Phenotyping Winter Dormancy in Switchgrass to Extend the Growing Season and Improve Biomass Yield

Switchgrass is a prominent bioenergy crop. Like most perennial warm season species, switchgrass undergoes growth suspension in winter as a surviving strategy in temperate climates to protect their meristems from cold injuries and dehydration, while storage organs below ground drive spring regrowth when conditions become favourable. In this paper, we describe a reliable phenotyping method for winter dormancy in switchgrass using various traits including regrowth height after clipping in early fall (FRH), senescence percentage, date of spring regrowth (SRD), and flowering date (FD). FRH and senescence percentage appear to be reliable indicators of the onset of winter dormancy, whereby accessions that initiated dormancy early have a low FRH and a high senescence percentage. Even though it is difficult to have an exact assessment of the duration of dormancy because it is hard to determine with precision the date of growth suspension, SRD can be used as a surrogate indicator of the duration. Flowering date showed low correlations with all the traits and biomass yield suggesting that it may not be a reliable indicator for winter dormancy in switchgrass. Combining the variables FRH, senescence, and SRD in a selection index may provide a reliable tool to phenotype winter dormancy in switchgrass. The strong correlation of these variables with biomass yield makes them useful candidates for the manipulation of the duration of dormancy to increase the growing season and consequently improving biomass production. In southern regions with mild winters, it might be possible through intense selection to develop germplasm with much reduced dormancy or even non-dormant switchgrass germplasm.


Introduction
Seasonal changes have a great impact on plant development and crop production. For warm season grasses like switchgrass (Panicum virgatum), the transition from a long-day, hot summer to a short-day cold winter will induce a phenomenon called winter dormancy, defined by the lack of visible growth [1] [2].
It is an adaptive mechanism that plants evolved in their environment of origin to enable survival during threatening environmental conditions [3]. One way a perennial plant become dormant is by terminating meristem growth and becoming unresponsive to growth promoting factors [4]. The suspension of growth is essential because cold winter temperature can have an adverse effect on the normal functions of plants such as photosynthesis, cellular transport, and the ability to deal with reactive oxygen species [5], while impacting the plants indirectly through the formation of intracellular and extracellular ice. Intracellular ice crystals can expand the extracellular space and damage the cell structure, while extracellular ice formation can reduce the availability of water for absorption and thus leading to cell dehydration [6].
Dormancy has been defined and classified with regard to the initial physiological reactions leading to dormancy [7] and the external factors that trigger dormancy [8]. Lang [9] and Lang et al. [7] [10] proposed the terms ecodormancy, paradormancy, and endodormancy to describe three types of dormancy.
Ecodormancy is the inhibition of growth by temporary unfavorable environmental conditions. Paradormancy refers to the inhibition of growth by signals from distal organs. Endodormancy results from the inhibition of growth by internal bud signals. Rohde and Bhalerao [4] described endodormancy as "the inability to initiate growth from meristems under favorable conditions".
Winter dormancy in switchgrass is a type of endodormancy where plants sense the changes in the duration of photoperiod and temperature and become dormant. Dormancy is usually accompanied by senescence, a degeneration process resulting from programmed cell death [11]. In switchgrass like all perennials senescence is restricted to the above ground part of the plants and not the meristems and below ground structures that enable the plant to resume growth in spring [11]. Senescence is also accompanied by translocation of nutrients to below ground storage organs and reduction in metabolic activity of the crowns, rhizomes, and associated tiller buds that remain dormant throughout winter [12]. Delaying aerial senescence can lead to extended plant's growing season and significantly increases yield as long as the plant still undergo dormancy and nutrient remobilization [13].
The growth cycle of switchgrass can be partitioned in three major phases, winter dormancy, new regrowth in spring, and flowering in mid-summer [12] [ 14]. The reserves stored before dormancy will later drive regrowth in spring when conditions become favorable for growth [13]. Switching from the vegetative tiller meristems to reproductive tillers and flowering are driven by the perception of appropriate photoperiod and temperature signals [13]. Switchgrass upland ecotypes, which are widely found in northern latitudes, enter dormancy earlier than lowland ecotypes that are adapted mostly to southern parts of North America [15]. Lowland ecotypes remain vegetative longer than the upland ecotypes and thus have higher yield, particularly when grown in southern locations [16]. However, despite being low biomass yielders, the upland ecotypes possess the advantage of being more winter hardy and more resistant to cold temperature [15]. Lowry et al. [17] reported that the upland ecotypes can grow in hardiness zones 2 -7, while lowland ecotypes are limited to the southern zones of 6 -10, but both ecotypes can be found in the transition zone [18]. Eight regional gene pools or cultivar deployment zones were described based on adaptation to different photoperiod and temperatures. These largely differ in the time of spring emergence and flowering, cold and heat tolerance, and the onset of winter dormancy [18] [19].
The principal use of switchgrass since the 1940s has been for pasture and grazing in the Great Plains and eastern region of North America [18].  [26] for Watkinsville, Georgia are summarized in Figure 1. The highest temperature was recorded in July (32.0˚C) and the longest day occurred in June (14.4 h), while the lowest temperature and shortest day were observed in January (0.2˚C) and December (9.9 h), respectively.

