Journal of Experimental Botany, Vol. 73, No. 12 pp. 4236–4249, 2022 https://doi.org/10.1093/jxb/erac144 Advance Access Publication 6 April 2022 This paper is available online free of all access charges (see https://academic.oup.com/jxb/pages/openaccess for further details) RESEARCH PAPER Phenological optimization of late reproductive phase for raising wheat yield potential in irrigated mega-environments Pengcheng Hu1,2, , Scott C. Chapman2, , Sivakumar Sukumaran3, Matthew Reynolds3 and Bangyou Zheng1,*, 1 CSIRO Agriculture and Food, Queensland Biosciences Precinct, 306 Carmody Rd, St Lucia, Queensland 4067, Australia 2 The University of Queensland, School of Agriculture and Food Sciences, St Lucia, Queensland 4072, Australia 3 International Maize and Wheat Improvement Centre (CIMMYT), Carretera México-Veracruz Km 45, El Batán, Texcoco, México, CP 56237, Mexico * Correspondence: bangyou.zheng@csiro.au Received 11 November 2021; Editorial decision 30 March 2022; Accepted 4 April 2022 Editor: Ariel Vicente, CONICET – National University of La Plata, Argentina Abstract Increasing grain number through fine-tuning duration of the late reproductive phase (LRP; terminal spikelet to an- thesis) without altering anthesis time has been proposed as a genetic strategy to increase yield potential (YP) of wheat. Here we conducted a modelling analysis to evaluate the potential of fine-tuning LRP in raising YP in irrigated mega-environments. Using the known optimal anthesis and sowing date of current elite benchmark genotypes, we applied a gene-based phenology model for long-term simulations of phenological stages and yield-related variables of all potential germplasm with the same duration to anthesis as the benchmark genotypes. These diverse genotypes had the same duration to anthesis but varying LRP duration. Lengthening LRP increased YP and harvest index by increasing grain number to some extent and an excessively long LRP reduced YP due to reduced time for canopy construction for high biomass production of pre-anthesis phase. The current elite genotypes could have their LRP ex- tended for higher YP in most sites. Genotypes with a ratio of the duration of LRP to pre-anthesis phase of about 0.42 ensured high yields (≥95% of YP) with their optimal sowing and anthesis dates. Optimization of intermediate growth stages could be further evaluated in breeding programmes to improve YP. Keywords: Breeding, crop model, harvest index, mega-environment, phenology, spring wheat, yield potential. Introduction global demand for food security in the near future (Foulkes et al., 2011; Reynolds et al., 2011). The yield of wheat, a major Yield potential remains a principal breeding target as it is di- grain food crop, is arguably limited by the grain sink strength rectly related to the actual on-farm production (Acreche et al., during grain filling (Miralles and Slafer, 2007). Raising yield 2008; Fischer and Edmeades, 2010). A substantial rise in grain potential subsequently requires further improvement in the yield potential is required to meet the continuously growing sink capacity through increasing grain number per unit land Abbreviations. AGDWAN, aboveground dry weight at anthesis; AGDWPM, aboveground dry weight at physiological maturity; GN, grain number; GY, grain yield; HI, harvest index; LRP, late reproductive phase; ME, mega-environment; TTTA, duration of the late reproductive phase (from terminal spikelet to anthesis) expressed in thermal time; YP, yield potential. © The Author(s) 2022. Published by Oxford University Press on behalf of the Society for Experimental Biology. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. Downloaded from https://academic.oup.com/jxb/article/73/12/4236/6564125 by guest on 20 July 2022 Phenological optimization of late reproductive phase for raising yield potential | 4237 area and/or their average grain weight (Borrás et al., 2004; al., 2007). For instance, Borràs-Gelonch et al. (2012) found Miralles and Slafer, 2007). Grain yield is largely determined by that several quantitative trait loci (QTLs) had different effects grain number, with grain weight having a much smaller effect on two pre-anthesis phases in two populations of spring lines, on yield variations (Fischer, 1985), as grain growth is normally and several of these QTLs were significant for only one of not limited by the availability of photosynthetic assimilated es- the two phases. This relative independence among pre-anthesis pecially in high yielding genotypes in favourable environments phases makes it possible to fine-tune and optimize the dura- (Borrás et al., 2004; Fischer, 2007; Peltonen-Sainio et al., 2007). tion of individual phases for raising yield potential (Halloran The timing of phenological stages during the transition from and Pennell, 1982; García et al., 2011). It is important to em- vegetative growth through spike development to anthesis is phasize that tuning the pre-anthesis phases should not change anticipated to affect grain number and yield potential. the timing of anthesis, as it is a major breeding objective of The phenological development in wheat from emergence to wheat adaptation to a target environment and avoids abiotic anthesis can be divided into vegetative phase (from emergence stresses (e.g. frost, heat, and drought) (Miralles and Slafer, 2007; to floral initiation), early reproductive phase (from floral initia- Zheng et al., 2012; Kamran et al., 2014). The optimal anthesis tion to terminal spikelet initiation) and late reproductive phase period of a site is largely related to the location-specific envi- (LRP; from terminal spikelet to anthesis) (Slafer and Rawson, ronment (e.g. temperature, water, radiation, and frost and heat 1994). The LRP coincides with the rapid growth of spikes and stresses) rather than genotype (Flohr et al., 2017; Hunt et al., the development processes of florets while stem internodes 2019; Chen et al., 2020; Hu et al., 2021). The anthesis time elongate, which determine the survival of floret primordia to of modern elite genotypes has been tuned to match the op- be fertile during anthesis and hence the final grain number timal anthesis period in most wheat production environments (Kirby, 1988; Slafer and Rawson, 1994). The duration of LRP (Miralles and Slafer, 2007). For a given target environment (a may affect grain number through the floret mortality and the combination of location, season, and sowing date), there may possibility of floret primordia becoming fertile florets (Kirby, exist diverse genotypes that flower within the optimal anthesis 1988; González et al., 2003; Miralles and Slafer, 2007). A longer period but vary in the phenological development patterns of LRP allows (i) spikes to accumulate more assimilates for larger the pre-anthesis phase and yield potential of wheat. A study spike dry weight at anthesis, which is positively correlated with on identifying the optimal development pattern of the wheat floret survival and fertility (Fischer, 1985; González et al., 2011; pre-anthesis phase for the target environment is particularly Gonzalez-Navarro et al., 2016), and/or (ii) floret primordia to important for raising yield potential while minimizing the risk have more time to develop and achieve the stage of fertile of abiotic stresses through gross changes in the timing of vege- floret (Miralles et al., 2000; Gaju et al., 2009; Gonzalez-Navarro tative and reproductive development. et al., 2016). The duration of LRP is therefore a major de- The selection of wheat genotypes with the optimal devel- terminant of yield potential (García et al., 2011). Optimizing opment patterns of pre-anthesis for the target environment the phenological development pattern of pre-anthesis through could be conducted through field experiments across mul- tuning the onset of LRP (i.e. terminal spikelet) may contribute tiple growing seasons with diverse genotypes and sowing to increasing grain number and then rising wheat yield poten- dates. Numerous studies supported the idea that modifying tial (Slafer et al., 2001). the development pattern of pre-anthesis resulted in changes The duration of pre-anthesis phases (i.e. from emergence in the number of fertile florets and then grain number under to anthesis) is determined by the sensitivity of the genotype controlled