alter patterns of development and associated traits. Eps genes are a group of completely different genes spread over the genome [1], only having in common what defines them: producing a relatively minor effect on time to heading/anthesis when requirements of photoperiod and vernalisation are fully met [11]. At least in part, the ambiguity of the results reported in the literature may well reflect the fact that the mechanism of each Eps gene could be quite different from one another, beyond the fact that all affect time to heading/anthesis (by definition). Final leaf number (FLN), phyllochron, plastochron, spikelet number, floret development patterns and developmental phases can be affected or not depending on the specific Eps gene studied [6][7][8][12][13][14][15]. Therefore, it is important to study and evaluate each novel Eps along with their interactions with the other genetic factors (involved in complex network of genes altering the life cycle of wheat) and the environment. The former interaction would identify the kind of genetic background in to which a particular Eps can be introgressed, resulting in optimum function. Determining to what degree the environmental condition affect the effects on phenology of a particular Eps and its interaction with the background is important if we are to predict the environment in which such a genetic combination can confer maximum benefit. For instance, in the companion paper, we showed that the environment affected the magnitude of effects of a novel Eps gene (Eps-7D) on phenology and spike fertility traits (Basavaraddi et al., submitted); moreover, in a recent paper [15], we showed that these effects of Eps-7D were affected by its interaction with another Eps gene (Eps-2B). Quantifying the interactions with major genes coarse-tuning wheat development is relevant to determine whether it is potentially useful to exploit a particular Eps gene in a certain wheat background.
The interaction between Eps and Ppd genes was evaluated in a previous study, which showed that the two genes have an additive interaction [16]. Similarly, another recent study has shown differential effect of Eps on heading time in presence of either Ppd sensitive or insensitive alleles [12]. These studies suggest the importance of understanding not just the independent effects of any new Eps gene discovered but also its interaction with Ppd alleles. The outcome of the above-mentioned studies is limited to the Eps × Ppd interaction effect on time to heading/anthesis. Such interactions were not studied yet for the recently identified Eps-7D gene.
The present study is focused on understanding the interaction effect of newly identified Eps-7D QTL with Ppd-D1 (the photoperiod-sensitivity gene most frequently shown to have the strongest effect; [17][18][19][20]) on phenology and related traits as well as on the initiation of primordia (leaves, spikelets and florets). The study was conducted in growth chambers under short day (12 h), otherwise the effect of Ppd-D1 (and its interactions with Eps-7D) would have been unnoticed, at three temperature regimes (9, 15 and 18 • C) to evaluate to what degree the effect of Eps-7D and interaction with Ppd-D1 is altered by these factors.

Phenology
Both temperature conditions (cf. different temperature regimes within the same panel of Figure 1) and the allelic status of the Ppd-D1 gene (cf. Figure 1A,B for the same temperature regimes) affected noticeably the time to anthesis of near isogenic lines (NILs) grown under short days. Furthermore, the sensitivity of time to anthesis to temperature was affected by the photoperiod sensitivity gene in the background, particularly for the difference between 15 and 18 • C (the difference between 9 and 15 • C was similar under either of the Ppd-D1 allele). That is, the difference in time to anthesis between 15 and 18 • C was much larger when the photoperiod allele was sensitive than when it was insensitive (cf. differences between these temperatures in Figure 1A,B). The same interaction is clear when comparing time to anthesis of lines with Ppd-D1b and Ppd-D1a alleles (cf. panels A and B of Figure 1). The difference was clearly larger at 9 and 15 • C than at 18 • C. Indeed, at 18 • C the difference between Ppd-D1b and Ppd-D1a was only 7 d (averaging across both Eps-7D alleles), but at that temperature, 7 d are equivalent to ca. 125 • C d. Therefore, sensitivity was still exhibited to the photoperiod at this high temperature, although it was less than the sensitivity at lower temperatures (at 9 • C, the difference between Ppd-D1b and Ppd-D1a lines was almost 20 d, equivalent to 175 • C d).
Plants 2021, 10, 533 3 of 13 7D-late allele, which always showed a delayed time to anthesis compared to Eps-7D-early, but the magnitude of these differences as well as the statistical significance clearly depended on the photoperiod sensitivity allele in the background (cf. Figure 1A, B at each temperature regime). The differences in time to anthesis between lines with the Eps-7Dlate and -early alleles were larger with Ppd-D1b in the background (ca. 7, 15 and 17 d at 18, 15 and 9 °C, respectively) than with Ppd-D1a (ca. 5, 4 and 7 d at 18, 15 and 9 °C, respectively). As expected, higher temperatures (cf. different temperature regimes within each of the panels of Figure 2) and the action of Ppd-D1a (cf. Figure 2A vs. Figure 2B and Figure  2C vs. Figure 2D for the same temperature regimes) reduced the duration of both phases.
