Grain yield penalties may occur in perennials due to resource tradeoffs between reproduction and regrowth and subsequent winter survival. A modeling study concluded that perennial wheat would be economically viable in Australia if it achieved 40% of annual wheat grain yields, combined with at least 800 kg ha−1 of additional forage contribution per year (beyond the forage production of the annual) (Bell et al., 2008). This prompts the question: can perennial cereals, at their current stage of development, achieve the threshold level of 40 to 60% yield level relative to annuals? Forage production, particularly early in the season, may be an important additional contribution of perennial cereals, thus it is necessary to assess biomass as well as grain production. Finally, an agronomic assessment of perennial cereals will need to include an understanding of their phenology. Differences in phenology between perennial cereals and their annual relatives could affect early season grazing potential, susceptibility to pathogens (Emrich et al., 2008), and vulnerability to extreme weather, but little information is available on the phenology of emerging perennial cereal species.
To understand the agronomic potential of perennial cereals, they must be studied over multiple years, as grain production may change with stand age. Perennial cereals might display delayed reproductive investment, similarly to many woody perennials: in this case, plants would show low reproductive investment the first year (the establishment year), but in future years the established root and crown reserves would allow plants to begin growth earlier, grow larger, and show marked increases in seed production (Jackson and Jackson, 1999). Alternatively, yield decreases might be observed due to the buildup of soil pathogens, short plant life span, or the proliferation of weak and unproductive tillers. Yield declines have been observed in grasses such as smooth brome (Bromus inermis Leyss.), tall fescue (Festuca arundinacea, Schreb.), and Kentucky bluegrass (Poa pratensis L.) (Loch et al., 1999; Fulkerson, 1980; Fairey and Lefkovitch, 2001), although a number of other cool season forage grasses show stability in seed yields over two to six seasons (Chastain et al., 1997; Canode and Law, 1978; Mueller-Warrant and Caprice Rosato, 2002; Fulkerson, 1980) or even increases as demonstrated for fairway crrested wheatgrass [Agropyron cristatum (L.) Gaertn.] and red fescue (F. rubra L.)(Canode, 1968). To test the hypothesis that yields will decrease over time, it is necessary to study perennial cereals over at least 2 yr.
Perennial wheat and rye are still in the process of development, and there are few studies of grain yield produced under agronomic conditions. Early work on perennial wheat was done in the former Soviet Union (Tsitsin, 1939), but little of this data has been easily accessible to researchers in other countries. More recently, first year grain yield was measured in a study of 31 perennial wheat genotypes in eastern Washington (Murphy et al., 2010), finding yields of up to 93% of annual wheat in the highest yielding accession. Grain yield was studied in approximately 90 perennial wheat derivatives over 2 yr in Australia (Hayes et al., 2012) and in earlier cultivars of perennial rye were studied over 2- or 3-yr periods: “Permontra” (Reimann-Philipp, 1986; Weik et al., 2002) and the newly developed ACE-1 (Acharya et al., 2003, 2004). These studies generally found perennial rye yielded approximately 55 to 60% of annual rye, and that grain yield of perennial wheat lines were highly variable, ranging between 2 and 135% of annual wheat among those lines that showed appreciable regrowth. However, these multi-year studies did not explicitly separate the effects of calendar year (reflecting year to year weather fluctuations) from plant age (reflecting possible effects of senescence, metabolic tradeoffs, changes in allocation, changing energy and nutrient status of the plants, and other age-related phenomena). Our study makes a novel contribution to the growing literature on perennial cereals by observing 1-yr-old plants in two different years (achieving replication in time), in a new and different environment, and by comparing 1- and 2-yr-old plants within a single year (thus separating the effects of plant age and calendar year). We were thus able to consider how genotype/cultivar, calendar year, and plant age each affected plant yield, biomass, and phenology.
Between 2008 and 2010, we assessed the agronomic potential of perennial wheat and perennial rye in southwest Michigan. We measured parameters relating to biomass, yield, and phenology to document how these plants compare, over time, to their annual relatives. The key hypotheses we tested were: (i) 1-yr-old perennial wheat and rye yield lower than their annual analogs, (ii) 2-yr-old perennial wheat and rye produce lower biomass and lower seed yields, compared to first-year plants, and (iii) 2-yr-old perennial wheat and rye initiate earlier spring growth and flowering, relative to 1-yr-old perennial and annual genotypes.
