Post by Admin on Mar 30, 2023 20:17:06 GMT
Convergent adaptation of bread wheat to early flowering across Eurasia
To further characterize these adaptation-associated genes, we focused on genes relevant to flowering time because flowering time is agriculturally important for crops and commonly deemed the most critical trait determining plant adaptation46. Remarkably, we found that the gene Ppd-D1 exhibited convergent adaptation to early flowering and best showcased local adaptation of bread wheat. Ppd-D1 on chromosome 2D is the primary determinant of photoperiod response in bread wheat. Dysfunctional Ppd-D1 exhibits photoperiod insensitivity and early flowering phenotypes, which is crucial to the adaptation of bread wheat to global environments47. A total of three loss-of-function alleles of Ppd-D1 have been identified so far in wheat populations (Fig. 4d), two of which are causal genetic variants, including a ∼2kb deletion at upstream, and a 5-bp deletion in gene exons47,48. The XP-CLR analysis between IA and SH populations identified selective footprints on Ppd-D1, predicting an increased frequency of causative alleles of Ppd-D1 in the SH population (Fig. 4e). However, the two causative alleles did not exist in SH landraces (Fig. 4f and Supplementary Fig. 35). Instead, we found a novel stop-gain mutation of Ppd-D1 in the SH landraces, particularly enriched in the population from the Tibetan Plateau (Fig. 4f and g). We speculated that the stop-gain allele helped adapt bread wheat to the short growing season in high-altitude and low-temperature areas (>3,000 m). To test the hypothesis, we divided the SH landraces into high-altitude and low-altitude subpopulations and performed XP-CLR analysis (Supplementary Table 29). The result showed that the XP-CLR score (99.75% quantile) on Ppd-D1 became more significant when compared with the score (97.61% quantile) between IA and SH landraces, indicating Ppd-D1 is involved in high-altitude adaptation (Fig. 4e and Supplementary Fig. 36). Furthermore, we found a strong correlation between the allele frequency of the stop-gain mutation and the average altitude of subpopulations from SH landraces (r2 = 0.778, Fig. 4h), showing the causal effect of the stop-gain mutation in the high-altitude adaptation of bread wheat. Taken together, the three causative alleles of Ppd-D1 complement each other in geographic distribution, with the stop-gain mutation in South Asia, the ∼2kb deletion in East Asia, and the 5-bp deletion in Europe (Fig. 4g, i, and j), illustrating a highly diverse but convergent adaptation of bread wheat across Eurasia through its changing flowering time.
Population size fluctuation of wheats
Compared with cultivars, crop wild relatives has received relatively little attention from the evolutionary perspective4. To decipher the population dynamics of wild wheats, we reconstructed the history of effective population size (Ne) of Triticum-Aegilops species using SMC++29. We found that Ne of Aegilops subspecies, including strangulata, tauschii (Ae. tauschii ssp. tauschii, DD), and speltoilds (Ae. speltoides, BB/SS), appeared to decline constantly in the last 100 thousand years (Supplementary Fig. 37). In contrast, all the Triticum species experienced a marked population size expansion during the Holocene. For bread wheat and early domesticates, such as domesticated einkorn, domesticated emmer, and free-threshing tetraploids, the population growth may reflect the cultivation history of these populations6; whereas for wild wheats that had never been domesticated, such as wild einkorn (T. monococcum ssp. aegilopoides, AA), urartu (T. urartu, AA), and wild emmer (Fig. 1 and Supplementary Fig. 37, 38), such population growth may result from their mixed growing with early domesticates for thousands of years24,49. Strikingly, we found a ubiquitous population contraction right after the population growth for all the Triticum populations except for modern cultivars of bread wheat. The population decline occurred sequentially with ploidy levels—the diploids came first, then the tetraploids, and lastly, the hexaploid landraces. Intriguingly, the rise and fall of Ne of diploids, tetraploids, and hexaploids complemented each other in the Holocene timeline even without impact from drastic climate change of glacial periods (Supplementary Fig. 37). Archeological studies showed that domesticated einkorn and domesticated emmer thrived since the Neolithic Age until they were gradually replaced by durum wheat (T. turgidum ssp. durum, AABB), spelt, and bread wheat during the Bronze Age (∼5,000 BP - ∼3,000 BP)6,24. The Ne fluctuation of wheats coincides with the shifts of human food choice from einkorn and emmer wheat to bread wheat. Despite largely being a natural evolutionary process, the population size decline of wild wheats is disturbing—the Ne of diploids and tetraploids in Triticum was reduced by 81.70% in the past two thousand years (Fig. 5a).