The effect of Eps-7D on time to anthesis was result of a 'domino effect' caused by the late allele delaying time to TS as well as the duration of the LRP (Figure 2). Then again, the magnitude and significance of these changes was altered by the Ppd-D1 allele in the background, the differences between lines with contrasting Eps-7D alleles being larger and more significant when having the Ppd-D1b than the Ppd-D1a allele in the background ( Figure 2).
Which of the two phases composing time to anthesis was most affected by the action of the Eps-7D gene depended on the Ppd-D1 allele in the background. When the background had the photoperiod-sensitive allele, when the effect of Eps-7D was largest ( Figure  1), its effect on time to TS was evidently weaker than on the duration of the LRP (cf. Figure  2A,C). On the other hand, when the background had the Ppd-D1a allele, the overall effect of Eps-7D was smaller ( Figure 1) and did not consistently affect any of the two component phases considered (cf. Figure 2B   The effect of Eps-7D on time to anthesis was consistent in that the presence of Eps-7D-late allele, which always showed a delayed time to anthesis compared to Eps-7D-early, but the magnitude of these differences as well as the statistical significance clearly depended on the photoperiod sensitivity allele in the background (cf. Figure 1A,B at each temperature regime). The differences in time to anthesis between lines with the Eps-7D-late and -early alleles were larger with Ppd-D1b in the background (ca. 7, 15 and 17 d at 18, 15 and 9 • C, respectively) than with Ppd-D1a (ca. 5, 4 and 7 d at 18, 15 and 9 • C, respectively).
As expected, higher temperatures (cf. different temperature regimes within each of the panels of Figure 2) and the action of Ppd-D1a (cf. Figure 2A vs. Figures 2B and 2C vs. Figure 2D for the same temperature regimes) reduced the duration of both phases.
The effect of Eps-7D on time to anthesis was result of a 'domino effect' caused by the late allele delaying time to TS as well as the duration of the LRP ( Figure 2). Then again, the magnitude and significance of these changes was altered by the Ppd-D1 allele in the background, the differences between lines with contrasting Eps-7D alleles being larger and more significant when having the Ppd-D1b than the Ppd-D1a allele in the background ( Figure 2).
Which of the two phases composing time to anthesis was most affected by the action of the Eps-7D gene depended on the Ppd-D1 allele in the background. When the background had the photoperiod-sensitive allele, when the effect of Eps-7D was largest (Figure 1), its effect on time to TS was evidently weaker than on the duration of the LRP (cf. Figure 2A,C). On the other hand, when the background had the Ppd-D1a allele, the overall effect of Eps-7D was smaller ( Figure 1) and did not consistently affect any of the two component phases considered (cf. Figure 2B,D).
Consequently, the relationship between time to anthesis with its component phases, time from seedling emergence to TS and time from then to anthesis revealed that the Eps-7D effect on time to anthesis was driven by its effect on the LRP ( Figure 3B) far more than by its effect on time to TS ( Figure 3A). This is not only because of the proportion of the variability in the effect of Eps-7D on time to anthesis explained by the effects on each of the two phases, but also by the actual magnitude of the effects (note that the abscissa scale in Figure 3B is three-fold that of Figure 3A). Consequently, the relationship between time to anthesis with its component phases, time from seedling emergence to TS and time from then to anthesis revealed that the Eps-7D effect on time to anthesis was driven by its effect on the LRP ( Figure 3B) far more than by its effect on time to TS ( Figure 3A). This is not only because of the proportion of the variability in the effect of Eps-7D on time to anthesis explained by the effects on each of the two phases, but also by the actual magnitude of the effects (note that the abscissa scale in Figure 3B is three-fold that of Figure 3A). Consequently, the relationship between time to anthesis with its component phases, time from seedling emergence to TS and time from then to anthesis revealed that the Eps-7D effect on time to anthesis was driven by its effect on the LRP ( Figure 3B) far more than by its effect on time to TS ( Figure 3A). This is not only because of the proportion of the variability in the effect of Eps-7D on time to anthesis explained by the effects on each of the two phases, but also by the actual magnitude of the effects (note that the abscissa scale in Figure 3B is three-fold that of Figure 3A).

