The genetic basis for inflorescence variation between foxtail and green millet (poaceae)

Abstract

Grass species differ in many aspects of inflorescence architecture, but in most cases the genetic basis of the morphological difference is unknown. To investigate the genes underlying the morphology in one such instance, we undertook a developmental and QTL analysis of inflorescence differences between the cereal grain foxtail millet and its presumed progenitor green millet. Inflorescence differences between these two species are the result of changes in primary branch number and density, spikelet number, and bristle (sterile branchlet) number; these differences also account for inflorescence variation within the clade of 300+ species that share the presence of bristles in the inflorescence. Fourteen replicated QTL were detected for the four inflorescence traits, and these are suggested to represent genes that control differences between the species. Comparative mapping using common markers from rice and maize allowed a number of candidate genes from maize to be localized to QTL regions in the millet genome. Searches of regions of the sequenced rice genome orthologous to QTL regions on foxtail millet identified a number of transcription factors and hormone pathway genes that may be involved in control of inflorescence branching.

Figure 1.—

Developmental stages and mature morphology of the inflorescence in green millet (S. viridis). (A) Inflorescence meristem, with sheathing leaf base. (B) Inflorescence meristem initiating primary branch primordia. (C) Inflorescence with many primary branches, each of which has produced up to four more orders of branching. The higher-order branch primordia at the distal end of the inflorescence are differentiating into spikelets and bristles. (D) A piece of an inflorescence, showing a number of primary branches with differentiating spikelets and bristles. (E) A mature inflorescence showing spikelets and bristles. (F) A mature plant of green millet, with multiple tillers and axillary branches, each tipped with an inflorescence. (G) A primary branch at maturity showing a mature spikelet with several bristles and two rudimentary spikelets. M, meristem; 1, primary branch; spk, spikelet; br, bristle; rdspk, rudimentary spikelet. Bars, 50 μm (A and B), 500 μm (C), 50 μm (D), 5 mm (E), 7 cm (F), and 500 μm (G).

Figure 2.—

Developmental stages and mature morphology of the inflorescence in foxtail millet (S. italica). (A) Inflorescence of foxtail millet with many primary branches, each of which has initiated further orders of branching. (B) Distal end of inflorescence, showing initiation of primary and secondary branches. (C) Basal end of inflorescence showing initiation of primary, secondary, and tertiary branches. (D) Primary branch late in development showing many orders of branching (>8). Spikelets and bristles are just starting to differentiate. (E) Mature inflorescence of foxtail millet with very tightly packed spikelets and short bristles. (F) A plant of foxtail millet, with three tillers but no axillary branches. 1, 2, and 3, primary, secondary, and tertiary branches. Bars, 500 μm (A), 50 μm (B–D), 2 cm (E), and 15 cm (F).

Figure 3.—

Electron micrograph and line-drawing interpretation of the orders of branching in a primary branch of Setaria. In early development, each order of branch axes has approximately one spikelet and one bristle; 1, 2, 3, 4, and 5 denote primary, secondary, tertiary, quaternary, and quinternary branch axes, respectively; circles represent spikelets; asterisks represent bristles. Bar, 50 μm.

Figure 4.—

Histograms showing distribution of F_2:3 means (untransformed) for the four phenotypic traits.

Figure 5.—

Differences in phenotypic means (untransformed) for green and foxtail millet parents at high and low planting density.

Figure 6.—

Transgressive segregation in trials 2 (high density) and 4 (low density) as evidenced by the greater range of hybrid population means vs. the parental means (data natural log transformed). PBN, primary branch number; PBD, primary branch density; SPK, spikelet number per primary branch; BR, bristle number per primary branch.

Figure 7.—

Millet genome map, showing all replicated high- and low-density QTL at the chromosome significance level of P < 0.05. Replicated QTL that are also significant at the genome level of P < 0.05 are indicated by an asterisk. High-density trials are indicated by solid bars and low-density trials by open bars. Length of the bar denotes the 1-LOD confidence interval. Joint QTL, estimated from the individual trials and significant at the genome level P < 0.05, are indicated by dashed lines. Arrows within the confidence interval indicate the position of the highest LOD score; direction of arrow indicates direction of effect, with left indicating a decrease in the phenotypic trait value and right an increase. Markers that were added to enable comparative mapping with the maize genome are indicated in blue; original markers used in the QTL analysis are in black. Approximate gene positions are indicated by bd1, branched silkless1; ts4, tasselseed4; ba1, barren stalk1; zfl1, zea leafy1; bif2, barren inflorescence2; tb1, teosinte branched 1.