or field conditions, through environmental manip- to vernalization (cold temperature), photoperiod (day length), ulation of temperature and photoperiod and/or genetic ma- and average temperature (earliness per se) (Rousset et al., 2011). nipulation of the sensitivity to these environmental factors (e.g. The sensitivity to vernalization and photoperiod is mainly de- González et al., 2003, 2005; Borràs-Gelonch et al., 2012; Sanna termined by the allele presence of vernalization and photo- et al., 2014; Gonzalez-Navarro et al., 2016; Basavaraddi et al., period genes, respectively. The allele presence of these genes 2021). However, these studies were mostly aimed at explor- enabled wheat genotypes to be classified as either a spring or ing the underlying physiological basis and limited to certain winter type. Wheat plants (including completely ‘spring’ type) growing seasons, genotypes, and environments. To the best of generally require a vernalization event to flower, followed by our knowledge, rarely has a study focused on the identification exposure to a lengthening photoperiod. The vernalization re- of the genotypes with the optimal development patterns of quirement of wheat is determined mainly by homoeologous pre-anthesis for further raising yield potential in target envi- VRN1 genes, with the winter alleles of VRN1 genes delaying ronments. Crop models could be used to augment empirical anthesis time of spring type instead of stopping anthesis (Eagles experiments to evaluate the potential of fine-tuning the devel- et al., 2009; Rousset et al., 2011). The duration of each in- opment pattern of pre-anthesis in improving yield potentials dividual pre-anthesis phase may be partially independent of of wheat through the simulation of genotype × environment the others, as they seem to be under different genetic control × management interactions (Chenu et al., 2017; Cooper et al., and hence differ in sensitivity to vernalization, photoperiod, 2021). Some crop models (e.g. APSIM-Wheat and DSSAT- and temperature (Slafer and Rawson, 1994; Whitechurch et Wheat) assume that more production of aboveground dry Downloaded from https://academic.oup.com/jxb/article/73/12/4236/6564125 by guest on 20 July 2022 4238 | Hu et al. matter at anthesis leads to greater grain number without CIMMYT genotypes at 70 sites across irrigated MEs (Hu et al., 2021). looking specifically at the spike growth and grain setting More information on the gene-based phenology model can be found during LRP (Jones et al., 2003; Holzworth et al., 2014). in Zheng et al. (2013) and Hu et al. (2021). Grain number in APSIM is estimated by a relationship between grain The objectives of this study were to use a gene-based phe- number and aboveground dry weight at the phenological stage of anthesis nology model (APSIM-Wheat-G; Zheng et al., 2013) to (i) (Holzworth et al., 2014; Zheng et al., 2014). The value of aboveground explore the response of yield-related variables (e.g. grain dry weight is potentially influenced by the dynamics of canopy develop- number and yield, aboveground biomass, and harvest index) ment interacting with phenological stages. Final grain yield through to to the varying duration of LRP without altering the anthesis maturity is then affected by the crop growth rate, potential grain size, and re-translocation of pre-anthesis biomass in stem and spike. In this study, date, and (ii) evaluate the potential of fine-tuning the develop- the phenology parameters affect the value of aboveground dry weight at ment patterns of pre-anthesis in raising yield potentials at 70 anthesis and grain number set, although there can be additional impacts representative sites of irrigated mega-environments (ME1 and on re-translocation, etc. That is to say, this study isolated the phenological ME5) for spring wheat. dynamics while appreciating multiple other physiological mechanisms af- fecting the establishment of grain number and size. The APSIM-Wheat-G model was used to simulate phenological Materials and methods timing and grain yield of spring wheat using 34 years of climatic data (1985–2018) at the 70 sites in irrigated MEs (Fig. 1) as described by Hu et Simulation of phenological phases and grain yield in irrigated al. (2021). Briefly, these sites were used by the Wheat Yield Collaboration Yield Trial (WYCYT) and the Elite Spring Wheat Yield Trial (ESWYT), mega-environments which were international multi-environment breeding nurseries that tar- The International Maize and Wheat Improvement Center (CIMMYT) geted the irrigated MEs (Sharma et al., 2012; Sukumaran et al., 2017). has used the concept of mega-environment (ME) to target germplasm Daily weather data for these sites were derived from the gridded NASA development of wheat and distribute international nurseries for testing POWER dataset with a spatial resolution of 0.5° (approximately 50 km, breeding lines regarding yield potential and adaptation to MEs. A ME is depending on latitude; Stackhouse et al., 2011). The maximum distance to a subset of unnecessarily contiguous areas that share similar quantitative the centre of the closest grid cell was less than 18 km. The soil data were and geospatial criteria including climate and soil characteristics, crop- derived from the ISRIC WISE soil dataset (Batjes, 2012), which is a de- ping system requirements, and biotic and abiotic stresses (Rajaram et al., tailed geo-referenced global soil profile database with a spatial resolution 1995; Hodson and White, 2007). ME1 is defined as low rainfall, optimal of 5 arc-minute (approximately 10 km, depending on latitude) and was irrigation, and highly productive spring wheat environments, and ME5 converted to compatible format with APSIM. is characterized as warm and humid tropical or subtropical regions that For each site, simulations were made for virtual genotypes and a real may also need to be irrigated (see Supplementary Fig. S1; Hodson and benchmark genotype of the site. The virtual genotypes were generated White, 2007; Braun et al., 2010). Spring wheat in these irrigated MEs by the exhaustive combinations of cultivar-specific parameters control- is autumn-sown and about 50 million ha of wheat was grown in 2014 ling time to anthesis in APSIM-Wheat, including vernalization sensitivity (Crespo-Herrera et al., 2017). (Rv), photoperiod sensitivity (Rp) and target thermal time from floral ini- A gene-based phenology module (Zheng et al., 2013) integrated into tiation to anthesis (TTFI,FL). Of these parameters, Rv and Rp ranged from 0 the widely used crop model APSIM-Wheat (version 7.6; Holzworth et to 5 at 0.1 intervals and TTFI,FL varied from 250 to 950 degree days (°Cd) al., 2014) was used to simulate the phenological stages and grain yield at an interval of 25 °Cd. Subsequently, a total of 75 429 virtual genotypes of sites in irrigated MEs. Briefly, the gene-based phenology module were simulated at each site. The benchmark genotype of a given site was related the sensitivity to vernalization (Rv) and photoperiod (Rp) in derived from our previous study (Hu et al., 2021), which used a model- APSIM-Wheat to the number of sensitive alleles of the VRN1 (i.e. ling approach (also with the APSIM-Wheat-G model) to identify the op- Vrn-A1, Vrn-B1, Vrn-D1) and PPD1 (i.e. Ppd-D1) genes, with 0 for timal anthesis and sowing date of the site with the benchmark genotype, the spring or photoperiod insensitive alleles and 1 for the winter or considering the effects of frost and heat stresses on yield. The benchmark photoperiod sensitive alleles. Linear functions were used to simulate genotype corresponded to the highest long-term (1985–2018) mean the relationships between the weighted numbers of VRN1 or Ppd-D1 yield of that site across different combinations of sowing dates (31 sowing alleles (weighting and summing the values of 0 or 1 at each locus) dates within a 5-month sowing window) and parameterized genotypes and Rv or Rp (Zheng et al., 2013). The integrated model (hereafter, (77 elite genotypes) from the WYCYT and ESWYT. More information APSIM-Wheat-G) requires inputs including daily weather data, man- on the parameterization of these genotypes with phenological and ge- agement practices, soil characteristics, and genetic information for cul- netic data and modelling analysis of optimal anthesis