Advances in cereal-based agricultural systems over the 40-year period from 1960 to 2000 helped to effectively reduce the global percentage of chronically malnourished people from 60 to 17%Reference Borlaug1. The elevated yields inherent in these high-input ‘Green Revolution’ systems also alleviated ecological land-use problems by reportedly keeping 1.2 billion hectares of largely fragile ecosystems out of agricultural productionReference Borlaug1. However, modern farming systems based on annual cropping systems are widely considered to be a contributing factor to multiple environmental problems, including the loss of genetic diversity and biological diversity, soil erosion and pollution from fertilizers and pesticidesReference Tilman, Cassman, Matson, Naylor and Polasky2, Reference Diaz and Rosenburg3. For example, coastal dead zones, exacerbated by runoff of nitrogenous fertilizers, have spread exponentially since the 1960s, the beginning of the Green Revolution era, and now total over 245,000 km2 in areaReference Diaz and Rosenburg3. Additionally, water erosion alone causes approximately 31.5 t ha−1 of soil loss per year in wheat-based cropping systems in the Palouse region of Washington StateReference Reganold, Elliott and Unger4. Due to the potential ecological benefits of growing perennial-based cereals as a solution to these global agroecological problems, research into the development and agronomic production of perennial cereal crops is emerging across the US, China and AustraliaReference Cox, Glover, Van Tassel, Cox and DeHaan5–Reference Jordan, Boody, Broussard, Glover, Keeney, McCown, McIsaac, Muller, Murray, Neal, Pansing, Turner, Warner and Wyse7.
The large increases in the percentage of people suffering from micronutrient malnutrition over the past four decades coincide with the global expansion of high-yielding, input-responsive cereal cultivarsReference Welch and Graham8, Reference Welch9. Although grain yields have significantly increased post-Green Revolution, global food systems are not providing people with sufficient micronutrientsReference Welch9–Reference Abeledo, Calderini and Slafer12. Currently, over 40% of the world's population is micronutrient deficient; the dietary intake of iron (Fe) of more than two billion people worldwide is inadequateReference Branca and Ferrari13–Reference Stoltzfus17. Notably, the mineral concentration of annual wheat varieties has shown a steady general decline over the past 50+ yearsReference Murphy, Reeves and Jones18.
Improving the nutritional value of wheat has potential to be realized through wide crosses utilizing wild wheatgrass speciesReference Monasterio and Graham19–Reference Chhuneja, Dhaliwal, Bains and Singh22. Although this study focuses on perennial wheat lines with Thinopyrum elongatum in the pedigree, multiple wild species in addition to T. elongatum can be used as parents in the development of perennial wheat. These species can be chosen based on chromosome number, and on traits including local adaptation, disease resistance, drought tolerance and growth habitReference Murphy, Carter, Zemetra and Jones23. Wild relatives of wheat confer beneficial characteristics to the development of perennial wheat; however, they can also be a source of negative traits, particularly those relating to end-use quality and grain threshability.
In addition to addressing the important agronomic traits of grain yield and regrowth after harvest, challenges inherent in the development of perennial wheat include selection of cultivars with enhanced nutritional value, acceptable baking and milling quality and free threshing grain. The objectives of this study were to evaluate grain concentration of calcium (Ca), copper (Cu), iron (Fe), magnesium (Mg), manganese (Mn), phosphorus (P) and zinc (Zn); several end-use quality traits important to baking and milling properties; and grain threshability in 31 perennial wheat lines and two annual wheat cultivars in three locations in Washington State. Each location represents a potential target environment for perennial wheat due to excessive soil loss from wind and/or water erosion. We discuss aspects of the current state of perennial wheat development and identify key traits to target for genetic improvement.
Materials and Methods
Mineral and quality analyses
Mineral analyses of Ca, Cu, Fe, Mg, Mn, P and Zn were performed at the Grand Forks Human Nutrition Research Center in North Dakota using methods reported by Murphy et al.Reference Murphy, Reeves and Jones18. Mineral analyses were conducted on all genotypes at each location. End-use quality traits, including test weight, kernel hardness and whole wheat protein were measured on all perennial lines from each location post-harvest according to AACC Approved Methods 55-10, 55-31 and 46-30, respectively24. Quality traits including flour yield, flour ash, flour protein, mixograph absorption, mixing time and loaf volume of pan bread were determined on both annual wheat cultivars and on five randomly chosen perennial breeding lines from Ritzville and Pullman. Ash and protein content, mixograph absorption, mixing time and bread baking quality were determined according to AACC Approved Methods 08-01, 46-30, 54-40A and 10-10B, respectively24. Seed size and weight were measured on all lines in Kahlotus and Ritzville. Kernel hardness was tested on all genotypes at each location and test weight was evaluated at Ritzville and Pullman.
