Abstract
Wheat (Triticum aestivum L.) is the most widely grown of any crop and provides one-fifth of the total calories of the world's population. Since the 1960s, increases in productivity have been achieved as a result of wide-scale adoption of Green Revolution technologies. However, in spite of growing demand, the challenges of increasing production to feed an estimated world population of 9 billion in 2050 are still considerable. Due to the increased demand, it is estimated that food production must be increased by about 50% by the year of 2050. Improving wheat productivity through developing cultivars with high yield potential and with high adaptability to specific environments is the key objective in the wheat breeding programs worldwide to fill the gap between the production and the demand. The overall aims of the present study were to: (i) investigate the physiological basis of yield potential progress from 1966 to 2009 in spring bread wheats released at the International Center for Maize and Wheat Improvement (CIMMYT) in the irrigated high potential environment of NW Mexico, (ii) investigate the physiological basis of effects of the tiller inhibition Tin1A gene on ear-fertility traits and yield potential and interactions with plant density in NW Mexico and UK environments in lines of a doubled-haploid (DH) population segregating for Tin1A/non-Tin1A alleles and (iii) identify breeding targets for new cultivars with higher yield potential. Four experiments were conducted in NW Mexico at the CIMMYT research station at Ciudad Obregon. Two of these experiments studied a set of 12 historic CIMMYT spring wheat cultivars released from 1966 to 2009 in 2008/9 and 2009/10. The other two experiments examined selected lines from a doubled-haploid (DH) population derived from a cross between CIMMYT spring wheat L14 and UK winter wheat Rialto contrasting for the presence/absence of the TinlA allele for tiller inhibition and their interaction with seed rate in 2008/9 and 2009/10. In addition, two other field experiments were conducted in the UK, one in 2008/09 at KWS UK Ltd in Thriplow, Hertfordshire and one in 2009110 at the University of Nottingham Farm, Sutton Bonington campus, Leicestershire. The plant material for both of these experiments was selected lines from the CIMMYT spring wheat advanced line Ll4 (+Tin1A allele) x UK winter wheat Rialto (-TinlA allele) DH population and the Rialto parent. In the experiment at Thriplow in 2008/09 the DH lines were examined at one seed rate and in the experiment at Sutton Bonington in 2009/10 at two seed rates. At the CIMMYT site in 2008/9 and 2009/10, a randomized complete block design was implemented with four replications for the experiments examining the CIMMYT wheat historic releases and a split-plot randomised complete block design with three replications was implemented for the experiments examining the +/- Tin1A DH lines, with three seed densities (50, 150 and 450 seeds per square metre); seed rates were randomized on main plots and eight genotypes randomized on sub-plots. At the UK site, in the KWS experiment, 24 DH lines (12+Tin1A allele) and (12-TinlA allele) from the L14 x Rialto population were used. There was only one seed rate (300 seeds m-2) and a completely randomised design in three replicates was implemented. The same 24 DH lines were examined in the experiment at the SB site, at two seed rates (40 and 320 seeds m-2) in a split plot randomised complete block design in three replicates. Seed rate was randomized on main plots and DH lines were randomized on sub-plots. In all experiments examining the DH lines of the Ll4 x Rialto population, lines were selected in pairs so that the two groups of +Tin1A and -Tin lA lines were approximately balanced for flowering time and plant height, i.e. every +TinlA line has a non-Tin1A pair with similar height and flowering date. Plots were sampled for destructive measurements of dry weight and DM partitioning and ear-fertility traits at four stages in the historic experiments at (GS3l, GS39, GS61+7d and at maturity) and at two stages in the DH population experiments (GS61+7d and at maturity). The water soluble carbohydrate (WSC) content of the stems plus attached leaf sheaths was also measured at GS61+7d and at maturity. In the historic experiments, at GS 61+14 days, a degraining treatment was implemented by removing all spikelets from one side of the ear (i.e. ca. 50% of the spikletes) in the histories experiment. Non-destructive measurements were taken for stomatal conductance, canopy temperature, fractional photosynthetically active radiation (PAR) interception and normalized difference vegetative index at various dates both pre- and post-anthesis in the historic experiments. In the experiments examining the set of 12 historic CIMMYT spring wheat releases, results showed that from 1966 to 2009 the linear rate of genetic gain in yield potential was 32 ha-1 yr-1 (0.59 % yr-1) (r = 0.76. P = 0.01). Yield progress was primarily associated with harvest index (percentage above-ground DM as grain DM) in the period from 1966 until about 1990 increasing from 43% to 49%, but deceased with year of release thereafter. A non-linear genetic gain in AGDM was evident over the 43-yr period with AGDM increasing from about 1990 from which point it increased rapidly to 2009. There was no association between genetic progress in grain yield and grain number per m2 in this set of 1 cultivars; a small increase in ears per m2 was counteracted by a decrease in grains per ear. However, grain weight tracked the improvement in yield potential over the 43-year period with a linear increase of 0.23 mg yr-1. No change was found in rachis length with plant breeding; however, number of fertile