and sowing dates at tivars (i.e. alleles of VRN1 and Ppd-D1 genes). The gene information 70 sites of irrigated MEs can be found in Hu et al. (2021). was used in the gene-based phenology module within the model to The simulated wheat crop was sown at the optimal sowing date of the predict the wheat phenology. Compared with the default APSIM- site (Fig. 1) at a depth of 50 mm with a plant density of 250 plants m−2. Wheat, the APSIM-Wheat-G model has two modifications related to The optimal sowing date could ensure the benchmark genotype of the phenology modelling: (i) the interaction between vernalization and site flowers at the optimal anthesis date (see Supplementary Fig. S2) for photoperiod effects was modelled by a multiplicative function (Weir the highest long-term mean yield while minimizing the frost and heat et al., 1984) instead of minimum function; and (ii) the photoperiod stresses. Irrigation of 15 mm was applied at the sowing date to ensure effect was extended to the anthesis stage as the photoperiod effects on emergence after sowing. The wheat crop was simulated with no water pre-anthesis stages were widely reported (Slafer and Rawson, 1994). and nitrogen limitations, as (i) ME1 and ME5 are environments where The APSIM-Wheat-G model has demonstrated its prediction perfor- wheat grows under near-full irrigation or sufficient rainfall, such that mance for wheat phenology by combining genotypic and phenotypic water is never a limiting factor (Hodson and White, 2007); (ii) water and data, which obtained a root mean square error (RMSE) value of 4.3 nitrogen deficits were rarely observed in these trials as they were well d for predicting 4475 observations of heading dates of 179 Australian irrigated and fertilized according to local management practices; and (iii) genotypes at 79 sites across the Australian wheatbelt (Zheng et al., 2013) this study focused on evaluating the potential of raising yield potential by and an RMSE of 5.5 d when validating with 1591 observations from 77 tuning phenological development patterns of pre-anthesis. Downloaded from https://academic.oup.com/jxb/article/73/12/4236/6564125 by guest on 20 July 2022 Phenological optimization of late reproductive phase for raising yield potential | 4239 Fig. 1. The geographical distribution of optimal sowing date of representative sites in irrigated mega environments. The optimal sowing date was required by the benchmark genotype in the 70 sites of irrigated mega-environments to flower at the optimal anthesis date (Supplementary Fig. S2) for the highest long-term mean yield (Hu et al., 2021). Evaluation of the effects of tuning the late reproductive phase Å ã on yield formation T T min+Tmax u= − T 2 base (1) Three phenological stages were outputted from each APSIM-Wheat-G simulation, including emergence (DC10; Zadoks et al., 1974), terminal  Tu, 0 < Tu < 26 spikelet (DC31), and anthesis (DC65). Five variables related to yield ∆TT 26 formation were also outputted, two of which are determined at an- i =  8 (34− Tu) ,26 < Tu ≤ 34 (2) thesis (i.e. aboveground dry weight at anthesis (AGDWAN) and grain 0, Tu ≤ 0 or Tu > 34 number (GN)), and the other three are determined at maturity (i.e. ∑ngrain yield (GY), aboveground dry weight at physiological maturity (AGDWPM), and harvest index (HI)). The grain weight was excluded as TTp = ∆TTi (3) the preliminary results of this study showed that grain weight remained i=1 constant with varying duration of LRP across sites, which was con- TT R TA sistent with the literature suggesting that grain weight is conservative as TA/EA= (4) TT grain growth was normally not limited by the availability of photosyn- EA thetic assimilates during the grain filling phase in diverse environments Tu is the thermal unit, Tbase is the base temperature (°C), i is the i-th and genotypes (Fischer, 1985, 2007; Borrás et al., 2004; Peltonen-Sainio day within LRP (p=TA) or the pre-anthesis phase (p=EA), and n is the et al., 2007). number of days within a phase. To evaluate the effects of fine-tuning LRP on yield potential, for each To quantify the effects of the variation in TTTA on these variables site we selected genotypes that had the same long-term (1985–2018) related to yield formation, the variability of AGDWAN, GN, AGDWPM, mean anthesis date (i.e. the same duration of pre-anthesis) as the bench- GY, and HI across the virtual genotypes with the same duration of mark genotype, with a difference of no more than 0.5 d. The duration pre-anthesis was calculated using the coefficient of variation (CV, %) of a phenological phase was expressed as thermal time and calculated at each site. To quantify the relationships between RTA/EA and the five from key growth stages of APSIM outputs. Daily thermal time (ΔTTi) variables of the virtual genotypes, Spearman’s correlation (r) was cal- was calculated with the daily minimum (Tmin) and maximum (Tmax) culated at a significance level of 0.05 (i.e. α=0.05) for each site. The temperatures from climate records and three cardinal temperatures of simulated long-term mean GY of a genotype was considered as its wheat, i.e. 0 °C (base), 26 °C (optimum), and 34 °C (maximum) as in yield potential as the wheat crop was simulated with no water and APSIM-Wheat (Eqs (1) and (2)) (Zheng et al., 2014). All the genotypes nitrogen limitations and flowered at the optimal anthesis date of the were assumed to have the same cardinal temperatures in this study while site to minimize the frost and heat stress (Hu et al., 2021). The virtual acknowledging that the cardinal temperatures may vary with genotypes genotype achieving the highest yield potential (GYph) was selected for and phenological stages in wheat (Slafer and Rawson, 1995; Porter and each site and the optimal TTTA and then RTA/EA obtained. GYph was Gawith, 1999). The duration of a phenological phase (TTp; Eq. (3)) was compared with the simulated yield potential of the benchmark geno- calculated as the summation of daily thermal time within the phase (p). type (GYpb) of the given site to quantify the percentage increase in GY The ratio of the duration of LRP (i.e. from terminal spikelet initiation (PGY, %; Eq. (5)) for quantifying the effect of tuning TTTA on raising to anthesis; TTTA) to the pre-anthesis phase (i.e. from emergence to an- yield potential. The optimal range of RTA/EA was estimated for each site thesis; TTEA) was calculated. Hereafter, the ratio was referred to as RTA/ and defined as the RTA/EA of genotypes that had the simulated long- EA (Eq. (4)). term mean GY ≥95% of the GYph (i.e. 5% of yield loss) of that site. Downloaded from https://academic.oup.com/jxb/article/73/12/4236/6564125 by guest on 20 July 2022 4240 | Hu et al. The optimal ranges of RTA/EA were used to estimate a ratio ensuring with a median of 8.0%. Variability of GY was from 1.2% to high yields across sites in irrigated MEs. In addition, the daily average 8.3% and its mean value was 3.1%. Variability of HI was in the temperature and average cumulative radiation of the pre-anthesis phase (Supplementary Fig. S1) were calculated to explore their relationships range of 7.5% to 14.2% with a mean value of 9.5%. Sites with with the CVs of these variables and RTA/EA corresponding to GYph. All a higher daily average temperature of the pre-anthesis phase the statistical analyses were implemented using customized R (R Core tended to obtain lower variabilities of these variables (Fig. 3), Team, 2019) scripts. with the average temperature closely correlated to variabili- ties of TTTA, GN, GY, and HI (all r>0.6) but with less effect GY P ph − GYpb GY= × 100 % (5) on AGDWAN and AGDWPM (r=0.33). Subsequently, ME5 sites GYpb tended to have lower variabilities of these variables than ME1 sites (Supplementary Figs S5, S6). The TTTA variability among sites was correlated with the variability of other variables (in- Results duced by the varying TTTA). Variability of TTTA showed strong Variability for variables of genotypes with the same effects on GN, GY, and HI variability (r>0.5). Variability of duration to anthesis AGDWAN and AGDWPM was only partially explained by the TTTA variability (r<0.4). The number of virtual genotypes with the same duration to an- thesis varied from 96 to 1541 across sites, with most sites having Responses of variables to the varying duration