Experimental design
In this study, perennial is defined as a plant exhibiting post-sexual cycle regrowth for a minimum of two cyclesReference Lammer, Cai, Arterburn, Chatelain, Murray and Jones25. All perennial lines used in this study were shown to be perennial in previous breeding trials. Thirty-one F5 perennial wheat breeding lines were derived from a T. elongatum/Chinese Spring/Madsen population using a modified-bulk pedigree selection method. Genotypes were selected as single plants from a second year perennial bulk population grown in Pullman, WA from 2000 to 2002, and seed increased during the 2002–2003 and 2003–2004 field seasons. Selection was based on regrowth after harvest, winter survival after regrowth and subsequent seed set.
The 31 perennial genotypes were grown with two annual hard red winter wheat cultivars, ‘Finley’ and ‘Bauermeister’, at three rainfed locations in Washington State (Kahlotus, Ritzville and Pullman) in 2005/06 using a randomized complete block design with four replicates per location, as described in Murphy et al.Reference Murphy, Lyon, Balow and Jones26. Briefly, plots were 2.5 m long and 1.25 m wide and consisted of four rows at 30 cm spacing at Ritzville and Kahlotus and seven rows at 18 cm spacing at Pullman. Seeding rate was approximately 45 kg ha−1 at Ritzville and Kahlotus and 85 kg ha−1 at Pullman. Plots were fertilized with 46.0 kg ha−1 of N at Ritzville and Kahlotus. In Pullman, 100.8 kg ha−1 of N, 22.7 kg ha−1 of phosphate and 17.0 kg ha−1 of sulfur were incorporated into the soil within a week of planting. These fertilizer treatments reflect the locally prevalent application rates for annual wheat.
Plots were harvested at grain maturity with a Hege plot combine (Niederlassung, Germany) with stainless steel sieves and cleaned with a Hege seed cleaner with stainless steel sieves. A threshability index (TI) was estimated by dividing initial grain yield before cleaning by grain yield after a uniform cleaning process. The cleaning process removes seed with tenacious glumes that were still attached to the rachis while conserving only the threshed grain.
Analysis of variance in PROC GLM (SAS Institute, Cary, NC) was used to analyze data threshability and mineral concentration and to test for genotype×location interactions. Location and genotype were considered fixed. Levene's test was used to test for homogeneity of variance across locations and normality was checked using the Shapiro–Wilk test in PROC Univariate (SAS Institute).
Fifteen soil subsamples were randomly collected to a depth of 30.5 cm and pooled for analysis at each location. Soil organic matter, pH, and available Cu, Fe, Mn, P and N were determined by the University of Idaho analytical soil testing laboratory.
Results
Grain mineral concentration
Micronutrient concentrations for Cu, Fe, Mn and Zn were 40, 24, 32 and 33% higher, respectively, in the perennial grain than in the annual grain (Table 1). Macronutrient concentration followed the same pattern, with Ca, Mg and P being 44, 23 and 30% higher, respectively, in the perennial lines (Table 1). Differences were found among genotypes for all minerals tested (P<0.001).
Grain yield (g plot−1), regrowth (% regrowing plants plot−1), threshability index (TI), thousand kernel weight (TKW) and mineral concentrations (μg g−1) for calcium (Ca), copper (Cu), iron (Fe), magnesium (Mg), manganese (Mn), phosphorus (P) and zinc (Zn) reported for annual wheat cultivars (A) and perennial breeding lines (P) across three locations in Washington State (Kahlotus, Ritzville and Pullman) in 2005/06. Trait mean and standard error (SE) are shown at the bottom of the table.
Soil organic matter was 1.6, 2.0 and 2.6 for Kahlotus, Ritzville and Pullman, respectively. Soil pH was 5.5% for Kahlotus and Ritzville and 5.1% for Pullman. Cation exchange capacity was 14, 15 and 29 cmol(+) kg−1 respectively for Kahlotus, Ritzville and Pullman. Available soil mineral concentrations of Cu, Fe, Mn, N (nitrate+nitrite), N (ammonia) and P for each location are shown in Figure 1. Available potassium (K) was 650, 720 and 190 μg g−1 for Kahlotus, Ritzville and Pullman, respectively.
