Introduction
Climate change will increase mean global temperature, but will also affect future rainfall patterns, with more intense and more frequent extreme precipitation events (Intergovernmental Panel on Climate Change, Reference Pachauri and Meyer2014). Indeed, during the 20th century precipitation increased by some 10–40% across northern Europe (Klein Tank et al., Reference Klein Tank, Wijngaard, Können, Böhm, Demarée, Gocheva, Mileta, Pashiardis, Hejkrlik, Kern-Hansen, Heino, Bessemoulin, Müller-Westermeier, Tzanakou, Szalai, Pálsdóttir, Fitzgerald, Rubin, Capaldo, Maugeri, Leitass, Bukantis, Aberfeld, Van Engelen, Forland, Mietus, Coelho, Mares, Razuvaev, Nieplova, Cegnar, López, Dahlström, Moberg, Kirchhofer, Ceylan, Pachaljuk, Alexander and Petrovic2002). Seed quality is influenced by environment during seed development and maturation (Gusta et al., Reference Gusta, Johnson, Nesbitt and Kirkland2004; Hampton et al., Reference Hampton, Boelt, Rolston and Chastain2013). Rainfall and temperature are the main environmental factors affecting crop yield and quality of wheat (Triticum aestivum L.) in the periods preceding harvest (Smith and Gooding, Reference Smith and Gooding1999; Landau et al., Reference Landau, Mitchell, Barnett, Colls, Craigon and Payne2000; Yadav and Ellis, Reference Yadav and Ellis2016).
Warmer temperatures in the UK improve wheat seed quality development (Sanhewe et al., Reference Sanhewe, Ellis, Hong, Wheeler, Batts, Hadley and Morison1996), whereas rainfall incident upon developing seed crops in cool, wet summers has long been reported to be deleterious to UK wheat seed quality (MacKay, Reference MacKay and Roberts1972). At the extreme, exposure of wheat varieties with low dormancy to heavy or prolonged rainfall results in visible sprouting (Flintham, Reference Flintham2000) and thus poor seed quality.
Simulated rainfall to developing wheat seed crops in the UK reduced the subsequent air-dry longevity of seeds harvested soon afterwards (Ellis and Yadav, Reference Ellis and Yadav2016). That study showed that damage to longevity could be detected from a single rainfall event, insufficient to promote sprouting, but also that this immediate damage was reversed in planta when harvest was delayed and seeds re-dried.
Post-harvest wetting of mature seed can benefit seed survival (Villiers and Edgecumbe, Reference Villiers and Edgecumbe1975). Seed hardening (a wetting and drying cycle) has long been applied by farmers in advance of sowing, but the effects of such wetting treatments (e.g. priming) have been reported to both reduce (e.g. Argerich et al., Reference Argerich, Bradford and Tarquis1989; Tarquis and Bradford, Reference Tarquis and Bradford1992) and improve longevity (e.g. Georghiou et al., Reference Georghiou, Thanos and Passam1987; Probert et al., Reference Probert, Bogh, Smith and Wechsberg1991). Priming post-harvest has also been shown to improve subsequent longevity when applied to immature seed, but to reduce longevity in seeds closer to maturity (Demir and Ellis, Reference Demir and Ellis1992a). Hence the effect of rainfall on longevity may vary with the stage of seed development.
We report here the effect of either rain shelter or additional simulated rainfall on seed quality, assessed by ability to germinate normally with or without desiccation and by longevity in air-dry storage, on seed crops of wheat in two years. The following null hypotheses were tested: (1) no effect of rainfall on subsequent seed quality assessed by both ability to germinate normally and by subsequent air-dry seed storage longevity, where the latter is the more sensitive to discriminate amongst high-quality seed lots in seed development studies (Pieta-Filho and Ellis, Reference Pieta-Filho and Ellis1991); and (2) no interaction with developmental stage. Accordingly, treatments were applied at different stages of, and for different durations during, seed development and maturation. Wheat seed quality improves throughout seed development and maturation, reaching maximum values at or approaching harvest maturity (Ellis and Pieta Filho, Reference Ellis and Pieta-Filho1992), where harvest maturity is typically 15% seed moisture content (but can vary from 12 to 20%, depending upon ambient relative humidity). Hence, the response to treatments was assessed against those developmental changes within control crops.
