India project summary June 2004 - June 2005
The project addresses improvement of animal productivity in crop-livestock systems in the driest rainfed parts of India where pearl millet (Pennisetum glaucum) is the only reliably productive cereal. Pearl millet stover is a major component of ruminant diets in these production systems. The project aims to use both marker-assisted and conventional plant breeding to genetically increase the nutritive value of pearl millet stover. In the present project reporting period the following three objectives were addressed a) produce hybrid parent lines with enhanced stover quality suitable for use in commercial hybrid seed production, b) determine the effects of individual genomic regions (so called "quantitative trait loci" or QTLs) controlling stover quality, and combinations of these in enhanced commercial hybrid parents, on in vitro stover quality and in vivo animal production, and c) promote the use of improved parental lines to public and private seed companies. The project commenced in June 2004 and in August 2004 stakeholders from various national and international public and private organizations were convened for the project planning meeting to discuss the objectives and activities of the project. There was strong support for the project and a consensus that the project has the potential to improve smallholder livestock production in India.

The 1st project year saw successful linking of laboratory quality traits for pearl millet stover with livestock productivity measurements. The in vitro true digestibility of stover determined after 48 hrs of incubation in rumen inoculum was found to be the single most important laboratory trait related to in vivo measurements in sheep that identify with animal performance, such as organic matter digestibility, organic matter intake, digestible organic matter intake and nitrogen balance. Other laboratory traits related to livestock-pertinent measurements were stover protein content, cell wall (NDF) content, cell wall digestibility and metabolizable energy content. Near Infrared Spectroscopy (NIRS) was successfully calibrated for these stover laboratory traits using pearl millet mapping population for ICMB 841 863B.

An improved pearl millet linkage map was constructed based on the ICMB 841 x 863B mapping population using newly developed SSR, EST-SSR and TRAP markers, which facilitated subsequent mapping of reliable QTLs for use in marker-assisted breeding. Major QTLs for several stover quality traits were identified on different pearl millet linkage groups from two different tester backgrounds and three different plant stover fractions. Consistent QTLs from this analysis will be used for marker-assisted backcross breeding. SSR, EST-SSR, and TRAP markers were screened against four recurrent parents and four donor parents to identify genomic regions for which marker polymorphism is sufficient for marker-assisted selection using QTL-flanking marker loci. Marker combinations for the different trait donor line x elite line cross combinations were selected for use in the marker-assisted backcross breeding program. New polymorphic markers were identified from this experiment and these will be included in the existing pearl millet linkage map. BC1F1 seeds were generated for different parent by parent combinations and a few hybrids were chosen for generation advancement. In addition, several late-generation breeding lines having introgressed stover quality QTLs from donor 863B in the genetic background of elite hybrid parental line ICMB 841 were identified.

In 15 commercial hybrids, stover protein content ranged from 6.1 to 8.1% (LSD = 0.12%), stover in vitro digestibility from 43 to 47% (LSD = 2.4%), and stover metabolizable energy content from 6.0 to 6.7 mega joule per kg (LSD = 0.4%). Digestible stover yield (stover yield times stover in vitro digestibility) per ha ranged from 1.0 to 1.8 tons. Digestible stover yield and grain yield were positively associated (r = 0.54; P = 0.04) suggesting that substantial cultivar-dependent variation in stover quality exists that can be exploited without detriment to grain yield.

The aim of this project is to improve the animal productivity of crop-livestock systems in India by increase the nutritive value of pearl millet stover. The vast majority of the ~10 m ha area on which pearl millet is cultivated in south Asia is rainfed, and dependant upon in-season precipitation. Since pearl millet is the only reliable productive cereal in the rainfed areas of northwestern India, and pearl millet stover accounts for over 30% of fodder resources available in this region, improvement of pearl millet stover yield and quality by genetic means is perhaps the only practical way to improve the productivity and economic returns of this rainfed crop-livestock production system without substantial public and/or private investments in irrigation and other infrastructure. This project was launched on July 2004 and a good progress has been made on various fronts. The following objectives were met during the second year of the project period 2005-2006: A) a better understanding of the genomic regions (Quantitative Trait Loci -- QTLs) controlling stover quality traits of pearl millet and exploiting these genomic regions to improve elite parental hybrid parental lines; and, B) realizing the positive effects of selection in pearl millet full-sib progenies for stover feeding value.

