In summary, all three variables—planting time, maturity class, and strategic irrigation—significantly influenced rice grain yield in the subtropical summer of central Queensland, Australia. Late planting exposed especially the late maturity varieties (i.e., japonica) to unfavourable low temperatures around flowering, and in general, early (i.e., indica) varieties performed better in early and late plantings. Strategic irrigation favoured late varieties if planted early or early varieties whether planted early or late. Final yields were influenced by specific effects of each of the three variables—planting time, maturity class, and strategic irrigation—on phenology, light capture and use in photosynthesis, rooting patterns, spikelet fertility, and other yield determining attributes, all of which also affected water productivity. These effects are discussed in detail below.
4.1. Influence of Planting Time, Maturity Class, and Strategic Irrigation and Phenology on Yields
Firstly, the late planting (i.e., after mid-January) in 2014 resulted in late varieties being subjected to low temperature at flowering and, under the rainfed treatment, to low water availability and to the ensuing exposure to terminal drought, which was more evident in late than early varieties. The similarity between drought- and cold-induced pollen sterility in rice has been reported earlier [27
], both stresses disturbing sugar metabolism and tapetum function in anthers. The lower soil moisture during flowering and grain filling of rainfed later varieties also hampers panicle exertion, spikelet fertility, and grain filling [37
]. However, yields for the later varieties were low even with strategic irrigation, which minimised the effect of terminal drought, but had no effect on the low-temperature induced sterility. In contrast, early varieties avoided the debilitating effects of low temperature around flowering and, to a lesser extent, low soil moisture during grain filling, hence yields exceeded those of late cultivars, and more so with strategic irrigation.
Others, such as [26
], have reported a similar yield penalty with exposure to low temperatures with late planting of sprinkler irrigated rice under south-eastern Queensland conditions (latitude 27.5 °S). Indeed, when rice is exposed to low temperatures prior to and at flowering, the fertility of panicles is reduced, as is grain set per panicle [38
]. Microspore and pollen grain formation, which are negatively affected by temperatures below 18 °C [39
], take place between 10 and 12 days before anthesis [40
] or 5 and 9 days before heading, and even short periods of low temperature (i.e., 15–18 °C for 8–10 d, or 12–13 °C for 3–5 d) at that time can negatively affect fertility because of reduced total pollen production. In our study, the late planting of 2014 exposed late varieties to low night temperatures not before, but around, heading (Figure 1
) (below 15 °C for 7 days from 101 DAS to 107 DAS and below 10 °C for 4 days and below 5 °C for 1 day within the same period). This resulted in less grain set with lower fertility percentage (18-35%, Figure 5
) and subsequent loss of yield potential compared with 2015 when temperatures around heading did not drop below 20 °C (Figure 1
) and percentage fertility was c. 50%. This loss of yield potential in 2014 accords well with the data presented by [41
] for irrigated dryland rice in NW Uzbekistan and with [40
], who reported that brief low temperature exposure at anthesis was almost as damaging as the same low temperature during microsporogenesis. An analysis of the likelihood of a two-day period of temperatures less than 13 °C in central Queensland, as occurred in April of 2014, with negative effects on spikelet fertility, shows that, over the past 35 years, such an occurrence in April is likely in 20% of the years, whereas in May, almost all years recorded such a low temperature for two consecutive days. Temperatures of 17 °C or less for five consecutive days occurred in c.
30% of years in April, and in all 35 years in May, illustrating the risks of low temperate impacts on fertility in late plantings. Hence the imperative of planting before the solstice to avoid the risks of low temperature around flowering. However, assuming 50 mm rainfall is necessary to allow for germination and 10 days’ growth [42
], successful sowing and crop establishment without strategic irrigation would only be possible in October in 22% of years over the period 1985–2019, in 26% in November, and in 45% in December (and 25% in January). However, only for sowings in December in 1990, November and December in 2010, and January in 2013 (i.e., four sowing in 35 years) were follow-up rains of 750 mm achieved (approximately that received with rainfall plus irrigation in both years of experiments and herein considered necessary for a rainfed crop), thus emphasising the need for strategic irrigation.