Soil bioavailability of copper (Cu), iron (Fe), manganese (Mn), nitrogen (N) with nitrites+nitrates, phosphorus and N in the form of ammonia at three locations (Kahlotus, Ritzville and Pullman) in Washington State.
Location had a significant effect on grain mineral concentration for all minerals tested. The perennial grain from Kahlotus had the highest level of all mineral nutrients except Mn (Fig. 2). Soil available Mn was accordingly lowest at this location (Fig. 1). All minerals were moderately to highly correlated with each other (P<0.05) (Table 2).
Effect of location (Kahlotus, Ritzville and Pullman) on micronutrient (A) and macronutrient (B) concentration (mg kg−1 dry weight) in perennial grain grown in Washington State. Lower-case letters above the bars indicate significant differences (P<0.05) among locations for each nutrient tested.
Correlations between grain yield, regrowth after harvest, threshability index (TI), thousand kernel weight (TKW) and mineral concentrations of calcium (Ca), copper (Cu), iron (Fe), magnesium (Mg), manganese (Mn), phosphorus (P) and zinc (Zn) among annual and perennial breeding lines. Correlations below the leading diagonal include the annual wheat cultivars; correlations above the leading diagonal represent the perennial breeding lines only. *P<0.05; **P<0.01; ***P<0.001.
Trait means across locations for each genotype are reported for grain yield, regrowth, TI, thousand kernel weight (TKW) and the seven minerals tested (Table 1). To estimate total mineral content of Cu, Fe and Zn per unit area for each genotype, we multiplied mineral concentration (μg g−1) by grain yield per plot (g plot−1), and divided by 1000. The annual cultivars had a mean of 4.92 g plot−1 Cu (range=4.65–5.18), 45.0 g plot−1 Fe (range=44.1–46.0), and 26.38 g plot−1 Zn (range=26.02–26.73). The perennials had a mean of 3.58 g plot−1 Cu (range=2.89–4.90), 25.9 g plot−1 Fe (range 20.0–37.1) and 17.19 g plot−1 Zn (range=13.05–22.56).
Grain yield per plot was negatively correlated with grain mineral concentration for all minerals when the annual cultivars were included in the analysis (Table 2). When the correlation analysis included only the perennial breeding lines, no associations were found between grain yield and mineral concentration for any minerals (Table 2). When annuals were included in the analysis, TKW was negatively correlated with Ca, Cu and Zn; however, in the absence of the annual cultivars, TKW was positively associated with Mg, Mn and P (Table 2).
End-use quality
The perennial lines produced smaller and lighter kernels than the annual cultivars, resulting in decreased test weight (Tables 3 and 4). The perennial lines generally exhibited lower flour yield, loaf volume and mix time; greater whole wheat protein and flour protein content; and, similar ash content of flour compared to the annual cultivars (Table 3). Kernel hardness of the perennial lines was 47.8% lower than what is typically observed in hard wheat, while protein content of the perennial lines was 3.5–4.5% greater than the annual hard wheat cultivars (Table 4).
End-use quality characteristics including test weight (Twt), single kernel hardness (SKHard), whole wheat protein (UWprt), flour yield (FYeld), flour ash (Fl Ash), flour protein (FProt), mixture absorption (MAbs), mix time (MTime) and loaf volume (LVol), measured for the two annual wheat cultivars (‘Bauermeister’ and ‘Finley’) and four perennial breeding lines in two locations (Pullman and Ritzville).
Quality characteristics, including test weight, percent whole wheat protein, single kernel hardness, thousand kernel seed weight (g) and single kernel seed size (mm), were compared between annual wheat cultivars and perennial breeding lines grown at three locations in Washington State (na=data unavailable for these tests in respective location).
Whole wheat protein content ranged from 14.0 to 16.5% in Pullman and 15.2 to 17.6% in Ritzville for the perennial lines. For the annual cultivars, ‘Bauermeister’ had 11.6 and 12.7% whole wheat protein in Pullman and Ritzville, respectively, and ‘Finley’ had 12.4 and 12.7% whole wheat protein at the respective locations (Table 3). The annual hard red cultivars had greater kernel hardness, seed weight and seed size than the perennial lines (Table 4).