Fig. 5
The population size fluctuation of wheats from the past to the future.
a, Holocene population dynamics in wheats. The top of this figure depicts the Ne for seven populations, and the bottom of this figure is the relative Ne proportion of each population. b, Genetic offset (GO) of bread wheat landrace based on 2040-2060 RCP8.5 and 2080-2100 RCP8.5 projections. c, Genetic offset of bread wheat landrace in six geographical regions, corresponding to b. The median and quartiles with whiskers reaching up to 1.5 times the interquartile range are shown in boxplots. d, Species distribution models (SDMs) projected the geographical range of wild emmer and strangulata populations in the present and future (2040-2060 and 2080-2100). Red dots pointed to the location of the samples in VMap1.1 and the USDA website (https://npgsweb.ars-grin.gov/gringlobal/search). The green shaded areas are suitable predicted regions for planting.
Rapid climate change is likely to impact the biodiversity of wheats profoundly50. To evaluate the adaptive capacity of Triticum-Aegilops species, we conducted biogeographical modeling to predict the response of wheats to the future climate. For bread wheat, we used a tree-based machine learning approach, gradient forest, to model allele frequency of genome-wide SNPs from 13 populations (Supplementary Fig. 30 and 39) with 19 bioclimatic variables. The adaptation-associated SNPs identified previously (Supplementary Fig. 33) presented a much faster turnover rate along environmental gradient than the randomly chosen SNPs (Supplementary Fig. 40). Using the adaptation-associated SNPs, we predicted the shift of allele frequency, namely genetic offset, of local landraces between present and future climates during 2040-2060 and 2080-2100. Local bread wheat populations showed varying degrees of genetic offset, with the highest value appearing in regions of the Indus Valley and Inner Asia, indicating that wheat production in the two regions is the most vulnerable to climate change (Fig. 5b, c and Supplementary Fig. 41). Since we did not have a large enough sample size to model allele frequency of individual wild wheat populations, we used Species Distribution Modeling (SDM)51 to predict the future habitats of wild wheats. Overall, we observed either a contraction of wild wheats’ habitats or shifting of their geographical ranges to the north (Supplementary Fig. 42-45). As such, two of the critical progenitors of bread wheat, wild emmer and strangulata (Fig. 1c), clearly showed the projected change of species distribution (Fig. 5d). It is worth noting that wild emmer, which is the ultimate source of genetic diversity of bread wheat52, may become a threatened species requiring conservation in a few decades.
To further characterize these adaptation-associated genes, we focused on genes relevant to flowering time because flowering time is agriculturally important for crops and commonly deemed the most critical trait determining plant adaptation46. Remarkably, we found that the gene Ppd-D1 exhibited convergent adaptation to early flowering and best showcased local adaptation of bread wheat. Ppd-D1 on chromosome 2D is the primary determinant of photoperiod response in bread wheat. Dysfunctional Ppd-D1 exhibits photoperiod insensitivity and early flowering phenotypes, which is crucial to the adaptation of bread wheat to global environments47. A total of three loss-of-function alleles of Ppd-D1 have been identified so far in wheat populations (Fig. 4d), two of which are causal genetic variants, including a ∼2kb deletion at upstream, and a 5-bp deletion in gene exons47,48. The XP-CLR analysis between IA and SH populations identified selective footprints on Ppd-D1, predicting an increased frequency of causative alleles of Ppd-D1 in the SH population (Fig. 4e). However, the two causative alleles did not exist in SH landraces (Fig. 4f and Supplementary Fig. 35). Instead, we found a novel stop-gain mutation of Ppd-D1 in the SH landraces, particularly enriched in the population from the Tibetan Plateau (Fig. 4f and g). We speculated that the stop-gain allele helped adapt bread wheat to the short growing season in high-altitude and low-temperature areas (>3,000 m). To test the hypothesis, we divided the SH landraces into high-altitude and low-altitude subpopulations and performed XP-CLR analysis (Supplementary Table 29). The result showed that the XP-CLR score (99.75% quantile) on Ppd-D1 became more significant when compared with the score (97.61% quantile) between IA and SH landraces, indicating Ppd-D1 is involved in high-altitude adaptation (Fig. 4e and Supplementary Fig. 36). Furthermore, we found a strong correlation between the allele frequency of the stop-gain mutation and the average altitude of subpopulations from SH landraces (r2 = 0.778, Fig. 4h), showing the causal effect of the stop-gain mutation in the high-altitude adaptation of bread wheat. Taken together, the three causative alleles of Ppd-D1 complement each other in geographic distribution, with the stop-gain mutation in South Asia, the ∼2kb deletion in East Asia, and the 5-bp deletion in Europe (Fig. 4g, i, and j), illustrating a highly diverse but convergent adaptation of bread wheat across Eurasia through its changing flowering time.