Secondly, and in contrast to the late-planted 2014 crop, rice varieties during the earlier-planted 2015 crop were not exposed to low temperature before, during, or beyond heading (Figure 1
). Nevertheless, in the rainfed treatment, all varieties were exposed to lesser water availability and yields of rainfed varieties were on average only 33% of those with strategic irrigation (Table 1
). When comparing late and early flowering varieties, in the absence of a late cold stress and with strategic irrigation, their yields were more similar (4.56 t/ha for early and 3.93 t/ha for late varieties). However, grain yields of rainfed late varieties were considerably less than those with strategic irrigation (83% less), indicative of the drought effect on their yields (Table 1
). The effect was more likely related to drought effects on grain set, for at least two reasons. First, there was no significant effect of strategic irrigation on straw yield (although, in 2015, the absolute effect was a 24% increase with strategic irrigation, and a concomitant strong influence on LAI at flowering, Table 2
), and second, grain filling should not have been influenced by strategic irrigation, for there was no irrigation applied after the 98 and 99 DAS cyclone (although 1000 grain weight was 11% greater with strategic irrigation, possibly because of the greater reallocation of pre-flowering stem carbohydrate reserves to growing grains). Similar yields of late flowering varieties to early flowering varieties under strategic irrigation in the semiarid tropics of central Queensland were possible, provided that cold stress is avoided.
Thirdly, differences in flowering time between varieties in both years within each of the two maturity classes were negligible and were not responsible for differences in yield between varieties within a maturity class.
Fourthly, the irrigation treatment did not influence days to flowering (data not presented), in line with the data of [43
], who found flowering to be delayed by only 3 days by drought. This is in contrast to the findings of [44
], who reported that drought stress has a strong (by up to 10 days) delaying effect on flowering time.
As a facultative short-day plant [45
], the rice crops would be expected to differ in their time to flower between years with different sowing dates, with fewer days to flower as daylengths shorten when sown after the solstice. Indeed, in the hot tropics, where temperature is relatively constant throughout the year, days to flower is entirely dependent upon daylength [46
]. However, in the semi-arid tropics, with a marked cool season, days to flower is not only dependent upon daylength, but also on temperature; cooler temperatures slow development according to the universal temperature effect [37
]. As a consequence, the late planting in 2014 would have subjected plants to shorter and shortening daylengths before and at the time of panicle initiation (c. 12 h, at 6–8 weeks after sowing) compared with a relative stable c. 13.5 h at the summer solstice, 6–8 weeks after sowing in 2015, and indeed, for early varieties, flowering was earlier by 11 days in 2014 (Table 7
). However, for late varieties, those generally considered as more photoperiod sensitive [47
], the reverse was evident (i.e., 8 d later in 2014). This may have been because of the greater low temperature sensitivity of the later indica
varieties, for it is well known that japonica
can develop at temperatures 2–3 °C lower than indica
], although [47
] report little difference in temperature sensitivity between late and early varieties. Later flowering indica
varieties have a longer basal vegetative period (BVP), that is, the duration from sowing to the start of photosensitivity [49
], in part responsible for their ‘lateness’, and lower temperature during this period might have delayed development even further, even though indica
varieties in general are more photosensitive and would have had a shortened photosensitive phase. Likewise, the period after the photosensitive period [49
] might have been lengthened as a result of low temperatures, resulting in a longer duration to flower than in the pre-solstice sowing. Thus, the low temperature delay was greater than the short-day hastening to flowering. The number of days to flower was less for early varieties in 2014 (the planting with shorter daylength overall), and the growing degree days to flowering were also less and much less for the early varieties. Earlier sowing (before the summer solstice in 2014/15) would be expected to delay flowering, leading to a greater number of growing degree days (GDDs) for a flowering event than if sown after the solstice. This was supported by data from our study for both sets of varieties (Table 7
). The greater difference of GDDs in 2014 between early varieties and late varieties can be attributed to the interaction between the greater sensitivity of late varieties to the low temperature conditions caused by late planting as compared with the response to temperature in the 2015 sowing. Further studies on the physiological responses of this germplasm to photoperiod and temperature may reveal the mechanisms for control of flowering date and interaction between them for Queensland. As [39
] reported in their study in the Riverina region of Australia, the growth duration of the crop was shortened and intercepted radiation was lessened when rice sowing was late (i.e., around or after the summer solstice), and they suggest choosing and sowing new varieties with a greater daylength sensitivity, leading to a greater likelihood that reproductive development take place before any later debilitating effect of low temperature, and thus availing of a wider planting window. A similar response to this was observed in early varieties in Alton Downs, Queensland and the early varieties are thus more suitable for a wider planting window, reducing year to year variation in flowering date and the demand for water during crop growth.