Materials and methods
Plant material
Crops of wheat (T. aestivum L.) cv. Tybalt were grown from March to August 2011, and similarly in 2012, in the field at the University of Reading, Crop Research Unit (CRU), Sonning (51°30′N, 00°54′W), in split-plot designs. The main plots (each 7.5 m × 2.5 m) were the open, sheltered (protection from rainfall), or wetted (additional simulated rainfall) treatments allocated at random within each block, with serial seed harvest dates as subplots. Full details of agronomy, weather, rain shelter construction, dynamics of seed filling and desiccation, and resultant crop quality are provided by Yadav and Ellis (Reference Yadav and Ellis2016). Rain shelter was provided by polythene covers (2 m × 5 m, 1.5 m tall) – twice the height of the crop and open at the sides to aid air circulation and reduce the possibility of temperature build-up above the crop canopy. Simulated rainfall treatments were provided to an area of 1.5 m × 4 m marked by bamboo canes at the centre of those plots. The developing ears were wetted with (tap) water equivalent to 2.5 cm rainfall on each occasion, in two halves with a 30-min gap, as described by Ellis and Yadav (Reference Ellis and Yadav2016).
The 2011 investigation comprised 22 main plots (11 treatments in each of two blocks). One treatment was untreated and always open to rainfall (C1O; open control) and one covered by full rain shelter from 50% anthesis until the last harvest (C2S; sheltered control). Four treatments were provided with rain shelter for 14-d periods at 0–14 d after 50% anthesis (DAA) (S1), 14–28 DAA (S2), 28–42 DAA (S3) or 42–56 DAA (S4). A further five treatments were provided with simulated rain: four were ear-wetting treatments at 7 DAA (W1), 21 DAA (W2), 35 DAA (W3) or 49 DAA (W4); the fifth was wetted four times, i.e. at 7, 21, 35 and 49 DAA (W5; wetting control). Samples from treatments subjected to ear wetting that day were drawn 30 min after treatments ended.
Samples of seeds from each treatment were harvested serially by cutting 100 ears from about 0.5 m2 with scissors at about 1–2 cm below the spikes on up to nine occasions (depending upon when a treatment began) during seed development and maturation, from 14 DAA on 23 June and then 25, 32, 39, 46, 53, 60, 67 and finally 74 DAA. The first two harvests (14 and 25 DAA) comprised six treatments, those on 32 and 39 DAA nine, and thereafter all 11 treatments from each block.
In 2012, there were 20 main plots (ten treatments in each of two blocks). One treatment was always open to rainfall (C1O; open control) and one covered by full rain shelter from 50% anthesis until the last harvest (C2S; sheltered control). The three temporary rain-shelter treatments were applied later in development than in 2011 and also differed in duration: S1 (42–70 DAA), S2 (42–56 DAA) and S3 (56–70 DAA). The five ear-wetting treatments were also applied later than in 2011: at 42 DAA (W1), 49 DAA (W2), 56 DAA (W3), 63 DAA (W4) or 70 DAA (W5). Samples of seeds from each treatment were harvested serially on up to ten occasions during seed development and maturation: the first four harvests (14, 21, 28 and 35 DAA) comprised two treatments, those on 42, 49, 56, 63 and 70 DAA 5, 6, 8, 9 and 10 treatments, respectively, and an additional sample was drawn 30 min after wetting treatments ended from each block. In the case of W5, the last wetting treatment, a sample was also taken at 77 DAA in order to provide a sample 7 d after wetting for this treatment also. In 2012, samples were taken shortly before each wetting as well as 30 min after treatments ended. The results for ability to germinate and subsequent longevity for samples harvested 30 min after wetting treatments ended are designated W (af) to distinguish these treatments’ immediate effect from results immediately before wetting.