Threshability
The mean TI in the annual cultivars was 0.97. This was considerably higher than the perennial lines, which ranged from 0.63 to 0.89, with a mean of 0.75 (Table 1). A TI of 1.00 would indicate that all the glumes had been separated from the grain during harvest. A TI of 0.0 would indicate that the grain had very tenacious glumes and the seed would not separate from these glumes through standard mechanical harvest. A significant genotype×location interaction was found for TI (P<0.0001). TI was positively associated with yield and negatively correlated with regrowth when the annual cultivars were included in the analysis (Table 2). No phenotypic associations were found between either TI and yield or TI and regrowth when only the perennial lines were included in the analysis (Table 2).
Discussion
Grain mineral concentration
Cereal crops have the potential to provide a significant increase in the overall micronutrient availability for much of the world's population without access to diverse food cropsReference Graham, Welch and Bouis27. The increased mineral nutrient concentration in the perennial grain is most likely derived from wheatgrass species T. elongatum. Accordingly, Uauy et al.Reference Uauy, Distelfeld, Fahima, Blechl and Dubcovsky20 found increased Fe and Zn concentrations in synthetic hexaploids developed from tetraploid durum and the diploid wheat ancestor, Aegilops tauchii. The elevated concentrations of micronutrients in the perennial grain may be a result of improved scavenging ability of the larger roots of the perennial linesReference Glover28; however, the total mineral content per unit area of cultivated land is significantly lower in the perennial grain than in the annual grain. This is predominantly due to the much higher grain yield in the annual cultivars.
Another reason for the enhanced micronutrient concentration in the perennial grain may be that the smaller grain size is directly related to the increased mineral concentration. This is known as the ‘dilution effect’. The mineral concentration of the grain would decrease if a corresponding increase in grain size was due only to an increase in the endosperm and not the bran or germ, where the majority of the minerals are located. We found significant differences in TKW among the perennial lines (Table 1), and TKW was positively correlated with mineral concentration for Mg, Mn and P (Table 2). No relationship was found between TKW and the mineral concentrations of Ca, Cu, Fe or Zn among the perennial lines (Table 1). When the annual cultivars were included in the correlation analysis, however, TKW was negatively associated with Ca, Cu and Zn (Table 2).
This provides contradictory evidence regarding the dilution effect theory and is in part based on the issue of whether comparing the grain composition between annuals and perennials is an appropriate test of the dilution effect theory. For example, perennial and annual wheat should likely be considered two different crop species (much the same as wheat and triticale), as they differ in chromosome number and designation. Therefore, correlations among the perennial lines only (annuals excluded) may provide the most accurate account of the actual relationships between TKW and mineral concentration. Future research is needed to fully understand the roles that scavenging ability and the dilution effect play in achieving perennial grain with enhanced mineral nutrition and improved grain yield.
Grain mineral concentration is dependent to a great extent upon various soil properties, including soil organic matter, pH and the bioavailability of minerals in the soilReference Wei, Hao, Shao and Gale29–Reference Sims31. Soils with a low pH have been shown to reduce uptake of the macronutrients Ca and Mg and to increase uptake of the micronutrients Zn, Mn and FeReference Fageria and Zimmermann32. The lack of a positive correlation between soil nutrient availability and grain nutrient concentration across all cultivars indicates that other environmental factors such as drought may have contributed to higher grain nutrient concentrations. In soybeans, plant stress induced by drought resulted in an increase in the accumulation of P, K, Ca, Mo, Mn, Cu and Zn, and may be an important response in drought stress toleranceReference Samorah, Mullen and Cianzio33. The more drought-prone environments represented by Kahlotus and Ritzville generally showed higher levels of micronutrients in the grain than Pullman, the region with higher annual precipitation (Fig. 2).
End-use quality
Despite the much higher protein content compared to the annual cultivars, the perennial lines produced a smaller loaf volume of bread (Tables 3 and 4), indicating the weak protein strength and inferior gluten quality of these perennial wheat lines for baking bread. P-0021 had the highest whole wheat protein among the two annual cultivars and four perennial lines tested in Ritzville and Pullman (Table 3). Though the protein is not of bread baking quality, its enhanced concentration in the grain could have a significant dietary impact for human and livestock consumption. High protein livestock feed is important for many animals and perennial wheat could contribute to local production of high protein grain to complement mixed legume species in regions where soybeans are unable to grow, expensive to import or unavailable.
With the exception of one line (P-0009), most of the perennial lines showed similar water absorption and shorter mixing time in comparison to the annual cultivars. The relatively low flour yield found in the perennial lines was probably due to the low test weight and smaller kernels of the grain. These traits are important for baking quality and the variation among the perennial lines for these traits indicates the potential for genetic improvement.