4.2. Effects of Planting Time, Maturity Class, and Strategic Irrigation on Light Capture and Use and the Influence on Yields
Even though strategic irrigation had next to no effect on time to heading, it did have a strong positive effect on leaf area expansion and LAI at flowering and on yield, more so in 2015 than in 2014 (Table 1
and Table 2
). The later (6 d) flowering and thus longer growth duration of late varieties with the later 2014 planting (Table 7
) led to greater LAI than in 2015 (Table 2
), but because of the overriding effect of fertility on yield in 2014, subject to low temperature and in the rainfed treatment to low water availability, there was no relationship across years between LAI and yield. In 2015, when fertility of the late varieties was not suppressed by low temperature, yield was related to LAI (r
= 0.79 **). This is in agreement with [50
], who reported that LAI measured during flowering was directly related to rice grain yield. Differences between varieties in LAI within the early or late flowering group were not due to maturity within a class, but due most likely to genetic differences, and differences between varieties in LAI between years were not related to yield. For example, varieties AAT 3, 4, 6, 19 had greater LAI in 2015 (Table 2
), but yields were the same in 2015 and 2014 (Table 1
), and others, e.g., AAT 17, had similar LAI at flowering in each year, but differed in yield between years.
Certainly, strategic irrigation raised soil moisture compared with rainfed conditions prior to flowering (e.g., Figure 2
); therefore, the positive effect of strategic irrigation on LAI measured at flowering (and presumably on pre-flowering biomass) was significant (Table 2
). Irrigation likewise significantly raised the number of effective tillers, which ultimately contributed to the higher LAI, but irrigation did not influence the total number of tillers (6.6 tillers per plant for rainfed and 6.8 tillers per plant with strategic irrigation).
Flag leaf area is considered important for yield as it is an important factor determining the photosynthetic output by influencing the photosynthetic area; indeed, [52
] reported a significant correlation between yield and flag leaf area, but such a relationship was not noted in our experiments. Smaller leaves had higher A during the flowering stage (r
= −0.38 ***). Across years, however, flag leaf area was positively correlated with quantity of strategic irrigation (r
= 0.52 ***) and negatively with instantaneous WUE (r
= −0.50 ***) during the flowering stage.
Across years, varieties, and irrigation treatments, there was significant positive correlation between yield and gas exchange parameters measured at flowering, e.g., for A (r
= 0.60 ***), gs (r
= 0.50 ***), and E (r
= 0.40 ***). Similar correlations were evident when analysing the within-year data, more so for 2015. For example, the correlation between yield and A was r
= 0.64 *** for 2014 and r
= 0.74 *** for 2015, although the effect was mainly due to differences between with or without strategic irrigation. We did not find intrinsic differences between the late flowering japonica
and early flowering indica
varieties in rates of photosynthesis, in agreement with [53
], but in contrast with the results of [54
]. Centritto et al. [7
] reported that drought stress is significantly correlated with the effects on A, as varieties with higher photosynthesis and conductance, presumably because they could access more soil water reserves, were more productive under all moisture conditions. Indeed, being able to maintain mesophyll, and not stomatal, conductance under water deficits largely determined tolerance to drought [55
]. One of our promising high yielding varieties, AAT 3, had a higher A during the flowering and grain filling period in both strategic irrigation and rainfed treatments in both years. Enhanced instantaneous water use efficiencies in our study across years, irrigation treatments, and varieties were associated with a high A (r
= 0.79 ***) rather than a low E (r
= 0.02 ns). The higher WUE was thus due to higher A than a reduced E, akin to the ‘capacity’ types of [25
]. The higher WUE at the cooler time of the year in 2014 when WUE was measured (at flowering time, Figure 1
) was presumably due to the lower vapour pressure deficit (VPD), but in contrast, later varieties that would have flowered in a slightly lower temperature and VPD regime than earlier varieties had similar or lower WUE. QTLs have been identified for rice photosynthesis rates using an indica
double haploid cross [56
], a good starting point for selecting for high rates of photosynthesis. In a study using lines from an introgression line population [57
], virtual ideotypes improved A and transpiration efficiency (TE, i.e., instantaneous water use efficiency) by 17 and 25%, respectively, over the best investigated, and their analysis showed the possibility of simultaneously improving A and TE.