MATERIALS AND METHODS
We conducted this study at the W. K. Kellogg Biological Station located in southwest Michigan, 50 km east of Lake Michigan (42°24’ N, 85°24' W, elevation 288 m) on soils developed from glacial outwash. Soils at the site are of the Kalamazoo (fine-loamy, mixed, semiactive, mesic Typic Hapludalfs), Oshtemo (coarse-loamy, mixed, active, mesic Typic Hapludalfs), and Miami (fine-loamy, mixed, active, mesic Oxyaquic Hapuldalfs) series (Crum and Collins, 1995). The area receives about 900 mm of precipitation annually (based on a 24-yr average before 2009), with approximately half as snow, and the mean annual temperature is 9.7°C.
Eight accessions (either breeding lines or named varieties) were involved in the study. The annual winter wheat “checks” included Frankenmuth (PVP 8000165), and Pioneer 25R37 (PVP 0020232). Frankenmuth is an older cultivar commonly grown in the 1980s and 1990s (e.g., Huebner et al., 1999) and has been used as a benchmark for yield and quality in breeding studies at Michigan State University: Pioneer 25R37 is a newer cultivar that was widely grown in the mid-2000s based on high performance in Michigan (J.M. Lewis, personal communication, 2009). Four perennial wheat accessions were included (T. aestivum ‘Chinese Spring’ × Thinopyrum elongatum (Host) D.R. Dewey//T. aestivum ‘Madsen’), all of which formed part of a set of 20 accessions contributed by Washington State University in 2005, and had been involved in previous studies in Washington (Murphy et al., 2009, 2010). We selected 4 accessions out of the 20 based on differences in harvest index and morphology displayed in a 2008 pilot observational trial at the Michigan State University Kellogg Biological Station. The P3 showed a relatively low harvest index (as well as more “grass-like” spikelet morphology) while P11 had the highest harvest index combined with spikelets resembling annual wheat: P15 and P19 had intermediate harvest index and morphology. Also included were an annual rye cultivar (Secale cereale ‘Wheeler’), and a perennial hybrid rye cultivar (Secale cereale × montanum Guss. ‘Rival’).
This study was laid out as a split plot randomized complete block design (RCBD) with five blocks, using “planting year” as the whole-plot factor and “accession” as the split-plot factor. Planting year was an indication of when a particular plot was first planted (2008 or 2009). When comparing 1-yr-old plants in 2009 to 1-yr-old plants in 2010, planting year thus reflected year-to-year weather variation (and will be referred to in the analysis as “year”); when comparing older vs. younger perennial plants in 2010, planting year reflected the age of the plants, and will be referred to as “age” (since 1-yr-old plants had been planted in 2009, and 2-yr-old plants had been planted in 2008). Thus we were able to consider separately the effects of both year-to-year weather variation, as well as plant age (Shefferson and Roach, 2010). A similar approach (comparing a single cohort across years, and multiple age cohorts within years) has been used to study a close relative of an agricultural plant, sea beet (Beta maritima L.), in the greenhouse (Van Dijk, 2009). In the analysis, tables, and graphs, the term “2009” will be taken to refer to the entire October 2008 to September 2009 growing season, and “2010” to the entire October 2009 to September 2010 growing season.
One set of five blocks, with each of the eight accessions replicated once per block, was first planted in fall 2008, and subsequent sets in 2009 and 2010. Following harvest in August 2009, perennial accessions were allowed to regrow into a second season, while the annual plots were tilled under and replanted with the same annual cultivars in November 2009. Also during November 2009, a full second set of perennial plants was planted. Thus in 2010 the study included 2-yr-old perennial plants, 1-yr-old perennial plants, and 1-yr-old annual plants (Table 1). The subplots were each approximately 5.47 m long, and included six rows of plants with 19 cm between rows; there were 57-cm alleys between subplots and 90 cm between whole plots.
|Planting year||Measurement year||Species||Accession||Plant age at measurement|
|2009||annual wheat||Pioneer 25R37||1|
|2010||annual wheat||Pioneer 25R37||1|
Dates of field operations are given in Table 2. Plants were seeded at 1.5 million seeds ha−1 about 1.9-cm deep, using an Almaco small plot planter. The field was fertilized in the fall of 2008 with composted dairy manure (at the rate of 90 kg ha−1 N), while in the fall of 2009 we fertilized with approximately 102 kg ha−1 N in the form of blood meal (13% N). Plots were irrigated once in summer 2009, applying approximately 3.2 cm of water to stimulate regrowth. In October 2010, regrowth was extremely low and variable, with no accession showing more than 10% regrowth, and over half the plots showing zero regrowth. This might have been due to the warm spring coupled with hot and dry weather in the late summer, and intense weed pressure. Because of very poor regrowth, the experiment was ended in October 2010.