Population size fluctuation of wheats
Compared with cultivars, crop wild relatives has received relatively little attention from the evolutionary perspective4. To decipher the population dynamics of wild wheats, we reconstructed the history of effective population size (Ne) of Triticum-Aegilops species using SMC++29. We found that Ne of Aegilops subspecies, including strangulata, tauschii (Ae. tauschii ssp. tauschii, DD), and speltoilds (Ae. speltoides, BB/SS), appeared to decline constantly in the last 100 thousand years (Supplementary Fig. 37). In contrast, all the Triticum species experienced a marked population size expansion during the Holocene. For bread wheat and early domesticates, such as domesticated einkorn, domesticated emmer, and free-threshing tetraploids, the population growth may reflect the cultivation history of these populations6; whereas for wild wheats that had never been domesticated, such as wild einkorn (T. monococcum ssp. aegilopoides, AA), urartu (T. urartu, AA), and wild emmer (Fig. 1 and Supplementary Fig. 37, 38), such population growth may result from their mixed growing with early domesticates for thousands of years24,49. Strikingly, we found a ubiquitous population contraction right after the population growth for all the Triticum populations except for modern cultivars of bread wheat. The population decline occurred sequentially with ploidy levels—the diploids came first, then the tetraploids, and lastly, the hexaploid landraces. Intriguingly, the rise and fall of Ne of diploids, tetraploids, and hexaploids complemented each other in the Holocene timeline even without impact from drastic climate change of glacial periods (Supplementary Fig. 37). Archeological studies showed that domesticated einkorn and domesticated emmer thrived since the Neolithic Age until they were gradually replaced by durum wheat (T. turgidum ssp. durum, AABB), spelt, and bread wheat during the Bronze Age (∼5,000 BP - ∼3,000 BP)6,24. The Ne fluctuation of wheats coincides with the shifts of human food choice from einkorn and emmer wheat to bread wheat. Despite largely being a natural evolutionary process, the population size decline of wild wheats is disturbing—the Ne of diploids and tetraploids in Triticum was reduced by 81.70% in the past two thousand years (Fig. 5a).
Fig. 5
The population size fluctuation of wheats from the past to the future.
a, Holocene population dynamics in wheats. The top of this figure depicts the Ne for seven populations, and the bottom of this figure is the relative Ne proportion of each population. b, Genetic offset (GO) of bread wheat landrace based on 2040-2060 RCP8.5 and 2080-2100 RCP8.5 projections. c, Genetic offset of bread wheat landrace in six geographical regions, corresponding to b. The median and quartiles with whiskers reaching up to 1.5 times the interquartile range are shown in boxplots. d, Species distribution models (SDMs) projected the geographical range of wild emmer and strangulata populations in the present and future (2040-2060 and 2080-2100). Red dots pointed to the location of the samples in VMap1.1 and the USDA website (https://npgsweb.ars-grin.gov/gringlobal/search). The green shaded areas are suitable predicted regions for planting.
Rapid climate change is likely to impact the biodiversity of wheats profoundly50. To evaluate the adaptive capacity of Triticum-Aegilops species, we conducted biogeographical modeling to predict the response of wheats to the future climate. For bread wheat, we used a tree-based machine learning approach, gradient forest, to model allele frequency of genome-wide SNPs from 13 populations (Supplementary Fig. 30 and 39) with 19 bioclimatic variables. The adaptation-associated SNPs identified previously (Supplementary Fig. 33) presented a much faster turnover rate along environmental gradient than the randomly chosen SNPs (Supplementary Fig. 40). Using the adaptation-associated SNPs, we predicted the shift of allele frequency, namely genetic offset, of local landraces between present and future climates during 2040-2060 and 2080-2100. Local bread wheat populations showed varying degrees of genetic offset, with the highest value appearing in regions of the Indus Valley and Inner Asia, indicating that wheat production in the two regions is the most vulnerable to climate change (Fig. 5b, c and Supplementary Fig. 41). Since we did not have a large enough sample size to model allele frequency of individual wild wheat populations, we used Species Distribution Modeling (SDM)51 to predict the future habitats of wild wheats. Overall, we observed either a contraction of wild wheats’ habitats or shifting of their geographical ranges to the north (Supplementary Fig. 42-45). As such, two of the critical progenitors of bread wheat, wild emmer and strangulata (Fig. 1c), clearly showed the projected change of species distribution (Fig. 5d). It is worth noting that wild emmer, which is the ultimate source of genetic diversity of bread wheat52, may become a threatened species requiring conservation in a few decades.