Dormancy Phenotypic Data
Dormancy phenotypic data evaluated consisted of 1) Level of senescence (senes-

Data Analysis
The data was analyzed using the Proc Mixed procedure in SAS 9.4 for windows

Results
Fall regrowth height data showed a normal distribution with a slight left skewness, signifying a high proportion of lower FRH values in the population ( Figure   3(A)). Senescence % distribution was also normal but skewed to the right, indicating a high proportion of higher senescence values in the population ( Figure   3(B)). This is expected because most of the upland and intermediate genotypes enter dormancy and senesce much earlier than lowland accessions. The data distribution for SRD was normal but with some outliers in both tails (Figure 3(C)).
For FD, the distribution was less bell-shaped and left skewed, suggesting a wide window of flowering and a large variation of flowering dates in the population ( Figure 3(D)).
Mean FRH after clipping was significantly different (p < 0.01) between the 36 accessions and two checks (    (Figure 4(A)). The intermediate accessions exhibited higher regrowth height than the upland, but were lower than the lowland accessions (37.18 ± 1.59 cm) (Figure 4(A)).
Senescence percentage based on image analysis was significantly different (p < 0.01) between the accessions and showed a significant (p < 0.01) interaction between years and accessions (    NDVI values showed significant differences among accessions (p < 0.01) ( The date of initiation of spring regrowth (SRD) was significantly different among accessions (p < 0.01) and showed an interaction between years and accessions (p < 0.01) ( Table 2  and showed an interaction between years and accessions (p < 0.01) (  (Figure 4(D)).
Dry biomass yield was significantly different among accessions (p < 0.01) and there was no significant interaction with years ( Table 2). Because of the non-significant accession by year interaction, dry biomass yield was averaged across years ( Biomass yield showed a significant positive correlation with fall regrowth height after clipping (r = 0.6, p < 0.01) and flowering date (r = 0.44, p < 0.01), while it was negatively correlated with senescence % (r = −0.56, p < 0.01) and spring regrowth date (r = 0.62, p < 0.01) ( Table 9). Senescence % based on image analysis showed a highly significant correlation with NDVI (r = −0.92, p < 0.01). Fall regrowth height after clipping showed a low positive correlation with date of flowering (r = 0.27 p < 0.05) and high negative correlations with senescence % (r = −0.79, p < 0.01) and SRD (r = −0.46, p < 0.01). Senescence % was positively correlated with SRD (r = 0.58, p < 0.58) but showed a low negative correlation with FD (r = −0.39, p < 0.01).