|Tilled||13 Oct. 2008||7 Oct. 2009|
|Fertilized||22 Oct. 2008||20 Oct. 2009|
|Soil finished||24 Oct. 2008||19 Oct. 2009|
|Planted||25 Oct. 2008||9 Nov. 2009|
|Irrigation||3 Aug. 2009||5 Sept. 2010|
|Weeding 1||15–20 May 2009||1–5 May 2010|
|Weeding 2||7–14 June 2009||1–7 June 2010|
|Weeding 3||29 June–7 July 2009||25 June–2 July 2010|
|Weeding 4||30 Aug.–5 Sept. 2009||28 Aug.–1 Sept. 2010|
|Postharvest tillage||7 Oct. 2009||23 Oct. 2010|
On 20 Mar. 2009 and 16 Mar. 2010, about 1 wk after snowmelt, the number of growing plant stems per meter in each plot were counted (which estimated spring emergence for annuals and 1- yr-old perennials, and regrowth for the 2-yr-old perennials). Plant height was estimated by sampling six plants from each plot, at the same time that we counted plant stems. To estimate flowering date, plants were observed once every 4 d from 20 May to 23 July 2009 or from 1 May to 6 July 2010. We estimated the percentage flowering at each date and calculated a 50% flowering date through linear interpolation. In the analysis, “first-year plants” (1 yr) denotes plants during their first growing season. “Second-year plants” (2 yr) denotes plants during their second full growing season.
Harvest during the 2009 and 2010 growing seasons took place 6 to 7 wk following flowering for each accession. Harvest dates in 2009 were 23 July for perennial rye, annual rye, and annual wheat, and 13 August for perennial wheat. Harvest dates in 2010 were 10 July for annual wheat, 17 July for annual and perennial rye, and 9 August for perennial wheat. Yield was measured by hand harvesting 4 m lengths of two central rows per plot, cutting at the base and separated into seed heads, leaves, and stems. Yield components were dried at 65°C for 4 d (to constant weight) and weighed. Because a pilot study had found that perennial wheat threshed poorly using standard equipment, we used a customized tabletop seed cleaner at Michigan State University to determine the grain percentage of seed head mass. Grain recovery was assessed for 70-g subsamples of unthreshed seed heads for each accession. At the time of harvest, we counted reproductive tillers on the harvested rows. Following seed cleaning we measured the thousand-kernel weight on bulked seed samples of each accession and divided by 1000 to obtain kernel mass. In October of each year, four 20 cm2 microplots in each plot were evaluated for post-sexual cycle regrowth (PSCR) by determining the following: (number of visibly regrowing green leaves)/(number of reproductive tillers at harvest) × 100.
To determine the effect of year-to-year variation on each parameter of interest, we compared performance of 1-yr plants in 2009 to 1-yr plants in 2010 (including annual wheat and rye checks). Our model was a split plot RCBD using calendar “year” as the main plot factor, “accession” as the subplot factor, calendar “year × accession”, and block. Following the overall analysis, we then held year constant to determine the effect of accession differences within each year. To determine the effect of plant age on each parameter, we compared 1-yr plants (perennial accession only) in 2010 to 2-yr plants within the same year. The model was a split plot RCBD using plant “age” as the main plot factor, accession as the subplot factor, “age × accession”, and block. To determine differences between accessions within each age class, we held age constant and considered the effect of accession. All analyses were done using the MIXED procedure in SAS 9.2 (SAS Institute, 2008) followed by planned contrasts using protected LSD.
Analysis of kernel mass did not include a block effect since these were measured after bulking seed from all blocks. Mass-related parameters were log-transformed to improve normality, and threshability and harvest index were logit-transformed to meet assumptions of ANOVA. Other variables were untransformed.