A full set of mapping population progenies based on the cross (ICMB 841 863B) was used in a QTL mapping exercise to get a better understanding of the behavior of the genomic regions controlling several stover quality traits in pearl millet. Near Infrared Spectroscopy (NIRS) was successfully calibrated for estimating all of the laboratory-based stover quality traits under study. Consistent QTLs were identified from the full set of the mapping population for in-vitro organic matter digestibility (IVOMD), metabolizable energy (ME), sugar content, and nitrogen content (NDM) as well as for yields of stover and grain. The results of this exercise agreed with the previous findings, when a subset of the same mapping population was used to map stover quality QTLs. Several QTLs identified in this study were consistent in their expression and controlled significant portions of observed phenotypic variation for more than one trait (although in many cases these traits were auto-correlated, so this was expected). Another important outcome of this mapping exercise was determination that a genomic region controlling a substantial portion of observed variation on drought tolerance was also responsible for controlling stover quality traits, and that the same parent (863B) contributed favourable alleles in this genomic region for both of these traits. Thus, transferring this stover quality QTL from 863B to ICMB 841 and other elite hybrid parental lines should also improve the terminal drought tolerance of their hybrids. Considering the geographical regions where dual-purpose pearl millet genotypes are grown in India, simultaneous improvement of these traits would be highly beneficial.

Marker-assisted backcrossing of stover quality QTLs from donor parents into several elite hybrid parental lines is now underway. BC2F1 progenies carrying single-QTL introgressions were identified in different trait donor elite parent combinations using polymorphic QTL-flanking molecular markers. The selected progenies will now be advanced to the BC3F1 generation.

Recurrent selection experiment was carried out using 256 full-sib progenies produced by plant - plant crosses of typical plants in widely-adapted pearl millet variety ICMV 221. They were classified into four groups based agronomic characters, grain yield and stover yield, and these groups random-mated to produce: 1) a high grain yield experimental variety, 2) a dual-purpose experimental variety, 3) a high nitrogen experimental variety, and 4) a high digestibility experimental variety. These four experimental varieties and a sample of the unselected control variety (ICMV 221) were then used for field and animal feeding experiments. Results of the first cycle of recurrent selection showed the presence of considerable genetic variation between the experimental varieties for yields of grain and stover, as well as for stover quality traits and animal performance. Substantial ranges were observed for grain (1.5-fold difference) and stover (1.8-fold difference) yields, stover protein content (4.3% to 8.6%), and yield of digestible stover (which is the product of stover yield and stover in vitro digestibility and ranged from 1132 to 2388 kg ha-1). Total biomass yield was highest in the dual-purpose experimental variety followed by the control and the grain types. Stover yields differed significantly between the varieties but grain yield and harvest index did not. The fact that no significant difference was found for grain yields between the five varieties suggests that genetic enhancement of stover yield and quality would not necessarily adversely affect grain yield in this relatively "uncivilized" crop. Grain yield was numerically lowest in the high digestibility variety. In terms of combined grain and stover yield, the dual-purpose variety appeared preferable. Two stover quality traits were directly targeted by the selection process, high nitrogen content and high digestibility, and the results observed were consistent with the intentional design of the experimental varieties. Stover of the high nitrogen variety indeed had the highest nitrogen content; while the high digestibility variety had the highest in vitro digestibility and lowest cell wall (NDF), cellulose (ADF) and lignin contents. If the control variety is taken as the point of departure (54.4%), one cycle of selection has resulted in an increase in stover digestibility of 3.1 percentage units, which seems remarkable.