Samples were harvested between 7 a.m. and 9 a.m. and seeds threshed from ears by hand. Throughout the study, seeds from the same treatment but from different blocks were analysed separately. With the exception of samples drawn to determine the moisture content [results reported previously by Yadav and Ellis (Reference Yadav and Ellis2016)] and ability to germinate of freshly harvested seed, samples were dried to 10–14% moisture content (wet basis) in a drying cabinet maintained at 15–17°C with 12–15% relative humidity. Decline in seed sample weight was monitored as a proxy for moisture content during drying. Drying period varied between extremes of 2 and 21 d amongst samples, depending upon stage of development and hence initial seed moisture content. Samples were then drawn to determine seed moisture content and ability to germinate after drying. The remainder of each sample was sealed in a laminated-aluminium-foil bag (Retort laminate, Moore and Buckle Ltd, St Helens, UK) and stored temporarily at −20°C.
Germination test
Ability to germinate was tested for two replicates of 50 seeds each between moist, rolled paper towels (Kimberley Clark Professional 6803 HOSTESS, Natural, 24 × 35 cm, Greenham Sales, UK) in an incubator at 10°C. Tests were monitored for germination at 7-d intervals up to 28 d. Seedlings were evaluated as normal, abnormal or dead seeds according to the ISTA rules (International Seed Testing Association, 2011). Seeds that had not germinated after 28 d were pricked in order to break any possible dormancy and tests continued until all seeds had either germinated or were no longer fresh (i.e. were dead).
Seed moisture content
Seed moisture content was determined using the two-stage or the single-stage high-constant-temperature-oven method (International Seed Testing Association, 2011), depending upon expected moisture content; two 100-seed replicates were used in place of two 4- to 5-g samples, due to limited seed supply.
Seed storage
Seed storage longevity was determined in a constant hermetic environment of 40°C with c. 15% moisture content. Seed packets were withdrawn from storage at −20°C about 4 months after harvest. They were first exposed to laboratory temperature within the sealed packets for 24 h to avoid condensation of moisture on seed surfaces. Seed moisture content was then estimated indirectly using a non-destructive equilibrium relative humidity (erh) water activity meter (65% relative humidity at 20°C being the common target for all the seed samples). In order to adjust seed moisture content to 15% (± 0.5%), samples were weighed and either placed in a muslin bag to dry at 15°C, 15% relative humidity, or humidified at 20°C above deionized water, depending upon estimated initial moisture content. This adjustment was controlled by weighing repeatedly and determining erh until the desired weight/erh was reached. To enable moisture to equilibrate within and amongst samples, the separate muslin bags were placed together for 15 d in a sealed container at 2–4°C.
The moisture content of each seed sample was then determined using the high-constant-temperature-oven method (International Seed Testing Association, 2011). Estimates ranged between extremes of 14.1 and 15.7% in 2011 (mean 15.0%) and 14.2 and 15.6% in 2012 (mean 14.7%). Ten subsamples of 100 seeds from each sample were sealed in separate laminated-aluminium-foil packets and stored in an incubator maintained at 40°C. One sample from each treatment was withdrawn from storage after different periods (0–49 d) of experimental storage and tested for ability to germinate. Seed survival curves in hermetic storage at 40°C with 15% moisture content were fitted by probit analysis for each sample, in accordance with the seed viability equation (Ellis and Roberts, Reference Ellis and Roberts1980):
$$v = K_i - p/\sigma $$
where v is probit percentage viability after p days in storage in a constant environment, K i is a constant specific to seed lot (equivalent to initial probit viability) and σ is the standard deviation of the frequency distribution of seed deaths in time (d) using Genstat (13th edition, VSN International Ltd, Hemel Hempstead, UK). The product of K i and σ is the period for viability to decline to 50% (p 50). The ability of a seed to produce a normal seedling was the criterion of survival. Genstat was also applied to compare seed-survival curves.