The overall model for one of the response variables (threshed yield) was, for example:
The pattern of precipitation in the fall of 2008 and 2009 was similar, with a dry November (50% below the 24-yr average), but otherwise well distributed precipitation (Fig. 1). Winter temperatures may have affected survival, as evidenced by the lower springtime emergence in 2010 following a very cold January (Fig. 1). Early spring precipitation was wet in 2009 (54% above the 24-yr average) and dry in 2010 (25% below the 24-yr average). However, there was no difference between the years in terms of early spring growth of perennial cereals, as indicated by March plant height measurements (Table 3). The later spring (May and June) was much wetter in 2010 than in 2009, with 120% higher precipitation (Fig. 1); overall, the spring was consistently warmer in 2010 than in 2009, with an average difference of 1.8°C over the March–June period. Thus the 2010 spring was warmer, drier early on but much wetter in the later spring, as compared to 2009.
||Early season height
|stems m−1||day of year||cm||%|
|Pioneer 25R37 wheat||59.1||40.5||157.6||157.9||7.9||8.0||49.5||67.3||–||–|
|–Accession × year||1.30||22.54***||2.71*||2.47*||2.26|
Yield of First-Year Perennials and Annuals
Threshed grain yield of 1-yr plants was similar between 2009 and 2010, with no overall main effect of year (Table 4). The grain yield of the annual cereals was moderate, averaged across 2 yr: 2.41 Mg ha−1 for Pioneer 25R37, 2.94 Mg ha−1 for Frankenmuth, and 1.83 Mg h−1 for Wheeler rye. These yields are typical for small grains at our location (Smith et al., 2008), and reflect the low organic matter, coarse soil type as well as weather conditions. Two accessions (P19 and P15) yielded 40% higher in 2010 compared to 2009, while all other accessions did not differ between the 2 yr (Fig. 2). Perennial wheat and perennial rye plants consistently yielded lower than their annual counterparts (Table 4). The 1-yr perennial wheat plants, averaging across accessions, consistently yielded approximately 48% of annual wheat in 2009 and 2010 (Table 4, Fig. 2). Similarly, perennial rye, on average yielded 74% of annual rye and response was similar in both years of the study (Fig. 3). All four perennial wheat accessions performed the same in 2009, but in 2010 the 1-yr P19 and P15 plants achieved approximately 81% higher yields than the lowest yielding line, P3 (Fig. 2).
|Accession||Threshed yield||Threshability||Tiller no.||Kernel no.||Kernel mass||Biomass||Harvest index|
|Mg ha−1||%||tillers m−2||no. tiller−1||mg||Mg ha−1||%|
|Pioneer 25R37 wheat||2.41||83.8||60.1||21.4||36.7||4.43||54.4|
|–Accession × year||2.25*||–||3.15**||2.38*||18.09***||3.30**||1.85|
The lower yields in perennial cereals compared to their annual relatives reflected lower ratio of grain to chaff, lower kernel mass, lower harvest index, and (in perennial wheat) a lower density of reproductive tillers (Table 4). In contrast, they did not reflect lower biomass production or lower number of seeds per tiller. The percentage of grain in seed heads was lower in perennial cereals than in annuals: 82 to 86% for perennial wheat relative to annual wheat, and 77 to 85% for perennial rye compared to annual rye (Table 4). Kernel mass explained much of the perennial vs. annual yield differences. Kernels of 1 yr perennial wheat plants (averaging across accessions) were 31% smaller than annual wheat in 2009 and 56% smaller in 2010, as indicated by a linear contrast between the 2 yr (p = 0.001). Similarly, kernel mass of 1-yr perennial rye was 24% lower than annual rye in 2009, and 20% lower in 2010 (p = 0.038).
Reductions in the number of reproductive tillers contributed to lower seed yield in perennial wheat, but not in perennial rye. Indeed, perennial rye had 50% more reproductive tillers than annual rye in 2009 and equivalent number of reproductive tillers in 2010 (Table 4). Harvest index also differed among genotypes: 1-yr perennial wheat had a harvest index 43% lower than annual wheat, while 1-yr perennial rye had a harvest index that was 23% lower than annual rye (Table 4). Total biomass, by contrast, was higher in perennial rye compared to annual rye in 2010, while perennial wheat biomass was equivalent to annual wheat that year. This suggests that differences in biomass were not an important contributor to the lower grain yields in perennials; rather, annuals appear to allocate a greater fraction of biomass to seed reproduction and thus had a higher harvest index than perennials (Table 4).