Results indicated that considerable scope exists within elite open-pollinated pearl millet varieties for selection of experimental varieties having contrasting grain and stover characteristics.

Mid-term Project Review and Planning Meeting:
This project was launched on July 2004 had reached the halfway mark on November 2006. During 20-22 November 2006, nine public and private-sector pearl millet research and development specialists from India joined external reviewer Prof. Peter Langridge and 15 members of the ICRISAT and ILRI teams based at Patancheru for a mid-term review of progress and problems during the first 29 months of the on-going pearl millet stover improvement project, and to plan research and dissemination activities for the remaining 31 months. Stakeholders from national agricultural universities, cooperatives and private sector seed companies participated actively in this project review and planning meeting. Participants expressed general satisfaction with the progress achieved to date. Constructive comments and suggestions were noted and a few activities have been modified accordingly.

Marker-Assisted Breeding:
During the reporting period June 2006-June 2007, quantitative trait locus (QTL) introgression lines have been advanced by two generations from BC2F1 to BC4F1 through marker-assisted backcross breeding. BC4F1 progenies harbouring single-QTL introgressions were obtained in different trait donor elite recurrent parent line combinations using QTL-flanking polymorphic SSR marker allele data. The selected QTL introgression progenies will be advanced to BC5F1/BC4F2 generation during the coming year.

Simultaneously, progenies of segregating BC3F2 populations were raised and screened to identify QTL introgression homozygotes for various stover quality trait QTLs in several different elite recurrent parent line genetic background. The selected BC3F2 plants were then selfed and crossed with pollen from an appropriate restorer line and the resultant hybrid seeds were harvested for multi-location testing during the 2007 rainy season, while the selfed seed (BC3F3 generation) was harvested for future use in developing cytoplasmic male-sterile counterparts of the stover quality QTL introgression lines that produce hybrids having improved performance compared to their recurrent parent during the 2007 rainy season trials.

NIRS-based Recurrent Selection:
During the reporting period the first livestock feeding trial was conducted to compare livestock performance (live weight gain ) in sheep fed pearl millet straw samples (of 500 kg each) produced during the 2006 rainy season from several different versions of released pearl millet composite ICMV 221. These versions of ICMV 221 had been reselected for grain and stover yield and quality parameters using NIRS-predicted estimates of full-sib progeny straw digestibility, metabolizable energy content based on straw samples produced in multi-location field trials. The dual-purpose version of ICMV 221, which has higher yield of NIRS-predicted digestible dry matter, performed best in this feeding trial. Seed of this variety was multiplied and supplied to ILRI collaborators in Ethiopia (where the original version of ICMV 221 is being considered for cultivar release) for testing during the 2007 rainy season. Similar full-sib progeny trials were conducted in the 2006 rainy in the genetic backgrounds of two pearl millet composites adapted to the arid zone of northwestern India. The trial sets conducted at ICRISAT-Patancheru were successful, but mid-season drought stress at trial locations in Rajasthan prevented us from generating the required field data sets from those sites. Parental progenies for reselected versions of these two additional composites were chosen based on the data sets available from Patancheru, and the progenies recombined in the 2006/07 off-season nursery to produce seed sufficient for livestock feed trial straw sample production in the 2007 rainy season.

Objective 1: To produce hybrid parent lines with enhanced stover quality suitable for use in commercial hybrid seed production.

Quantitative trait locus (QTL) introgression lines of BC4F1 were advanced to produce BC4F2-BC5F1-BC5F2-BC6F1 generations through marker-assisted breeding. Progenies of segregating BC4F2/BC5F2 generations were screened to obtain QTL homozygote lines for various stover quality traits. Selected lines in BC4F2/BC5F2 were pollinated by appropriate restorer line and the resultant hybrid seeds were harvested for multi-location testing. Improved hybrid versions derived from BC4F2 have been tested in replicated plots at Patancheru and seed multiplication is in progress. Ten improved BC3F2 hybrids for stover quality traits were evaluated in small plots at Patancheru and Hisar. The performance of the improved hybrids for stover quality traits (metabolizable energy, sugar content, organic matter digestibility) and yield of grain and stover were better than that of their corresponding original versions. Seeds of the improved-hybrid versions have been multiplied for livestock feeding trials at multi-locations. Forty-five testcross hybrids produced by crossing 15 lines to 3 diverse pollinators were evaluated in Kharif 2006 and 2007 at different locations in India. Within and across-environments analysis showed good general combining ability for grain, and stover yields and stover quality traits. Significant and positive specific combining ability for nitrogen, sugar, grain and stover yields indicated possibilities to develop improved hybrids by exploiting heterosis.