Results
Fifty per cent anthesis occurred on 8 June 2011 (84 d after sowing) or 18 June 2012 (95 d after sowing), when the first rain shelters were installed (e.g. C2S). Sheltered plots provided mean temperatures 0.8–0.9°C warmer than control plots in both years, with 90.2% direct light and 84.6% photosynthetically active radiation. Both years provided post-anthesis environments that were wetter and cooler than long-term site means, with 2011 cooler and wetter than 2012 on average.
The treatments affected seed-filling duration in 2011, only, and period to harvest maturity in both years (Table 1). In 2011, seed-filling durations were reduced by rain shelter throughout (C2S) or late within this period (S3). Durations from anthesis to (almost) harvest maturity were reduced by shelter throughout (C2S in 2011 and 2012) or during late maturation drying (S1, S3 in 2012) but were only increased by multiple simulated rainfall events (W5 in 2011). Seed moisture contents were 2–5% greater 30 min after ear wetting, the only exception being a 10% increase in W5, 2011 at 21 DAA, with control values regained in subsequent harvests (Yadav and Ellis, Reference Yadav and Ellis2016).
Effect of rain shelter or ear wetting on durations to mass maturity and 20% moisture content (i.e. approaching harvest maturity) in wheat cv. Tybalt in 2011 and 2012 (from Yadav and Ellis, Reference Yadav and Ellis2016)
1Open control: neither rain shelter nor ear wetting provided.
Ability to germinate
No pre-harvest sprouting was detected in any treatment. Ability to germinate normally improved greatly amongst early harvests, when it also benefited from post-harvest drying (Figs 1, 2). From 32 to 74 DAA in 2011, all treatments provided 100% normal germination for dried seed samples (Fig. 1c, d). Ability to germinate without post-harvest drying was slightly lower during this period, but also more variable; a marked decline (P < 0.001) occurred from 39 to 53 DAA, with subsequent reversal from 53 to 67 DAA in 2011 (Fig. 1a, b). This contrasted with 100% normal germination throughout for dried seeds (Fig. 1c, d). Ability to germinate normally was also greater with post-harvest drying in 2012 from 35 to 70 DAA (Fig. 2), but in this case slightly more variation was apparent amongst treatments, with lower values during S3 shelter (56–70 DAA; Fig. 2c) and 30 min after ear wetting ended in W2 (49 DAA) and W3 (56 DAA) (Fig. 2d).
Ability of fresh (a, b) or dried (c, d) seeds of wheat cv. Tybalt to germinate normally when tested at 10°C after harvesting at different times during seed development when grown (a, c) under rain shelter (C2S, S1, S2, S3, S4) or (b, d) with ear wetting (W1, W2, W3, W4, W5) compared with a control (C1O; no shelter or ear wetting) in 2011. C2S is the equivalent shelter control, with rain shelter provided from 50% anthesis until the last harvest.
Ability of fresh (a, b) or dried (c, d) seeds of wheat cv. Tybalt to germinate normally when tested at 10°C after harvesting at different times during seed development when grown (a, c) under rain shelter (C2S, S1, S2, S3) or (b, d) with ear wetting [W1, W2, W3, W4, W5, W (af)] compared with a control (C1O; no shelter or ear wetting) in 2012. C2S is the equivalent shelter control, with rain shelter provided from 50% anthesis until the last harvest. Note that W (af) is not a separate treatment: it represents the results of samples harvested 30 min after ear wetting, once, on the day shown (to distinguish it from that treatment's harvest just before wetting that day).
Longevity
The seed-survival curves were sigmoidal, conformed to negative cumulative normal distributions and were described well by equation (1). In each year, seed-survival curve comparisons showed significant differences in K i (P < 0.001) and σ (P < 0.001). The major differences in longevity (p 50) resulted from differences in estimates of K i . There was a clear main effect of harvest date on longevity (P < 0.001). This main effect provided a somewhat erratic pattern of considerable improvement in longevity until 53 DAA, but with a plateau between 32 and 46 DAA in 2011 (Fig. 3) or 56 DAA in 2012 (Fig. 4). A further difference amongst years occurred after maximum longevity was first attained: longevity declined after 56 DAA in 2012, but was stable between 53 and 74 DAA in 2011.