Similarly, lower grain yields in perennials did not seem to be explained by lower number of seeds per reproductive tiller. Perennial rye, in both years, had an equivalent number of seeds per reproductive tiller as annual rye. Perennial wheat did not have a consistently lower number of seeds per tiller, and in fact in 2010 three of the four lines had equivalent numbers of seeds per reproductive tiller as Frankenmuth annual wheat, and 60 to 89% more than Pioneer 25R37. This is, importantly, not equivalent to saying that the perennial cereals have equal fertility to annual wheat and rye, since we did not count the number of sterile spikelets: if perennial wheat and rye have many more spikelets than their annual analogs, it is possible that lower fertility might co-exist with a comparable or higher number of seeds per tiller. Further work is needed to determine whether perennials in fact display lower fertility, in the sense of a lower ratio of mature seeds to total spikelets.
Effects of Plant Age on Yield and Components
Overall, 1- and 2-yr perennial plants in 2010 showed consistent yields, with no effect of plant age or age × accession interaction observed (Table 5). Second year perennial wheat lines had equivalent yields to 1-yr perennial wheat (1.42 Mg ha−1), and 2-yr perennial rye yielded equivalent to 1-yr perennial rye (1.44 Mg ha−1). Relative to annual counterparts, 2-yr perennial wheat yielded 53% of annual wheat (p < 0.0001), while 2-yr perennial rye yielded equivalent to annual rye. The only difference among accessions was that P19 and P15 produced 50% higher yields than P3 (Table 5).
|Accession||1 yr||2 yr||1 yr||2 yr||1 yr||2 yr||1 yr||2 yr||1 yr||2 yr|
|Mg ha−1||tillers m−1||no. tiller−1||Mg ha−1||%|
|–Plant age × accession||1.66||3.66*||2.01||2.14||6.42***|
The consistency of grain yields in 1- and 2-yr plants was also reflected in the lack of age-related effects on yield components. In only a few cases did individual accessions show changes in yield components between 1 yr and regrowth, for example, 2-yr plants (Table 5). Kernel mass, for example, showed an accession × age interaction effect, reflecting the fact that kernel mass decreased with age in a couple genotypes. Specifically, 2-yr perennial rye and P19 plants had 15% smaller kernels than 1-yr plants (Fig. 4). The other three perennial wheat lines showed no differences. Similarly, there were no consistent age-related trends in terms of biomass or harvest index (Table 5). Harvest index showed age × accession interactions: specifically, older plants of P11 and perennial rye had ∼16% lower harvest index than younger plants, while the opposite trend was seen in older P3, P15, and P19 accessions which had 16 to 33% higher harvest indices than younger plants. Biomass, on the other hand, differed among accessions but showed no age-related effects. Thus there were few clear trends in age effects on yield components, which helped to explain the overall consistency in grain yield between 1- and 2-yr plants.
Plant Growth, Post-Sexual Cycle Regrowth, and Phenology
Interestingly, a plant age effect was observed for early season growth. Plant height achieved by mid-March differed between 1- and 2-yr perennials, with the older plants being on average 110% taller (Table 6). In mid-March 2010, 1-yr perennial wheat was 22% shorter than annual wheat and 2-yr perennial wheat was 59% taller. A similar pattern was observed in perennial rye, where 1-yr perennial plants were 27% shorter than annual rye and 2-yr plants were 75% taller. Early season height differences were, to some extent, maintained later in the season: 2-yr-old perennials were overall 10% taller at anthesis than 1-yr-old perennials, with a 26% difference in the case of perennial rye (Table 6). Second-year perennial wheat and rye were 21 and 11% taller than annual wheat and rye in 2010, respectively. Height at flowering showed effects of year as well as plant age, with 1-yr perennial wheat and rye both being shorter in 2010 compared to 2009 (Table 3).
||Early season height
|Accession||1 yr||2 yr||1 yr||2 yr||1 yr||2 yr||1 yr||2 yr||1 yr||2 yr|
|stems m−1||day of year||cm||%|
|–Age × accession||1.46||1.79||2.61‡||9.66***||0.97|
Number of stems present in March 2010 was similar for both 1- and 2-yr perennial plants (Table 6) and there was no main effect of age, nor an age × accession interaction. This suggests that each perennial stand, as a whole, was able to fully replenish itself through regrowth (or there may have been some reseeding although harvest was conducted in a timely manner to prevent this occurrence). Regrowth in the fall of 2009 was vigorous for both perennial rye and P19 perennial wheat (over 100% regrowth, Table 3). P3 was the least vigorous line, at 50% regrowth. In the fall of 2010, however, regrowth was extremely poor: no accession achieved more than 10% regrowth, and 55% of plots showed zero detectable regrowth.