of the 100–250 genotypes meeting this criterion (see Supplementary late reproductive phase Fig. S3), depending on the intervals of parameter values used in this study. The simulated TTTA, AGDWAN, GN, AGDWPM, The simulated AGDWAN, GN, AGDWPM, GY, and HI showed GY, and HI varied across virtual genotypes with the same du- different responses to the varying TTTA of virtual genotypes ration to anthesis at each site (Fig. 2; Supplementary Fig. S4). with the same duration to anthesis across sites (Fig. 4). These For variables determined at anthesis, TTTA showed the highest sites represent diverse response patterns of these variables to variability across sites, its CV ranging from 18.4% to 29.6% the varying TTTA and different temperature and radiation with a median of 22.8%. Both AGDWAN and GN showed conditions. The daily average temperatures of the pre-anthe- lower variability, having a similar range of less than 10% with sis phase of these sites were 22.4 °C (Powarkheda), 22.8 °C a median of about 3%. For variables determined at maturity, (Gokulwadi Jalna), 13.1 °C (Texcoco), and 11.2 °C (Chiillan); AGDWPM variability across sites ranged from 4.6% to 16.8% Fig. 2. Distribution of coefficient of variation (%) of the duration of LRP (TTTA), aboveground dry weight at anthesis (AGDWAN), grain number (GN), aboveground dry weight at physiological maturity (AGDWPM), grain Fig. 3. Relationship between the daily average temperature of pre- yield (GY), and harvest index (HI) across 70 sites in irrigated mega- anthesis phase of benchmark genotypes and coefficients of variation (CV) environments. The coefficient of variation was calculated among virtual of the duration of the late reproductive phase (TTTA), aboveground dry genotypes with the same duration to anthesis for each site. The violin weight at anthesis (AGDWAN), grain number (GN), aboveground dry weight plots represent the distributions of coefficients of variation of variables. The at physiological maturity (AGDWPM), grain yield (GY), and harvest index (HI) Spearman correlations (r) between the coefficients of variation of TTTA and of virtual genotypes of spring wheat with the same duration to anthesis at other variables are shown at the top of the figure. sites of irrigated mega-environments. Downloaded from https://academic.oup.com/jxb/article/73/12/4236/6564125 by guest on 20 July 2022 Phenological optimization of late reproductive phase for raising yield potential | 4241 Fig. 4. Examples of the relationships of simulated long-term (1985–2018) mean aboveground dry weight at anthesis (AGDWAN), grain number (GN), aboveground dry weight at maturity (AGDWPM), grain yield (GY), and harvest index (HI) against the duration of the late reproductive phase of virtual genotypes with the same duration to anthesis at sites of irrigated mega-environments. The four sites were located at Powarkheda (India), Gokulwadi Jalna (India), Texcoco (Mexico), and Chiillan (Chile). the average cumulative radiations were 1293.3 MJ m−2 sites are demonstrated in Fig. 4, the other sites in this study (Powarkheda), 1254.4 MJ m−2 (Gokulwadi Jalna), 2067.3 MJ showed similar patterns. m−2 (Texcoco), and 1912.4 MJ m−2 (Chiillan) (Supplementary Both AGDWAN and AGDWPM were closely and negatively Fig. S1). AGDWAN and AGDWPM had a similar response to correlated with TTTA (Fig. 5; Supplementary Figs S7, S8) across the increased TTTA, i.e. they consistently decreased (as in most sites. The Spearman correlation coefficient (r) varied Powarkheda and Gokulwasi Jalna; Fig. 4A, C, F, H) or ini- from −0.99 to 0.68 for AGDWAN, with most of them (63 out tially increased to a maximum value and then decreased (as in of 70 sites) less than −0.8, showing that AGDWAN consistently Texcoco and Chiillan; Fig. 4K, M, P, R). Conversely, GN and decreased with TTTA increased (e.g. Fig. 4A, F). Some sites HI increased at first (as in Powarkheda and Chiillan; Fig. 4E, T) with lower average temperatures (latitude >34° and/or high or then decreased after increasing to the maximum value (as altitude) showed less correlation between AGDWAN and TTTA in Gokulwasi Jalna and Texcoco; Fig. 4G, J, L, O) when TTTA (−0.80.8) at 30 sites and so was GY at eight sites, indicating that tively. The span of the optimal range of RTA/EA varied between they increased with TTTA (e.g. Fig. 4B). GN and GY in the 0.13 and 0.21 across sites, and an RTA/EA of 0.42 could ensure other sites followed the pattern of initially increasing to a max- the highest yields at most sites (67 out of 70 sites). The RTA/EA imum value and then decreasing as TTTA increased (e.g. Fig. of benchmark genotype was within the optimal range of RTA/ 4D, Q). HI was positively and closely correlated with TTTA, EA at most sites (58 out of 70 sites). The difference of RTA/EA with r ranging from 0.90 to 0.99, showing that HI tended to between the GYph and the benchmark genotype of each site monotonically increase along with increased TTTA at most sites ranged from −0.07 to 0.19. The RTA/EA of the benchmark gen- (e.g. Fig. 4E, T). otype was normally less than that of the GYph (56 out of 70 sites). Consequently, the TTTA of benchmark genotypes should be extended to approach a larger RTA/EA at these sites for a Yield potential and harvest index raised by fine-tuning higher yield potential (Supplementary Fig. S12). the late reproductive phase Across years, the simulated mean GYph varied across sites, rang- ing from 3.6 to 11.7 t ha−1 (Supplementary Fig. S9A). ME1 Discussion sites had higher GYph than ME5 sites, varying from 5.1 to 11.7 Increasing grain number per square metre through fine-tuning t ha−1 with a mean of 6.9 t ha−1. ME5 sites achieved lower phenological development pattern of the pre-anthesis phase GYph ranging from 3.6 to 5.9 t ha−1 with a mean of 5.1 t ha−1. without altering the anthesis time was considered as a breeding Sites with higher latitudes or altitudes had the highest GYph strategy to increase the yield potential of wheat (González (>8 t ha−1). When compared with the simulated GYpb of sites, et al., 2005; Miralles and Slafer, 2007). In a previous paper (Hu the increase in yield potential ranged from 0.04 to 0.59 t ha−1 et al., 2021), it was shown that the gene-based model (APSIM- with a mean of 0.2 t ha−1 (Fig. 6A). The increases in yield po- Wheat-G) could be used to predict wheat anthesis time across tential of ME1 sites were slightly smaller than ME5 sites on av- irrigated MEs. The model was then used to identify the optimal erage, with a mean of 0.19 t ha−1 for ME1 sites and 0.22 t ha−1 anthesis and sowing date of the current elite genotype with the for ME5 sites. The percentage increase in the yield potential highest long-term mean yield at individual sites, considering varied between 0.7% and 9.6% across sites (Fig. 6B). ME5 sites the potential impacts of frost and heat stress. Based on these Downloaded from https://academic.oup.com/jxb/article/73/12/4236/6564125 by guest on 20 July 2022 Phenological optimization of late reproductive phase for raising yield potential | 4243 Fig. 6. Amount (A) and percentage (B) of increase in simulated (1985–2018) yield potential at 70 sites of irrigated mega-environments. The increases in yield potential were calculated by the comparison of the virtual genotype with the optimal duration of the late reproductive phase and the benchmark genotype of the site. The two genotypes had the same duration to anthesis but different duration of the late reproductive phases. results, this study used a modelling approach with APSIIM- which could favour more genotypes with the same duration Wheat-G to evaluate the potential of fine-tuning the duration to anthesis but distinct partitioning of the pre-anthesis phases. of LRP for further raising the yield potential of spring wheat Variabilities of these variables (medians of the CV of less than in irrigated MEs through increasing sink strength. 10%) were significantly less than the TTTA variability (a median There were multiple virtual genotypes with the same dura- of 22.8%), and their correlation coefficients with the TTTA var- tion to anthesis at each site (see Supplementary Fig. S3), which iability were less than 0.75 (Fig. 2), which suggested that the agreed with experimental studies (e.g. Whitechurch and Slafer, TTTA variability was partially converted to these variables. 