Improvement in subsequent air-dry seed storage longevity (p 50, days, provided by probit analysis), in hermetic storage at 40°C with c. 15% moisture content, during seed development in wheat cv. Tybalt in 2011, comparing amongst rain shelter (a) or ear-wetting treatments (b). Standard errors of estimates are shown (but are largely similar to or smaller than symbols). Control treatment results (C1O) are shown in both (a) and (b). Further details as for Fig. 1.
Improvement, and later decline, in subsequent air-dry seed storage longevity (p 50, days, provided by probit analysis), in hermetic storage at 40°C with c. 15% moisture content, during seed development in wheat cv. Tybalt in 2012, comparing amongst rain shelter (a) or ear-wetting treatments (b). Standard errors of estimates are shown (but are largely similar to or smaller than symbols). Control treatment results (C1O) are shown in both (a) and (b). The negative values for p 50 in the first two harvests are because initial viability was <50% (see Fig. 2). Further details as for Fig. 2.
The treatments had comparatively little effect on the development of longevity in 2011, other than that shelter throughout (C2S) may have been in advance of the control (C1O) early on, with close agreement amongst treatments at 53 DAA when maximum longevity was first attained (Fig. 3a, b). It is also noteworthy that longevity 21 d later at 74 DAA after all five wetting treatments (W1–W5) was not only slightly greater than that of the control (C1O) but also similar to that of shelter throughout (C2S).
Longevity for shelter throughout (C2S) improved in advance of the control (C1O) at 28 and 35 DAA in 2012 (Fig. 4a). Moreover, from 56 to 70 DAA longevity was greater from all shelter treatments than the control. Wetting reduced longevity immediately, as shown by comparisons between W (af) and C1O (Fig. 4b). However, W (af) is not one treatment per se but a compilation of the results of all wetting treatments 30 min after the treatments ended; after a further 7 or more days in planta all wetting treatments completed by then provided similar longevity at 56 DAA to the control, and only two subsequent wetting treatment observations were below control values (W2 at 63 DAA; W3 at 70 DAA).
To investigate the effect of the timing of ear wetting more closely, Fig. 5 provides a compilation of the 2012 results for longevity immediately before and 30 min after ear-wetting treatments ended, and after a further 7 d in planta for treatments applied up to 28 d apart late in maturation. Subsequent air-dry longevity was reduced 30 min after wetting in each of the five wetting treatments, to a similar extent in W2–W5 but to about half this extent in the earliest treatment, W1 (42 DAA); but 7 d in planta thereafter resulted in improvement to longevity in every treatment. These reductions and subsequent improvements in p 50 resulted from changes in the estimates of K i and, to a lesser extent, σ. For each ear-wetting treatment, the difference between the first and second samples (vertical broken line in Fig. 5) provides an estimate of the immediate (gross) damage to seed quality (as assessed by longevity) from simulated rainfall, while that between the second and third samples (solid line in Fig. 5) provides an estimate of the reversal of that damage over 7 d. The difference between the first and third samples provides an estimate of net damage from simulated rainfall, therefore. The magnitude of the initial damage to longevity from simulated rainfall and the subsequent improvement (difference between second and third samples), in (a) absolute terms and (b) relative to the gross (immediate) damage from wetting, showed clear patterns over developmental time amongst treatments about a fulcrum at 56 DAA (when longevity reached maximum values, Fig. 4). Gross damage showed a consistent pattern where W1 < W2 < W3 = W4 > W5; for the subsequent improvement in absolute terms, least was shown in W1 and W5, greatest in W2, and about a quarter of W2 in both W3 and W4. In relative terms, the improvement was similar to initial (gross) damage in W1, more than twice as great in W2, but just under half of the gross damage in W3–W5.
Damage from ear wetting to subsequent seed storage longevity (p 50) of wheat cv. Tybalt (dashed lines) 30 min after wetting, and subsequent improvement with natural re-drying for 7 d in planta (solid lines) from treatments at different times during seed development: W1 (closed squares), W2 (open squares), W3 (closed triangles), W4 (open triangles),W5 (closed diamonds). Data repeated from Fig. 4.