Flowering dates of 1-yr plants showed strong effects of accession (with perennials generally showing later flowering) as well as effects of calendar year (Table 3). The 1-yr perennial wheat generally flowered on 31 June in 2009 (26 d after annual wheat, p < 0.0001) and 19 June in 2010 (15 d after annual wheat, p < 0.0001), while the average flowering date was 11 June for perennial rye in both years (11–12 d after annual rye). The tendency of plants to flower earlier in 2010, due to the difference in weather, was one of the strongest and most distinct trends between the years, and was much more marked in the perennial species than in the annual species.
Flowering date in 2010 showed strong differences between 1- and 2-yr perennials (Table 6). When comparing the 1-yr and regrowth perennial plants in 2010, we found that flowering date was significantly affected by accession and by age, but not by the interaction of the two; thus all 2-yr perennials showed a similar shift in flowering date relative to 1-yr plants (approximately 7–11 d earlier). Two-year-old perennial wheat plants flowered on 13 June, 9 d after annual wheat (p < 0.0001), while 2-yr-old perennial rye plants flowered at approximately the same time as annual rye on 3 June. Perennial wheat accessions did not significantly differ in terms of flowering date. To sum up, the strong age effects on early-season plant height and flowering date reflect an overall “shift” in phenology towards earlier in the year. It is interesting that plant age, perennial vs. annual life history, and calendar year all seem to have effects of similar magnitude on flowering date (shifting by about 10 d in either direction).
Yield of One-Year-Old Plants
This is one of the first reports of yield potential for perennial wheat and rye that clearly separates effects of plant age from effects of year-to-year weather variation. This allowed testing of our hypotheses that first-year grain yields in perennial cereals would be lower than annuals, and that perennial grain yields would decline with increasing plant age. As expected, 1-yr yield in perennial wheat and perennial rye was consistently lower than their annual analogs: perennial rye yielded 72% of annual rye, whereas perennial wheat yielded 50% of annual wheat. Overall, 1-yr perennial cereal yields remained steady for both years of the study, despite widely varying weather (hotter, drier conditions in 2010 contrasted with 2009).
The grain yields we observed for perennial cereal genotypes relative to annuals were comparable to previous results for these improved accessions. A trial of 31 perennial wheat lines in Washington State included the four lines we studied, and reported yields that varied (as a proportion of annual wheat) from 28% (P15) to 51% (P19) (Murphy et al., 2010). A recent study in Australia compared more than 90 perennial cereal lines to the annual wheat cultivar Wedgetail, with about half grown as single rows (Hayes et al., 2012). Among the 40 lines which showed evidence of PSCR, grain yield averaged 34% of annual wheat. The generally lower yield potential than in our study was not surprising given that a very wide range of genotypes was included, including less developed accessions. The four perennial wheat lines we investigated were included in the Australia study, and achieved similar 1-yr grain yields to those found in our study (relative to annual wheat). Although no 2-yr seed yields were observed, 1-yr yields in Australia were approximately 30% of annual wheat for P3, 25% for P11, 70% for P15 and 50% for P19 (Hayes et al., 2012). In absolute terms, higher yields were achieved in Australia than in Michigan. P19, for example, produced 2.2 Mg ha−1grain yield in Australia, compared to 1.7 Mg ha−1 in Michigan.
Perennial rye yield was 1.3 to 1.4 Mg ha−1 in our study, about 60% of previously reported values (e.g., 2.2–2.4 Mg ha−1 in central Europe; Reimann-Philipp, 1986; Weik et al., 2002). However, our perennial rye yields were close to those of annual rye; this is suggestive that in our marginal yield environment perennial rye is able to preform well, relative to annual rye. Studies of the perennial rye cultivar ACE-1 in western Canada showed yields of 2.5 Mg ha−1, averaging over 1- to 3-yr-old plants; this was approximately 55% of annual rye yield at that site (Acharya et al., 2004). We do not have a clear explanation for why perennial rye yields are high relative to the annual analog, compared to perennial wheat. It may be that annual wheat has been subject to more intense selection for high yields than annual rye, and the annual wheat yield threshold is a harder target to meet. Alternatively, it could be that perennial rye is a more genetically stable hybrid than perennial wheat amphiploids and thus able to perform better, or possibly perennial wheat could suffer from inbreeding yield depression; clearly, further studies are required.