2002; Gonzalez-Navarro et al., 2016) and indicated that di- The response to varying TTTA differed in different variables verse phenological combinations of genotypes achieved the (Figs 4, 5). AGDWAN gradually decreased as the TTTA increased same timing to anthesis. This was because the duration of these at most sites (e.g. Fig. 4A and F), which was due to that longer phases may be partially independent of each other, as the sensi- LRP consequently shortened the vegetative and early repro- tivity of each phase to vernalization, photoperiod, and temper- ductive phases (from emergence to terminal spikelet) under a ature seems to be under different genetic controls (Slafer and given duration to anthesis. A shorter duration of vegetative and Rawson, 1994; González et al., 2003; Borràs-Gelonch et al., early reproductive phase produced a small canopy with lower 2012). The varying TTTA of virtual genotypes with the same maximum leaf area index (see Supplementary Fig. S13), thus duration to anthesis induced variations in variables related to reducing the amount of biomass production at anthesis (Borràs- yield formation (i.e. AGDWAN, GN, AGDWPM, GY, and HI). Gelonch et al., 2012; Gonzalez-Navarro et al., 2016). AGDWPM Variabilities of these variables appeared to be negatively associ- similarly responded to TTTA as AGDWAN (e.g. Fig. 4C, H) since ated with the daily average temperature of pre-anthesis (Fig. 3), genotypes shared the same grain filling process. Lengthening perhaps because the benchmark genotypes of sites with mild LRP is usually recognized as an avenue to increase GN and temperatures tended to have a longer duration to anthesis, then GY (Miralles and Slafer, 2007). Our study confirmed this Downloaded from https://academic.oup.com/jxb/article/73/12/4236/6564125 by guest on 20 July 2022 4244 | Hu et al. Fig. 7. Amount (A) and percentage (B) of increase in simulated (1985–2018) harvest index corresponding to the highest yield potential of sites in irrigated mega-environments. The increases in harvest index were calculated by the comparison between the virtual genotype with the optimal duration of the late reproductive phase and the benchmark genotype of the site. The two genotypes had the same duration to anthesis but different durations of the late reproductive phases. idea with GN increasing as the TTTA lengthened (e.g. Fig. 4B) advancing the onset of stem elongation. The early onset of stem at some sites (30 out of 70 sites), which was because lengthen- elongation may increase the spring frost risk during the sensi- ing LRP allowed more assimilates to be diverted to the growing tive LRP, which deforms young spikes or reduces the tiller and spikes during LRP for an increased spike dry weight at an- spike number and then GN (Fletcher and Cullis, 1988; Martino thesis, which resulted in higher number of fertile florets and and Abbate, 2019). GY was closely correlated to GN, so the re- GN (Fischer, 1985; Slafer et al., 1990; Miralles and Slafer, 2007; sponse of GY to varying TTTA was similar to that of GN, i.e. GY Gaju et al., 2009). However, as the TTTA further lengthened, decreased after reaching a maximum as the TTTA increased (Fig. GN experienced a decline after achieving a maximum value 4D, S) at most sites (62 out of 70 sites). The strong correlation (e.g. Fig. 4Q) at other sites, and this means that lengthening between GY and GN implied that there was no effect of LRP LRP for higher GN failed to compensate for the biomass pro- variations on individual grain weight in this study. This was in duction of a small canopy caused by shortened vegetative phase. agreement with previous studies, which concluded that wheat It was reported that a longer LRP was not always related to has a high degree of homeostasis in final grain weight as grain a higher spike dry weight and then higher GN, as it might growth is mostly sink-limited during the grain filling phase hamper canopy growth and/or biomass partitioning to spike (i.e. the assimilate availability exceeds the demand of growing (Gonzalez-Navarro et al., 2016) or there might be a trade-off grains) under broad combinations of growing environments between duration and the rate of spike growth (González et al., and cultivars (Fischer, 1985, 2007; Borrás et al., 2004; Peltonen- 2005). This experimental result is apparently captured at least Sainio et al., 2007). HI tented to increase with TTTA, which was in part in the simulations presented here. Further, lengthen- mainly attributed to decreasing AGDWPM and then followed ing the LRP without altering the timing of anthesis requires by increasing GY. This also showed that the trade-off between Downloaded from https://academic.oup.com/jxb/article/73/12/4236/6564125 by guest on 20 July 2022 Phenological optimization of late reproductive phase for raising yield potential | 4245 Yield potential could be raised by selecting or breeding genotypes with optimal phenological development patterns of pre-anthesis across sites of irrigated MEs (Fig. 6). The simulated highest yield potentials on the ME1 and ME5 sites were 6.9 t ha−1 and 5.1 t ha−1 on average, respectively (see Supplementary Fig. S9A), which represented a 3.0% and 4.6% increase on av- erage, respectively, compared with the yield potential of the simulated benchmark genotypes of the sites. It should be noted that the yield potential was the further improvements on that of the benchmark genotype of the site, which was bred to raise the yield potential or selected from the current elite lines of spring wheat in irrigated environments (Trethowan et al., 2003; Sharma et al., 2012; Sukumaran et al., 2017). The benchmark genotype was well adapted to the given environment for real- izing its yield potential with the known optimal sowing and anthesis date of the site, which was evaluated by a comprehen- sive modelling analysis of genotype, environment, and manage- ment in the previous paper (Hu et al., 2021). Therefore, it was expected that the relatively low percentage increase in yield potential (about 4% on average) was estimated in this study by fine-tuning LRP when compared with the current high- yielding genotype (i.e. the benchmark genotype). On the other hand, the simulated yield potential of the benchmark genotype was normally within the range of ≥95% of the highest yield potential of the site (58 out of 70 sites; Fig. 8). The HI corre- sponding to the highest yield potential increased by 8.9% on average across sites as compared with that of the benchmark genotypes (Fig. 7), which combined with the improved yield potential implied that the genotype with optimal duration of LRP might have an appropriate balance between source and sink and succeed in more fully exploiting assimilation capacity (Reynolds et al., 2009). The differences of RTA/EA between these virtual genotypes with the highest yield potentials and corresponding bench- mark genotypes suggested the strategy of fine-tuning LRP for further raising yield potential at individual sites (Fig. 8). The results showed that the duration of LRP of spring wheat (represented by the benchmark genotype) should be ex- tended without altering anthesis time for high yield potential at most sites (56 out of 70 sites; see Supplementary Fig. S12). This agrees with experimental studies, which concluded that extending the LRP of wheat could raise yield potential by increasing GN without offsetting grain weight (Miralles and Fig. 8. Optimal spans of the ratio of the duration of the late reproductive Slafer, 2007). The virtual genotypes with the highest yield to pre-anthesis phases (RTA/EA) for the highest yields (≥95% of the highest potentials of the sites had varying RTA/EA ranging from 0.29 yield potential) of 70 sites in irrigated mega-environments. Rectangles are the optimal ranges of the ratio of the highest grain yields of the sites. Blue to 0.56 (Supplementary Fig. S9C), but an RTA/EA of about segments are the R 0.42 could realize at least 95% of the highest yield poten- TA/EA of the genotypes with the highest yield potentials and the grey circles are the RTA/EA of the benchmark genotypes. The red tial (with a risk of 5% yield loss) across most sites (67 out dotted line is the RTA/EA ensuring the highest yields across most sites (67 of 70 sites; Fig. 8) with their optimal sowing and