Discussion
The general temporal pattern of development in the ability of seeds to germinate during seed development and maturation was similar in the two years, and largely as expected in wheat (Ellis and Pieta Filho, Reference Ellis and Pieta-Filho1992). Onset of ability to germinate and of desiccation tolerance was apparent from the first harvest (14 DAA) early in the seed-filling phase (Figs 1, 2), at which time seeds were only about 10% filled (Ellis and Yadav, Reference Ellis and Yadav2016), with full or close-to-full ability to germinate achieved 70–80% through the seed-filling phase – and so, well before mass maturity [end of the seed-filling phase (Ellis and Pieta-Filho, Reference Ellis and Pieta-Filho1992)]. Post-harvest drying early in seed development promoted the ability to germinate. Such improvements from drying immature seeds are well known (Dasgupta et al., Reference Dasgupta, Bewley and Yeung1982; Bewley and Black, Reference Bewley and Black1994). The effect was equivalent to several days of further development in planta, which drying ex planta mimics to a certain extent. No decline in ability of the dried seeds to germinate normally was detected amongst harvests later in development and maturation in either year, despite sampling until 74 DAA (extremes of 32 d after mass maturity, or 13 d after seeds dried to 20% moisture content; Table 1).
Testing the ability to germinate at 10°C, combined with pricking, minimized the limitation of germination by dormancy, as expected in temperate cereals (Ellis et al., Reference Ellis, Hong and Roberts1987). Nevertheless, the results for freshly harvested seeds, harvested between 39 and 67 DAA in 2011, showed first a decline and then an increase in ability to germinate, whereas seeds first dried showed 100% germination throughout this period (Fig. 1). This represents late induction and subsequent loss of wheat seed dormancy (Mitchell et al., Reference Mitchell, Armstrong, Black, Chapman and Hebblethwaite1980; Gooding et al., Reference Gooding, Uppal, Addisu, Harris, Uauy, Simmonds and Murdoch2012). In sorghum (Sorghum bicolor L.), Fenner (Reference Fenner1991) noted that plant water status during seed filling affected dormancy. No consistent effect on dormancy induction and loss pattern was detected here, however, with the extreme opposite treatments of shelter (S1, S2) or wetting (W1, W2, W3) during seed filling both providing greater dormancy than the control (Fig. 1).
Seed longevity can be a more sensitive indicator of differences in seed quality than ability to germinate amongst high-quality seed lots (Ellis and Roberts, Reference Ellis and Roberts1981). This was the case here, with substantial differences in seed longevity identified (Figs 3 and 4) amongst the high-viability samples (Figs 1, 2) harvested from 25 (2011) or 35 DAA (2012) onwards. Prior to this, both approaches provided a consistent increase in seed quality but, whereas ability to germinate then reached maximum values, longevity showed continued improvement thereafter, albeit not linear, with further development. Subsequent longevity first reached maximum values during the maturation drying phase at 53–56 DAA (Figs 3 and 4), some 5–11 (2011) or 8–14 d (2012) after mass maturity and 9–23 (2011) or 1–18 d (2012) before seed moisture content had declined in planta to 20% (Table 1). This confirms earlier studies in barley (Hordeum vulgare L.) and wheat, in which subsequent longevity continued to improve, and considerably so, beyond the end of the seed-filling phase (Pieta Filho and Ellis, Reference Pieta-Filho and Ellis1991; Ellis and Pieta Filho, Reference Ellis and Pieta-Filho1992; Sanhewe et al., Reference Sanhewe, Ellis, Hong, Wheeler, Batts, Hadley and Morison1996; Ellis and Yadav, Reference Ellis and Yadav2016).