The overall low yields of perennial cereals reflected reproductive allocation and kernel mass. This is consistent with previous studies of perennial sorghum, which showed 35% lower kernel mass and 16% lower reproductive allocation than annual sorghum (Piper and Kulakow, 1994) as well as ACE-1 perennial rye which had lower harvest index but greater biomass than annual rye (Acharya et al., 2004). Modest grain yields in perennials may also have reflected low initial plant population density. In both years the perennial lines generally had fewer growing plant stems in the early spring, as well as (in perennial wheat) fewer reproductive tillers, relative to annuals. Interestingly, the shorter period for pre-winter growth in the fall of 2009, does not appear to have reduced yields in 2010 relative to 2009. Growing conditions before vernalization can have a very strong effect on subsequent year yields in cool-season perennial grasses (Chastain and Young, 1998), but such effects were not seen in our study.
Yield of Two-Year-Old Plants
Grain yield and yield components in perennial wheat remained generally consistent over time, with few detectable differences between 1- and 2-yr-old plants; this contradicted our hypothesis. Hayes et al. (2012) found that 2-yr perennial wheat grain yield was highly variable among the perenniating lines, from nil to markedly higher relative to 1-yr plants (depending on the accession). However, they did not observe sufficient regrowth to monitor 2-yr plants in the perennial wheat lines that we tested.
We found that perennial rye maintained equivalent yields in 2-yr plants relative to 1 yr. This is consistent with an early study of a perennial rye cultivar Permontra (Reimann-Philipp, 1986). Our findings are in conflict with a 73 to 88% decline in yield observed in a 2002 field study; however, regrowth in that experiment may have been compromised by heavy weed pressure in Year 2 (Weik et al., 2002).
Overall, yields were maintained at comparable levels over 2 yr in both of the perennial cereals studied, an encouraging result. Field studies conducted on perennial forage grasses can further illuminate the potential for second-year seed yield, although these forage grasses have not been selected for high allocation to reproduction. A study of mission grass [Pennisetum polystachyon L. (Schult.)] and gamba grass (Andropogon gayanus Kunth.), involving spaced plants, found increases in seed yield between Years 1 and 2, followed by a decrease in subsequent years (Mishra and Chatterjee, 1968). Seed yields doubled from the first to the second year in desert wheatgrass [Agropyron desertorum (Fisch. ex Link) Schult.; Canode and Law, 1978] and also increased in red fescue (Festuca rubra L.; Canode, 1968). In contrast, yields of Kentucky bluegrass (Poa pratensis L.; Chastain et al., 1997) and orchardgrass (Dactylis glomerata L.; Fulkerson, 1980) initially remained stable, whereas Russian wildrye (Elymus junceus Fisch.; Lawrence and Ashford, 1964) and timothy (Phleum pratense L.; Fulkerson, 1980) yields declined rapidly. Species appear to markedly influence plant age effects on yield, and our study provides new information concerning second year yield of advanced perennial cereal lines.
Biomass and Allocation
As yet, little is known about the biomass production potential of perennial cereals, in spite of the fact that increased forage production could be an important secondary product of these species (Bell et al., 2008). In our study biomass was equivalent or higher in perennial wheat and rye, compared to annual analogs, while perennials allocated a lower fraction of biomass to reproduction. However, biomass was highest in an annual wheat: Frankenmuth is an older annual wheat cultivar that is used for straw production as well as grain (Table 4). Biomass remained comparable between 1- and 2-yr-old perennial plants (Table 5). This contrasts with a field study in western Canada that found biomass declined in perennial rye from 14 to 7.5 t ha−1 between the first and second year, where multiple harvests were conducted each year (Acharya et al., 2003).
In general, biomass yields increase over the initial years of production in perennial grasses: for example, miscanthus (Miscanthus spp.) biomass production tripled between the first and second year (Clifton-Brown and Lewandowski, 2002) and similar increases in yield were observed over 3 yr in giant reed (Arundo donax L.) (Mantineo et al., 2009) and in switchgrass (Schmer et al., 2010). We found that reproductive allocation, reflected by harvest index, remained constant between 1- and 2-yr-old plants. This is consistent with the hypothesis that age has little effect on seed production or reproductive allocation in these perennial cereals. Perennial forage grasses may show increases in harvest index between the first and second years (e.g., tall fescue, Festuca arundinacea Schreb.) or decreases (Kentucky bluegrass, Poa pratensis L.), depending on the species (Fairey and Lefkovitch, 2001).