4.3. Effects of Planting Time, Maturity Class, and Strategic Irrigation on Root Growth and the Influence on Yields
Root properties represented by greater root length density and deeper root systems are considered as target traits for drought tolerance [58
]. Chang and Vergara [59
] reported that a long and deep root system correlated with drought tolerance in upland or aerobic rice varieties. In our study, in 2015, rooting traits were quantified, and all varieties were deep rooted and had root systems reaching deeper than 60 cm.
Root characteristics such as root dry weight (RDW) at 0–15 cm were more closely correlated with yield (RDW and yield, r
= 0.48 ***), HI (RDW and HI, r
= 0.60 ***), and water productivity (RDW and water productivity, r
= 0.80 ***) when we analysed the combined data for strategic irrigation and rainfed, but not when analysing the responses under rainfed conditions only. The strong correlation between irrigation applied and RDW at 0–15 cm (r
= 0.80 ***) is directly due to the effect of water supply to that region. At 0–15 cm, roots of the irrigated crop could access more moisture supplied through surface drip irrigation, receiving 1.89 ML/ha more than from rainfed conditions in 2015 (Figure 1
). This favoured yield for the irrigated crop by confining and expanding more of its roots for better water extraction in the topsoil (at 0–15 cm, Figure 3
), resulting in a significant contribution to yield. This agrees with data of [60
], who showed that, when watered or re-watered from the surface, rice plants in pots took up very little water from the subsoil. They suggested that water uptake for the subsoil occurred only when water in the topsoil was less than −190 kPa. Such development plasticity triggered by mild drought stress is important in enhancing the efficiency of water uptake and biomass production in rice [61
Under rainfed conditions, there was no significant correlation between root characteristics and any of varietal grain yield, harvest index, or biomass, in contrast to the data of [62
], even though plants invested more overall in more extensive deep roots when rainfed (Figure 3
4.4. Effects of Planting Time, Maturity Class, and Strategic Irrigation on Spikelet Fertility, and Other Yield Determining Attributes and the Influence on Yields
The number of spikelets per panicle and the number of panicles per plant are determined well before flowering, but the fertility of spikelets, in particular functional pollen, is determined during the 10 days before and around flowering. Strategic irrigation did improve the number of spikelets per panicle (Table 5
), in accordance with the data of [63
], and averaged over both years increased spikelet fertility of all varieties, on average by 36%.
With the later, cooler, planting in 2014, on average, the total number of effective tillers per plant and number of spikelets per panicle were greater than in 2015, the latter by three times. Large numbers of spikelets per panicle have been reported to reduce the amount of pollen per anther [64
], which might also reduce the fertility of spikelets. Moreover, late flowering varieties, across both years and irrigation treatments, although having more spikelets per panicle than early varieties, had greater numbers of unfilled grains per panicle (data not presented) and a lower harvest index (Figure 4
). Early varieties had slightly higher spikelet fertility in 2014 than in 2015 (Figure 5
). Early varieties in 2014 flowered at 76 DAS and received rainfall up until 85 DAS (during flowering), whereas in 2015, early varieties flowered in 87 DAS and received rainfall only until 83 DAS. The next rainfall in 2015 was at 97 DAS, hence flowering and post-flowering drought stress may have reduced fertility somewhat in 2015. For late varieties in 2014, a shortage of rainfall during the flowering stage (Figure 1
) created drought, as evidenced by the rainfall record of only 11.6 mm from 100 to 130 DAS. In contrast, in 2015, 174.4 mm of rainfall was recorded for the same growth stage.