Longevity remained high, possibly improving a little, from 53 DAA until the last harvest at 74 DAA in 2011 (Fig. 3) but declined, more or less linearly and consistently, between 56 and 70 DAA in 2012 (Fig. 4). A somewhat similar difference between years was reported in a study with barley: longevity declined immediately after improvement ended in 1988, whereas in the warmer, drier 1989, longevity was maintained at high values for around 14 d, before then declining (Pieta Filho and Ellis, Reference Pieta-Filho and Ellis1991). Decline in longevity after attaining maximum values was detected in wheat in both 2008 and 2009, and was steeper in the wetter and marginally (only 0.1°C) cooler 2008 than in 2009 (Ellis and Yadav, Reference Ellis and Yadav2016). However, the potential assumption that warmer, drier conditions reduce decline in longevity with delay to harvest is not borne out by the current results, because August 2011 was appreciably wetter and cooler than August 2012 (Yadav and Ellis, Reference Yadav and Ellis2016). Hence, the assumption that the wetter the environment, the more rapidly seeds deteriorate post peak-longevity is not supported consistently by these inter-annual comparisons.
A similar lack of clarity for the effect of rainfall is provided by comparison within years between shelter and ear-wetting treatments amongst later harvests. In 2011, longevity in the ear-wetting treatments (at 7–49 DAA) from 54 DAA onwards was similar to, or perhaps greater than, that of the control (Fig. 3b), but this was also the case for the rain shelter treatments (Fig. 3a). In 2012, the later harvests (56 DAA onwards) provided consistently greater longevity with shelter than the control (Fig. 4a) and while longevity 30 min after wetting (at 42–70 DAA, and so applied later than 2011) was reduced, thereafter most wetting treatments provided longevity similar to that of the control (Fig. 4b). Differences in longevity amongst treatments were slightly greater in 2012 (Fig. 4) than in 2011 (Fig. 3), perhaps because the majority of treatments were applied later, deliberately, during seed development and maturation in 2012. The greater effect of treatments applied later tallies with differences detected amongst previous years (Ellis and Yadav, Reference Ellis and Yadav2016), but the comparison was confounded with that between field and protected environments.
Generally, therefore, it would appear that rainfall events result in gross damage to subsequent longevity at all stages of development and maturation, but that the damage is reversed in planta in whole or in part thereafter, such that there is often little or no net damage. However, net damage may be detected from rainfall in the period approaching harvest maturity and thereafter (W3–W5, Fig. 5), especially if rainfall is repeated (Ellis and Yadav, Reference Ellis and Yadav2016). Moreover, there is a clear trend with respect to developmental stage: simulated rainfall early–mid development is less damaging initially and is more likely to be entirely reversed thereafter in planta, than that from late maturation drying onwards. Hence, net damage to seed quality from rainfall is more likely in the period approaching harvest maturity, and thereafter, than earlier in seed development. The comparatively limited net damage from ear wetting may help to explain the absence of any consistent effect of rain shelter versus wetting and control treatments amongst the final samples harvested between the years. It also agrees with 2008 and 2009 field results in which simulated rainfall early in development delayed the pattern of seed quality improvement but only by a few days, and the more dramatic damage, but largely reversible, from treatment close to harvest maturity in a tunnel house in 2010 (Ellis and Yadav, Reference Ellis and Yadav2016).
We suggested previously that damage to longevity from rainfall at harvest maturity, and subsequent reversal in planta, might be associated with changes to the glassy state – which may affect air-dry seed survival (Buitink and Leprince, Reference Buitink and Leprince2004; Walters, Reference Walters2015), because seeds were at 14–19% moisture content, the gross damage occurred rapidly and subsequent recovery was within 24 h (Ellis and Yadav, Reference Ellis and Yadav2016). While we continue to support that argument for rainfall events close to harvest maturity in very dry seeds, the damage and reversal detected here over a longer period of 7 d and, moreover, earlier in seed development and maturation (W1–W2, Fig. 5) coincides with periods when both oligosaccharides and low molecular weight heat-stable proteins are accumulating [albeit with oligosaccharide accumulation more prevalent during development and heat-stable protein accumulation more so during maturation drying (Sinniah et al., Reference Sinniah, Ellis and John1998)], and both benefit subsequent air-dry seed survival (e.g. Crowe et al., Reference Crowe, Crowe and Chapman1984; Galau et al., Reference Galau, Hughes and Dure1986; Leopold, Reference Leopold, Alscher and Cumming1990). A recent network analysis of the co-expression of a considerable number of regulatory genes associated with seed longevity, during seed development and maturation in Medicago truncatula L., confirms there is a substantial temporal pattern of expression of different genes, but that gene expression can be detected quite late when moisture content is only c. 20% (Righetti et al., Reference Righetti, Vu, Pelletier, Vu, Glaab, Lalanne, Pasha, Patel, Provart, Verdier, Leprince and Buitink2015). Hence, it is possible that the drivers of both the damage to longevity from rainfall and its subsequent reversal may differ depending upon when during seed development and maturation the rainfall event occurs. If so, that might explain the differences in the magnitude of the gross and net responses at different stages of seed development.