Production of forage biomass may be a critical dual benefit of perennial cereals, enhancing economic viability (Bell et al., 2008). The preliminary findings here indicate moderately taller plants but minimal biomass produced by perennial wheat and rye lines, relative to annuals. This is suggestive that early-season grazing could be a viable option for perennial cereals, if they are successfully selected for greater biomass production. Overall, we acknowledge limitations in seed quantity and agronomic knowledge related to these new perennial cereal accessions may have limited the biomass potential expressed in our study.
The high rate of regrowth in the fall of 2009 was encouraging, given that previous work on these lines in Washington had found only 40% PSCR in the fall (Murphy et al., 2010). We note that initially these accessions were selected for vigorous PSCR as isolated plants; however, regrowth within dense population density stands (as required for agronomic performance) has proven less reliable. Hayes et al. (2012) found that only a minority of 90 perennial cereal accessions studied showed regrowth, and specifically almost zero PSCR was observed in Australia for the four perennial wheat lines we studied (ratings of 1.0 for P3, 0 for P11 and P19, and 0.6 for P15, compared to 7.2% for perennial rye). Overall, we found strong year-to-year variation in the PSCR ability of these perennial plants. The unusually hot and dry weather in 2010 may have been an important factor underlying the failure to regrow (Fig. 1). The extreme variability of regrowth in perennial wheat and rye indicates that further breeding efforts should prioritize vigorous regrowth and reliable perenniality.
Our findings regarding phenology were consistent with the general trend for perennial grasses to flower later than closely related annuals (Garnier et al., 1997). We also found strong age-related effects on flowering. Older perennial wheat flowered 10 d later than younger ones, whereas in perennial rye age had no effect on flowering date (Table 6). In a study of sea beet, for each increase of 1 yr in age, plants flowered approximately 1.3 d later (Van Dijk, 2009). Weik and colleagues (2002) found that 2-yr-old perennial rye plants flowered 2 to 7 d later than 1-yr-old plants. However, older plants flowered earlier in perennial Lupinus populations (Bishop and Schemske, 1998). Our study suggests that calendar year, perennial growth habit, and plant age can all exert effects of similar magnitude on flowering date (ca. 1–2 wk) which makes separating age and year effects important for future studies.
The phenological differences between perennial and annual cereals observed could affect the agronomic potential of perennial wheat and rye. The later flowering dates of perennial cereals could either increase susceptibility or grant some degree of protection from pathogens, depending on summer weather patterns. Climate change could also affect perennial wheat yields negatively in regions where summers are becoming warmer and drier. The deeper root systems of perennial cereals might, conversely, help with drought survival at certain portions of the flowering period.
Yield potential of perennial cereals did not decline over multiple years, and perennial wheat and rye consistently produced more than 50% of their respective annual counterparts. A threshold of 50% annual yield has been proposed for economic feasibility of perennial cereals in the Australian context. For perennial wheat to be economically feasible in places other than marginal lands, however, modest seed yields will need to be coupled with increased biomass production for forage, and our experiment did not find evidence of high forage biomass potential. Clearly, growth, biomass, and yield potential of perennial cereals requires in-depth investigations through longitudinal studies in diverse ecological and socioeconomic contexts.
Overall, perennial cereal yields were maintained at 50 to 73% of annual yields; this held across diverse weather conditions, and for the establishment year as well as a regrowth year. Perennial wheat and rye thus appear to be nearing the threshold of being viable crops although reliable regrowth remains a challenge at our site. However, our study provides evidence for substantial regrowth and perenniality for only 1 yr. The initial objective had been to continue the study for a longer time period: however, poor regrowth in fall 2010 allowed us to observe only 2 yr of growth and further studies are required. Plant breeding efforts are required that focus on regrowth vigor, increasing biomass as well as grain yield, and selecting for larger kernel size as a component of grain quality. The consistency of yield and yield components over multiple ages and years indicates that progress has been made on developing perennial wheat and rye accessions, and perennial rye shows potential as a grain crop for marginal areas.