Dynamics of Leaf Appearance and Spikelet Primordia Development
As expected, FLN did not show any clear trend with temperature while the rate of leaf appearance was strongly positively affected (Table 1). On the other hand, photoperiod sensitivity (i) affected, although slightly, FLN (lines with the sensitive allele-Ppd-D1binitiated on average 7.22 ± 0.17 leaves, while those carrying the insensitivity allele-Ppd-D1a-initiated 6.70 ± 0.14 leaves (Table 1)); and (ii) did not exhibit any clear effect on the rate of leaf appearance (averaging across temperatures and Eps-7D alleles, this rate was 0.089 ± 0.030 and 0.087 ± 0.026 leaves d −1 for lines with Ppd-D1a and Ppd-D1b alleles, respectively; Table 1). The effect of Eps-7D on FLN was very subtle in that the line with the Eps-7D-late allele had slightly higher FLN compared to the -early allele (Table 1). Thus, the delayed anthesis of the Eps-7D-late allele was not related to the effect on FLN. Indeed, there was a consistent reduction in the rate of leaf appearance due to the action of Eps-7D-late, particularly when the background had the Ppd-D1b allele, and the reduced rate of leaf appearance was the main reason for the delayed anthesis produced by Eps-7D-late.
Most leaf primordia were initiated before the onset of the experiment (as c. four leaves are already initiated in the embryo and c. two more leaves can be initiated from sowing to seeding emergence). Hence, we could only have a good record of the dynamics of spikelet primordia initiation. As described above (Figure 2), there was an interaction between Eps-7D and Ppd-D1 genes on the timing of TS: when the background had the sensitive Ppd-D1b allele, TS was delayed by the action of the Eps-7D-late allele, while that delay was not evident when the background had the insensitive Ppd-D1a allele.
The lines with the Eps-7D-late allele had c. one spikelet more than the line with the -early allele when the sensitive Ppd-D1b allele was present in the background under 18 and 15 • C, but not at 9 • C ( Figure 4A,C,E, insets). Furthermore, the Eps-7D gene did not affect the final number of spikelets when the insensitive allele Ppd-D1a was in the background ( Figure 4B,D,F, insets).
The differences in spikelet primordia produced by Eps-7D was mainly related to the effects on duration of spikelet initiation, as spikelets were initiated more or less at a constant rate under each temperature regime (although naturally, this rate increased with higher temperatures) (Figure 4). The effect of temperature on accelerating the rate of spikelet initiation did not compensate the effect on reducing the duration of spikelet initiation, and therefore, the total number of spikelets was higher at lower temperatures ( Figure 4). Plants 2021, 10, x FOR PEER REVIEW 6 of 13 The differences in spikelet primordia produced by Eps-7D was mainly related to the effects on duration of spikelet initiation, as spikelets were initiated more or less at a constant rate under each temperature regime (although naturally, this rate increased with higher temperatures) (Figure 4). The effect of temperature on accelerating the rate of spikelet initiation did not compensate the effect on reducing the duration of spikelet initiation, and therefore, the total number of spikelets was higher at lower temperatures ( Figure 4).

Spike Fertility and Dynamics of Floret Development
Spike fertility, measured as number of fertile florets per spike, was slightly yet noticeably affected by Eps-7D and the magnitude of the effect depended upon the Ppd-D1 allele in the background and temperature regime. Lines with the Eps-7D-late allele tended to have more fertile florets per spike than that of Eps-7D-early across all conditions ( Figure 5), but the magnitude and significance of these differences depended upon the Ppd-D1 allele in the background and the temperature ( Figure 5).

Spike Fertility and Dynamics of Floret Development
Spike fertility, measured as number of fertile florets per spike, was slightly yet noticeably affected by Eps-7D and the magnitude of the effect depended upon the Ppd-D1 allele in the background and temperature regime. Lines with the Eps-7D-late allele tended to have more fertile florets per spike than that of Eps-7D-early across all conditions ( Figure 5), but the magnitude and significance of these differences depended upon the Ppd-D1 allele in the background and the temperature ( Figure 5).