This research, particularly Fig. 5 here in combination with Fig. 4 of Ellis and Yadav (Reference Ellis and Yadav2016), confirms a long-held suspicion: during the period of seed quality development from the beginning of seed filling until close to harvest maturity, the improvement in seed quality detected is net improvement whereby both improvement and deterioration occur, but the former is (usually) by far the greater. Similarly, once maximum seed quality has been attained, it is net loss in quality that is detected with, again, improvement and deterioration possible, but in this case the latter is (usually) the greater. Hence, the difference between years during late maturation drying of no appreciable change in seed quality between 53 and 74 DAA in 2011 (Fig. 3) but consistent decline from 56 to 70 DAA in 2012 (Fig. 4): in the latter, deterioration was greater than improvement, whereas in the former they were almost equal, presumably, cancelling each other out. Certain earlier observations in rice (Oryza sativa L.) and tomato (Lycopersicon esculentum Mill.) may also be compatible with this interpretation. Longevity of a japonica rice at 32/24°C reached a plateau from mass maturity onwards during the maturation drying phase, whereas longevity continued to improve over this period at 28/20°C and in other (indica) cultivars in both regimes (Fig. 4a, Ellis et al., Reference Ellis, Hong and Jackson1993). We suggest that the former represents no net change, because deterioration and improvement were in balance, whereas in the latter, improvement exceeded deterioration. Similarly, tomato seed survival periods were stable over a 40-d period in planta, within fleshy fruits, from 55 DAA, 13 d after mass maturity, to 95 DAA (Fig. 3c, Demir and Ellis, Reference Demir and Ellis1992b) – again representing a possible close match between rates of seed deterioration and improvement.
Greater tolerance to heat and drought stress are priorities for the genetic improvement of wheat, in order to maintain future yield by adapting to climate change (Semenov et al., Reference Semenov, Stratonovitch, Alghabari and Gooding2014). In the UK at least, warmer summer temperatures from anticipated climate change will benefit wheat seed quality (Sanhewe et al., Reference Sanhewe, Ellis, Hong, Wheeler, Batts, Hadley and Morison1996). Wheat has long been reported to produce poor-quality seeds in cool, wet UK summers (MacKay, Reference MacKay and Roberts1972) and, similarly, poor-quality grain for bread-making purposes in wet UK summers (Smith and Gooding, Reference Smith and Gooding1999) as a result of germination (visible and so pre-harvest sprouting, or sensu strictu). The wheat cultivar selected for this study is of bread-making quality and shows strong dormancy, in order to maintain grain quality in wet UK summers (Yadav and Ellis, Reference Yadav and Ellis2016). Clearly, such strong dormancy also provides resilience in terms of high seed quality following rainfall events in wet summers. Hence, strong seed dormancy is an important character to include in breeding programmes to provide new feed, as well as quality, cultivars adapted to future north-west European climates, in order to safeguard high-quality wheat seed supplies.
Acknowledgements
We thank Caroline Hadley, Richard Casebow and Dr Tran Dang-Hong for their considerable help with this study.
Financial support
The Commonwealth Scholarship Commission provided a research studentship to G.Y. (grant code INCS-2010-170).
Conflicts of interest
None.
