Paleo-Green Revolution for rice

A central element of the Green Revolution was the widespread adoption of semidwarf rice cultivars (SRCs) that more than doubled worldwide rice production (1). The high-yield potentials of modern SRCs are attributed primarily to their improved harvest index, lodging resistance, and responsiveness to high inputs (primarily nitrogen and water) (1–3), contributing to their adoption in irrigated areas that occupy 57% of world rice lands (4). The short stature of SRCs is caused by sd1, a mutant allele in the gene encoding a key enzyme (gibberellin 20-oxidase) functioning in biosynthesis of the plant hormone gibberellic acid (GA) that greatly reduces content of the bioactive molecule GA₁ (5, 6).

In PNAS, Asano et al. (7) show that the SD1 locus (where the sd1 mutation resides) has been a target of human-mediated selection since prehistoric times. In a population of backcross inbred lines from a cross between an Oryza sativa paddy rice (subspecies japonica) variety Nipponbare and an upland rice (ssp. indica) variety Kasalath, a quantitative trait locus, qCL1, accounts for 20.9% of phenotypic variation in culm length. Study of 5,000 plants segregating at the qCL1 locus shows its phenotypic effects to result from at least two genes, with one (qCL1a) delimited to a 336-kb genomic region that included the SD1 locus and at least 40 other genes. The Nipponbare and Kasalath SD1 alleles differ by 2 aa-altering nucleotides. Plants transgenic for the respective alleles largely recapitulate the phenotypic difference in culm length. The Kasalath allele (SD1-GR) also has significantly higher catalytic activity than the Nipponbare allele SD1-EQ in the 13-hydroxylated pathway of GA synthesis leading to GA₁, which is consistent with its greater culm length.

Levels and patterns of genetic diversity among primitive rice cultivars and wild relatives at the SD1 locus strongly suggest that the japonica allele of qCL1a was selected during domestication, perhaps 10,000 y ago. All 40 accessions of a wild progenitor, O. rufipogon, as well as most ssp. indica accessions contain SD1-GR, whereas all 16 diverse japonica accessions contain the Nipponbare allele (SD1-EQ). A 4-kb region encompassing the entire SD1 sequence shows 98% reduction in genetic diversity in ssp. japonica relative to O. rufipogon, a 10-fold greater reduction than in random genomic fragments (ruling out a population bottleneck). Indeed, diversity is reduced in subspecies japonica but not indica in a 404-kb region surrounding the locus, consistent

The SD1 locus (where the sd1 mutation resides) has been a target of human-mediated selection since prehistoric times.

with a selective sweep during japonica domestication.

Semidwarfism and Environmental Adaptation

If SD1-EQ was so advantageous as to be selected to near-fixation in paddy (japonica) rice, why does it occur at very low frequency in upland (indica) rice? At other genetic loci, substantial differentiation between the two subspecies is well-known (8). Multilocus indica and japonica allele complexes are maintained by strong postreproductive barriers such as hybrid sterility and hybrid breakdown (9). Asano et al. (7) suggest that the low introgression of SD1-EQ from japonicas into indicas is probably caused by a reproductive barrier between the two subspecies in this particular region of the genome. Such a barrier might explain why not only qCL1a but also qsh1, another domestication gene only 1.9 Mbp away, are confined to japonicas.

Although a reproductive barrier may be a contributing factor, we suspect that the answer to this question may also be related to the observation that most modern semidwarf (sd1) rice cultivars perform poorly under the rainfed ecosystems of Asia and Africa that occupy ∼30% of world rice lands, where drought, submergence, low soil fertility, and other abiotic stresses are commonplace (10). Because GAs play key regulatory roles in plant growth, development, productivity, and adaptation to external stresses, the observed phenotypic effects of either quantitative variants such as SD1-GR and SD1-EQ or qualitative mutants such as sd1 (5) are likely to be a result of changed regulatory properties in GA signaling pathways. It may be important to further investigate the specific properties of the different alleles at the SD1 locus and how these properties alter GA regulated downstream genes/pathways that contribute to adaptation to different environments. If sd1 mutations were directly responsible for poor adaptation to rainfed upland conditions, then both sd1 itself and its qCL1a allele might be selected against in indica rice. Indeed, failure of the prior Green Revolution to adequately address food security in regions dependent on low-input production has been an important factor in asserting the need for a second Green Revolution (9, 11), particularly one that is aimed at rainfed areas of Asia and Africa. If our hypothesis is correct, such a second Green Revolution may need to take a fundamentally different approach from the first.

Identifying QTL Alleles

Beyond its agricultural importance, the isolation of qCL1a exemplifies how the study of discrete phenotypic mutants may expedite subsequent isolation of more subtle QTL alleles, a long standing idea (12) that may accelerate progress at identifying specific genes responsible for much of the genetic variation important to agriculture, evolution, and medicine. Although both the qualitative and quantitative alleles at this particular locus happen to be agronomically important, activities in many crops and botanical models over the past decade have led to the generation of large panels of mutants well-suited for reverse genetic approaches to gene functional characterization (13). The case of qCL1a suggests that en masse mapping of phenotypic mutants by forward genetics or resequencing (10) coupled with massively parallel resequencing in diversity panels may quickly reveal both subtle and qualitative mutants implicated in domestication and/or improvement of many crops.