spikelets per ear decreased since about 1990 and was associated with the decrease in grains per ear. There were statistically significant differences in above-ground DM production at all growth stages and a tendency to produce more biomass during the GS31 to GS61+7d phase with year of release. No differences amongst cultivars were found in the amount of radiation intercepted by the whole canopy from GS3l to GS6l+7 days. Although not conclusive, since Bacanora was an exception to the trend and radiation-use efficiency (above-ground biomass per unit PAR interception; RUE), there was a tendency for RUE to increase with year of release which was consistent with a positive association with crop growth rate (above-ground DM per m2 per day; CGR) and the trend for an increase in biomass accumulation during the stem-elongation phase with plant breeding. Although there was a trend for an increase in biomass accumulation from GS31 GS61+7d this was counteracted by a decrease in ear DM partitioning so that ear DM per m2 at GS61+7d and grains per m2 did not change with plant breeding. Results showed that the improvement in the individual grain weight from 1966 to 2009 in this set of cultivars was associated with improvements in the grain filling rate from 1966 to ca. 1990 and in the duration of grain filling from ca. 1992 to 2009. Averaging across years, there was a significant positive association between post-flowering canopy-temperature depression and grain yield. Fractional PAR interception by the canopy layers of the ear, flag leaf and the penultimate leaf was increased with year of release since about 1990. This increase in the fractional interception of PAR correlated significantly with the grain weight and grain yield amongst the 12 cultivars. Grain growth of the cultivars in this historic set was generally sink limited rather than source limited. There was no change in source-sink balance as indicted by grain growth responses to the degraining treatment with year of release. The percentage increase in grain weight in the manipulated ears ranged amongst cultivars from 0.5 to 13.2%, but differences between cultivars in the response to degraining were not statistically significant and the responses were not correlated with year of release. The results also indicated that potential grain weight has increased with plant breeding over the 43-year period, since the final grain weight of the grains in the degrained ears increased linearly with year of release. Overall the contribution of stem WSC to grain DM growth in the current study was relatively low (4 - 18%) which is consistent with the hypothesis that under the high radiation, modern cultivars are more likely to be sink than source limited. Results of TinlA experiments showed that under the UK environment, TinlA lines produced more grain yield than non TinlA lines under high seed densities. The main yield component explaining this was grains per m2. TinlA lines had a longer rachis, a wider ear, and more total and fertile spikelets per ear than non-TinlA lines. Non-TinlA lines produced more ears per plant than TinlA lines under low seed density; however, both groups of lines had similar values of ears per plant under high seed density. Non-TinlA lines produced heavier grains than TinlA lines either under high or low seed density, and individual grain weight was not affected by the increase of seed density. Under the Mexican environment, there was a slight increase in the yield of non-TinlA compared to TinlA lines, but only at low seed density. This increase was attributed mainly to heavier individual grains in non-TinlA than TinlA lines. Although Tinl A lines produced fewer ears m-2 than non-Tin 1A lines (-13.4, -12.5 and -11.3%, under seed densities of 50,150 and 450 seed m2, respectively). TinlA lines produced more grains m-2 at all seed densities and this resulted from a longer rachis, more total and fertile spike lets per ear and thus more grains per ear. There was a reduction in the grain yield in non-Tin1A lines with increasing seed density. However. TinlA lines continued to increase grain yield with increasing seed density, so that the two groups of lines yielded similarly at high plant density. This result therefore supported the hypothesis that Tin1A lines yield relatively better under high seed density than non Tin1A lines and may have a higher economic optimum seed density. Overall, in contrast to in the UK experiments, it seemed that under the high temperature and radiation environment in NW Mexico, possessing the Tin1A allele may not be an advantage compared to the non-TinlA allele even under high seed rates. Overall, several target traits were identified for future improvement of yield potential in CIMMYT plant breeding and worldwide in the present study. Genetic progress in grain yield potential in the CIMMYT spring wheat program from 1966 to 2009 of 37 kg ha-2 yr-1 was positively associated with above ground biomass and grain weight. The amount of radiation intercepted during the stem-elongation phase did not change with breeding; however, there was an apparent tendency for RUE to be increased with year of release. Potential grain weight also tracked increases in final grain weight. Therefore target traits as selection criteria in wheat breeding programs for yield potential should include a combination of traits favouring enhanced RUE during the stem elongation and potential grain weight. In addition, in winter wheat in the UK restricted tillering with the introduction of the TinlA allele offers scope to increase grains m-2 and grain yield, associated with more fertile spikelets per ear and grains per ear, in wheat crops with established plant densities in the range ca. 150 - 200 plant m-2.