anthesis out of 70 sites). dates. The RTA/EA of about 0.42 agreed with that of geno- types/cultivars of spring wheat under field conditions, which AGDWPM and HI suggested that pursuing larger AGDWPM or normally varied between 0.30 and 0.45 (Slafer et al., 2001; higher HI alone was not sufficient for GY gain (Aisawi et al., Whitechurch et al., 2007; Borràs-Gelonch et al., 2012; Sanna et 2015; Reynolds et al., 2017). al., 2014; Guo and Schnurbusch, 2015; Ochagavía et al., 2017). Downloaded from https://academic.oup.com/jxb/article/73/12/4236/6564125 by guest on 20 July 2022 4246 | Hu et al. Whitechurch et al. (2007) evaluated the variability in the du- Conclusions ration of LRP of 64 Argentine wheat cultivars, and their RTA/ EA varied from about 0.27 to 0.39 when sown at the rec- This study used a modelling approach to evaluate the potential ommended sowing date, which was smaller than the R of fine-tuning LRP of spring wheat in raising yield potential TA/EA (≥0.43) of the highest yield potentials at the three Argentina at sites of irrigated MEs. The aim was to isolate the phenolog- sites (Supplementary Fig. S9C). Further, Whitechurch et al. ical dynamics while appreciating that there are multiple other (2007) reanalysed the 20 wheat cultivars reported in previous physiological mechanisms affecting the establishment of grain literature, whose RTA/EA mostly ranged from about 0.27 to number and size. The simulation analysis demonstrated that 0.51. Guo and Schnurbusch (2015) measured the TTEA and diverse genotypes with the same duration to anthesis could TTTA of 12 German spring cultivars and the RTA/EA varied vary in duration of LRP at individual sites. Lengthening LRP between 0.37 and 0.42 with a mean of 0.40. Variability in to some extent could increase the yield potential and harvest RTA/EA was observed in populations of recombinant lines of index of wheat by increasing grain number, with genotypes of wheat as well. Borràs-Gelonch et al. (2012) reported TT the optimal duration of LRP achieving further increase (about EA and TTTA of 212 recombinant lines of Australian spring cul- 4% on average and up to 10%) in yield potential and harvest tivars and the calculated RTA/EA ranged between about 0.33 index (about 9% on average and up to 24%) compared with and 0.47. Similarly, another population of 100 recombinant those of the high-yielding benchmark genotypes of the respec- lines had RTA/EA of about 0.22 to 0.41 with a mean of 0.34 tive sites. Genotypes with a ratio of the duration of LRP to the under the long day and non-vernalized condition (Sanna et pre-anthesis phase of about 0.42 could ensure at least 95% of al., 2014). The range of optimal RTA/EA for high yields (≥95% the highest yield potential across most sites with their optimal of the highest yield potential) was an indicator of difficulty sowing and anthesis dates. The current elite genotypes could in the selection of the well-adapted genotypes for the site have their LRP extended for higher yield potential in most (Fig. 8), with a wide range suggesting more potential candi- sites. The results also implied that an excessively long duration dates for the target environment and vice versa. of LRP reduced yield potential due to a reduction in time Further work is required in terms of model development for canopy construction (with sufficient leaves and tillers) for and analysis. This study focused on the potential of fine-tuning high biomass production in the pre-anthesis stage. The study LRP in improving the sink strength and yield potential via suggested that fine-tuning pre-anthesis development patterns boosting grain number, but altering the duration of the LRP without altering anthesis time can raise wheat yield poten- may also change the potential grain weight (Calderini et al., tial by improving grain set, and therefore post-anthesis sink 1999; Calderini and Reynolds, 2000; Ugarte et al., 2007). Thus strength and HI. potential grain weight will be considered in future modelling analysis of fine-tuning LRP for increasing yield potential. The Supplementary data terminal spikelet stage was not explicitly simulated by the cur- rent ASPIM-Wheat model; mechanistic models with a molec- The following supplementary data are available at JXB online. ular and/or physiological basis to predict the terminal spikelet Fig. S1. The long-term (1985–2018) daily average tem- stage were developed (e.g. Brown et al., 2013), but more effort perature and average cumulative radiation of the pre-anthe- is required to improving their validity. Further, crop processes sis phase of benchmark genotypes at 70 sites in irrigated of reproductive growth and development (e.g. floret fertiliza- mega-environments. tion, grain setting, and grain growth) need to be more specif- Fig. S2. The geographical distribution of optimal flowering ically modelled to improve predictive performance (Brisson et date of the 70 representative sites in irrigated mega-environments. al., 2002; Messina et al., 2019). Particularly, the effects of plant Fig. S3. The distribution of the number of virtual geno- nitrogen status at anthesis on grain setting should be mod- types with the same duration to anthesis at 70 sites of irrigated elled as it was strongly correlated to wheat GN (Abbate et mega-environments. al., 1995; Jeuffroy and Bouchard, 1999; Lemaire and Ciampitti, Fig. S4. Variations in the duration of the late reproductive phase, 2020). Nitrogen deficiency before anthesis could reduce GN, aboveground dry weight at anthesis and grain number, aboveg- as it is related to the reductions in various components of grain round dry weight at physiological maturity, grain yield, and har- set (e.g. the number of spikes per square metre, spikelets per vest index of virtual genotypes of spring wheat with the same spike, and differentiated florets, floret survival, and fertility) duration to anthesis at 70 sites of irrigated meta-environments. (Peltonen-Sainio and Peltonen, 1995; Demotes-Mainard et al., Fig. S5. Coefficient of variation (%) in the duration of the 1999; Demotes-Mainard and Jeuffroy, 2001). In addition, the late reproductive phase, aboveground dry weight at anthesis and effects of post-anthesis nitrogen status on grain weight also grain number of virtual genotypes of spring wheat with the same should be considered in modelling grain yield formation and duration to anthesis at 70 sites of irrigated mega-environments. grain quality (Kimball et al., 2001; Nuttall et al., 2017; Zhao et Fig. S6. Coefficient of variation (%) in the aboveground dry al., 2020). weight at physiological maturity, harvest index and grain yield Downloaded from https://academic.oup.com/jxb/article/73/12/4236/6564125 by guest on 20 July 2022 Phenological optimization of late reproductive phase for raising yield potential | 4247 of virtual genotypes of spring wheat with the same duration to Funding anthesis at 70 sites of irrigated mega-environments. This research was supported by the International Wheat Yield Partnership Fig. S7. Spearman correlation between the duration of late (IWYP) as part of an IWYP project ‘A Genetic Diversity Toolkit to reproductive phase and two variables determined at anthesis, Maximize Harvest Index by Controlling the Duration of Developmental aboveground biomass and grain number, of virtual genotypes Phases’ via funding from the Grains Research and Development of spring wheat with the same duration to anthesis at 70 sites Corporation (GRDC; Grant No. CSP00205). of irrigated mega-environments. Fig. S8. Spearman correlation between the duration of late reproductive phase and three variables determined at maturity, Data availability aboveground biomass, harvest index, and grain yield, of virtual genotypes of spring wheat with the same duration to anthesis The data supporting the findings of this study are available from the cor- responding author (BZ) upon request. at 70 sites of irrigated mega-environments. Fig. S9. Spatial distribution of the simulated (1985–2018) highest yield potentials, the corresponding harvest index, and References the ratio of the duration of the late reproductive phase to pre- anthesis phase of genotype with the optimal duration of the late Abbate PE, Andrade FH, Culot JP. 1995. The effects of radiation and nitrogen on number of grains in wheat. The Journal of Agricultural Science reproductive phase across sites of irrigated mega-environments. 