Thus, in general, the effect of the Eps-7D gene was more noticeable when the photoperiodsensitive allele was in the background ( Figure 5, left panels), and that difference was largest (and significant) at 15 • C, smaller and insignificant at 18 • C and even smaller at 9 • C ( Figure 5, left panels). Thus, in general, the effect of the Eps-7D gene was more noticeable when the photoperiod-sensitive allele was in the background ( Figure 5, left panels), and that difference was largest (and significant) at 15 °C, smaller and insignificant at 18 °C and even smaller at 9 °C ( Figure 5, left panels).
As expected, beyond the effect of Eps-7D, lines with Ppd-D1b produced more fertile florets per spike than those with Ppd-D1a, although interacting with temperature as the sensitivity to photoperiod affected spike fertility at 15 and 18 °C, but not at 9 °C ( Figure 5). Eps-7D-late 16  As expected, beyond the effect of Eps-7D, lines with Ppd-D1b produced more fertile florets per spike than those with Ppd-D1a, although interacting with temperature as the sensitivity to photoperiod affected spike fertility at 15 and 18 • C, but not at 9 • C ( Figure 5).

Ppd-D1b
The differences in floret fertility are the final outcome of those in the dynamics of floret development. Floret 1 (i.e., the first floret developing in each spikelet and the one closest to the rachis; F1) reached stage 10 (fertile floret stage) in all lines, irrespective of the Eps-7D allele, the sensitivity to photoperiod given by the allele of Ppd-D1 in the background, and the temperature regimes (data not shown, as this floret is not responsible for differences in spike fertility). As the differential developmental rates are the bases for the higher spike fertility in the Eps-7D-late lines when grown at 15 • C with Ppd-D1b in the background ( Figure 5), we showed the dynamics of floret development for F2 and F3 in each of the three spikelets (apical, central and basal) considered individually ( Figure 6). To broadly illustrate the effects of the treatments (Eps-7D, Ppd-D1 and temperature) on the dynamics of floret development, a supplementary figure ( Figure S1) is available where we averaged the stages of development of these two floret positions (F2 and F3) of basal (third spikelet from the base), central (the middle spikelet) and apical (third spikelet from the TS) spikelets at each sampling time.

Discussion
The aim of this work was to study the Eps-7D × Ppd-D1 interaction under contrasting temperatures to evaluate to what degree the effects of the newly identified Eps-7D depended upon the photoperiod sensitivity in the genetic background. Naturally, we also obtained results on the effects of Ppd-D1 and the temperature on the processes studied. All these effects, occasionally commented in the description of the results, were commen-   Figure S1G).
Although for the average of the three spikelets it is not possible to appreciate clear differences, for this particular case of lines, having Ppd-D1b in the background and grown at 15 • C (in which the differences in spike fertility between the contrasting Eps-7D lines were largest; Figure 5C), we offered the rates of floret development of florets 2 and 3 at each spikelet individually ( Figure 6).
Floret 2 in Eps-7D-late lines when grown at 15 • C and with Ppd-D1b in the background clearly developed more than Eps-7D-early lines in all three spikelet positions considered ( Figure 6, left panels). In all spikelets, F2 reached the stage of fertile florets in all or almost all plants dissected when lines were Eps-7D-late, while only a minor proportion of them developed to stage 10 (of fertile florets) in Eps-7D-early lines ( Figure 6, left panels).
The third floret from the rachis aborted in the basal and apical spikelets of lines with either Eps-7D-late or -early allele, although lines Eps-7D-late allele did reach later stages of development in the apical spikelets ( Figure 6B). On the other hand, in the central spikelets, the differences between Eps-7D-late and -early lines became clear: whilst F3 in this spikelet finally aborted in all plants sampled and floret 3 was never fertile in Eps-7D-early lines, most plants of the lines carrying the late allele showed F3 reaching the stage of fertile floret ( Figure 6D).

Discussion
The aim of this work was to study the Eps-7D × Ppd-D1 interaction under contrasting temperatures to evaluate to what degree the effects of the newly identified Eps-7D depended upon the photoperiod sensitivity in the genetic background. Naturally, we also obtained results on the effects of Ppd-D1 and the temperature on the processes studied. All these effects, occasionally commented in the description of the results, were commensurate with what is well known in the literature [1,7,9,[20][21][22][23][24][25][26][27][28][29][30]. Thus, we will concentrate this discussion on the aim of the study: the effects of this new Eps-7D gene as well as the dependence on the sensitivity to photoperiod for having these effects.