Further improvements in grain characteristics, milling quality and protein strength of the perennial lines are needed before acceptable utilization as hard red hard wheat for baking is possible. The potential for perennial wheat in the soft white market class, however, is greater. Soft white wheat is the predominant market class in the Pacific Northwest, and is used for pastries, white Asian noodles, cookies, cereal flakes and flatbreads. In general, the quality of these products does not depend on protein strength and typical bread making characteristics like gluten quality and loaf volume. With this marketing option in mind, we have shifted a percentage of the perennial wheat breeding priorities to the selection of perennial lines with potential in the soft white market class.
Threshability and seed shattering
Thinopyrum species have tenacious glumes, a trait controlled by recessive mutations at the Tg loci, dominant modifying genes at the Q locus and additional modifying mutations at several other lociReference Dubcovsky and Dvorak34, Reference Jantasuriyarat, Vales, Watson and Riera-Lizarazu35. These tenacious glumes were transferred to a portion of the perennial progeny, making mechanical harvest less efficient. Though early domesticated wheat, with non-shattering, indehiscent spikes, have been found dating back to ~9250 years bpReference Tanno and Willcox36, the wild wheatgrass species we use in our perennial wheat breeding program have typically not been subject to artificial selection for either tenacious glumes or indehiscent spikes. Therefore, we anticipate selection for these traits to be of importance both during pre-breeding of wild species and selection of perennial wheat hybrids.
The TI range of 0.63–0.89 in the perennial lines indicates that the genes for threshability are segregating, and the population as a whole is tending to improve for this trait. Unlike selection for nutritional value and end-use quality traits, selection for increased threshability has potential for further improvement through the use of evolutionary breeding methodsReference Murphy, Lammer, Lyon, Carter and Jones37, Reference Suneson38. The significant genotype×location interaction for TI found in this study may be due to the environmental effect of abundant moisture at Pullman which reduced threshability.
We observed that plants with shorter spikes did not have the brittle rachis trait that we have found in many wild wheatgrass species and in perennial wheat hybrids with longer spikes (unpublished data). In the tetraploid emmer wheat, shattering is determined by the Br (brittle rachis) loci on chromosomes 3A and 3BReference Dubcovsky and Dvorak34, Reference Nalam, Vales, Watson and Riera-Lizarazu39. In rice, Li et al.Reference Li, Zhou and Sang40 showed that a single amino acid change is primarily responsible for the loss of shattering. The cultivated allele of the sh4 (shattering4) gene confers both a non-shattering trait and allows for easy threshability of the grain. This suggests the potential for a rapid removal of the shattering trait from the perennial wheat genome. The brittle rachis trait, though possibly exploited in the Neolithic era bridging the hunter/gatherer era and the dawn of agriculture through ground gathering of shattered cereal spikeletsReference Kislev, Weiss and Hartmann41, is a characteristic that does not conform to efficiencies needed for harvesting in modern agriculture.
Future Research in Perennial Wheat
Though the ability to transfer genes from perennial wild wheat species that confer the post-harvest regrowth trait is elegantly simple, there exist additional characteristics necessary for the successful establishment of perennial wheatReference Lammer, Cai, Arterburn, Chatelain, Murray and Jones25. These agronomic traits can vary by region and by agroecosystem and include disease resistance; drought, heat and cold tolerance; winter and/or summer dormancy; regrowth vigor and timing; and carbohydrate allocations in the root and crownReference Cox, Garrett, Cox, Bockus and Peters42–Reference DeHaan, Van Tassel and Cox44.
This study focused on the post-harvest characteristics of mineral concentration, end-use quality and grain threshability. In addition to multiple agronomic traits, baking and milling quality and grain threshability must be improved before widespread adoption of perennial wheat by farmers, millers and bakers will occur. Though not a requirement for the successful development of a perennial wheat cultivar, increased grain mineral concentration is a reportedly beneficial trait currently found in perennial wheat breeding lines that should continue to be exploited and selected for in subsequent perennial wheat breeding evaluations.
Acknowledgements
We gratefully acknowledge our farmer cooperators, Jim Moore and Mark Schoesler for their support, encouragement and use of their farm ground for this research. This project is indebted to the excellent technical support provided by Meg Gollnick, Kerry Balow and Steve Lyon. Funding for this study was provided by USDA-CSREES Fund for Rural America, USDA-CSREES Line Item Special Grant, The Land Institute and the Washington Wheat Commission.