The onset of drought during the flowering stage of late varieties had an effect on spikelet sterility and grain filling. The correlation across varieties years and treatments between spikelet fertility and grain yield was highly significant in both years (r
= 0.73 *** in 2014 and r
= 0.53 *** in 2015), as was the correlation between yield and harvest index (r
= 0.93 *** in 2014 and r
= 0.81 *** in 2015). Spikelet sterility of up to 73% has been reported by [65
], while [66
] reported up to 98% sterility due to terminal drought. Our values for early varieties were not so affected. Early varieties such as AAT 4 recorded up to 90% and 85% fertility under rainfed condition in 2014 and 2015, respectively, compared with late varieties such as AAT 15, which recorded 1.4% fertility under rainfed conditions in 2014 and 39% in 2015. The harvest index was, therefore, highly influenced by spikelet fertility (compare data in Figure 4
and Figure 5
). In addition to the effects on pollen fertility, the lower spikelet fertility under rainfed conditions could also be due to a reduced assimilate availability and slower panicle exertion due to water stress, leading to fewer grains setting and a high proportion of abortion, as reported by [67
The effect of planting time on 1000 grain weight was not significant (Table 6
), in contrast to the effect of planting time on number of numbers of effective tillers per plant (4.8 vs. 6.3 /plant, up by 31% in late planted 2014) and yield (2.32 vs. 2.80 t/ha, up by 21% in early planted 2015). Likewise, the effect of irrigation treatment on 1000 grain weight was conservative (strategic irrigation raised it, albeit significantly by 7%, from 22.4 to 24.0 g/1000 grains), compared with its effects on numbers of effective tillers per plant (4.6 vs. 6.5 /plant, up by 41%) and yield (1.405 vs. 3.715 t/ha, up by 164%). Late indica
varieties in general had lower 1000 grain weight (c. 21 vs. c. 27 g), as reported earlier in the literature (e.g., [69
]). The values were consistent across years and irrigation treatments and may reflect the contribution of remobilised non-structural carbohydrates from stems to developing grains, capitalising upon conditions that favour pre-heading growth [70
When there is the occurrence of late season drought during flowering and grain filling under rainfed conditions, as in 2014, early flowering varieties have an advantage over late varieties and escape the drought. The varieties with delayed flowering (i.e., with a long BVP and more sensitive to the cold) were more susceptible to drought stress and recorded greater decreases in grain yield and HI compared with early varieties (Table 1
and Figure 4
). Prolonged drought under rainfed conditions resulted in lower yield and a decreased HI in late varieties as compared with early varieties. A similar relationship of reduction in HI with the onset of terminal drought in late flowering varieties was reported, showing a subsequent reduction in yield. As expected, based on the differential responses of straw yield (raised by 22%) and aboveground biomass (raised by 55%) to strategic irrigation, strategic irrigation raised HI consistently in both years (from 0.22 to 0.36 in 2014 and from 0.29 to 0.46 in 2015), with late varieties benefitting more from strategic irrigation than early varieties, especially in the late planted 2014 (data not presented).
Yields with strategic irrigation in the current study of the semi-arid tropics of central Queensland were similar to those of [71
], who recorded grain yield of 4.0–5.7 t/ha under irrigated aerobic conditions in the dry season in the Philippines. During wet seasons with supplemental sprinkler irrigation [72
], centre pivot [73
], flooded irrigation [74
], piped irrigation [75
], or under rainfed condition [76
], yields of more than 8 t/ha have been recorded, albeit with soil moisture content maintained at close to field capacity, whereas our strategic irrigation treatments were allowed to dry out to temporary wilting before strategic irrigation was supplied.