We have completed two cycles of full sib (FS) recurrent selection for various stover quality traits in widely-grown open-pollinated variety ICMV 221 and in the Barmer (Rajasthan) landrace restorer population and one cycle in the Jakharana x SRCII (Rajasthan), but with selection done in two different locations. In ICMV 221, divergent selection for and against stover digestibility and nitrogen produced significantly (P < 0.10) different experimental varieties for both traits (in vitro analyses), averaged over three test environments, plus a dual-purpose (grain plus stover yield and quality) experimental variety that has significantly better stover quality than the original ICMV 221. Response to selection for stover quality in the Barmer population has not been demonstrable in two years of testing, probably because the variation among FS for quality in the original population was not significant. We completed one year of testing in the Jakharana x SRCII population, but results indicate that the C1 bulk, the high digestibility and high intake experimental varieties had significantly (P < 0.10) higher values for in vitro digestibility, metabolizable energy and sugar content.

Two of the ICMV 221 cycle 2 experimental varieties, plus the cycle 2 bulk have significantly better stover quality, and a greater yield of digestible stover dry matter than the original C0 bulk of the variety, without a change in either grain or stover dry matter yield, based on two years of testing. An in vivo stover quality comparison (using sheep as the test animal, with the millet stover fed without any supplementation) of the original C0 bulk, the C2 dual purpose variety and check hybrid HHB 67. The data indicated that the C2 dual purpose variety was better at maintaining animal weight (a daily weight loss of 18 gm per day per animal) that was the original C0 bulk (a loss 29 gm per day per animal) and very much better than the check hybrid (a daily weight loss of 68 gm per day per animal). Multiplying these figures by the numbers of animals in a farmer's herd and extrapolating this over the entire dry season, there would very likely be an incentive for a farmer to consider the improved version of ICMV 221.

Improved (C1) as well as original (C0) and versions of ICMV 221 was transferred to Ethiopia and used for initial seed multiplication in June 2008.

Objective 2 was decided against given the project's mid-term review (2006).

Objective 3: Promote the use of improved parental lines to public and private seed companies.

Variations in food-feed traits were investigated in ten popular commercial pearl millet hybrids grown at Patancheru in 2007. Stover fodder quality traits analysed were nitrogen, in vitro digestibility and metabolizable energy content. Highly significant (P < 0.0001) cultivars-dependent variations were observed for grain yields (range 2 855 to 4 216 kg/ha), stover yields (range 3 758 to 4 933 kg/ha), stover nitrogen (0.62 to 1.10%); stover in vitro digestibility (37.6 to 46.7%) and stover metabolizable energy (5.26 to 6.88 MJ/kg). The cultivars with the highest in vitro digestibility and metabolizable energy content in the straw had also the highest grain yield and the 3rd highest stover yield. No trade-offs were observed between stover in vitro digestibility and metabolizable energy contents and grain and stover yields. Out of the 10 cultivars GK 1044 with the highest stover digestibility (46.7%) and metabolizable energy (6.88 MJ/kg) and cultivars MLBH 267 with the lowest stover digestibility (37.6%) and metabolizable energy (5.26 MJ/kg) were selected to be re-grown on large plots in Kharif 2008 to be tested in vivo with sheep as sole feed and as major diet ingredient at the end of 2008, beginning of 2009. Parental lines for GK 1044 are identified by the seed company and will be investigated.