Discussion
Switchgrass production depends on environmental cues to synchronize growth with favorable environmental conditions, and dormancy is triggered when conditions are unfavorable in winter [17]. Photoperiod and temperature are important factors determining plant metabolism in warm season grasses. Most metabolic pathways such as photosynthesis, respiration, and growth processes are catalyzed by enzymes which activities are influenced by temperature, and thus are affected when temperature drops. Switchgrass growth is much greater under long days than short days [29] and flowering is usually delayed under long days.
Optimal harvest time for switchgrass has been debated, with most studies suggesting early fall season. Delaying harvest to winter may reduce biomass yield by up to 40% [30]. Johnson and Gresham [31] found a decrease in yield when switchgrass was harvested in spring compared to harvest in fall with significantly R. M. Razar, A. Missaoui    higher N, P, and S concentration in fall harvest compared to spring harvest [31].
Gamble et al. [32] reported lower biomass yield harvested in winter compared to late summer-fall and late spring. Concentration of N was not different between Journal of Sustainable Bioenergy Systems harvests but P and K concentrations were decreased enormously from late summer to late spring harvests [32]. The decrease in biomass yield reported in these studies might be the result of remobilization of nutrients from aboveground structures to belowground storage organs during the onset of dormancy in fall and senescence, leading to a low biomass yield for over-wintered switchgrass plants. The significant differences observed in our study among accessions in regrowth height after clipping in early fall, in addition to the differences in senescence are a clear indication that some genotypes continued growing and accumulating biomass under reduced temperatures and short photoperiod. Over the two years, the lowland accessions showed higher regrowth height and lower senescence than intermediate and upland accessions (Figure 2). This is a clear indication that the lowland accessions exhibit a delayed onset of seasonal dormancy. The high correlation between NDVI and senescence percent based on image analysis (r = −0.92) suggests that NDVI is an effective tool to quantify senescence and can be used as a substitute to the laborious and time-consuming image analysis procedure. Plants showing high senescence % will have a low NDVI reading and plants with low senescence will have a high NDVI rating.
The differences in spring regrowth date are also an indication of the variability in the duration of winter dormancy between accessions. The lowland accessions exhibited on average lower values for SRD suggesting that they exited dormancy much earlier than the intermediate and the upland accessions even though they remained actively growing much later in the fall as indicated by the regrowth height after clipping in early September. The earliest lowland accessions exited dormancy 12 days earlier than the earliest upland (day 64 vs. day 76 from January 1 st ). Spring regrowth initiation was negatively correlated with date for flowering, suggesting that the accessions that started growth early in spring had a later flowering time in the season. These accessions are predominantly lowland. On the other hand, flowering date showed a low correlation with regrowth height in the fall and senescence suggesting that flowering date may not be a reliable indicator of the onset of dormancy. Variability in the duration of the flowering period was reported, with northern ecotypes taking one week between the emergence of inflorescence to the beginning of anthesis, and southern ecotypes taking between 4 and 6 weeks [33]. Van Esbroeck et al. [34] found a delay in panicle emergence and a longer duration of panicle exertion in the upland switchgrass cultivar Cave-in-Rock when exposed to longer photoperiod and suggested that the delay was associated with increase in the period between the sequential emergences of leaves on the main stem of the plant. Flowering date also showed the lowest correlation with biomass yield compared to the other dormancy indicators. Fall regrowth height showed a high positive correlation with biomass yield, while spring regrowth initiation showed a strong negative correlation with biomass yield, suggesting that the accessions that underwent shorter durations of winter dormancy accumulated more biomass and therefore higher yield. Journal of Sustainable Bioenergy Systems Based on temperature and photoperiod charts (Figure 1), the ambient temperature in the study site declines below the favorable levels for warm season grasses around late August while daylight continues to decrease since early August.
Evidences from studies attempting to establish the optimal time for switchgrass harvest point to early fall as the right harvesting time based on maximum yield and nutrient recycling [30] [31] [32] [35] [36] [37]. These findings provide indirect evidence for the window of end of growth and seasonal nutrient remobilization known to be connected to the onset of endodormancy. In a study measuring seasonal changes in N content in switchgrass, Wayman et al. [35] found the highest N content in above ground structures being in June and started to decrease from September until late fall while N concentration in below ground structures increased starting in August. Piecing together all this information, we rationalize that the window to phenotype the onset of winter dormancy in switchgrass is most likely the end of August and early September when temperature and daylength fall below the thresholds favoring growth, and when plants start remobilizing nutrients to the belowground storage organs.
The application of this phenotyping procedure will allow selecting a suitable combination of switchgrass germplasm with very short duration of winter dormancy or even non-dormant provided incorporation of cold tolerance to overcome the occasional freeze. Since the growing degree-days in the southeastern US are higher than the northern regions, it might possible to implement a management system based on multiple cuts per year instead of the traditional one harvest system leading to a significant increase in feedstock biomass yield.
Future research work will focus on identifying a set of standard checks with varying degrees of dormancy that can be used to standardize dormancy ratings across breeding programs. Winter dormancy appears to be quantitative, therefore genetic mapping of the underlying genetic loci and development of genomic resources will make selection more efficient.

Conclusion
FRH and senescence percentage appear to be reliable indicators of the onset of winter dormancy, whereby an accession that initiated dormancy early will have a low FRH and a high senescence percentage. Even though it is difficult to have an exact assessment of the duration of dormancy because it is hard to determine with precision the date of growth suspension, SRD can be used as a surrogate indicator of the duration. Combining the three variables in a selection index may provide a reliable tool to phenotype winter dormancy in switchgrass. The strong correlation of these variables with biomass yield makes them useful candidates for the manipulation of the duration of dormancy to increase the growing season and consequently improving biomass production. In southern regions with mild winters, it might be possible through intense selection to develop germplasm with much reduced dormancy or even non-dormant switchgrass germplasm.
Incorporating cold tolerance would provide an insurance against potential ex-posure to infrequent low temperatures in the winter season of southern environments.