A clear interaction between Eps-7D and Ppd-D1 was evidenced, not only on the phenology (e.g., the delay in time to anthesis due to the late allele of Eps-7D in presence of sensitive Ppd-D1b was much stronger compared to that produced by the same allele on Ppd-D1a) but also in terms of rates of leaf appearance and of floret development. Some of these Eps-7D × Ppd-D1 interactions were evident at all temperatures considered (e.g., phenology), while others were clear only at particular temperatures (e.g., floret development and the resulting effects of Eps-7D on spike fertility). Interactions between Eps and Ppd genes were identified before [12,16] and could well be a major reason for the well acknowledged Eps × genetic background interaction [31,32]. This interaction is relevant as it means that one or other Eps gene could potentially be exploited in lines with contrasting sensitivity to the photoperiod. In this particular case, to exploit the effects of this new Eps-7D gene, it should be introgressed in lines that are highly sensitive to the photoperiod.
The fact that the Eps-7D × Ppd-D1 interactions would be clearer at some temperatures as opposed to others is also something that could have been expected as both Eps-7D (Basavaraddi et al., submitted), as well as other Eps [6,8,13,14,33,34], and Ppd genes [11,35,36] have shown to interact with temperature.
In the background of Ppd-D1b, in which the effects of Eps-7D were far more relevant, it was clear that this Eps gene affected phenology mainly during the LRP. This is relevant as it is during this phase that spike fertility is determined [21,[37][38][39][40][41][42]. Knowledge of genetic factors affecting particular phases of wheat phenology would help in tailoring not only time to anthesis but also the partitioning of this time into particular phases.
The effect of Eps-7D on FLN was subtle even under sensitive Ppd-D1 background where the Eps-7D-late significantly delayed phenology. This subtle effect is commensurate with the fact that Eps-7D-late mainly affected development during later stages, and in agreement with other Eps genes mainly affecting reproductive development [13,23]. Indeed, the effect of Eps-7D-late on the duration of the LRP (encompassing time to flag leaf emergence) was the consequence of this allele decreasing the rate at which leaves appeared, lengthening the phyllochron and consequently delaying the appearance of the flag leaf.
The lines carrying the Eps-7D-late tended to increase spike fertility consistently compared to those with Eps-7D-early, although the effect was important only when the background was photoperiod-sensitive and when the temperature was intermediate. This slight improvement in fertility at individual spikelet level was a result of extending the floret development period, which allowed labile florets (F2 and F3 in this case, depending on the particular spikelets considered) to reach the stage of fertile floret. This means that the mechanism by which the Eps-7D-late allele improved spike fertility was in line with what had been seen in other studies in the literature ( [14,15] and references quoted therein).
This Eps-7D interacted with temperature (Basavaraddi et al., submitted) and Eps-2B [15], and the interaction effect have been shown to be effective well beyond the duration of developmental phases. Our findings suggest that it is not only important to study every newly identified Eps gene (simply because although they are all named equally, the mechanisms by which they actually modify time to anthesis may be quite different), but it is also necessary to evaluate their functions under varying backgrounds of other important genetic factors affecting wheat development to ensure repeatability of their effect and their application. The Eps QTL studied here brought about subtle changes in time to anthesis, similar to most other Eps studies before [3,14,23,43], which naturally had a weak consequential effect on spike fertility, although under particular circumstances (Ppd-D1 in the background and temperature) these effects were relevant. This is important to be considered when particular Eps genes are to be exploited in wheat breeding.
As detailed in the companion paper (Basavaraddi et al., submitted), we determined not only the timing of the important growth stages (seedling emergence (DC10), onset of stem elongation (DC30), flag leaf, heading (DC39), and anthesis (DC65); following the Zadoks' scale [44]), but also periodically (i) the number of leaves appearing on the main shoot was recorded following the Haun scale [45], and with that, we described the pattern of leaf appearance, and (ii) the apex stages and number of primordia initiated-firstly analysing the apex as a whole, and secondly, after the TS stage we dissected the individual spikelets and determined the number and stage of floret primordia (following the apex scales of Kirby and Appleyard, [46] and floret score proposed by Waddington et al., [47]).
Finally, at anthesis, we counted the number of fertile florets at each spikelet of the spikes (the first spikelet being the most basal and the last being the terminal spikelet) to record the total number of fertile florets per spike.
Data were analysed considering a full factorial model using JMP Pro version 14.0 (SAS Institute Inc., Cary, NC, USA) through a two-way ANOVA; in addition, one-way ANOVA was used to determine the significance of differences between pairs of NILs carrying contrasting alleles of Eps-7D at each combination of Ppd-D1 allele in the background and temperature.