124, 351–360. Fig. S10. Durations of different pre-anthesis phases (°Cd) Acreche MM, Briceño-Félix G, Sánchez JAM, Slafer GA. 2008. of the virtual genotypes with the highest yield potentials of Physiological bases of genetic gains in Mediterranean bread wheat yield in spring wheat at 70 sites of irrigated mega-environments. Spain. European Journal of Agronomy 28, 162–170. Fig. S11. Relationship between the ratio of the duration of Aisawi KAB, Reynolds MP, Singh RP, Foulkes MJ. 2015. The physi- ological basis of the genetic progress in yield potential of CIMMYT spring the late reproductive phase to the pre-anthesis phase (RTA/EA) wheat cultivars from 1966 to 2009. Crop Science 55, 1749–1764. of the genotypes with highest yield potentials and their daily Basavaraddi PA, Savin R, Wingen LU, Bencivenga S, Przewieslik- average temperature and average cumulative radiation of the Allen AM, Griffiths S, Slafer GA. 2021. Interactions between two QTLs pre-anthesis phase of sites in irrigated mega-environments. for time to anthesis on spike development and fertility in wheat. Scientific Reports 11, 2451. Fig. S12. Strategies for fine-tuning the duration of the Batjes NH. 2012. ISRIC-WISE derived soil properties on a 5 by 5 arc-min- late reproductive phase of the benchmark genotypes to ap- utes global grid (ver 1.2). Wageningen, The Netherlands: ISRIC-World Soil proach the highest yield potential at 70 sites in irrigated Information. mega-environments. Borrás L, Slafer GA, Otegui ME. 2004. Seed dry weight response to Fig. S13. Spearman correlation between the duration of late source–sink manipulations in wheat, maize and soybean: a quantitative re- appraisal. Field Crops Research 86, 131–146. reproductive phase and the maximum leaf area index of virtual Borràs-Gelonch G, Rebetzke GJ, Richards RA, Romagosa I. 2012. genotypes of spring wheat with the same duration to anthesis Genetic control of duration of pre-anthesis phases in wheat (Triticum aesti- at 70 sites of irrigated mega-environments. vum L.) and relationships to leaf appearance, tillering, and dry matter accu- mulation. Journal of Experimental Botany 63, 69–89. Braun HJ, Atlin G, Payne T. 2010. Climate change and crop production. In: Reynolds MP, ed. Multi-location testing as a tool to identify plant response to Acknowledgements global climate change. Wallingford, Oxfordshire, UK: CABI, 115–138. We thank Dr Dean Holzworth (CSIRO) for sharing the APSoil files of Brisson N, Ruget F, Gate P, et al. 2002. STICS: a generic model for simu- lating crops and their water and nitrogen balances. II. Model validation for global soil profiles. Acknowledgment is made to the APSIM Initiative, wheat and maize. Agronomie 22, 69–92. which takes responsibility for quality assurance and a structured innova- Brown HE, Jamieson PD, Brooking IR, Moot DJ, Huth NI. 2013. tion programme for APSIM’s modelling software, which is provided free Integration of molecular and physiological models to explain time of anthesis for research and development use (see www.apsim.info for details). in wheat. Annals of Botany 112, 1683–1703. Calderini DF, Abeledo LG, Savin R, Slafer GA. 1999. Effect of tempera- ture and carpel size during pre-anthesis on potential grain weight in wheat. Author contributions The Journal of Agricultural Science 132, 453–459. Calderini DF, Reynolds MP. 2000. Changes in grain weight as a con- SCC: funding acquisition. BZ: project administration. SSC and BZ: su- sequence of de-graining treatments at pre- and post-anthesis in synthetic pervision. PH, SCC, and BZ: conceptualization. PH: methodology. PH: hexaploid lines of wheat (Triticum durum × T. tauschii). Functional Plant Biology 27, 183–191. software. PH: investigation. SS and MR: data curation. PH: formal anal- ysis. PH: visualization. PH: writing—original draft. PH, SCC, SS, MR, Chen C, Fletcher AL, Ota N, Flohr BM, Lilley JM, Lawes RA. 2020. Spatial patterns of estimated optimal flowering period of wheat across the and BZ: writing–review and editing. southwest of Western Australia. Field Crops Research 247, 107710. Chenu K, Porter JR, Martre P, Basso B, Chapman SC, Ewert F, Bindi M, Asseng S. 2017. Contribution of crop models to adaptation in wheat. Conflict of interest Trends in Plant Science 22, 472–490. Cooper M, Voss-Fels KP, Messina CD, Tang T, Hammer GL. 2021. The authors declare that there is no conflict of interest. Tackling G × E × M interactions to close on-farm yield-gaps: creating novel Downloaded from https://academic.oup.com/jxb/article/73/12/4236/6564125 by guest on 20 July 2022 4248 | Hu et al. pathways for crop improvement by predicting contributions of genetics and Hu P, Chapman SC, Dreisigacker S, Sukumaran S, Reynolds M, management to crop productivity. Theoretical and Applied Genetics 134, Zheng B. 2021. Using a gene-based phenology model to identify optimal 1625–1644. flowering periods of spring wheat in irrigated mega-environments. Journal Crespo-Herrera LA, Crossa J, Huerta-Espino J, Autrique E, Mondal of Experimental Botany 72, 7203–7218. S, Velu G, Vargas M, Braun HJ, Singh RP. 2017. Genetic yield gains in Hunt JR, Lilley JM, Trevaskis B, Flohr BM, Peake A, Fletcher A, Zwart CIMMYT’s international elite spring wheat yield trials by modeling the geno- AB, Gobbett D, Kirkegaard JA. 2019. Early sowing systems can boost type × environment interaction. Crop Science 57, 789–801. Australian wheat yields despite recent climate change. Nature Climate Demotes-Mainard S, Jeuffroy M-H. 2001. Partitioning of dry matter and Change 9, 244–247. nitrogen to the spike throughout the spike growth period in wheat crops Jeuffroy M-H, Bouchard C. 1999. Intensity and duration of nitrogen defi- subjected to nitrogen deficiency. Field Crops Research 70, 153–165. ciency on wheat grain number. Crop Science 39, 1385–1393. Demotes-Mainard S, Jeuffroy M-H, Robin S. 1999. Spike dry matter Jones JW, Hoogenboom G, Porter CH, Boote KJ, Batchelor WD, Hunt and nitrogen accumulation before anthesis in wheat as affected by ni- LA, Wilkens PW, Singh U, Gijsman AJ, Ritchie JT. 2003. The DSSAT trogen fertilizer: relationship to kernels per spike. Field Crops Research 64, cropping system model. European Journal of Agronomy 18, 235–265. 249–259. Kamran A, Iqbal M, Spaner D. 2014. Flowering time in wheat (Triticum Eagles HA, Cane K, Vallance N. 2009. The flow of alleles of important aestivum L.): a key factor for global adaptability. Euphytica 197, 1–26. photoperiod and vernalisation genes through Australian wheat. Crop and Pasture Science 60, 646–657. Kimball BA, Morris CF, Pinter PJ Jr et al. 2001. Elevated CO2, drought and soil nitrogen effects on wheat grain quality. New Phytologist 150, Fischer RA. 1985. Number of kernels in wheat crops and the influence of 295–303. solar radiation and temperature. The Journal of Agricultural Science 105, 447–461. Kirby EJM. 1988. Analysis of leaf, stem and ear growth in wheat from ter- minal spikelet stage to anthesis. Field Crops Research 18, 127–140. Fischer RA. 2007. Understanding the physiological basis of yield potential in wheat. The Journal of Agricultural Science 145, 99. Lemaire G, Ciampitti I. 2020. Crop mass and N status as prerequisite covariables for unraveling nitrogen use efficiency across genotype-by-envi- Fischer RA, Edmeades GO. 2010. Breeding and cereal yield progress. ronment-by-management scenarios: a review. Plants 9, 1309. Crop Science 50, S-85–S-98. Martino DL, Abbate PE. 2019. Frost damage on grain number in wheat at Fletcher RJ, Cullis BR. 1988. Evaluation of chromosome substitution different spike developmental stages and its modelling. European Journal of lines of wheat (Triticum aestivum L.) for freezing injury suffered during the Agronomy 103, 13–23. stem elongation stage of development. Australian Journal of Agricultural Research 39, 111–128. Messina CD, Hammer GL, McLean G, Cooper M, van Oosterom EJ, Tardieu F, Chapman SC, Doherty A, Gho C. 2019. On the dynamic Flohr BM, Hunt JR, Kirkegaard JA, Evans JR. 2017. Water and temper- determinants of reproductive failure under drought in maize. in silico Plants ature stress define the optimal flowering period for wheat in south-eastern 1, diz003. Australia. Field Crops Research 209, 108–119. Miralles DJ, Richards RA, Slafer GA. 2000. Duration of the stem elon- Foulkes MJ, Slafer GA, Davies WJ, Berry PM, Sylvester-Bradley R, gation period influences the number of fertile florets in wheat and barley. Martre P, Calderini DF, Griffiths S, Reynolds MP. 2011. Raising yield po- Functional Plant Biology 27, 931–940. tential of wheat. III. Optimizing partitioning to grain while maintaining lodging resistance. Journal of Experimental Botany 62, 469–486. Miralles DJ, Slafer GA. 2007. Sink limitations to yield in wheat: how could it be reduced? Journal of Agricultural Science 145, 139–149. Gaju O, Reynolds MP, Sparkes DL, Foulkes MJ. 2009. Relationships between large-spike phenotype, grain number, and yield potential in spring Nuttall JG, O’Leary GJ, Panozzo JF, Walker CK, Barlow KM, wheat. Crop Science 49, 961–973. Fitzgerald GJ. 2017. Models of grain quality in wheat—a review. Field García GA, Serrago RA, Appendino ML, Lombardo LA, Vanzetti LS, Crops Research 202, 136–145. Helguera M, Miralles DJ. 2011. Variability of duration of pre-anthesis Ochagavía H, Prieto P, Savin R, Griffiths S, Slafer GA. 2017. Duration phases as a strategy for increasing wheat grain yield. Field Crops Research of developmental phases, and dynamics of leaf appearance and tillering, as 124, 408–416. affected by source and doses of photoperiod insensitivity alleles in wheat González FG, Miralles DJ, Slafer GA. 2011. Wheat floret survival as under field conditions. Field Crops Research 214, 45–55. related to pre-anthesis spike growth. Journal of Experimental Botany 62, Peltonen-Sainio P, Kangas A, Salo Y, Jauhiainen L. 2007. Grain 4889–4901. number dominates grain weight in temperate cereal yield determination: ev- González FG, Slafer GA, Miralles DJ. 2003. Grain and floret number in idence based on 30 years of multi-location trials. Field Crops Research 100, response to photoperiod during stem elongation in fully and slightly vernal- 179–188. ized wheats. Field Crops Research 81, 17–27. Peltonen-Sainio P, Peltonen J. 1995. Floret set and abortion in oat and González FG, Slafer GA, Miralles DJ. 2005. Pre-anthesis development wheat under high and low nitrogen regimes. European Journal of Agronomy and number of fertile florets in wheat as affected by photoperiod sensitivity 4, 253–262. genes Ppd-D1and Ppd-B1. Euphytica 146, 253–269. Porter JR, Gawith M. 1999. Temperatures and the growth and develop- Gonzalez-Navarro OE, Griffiths S, Molero G, Reynolds MP, Slafer ment of wheat: a review. European Journal of Agronomy 10, 23–36. GA. 2016. Variation in developmental patterns among elite wheat lines and R Core Team. 2019. R: A language and environment for statistical com- relationships with yield, yield components and spike fertility. Field Crops puting. Vienna, Austria: R Foundation for Statistical Computing. Research 196, 294–304. Rajaram S, Van Ginkel M, Fischer RA. 1995. CIMMYT’s wheat breeding Guo Z, Schnurbusch T. 2015. Variation of floret fertility in hexaploid wheat mega-environments (ME). In: Li ZS, Xin ZY, eds. Proceedings of the 8th revealed by tiller removal. Journal of Experimental Botany 66, 5945–5958. International Wheat Genetic Symposium, Beijing, China, 20–25 July, 1993. Halloran GM, Pennell AL. 1982. Duration and rate of development phases Beijing: China Agricultural Scientech Press, 1101–1106. in wheat in two environments. Annals of Botany 49, 115–121. Reynolds M, Bonnett D, Chapman SC, Furbank RT, Manès Y, Mather Hodson DP, White JW. 2007. Use of spatial analyses for global charac- DE, Parry MAJ. 2011. Raising yield potential of wheat. I. Overview of a terization of wheat-based production systems. The Journal of Agricultural consortium approach and breeding strategies. Journal of Experimental Science 145, 115–125. Botany 62, 439–452. Holzworth DP, Huth NI, deVoil PG, et al. 2014. APSIM – evolution Reynolds M, Foulkes MJ, Slafer GA, Berry P, Parry MAJ, Snape JW, towards a new generation of agricultural systems simulation. Environmental Angus WJ. 2009. Raising yield potential in wheat. Journal of Experimental Modelling & Software 62, 327–350. Botany 60, 1899–1918. Downloaded from https://academic.oup.com/jxb/article/73/12/4236/6564125 by guest on 20 July 2022 Phenological optimization of late reproductive phase for raising yield potential | 4249 Reynolds MP, Pask AJD, Hoppitt WJE, et al. 2017. Strategic crossing International bread wheat yield evaluation environments. Crop Science 43, of biomass and harvest index—source and sink—achieves genetic gains in 1698–1711. wheat. Euphytica 213, 257. Ugarte C, Calderini DF, Slafer GA. 2007. Grain weight and grain number Rousset M, Bonnin I, Remoué C, et al. 2011. Deciphering the genetics of responsiveness to pre-anthesis temperature in wheat, barley and triticale. flowering time by an association study on candidate genes in bread wheat Field Crops Research 100, 240–248. (Triticum aestivum L.). Theoretical and Applied Genetics 123, 907–926. Weir AH, Bragg PL, Porter JR, Rayner JH. 1984. A winter wheat crop Sanna G, Giunta F, Motzo R, Mastrangelo AM, De Vita P. 2014. Genetic simulation model without water or nutrient limitations. The Journal of variation for the duration of pre-anthesis development in durum wheat and Agricultural Science 102, 371–382. its interaction with vernalization treatment and photoperiod. Journal of Whitechurch EM, Slafer GA. 2002. Contrasting Ppd alleles in wheat: Experimental Botany 65, 3177–3188. effects on sensitivity to photoperiod in different phases. Field Crops Sharma RC, Crossa J, Velu G, Huerta-Espino J, Vargas M, Payne Research 73, 95–105. TS, Singh RP. 2012. Genetic gains for grain yield in CIMMYT spring bread Whitechurch EM, Slafer GA, Miralles DJ. 2007. Variability in the duration wheat across international environments. Crop Science 52, 1522–1533. of stem elongation in wheat and barley genotypes. Journal of Agronomy Slafer GA, Abeledo LG, Miralles DJ, Gonzalez FG, Whitechurch EM. and Crop Science 193, 138–145. 2001. Photoperiod sensitivity during stem elongation as an avenue to raise potential yield in wheat. Euphytica 119, 191–197. Zadoks JC, Chang TT, Konzak CF. 1974. A decimal code for the growth stages of cereals. Weed Research 14, 415–421. Slafer GA, Andrade FH, Satorre EH. 1990. Genetic-improvement effects on pre-anthesis physiological attributes related to wheat grain-yield. Field Zhao B, Niu X, Ata-Ul-Karim ST, Wang L, Duan A, Liu Z, Lemaire Crops Research 23, 255–263. G. 2020. Determination of the post-anthesis nitrogen status using ear crit- ical nitrogen dilution curve and its implications for nitrogen management in Slafer GA, Rawson HM. 1994. Sensitivity of wheat phasic development to maize and wheat. European Journal of Agronomy 113, 125967. major environmental factors: a re-examination of some assumptions made by physiologists and modellers. Functional Plant Biology 21, 393–426. Zheng B, Biddulph B, Li D, Kuchel H, Chapman S. 2013. Quantification of the effects of VRN1 and Ppd-D1 to predict spring wheat (Triticum aes- Slafer GA, Rawson HM. 1995. Base and optimum temperatures vary with tivum) heading time across diverse environments. Journal of Experimental genotype and stage of development in wheat. Plant, Cell & Environment Botany 64, 3747–3761. 18, 671–679. Zheng B, Chenu K, Doherty A, Chapman S. 2014. The APSIM- Stackhouse PWJ, Gupta SK, Cox SJ, Zhang T, Mikovitz JC, wheat module (7.5 R3008). Agricultural Production Systems Simulator Hinkelman LM. 2011. The NASA/GEWEX surface radiation budget release (APSIM) Initiative. https://www.apsim.info/wp-content/uploads/2019/09/ 3.0: 24.5-year dataset. Gewex News 21, 10–12. WheatDocumentation.pdf Sukumaran S, Crossa J, Jarquín D, Reynolds M. 2017. Pedigree-based Zheng B, Chenu K, Fernanda Dreccer M, Chapman SC. 2012. prediction models with genotype × environment interaction in multienviron- Breeding for the future: what are the potential impacts of future frost and ment trials of CIMMYT wheat. Crop Science 57, 1865–1880. heat events on sowing and flowering time requirements for Australian Trethowan RM, van GM, Ammar K, Crossa J, Payne TS, Cukadar bread wheat (Triticum aestivium) varieties? Global Change Biology 18, B, Rajaram S, Hernandez E. 2003. Associations among twenty years of 2899–2914. Downloaded from https://academic.oup.com/jxb/article/73/12/4236/6564125 by guest on 20 July 2022