Crop Science
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Published online 1 January 2005
Published in Crop Sci 45:388-398 (2005)
© 2005 Crop Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Related articles in Crop Science
Right arrow Similar articles in this journal
Right arrow Similar articles in Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Web of Science (10)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Casler, M. D.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Casler, M. D.
Agricola
Right arrow Articles by Casler, M. D.
Related Collections
Right arrow Other Forage Crops
Right arrow Crop Genetics

Ecotypic Variation among Switchgrass Populations from the Northern USA

M. D. Casler*

USDA-ARS, U.S. Dairy Forage Research Center, Madison, WI 53706-1108

* Corresponding author (mdcasler{at}wisc.edu)


   ABSTRACT
TOP
 NOTES
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Switchgrass (Panicum virgatum L.) is a widely adapted warm-season perennial that has considerable potential as a biofuel crop. Broad species adaptation, natural selection, and photoperiodism have combined to create considerable ecotypic differentiation in switchgrass. The objective of this study was to characterize phenotypic variability among switchgrass ecotypes collected from prairie remnants in the northern USA. Thirty-eight switchgrass collections from 33 prairie-remnant sites and 11 switchgrass cultivars were evaluated for 2 yr at two locations (Arlington and Marshfield, WI) for nine variables: biomass yield, survival, dry matter, lodging, maturity, plant height, holocellulose, lignin, and ash. Autocorrelations, measuring spatial variation, and correlations between phenotypic distances and geographic distances were all nonsignificant. A small amount of variation for maturity, lodging, holocellulose, lignin, and ash could be attributed to latitude and/or longitude of the collection site. Populations from several of the westernmost collection sites clustered with cultivars from the Great Plains, suggesting an ecological basis for some of the phenotypic variation observed. However, there was a considerable amount of phenotypic variability between populations from collection sites in close proximity to each other. Hardiness zones (defined largely by temperature extremes) and ecoregions (defined largely by soil type and historic vegetation) partly define the phenotypic characteristics for many switchgrass populations collected from prairie remnants. Most switchgrass populations can be utilized for conservation and restoration projects throughout a combined ecoregion and hardiness zone without undue concern over contaminating, diluting, or swamping the local switchgrass gene pool.


   INTRODUCTION
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
SWITCHGRASS is a widely adapted warm-season perennial that has considerable potential as a biofuel crop. Switchgrass is capable of producing a high yield of biomass across a wide geographic range (Sanderson et al., 1996). Switchgrass is widely adapted and suitable for use on marginal, highly erodable, and droughty soils (Moser and Vogel, 1995). It has the potential of sequestering large amounts of atmospheric carbon in permanent grasslands (Sanderson et al., 1996). Switchgrass can also provide excellent nesting habitat for migratory birds (Paine et al., 1996). The combination of heat, cold, and drought tolerance allows switchgrass to grow in nearly all ecosystems east of the Rocky Mountains and south of Hudson Bay, ranging from arid conditions in the shortgrass prairie to marshland and open woodland (Hitchcock, 1951).

Switchgrass is a highly heterozygous, self-incompatible, and outcrossing species, characterized by a ploidy series from 2n = 2x = 18 to 2n = 12x = 108 (Nielsen, 1944) and two distinct cytotypes, upland and lowland (Hultquist et al., 1996). Upland and lowland cytotypes tend to be genetically and phenotypically distinct from each other (Gunter et al., 1996; Sanderson et al., 1996). Upland cytotypes tend to be adapted to the mid- and northern latitudes of the USA, while lowland cytotypes tend to be adapted to the southern USA (Brunken and Estes, 1975; Casler et al., 2004). Furthermore, there is genetic variation for adaptation within both upland and lowland cytotypes. Within cytotypes, strains from more northern latitudes tend to have higher relative biomass yield and survival at more northern sites, while southern strains show the opposite reaction (Casler et al., 2004). Genetic responses to latitude may be complex, resulting from genetic variation for photoperiodism, cold tolerance, or heat tolerance (Casler et al., 2004).

The strongly photoperiodic nature of switchgrass, combined with apparent genetic variation for heat and cold tolerance, indicates that switchgrass strains have a limited adaptation zone compared to the species as a whole. Some strains, such as ‘Cave-in-Rock’, an upland ecotype originating in southern Illinois, are adapted across a wide geographical region (Casler and Boe, 2003; Casler et al., 2004; Hopkins et al., 1995). Other strains have much more restricted adaptation zones, limited by unknown factors related to both latitude and longitude (Casler and Boe, 2003; Madakadze et al., 1998).

Despite the spread and duration of agriculture in eastern North America, there remain hundreds of remnant prairie sites, protected by public or private organizations (Hopkins et al., 1995; Hultquist et al., 1997). Most switchgrass cultivars are either seed increases of source-identified collections or products of a limited number of breeding cycles tracing to many of these remnant-prairie sites (Alderson and Sharp, 1994). Furthermore, most switchgrass cultivars derive from collections made in the Great Plains region of the USA, where switchgrass-dominated prairie remnants are larger and more frequent than in the eastern USA. New collections from the eastern half of the USA may be useful in breeding and selection of switchgrass cultivars for both biofuel and forage uses in this region. The objective of this study was to characterize phenotypic variability among switchgrass ecotypes collected from prairie remnants in the north central and northeastern USA.


   MATERIALS AND METHODS
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
A total of 78 switchgrass collections were made from 59 sites in Minnesota, Wisconsin, Michigan, Indiana, Ohio, and New York in 1997 and 1998. Some of the collection sites were sufficiently large or variable to warrant multiple collections from these sites. Multiple collections from a site were generally made when there was a significant change in soil type, aspect, or habitat. Seeds were stored at room temperature until December 1998. A sample of seed of each accession was chilled at 3°C for 3 wk and planted in plastic seedling tubes containing a 1:1 mixture of silt loam soil and peat moss. Seed dormancy problems limited the study to a total of 38 accessions from 33 sites (Table 1). In January 1999, seedlings of 11 cultivated switchgrass populations were germinated without pre-chilling.


View this table:
[in this window]
[in a new window]

  		Table 1. Latitude, USDA hardiness zones, and Bailey's ecoregion classification for 38 switchgrass populations collected on prairie-remnant sites in the northern USA.

 
In late May 1999, seedlings were transplanted to two field sites near Arlington, WI [Plano silt loam (fine-silty, mixed, mesic Typic Argiudoll); 43°20' N, 89°23' W] and Marshfield, WI [Withee silt loam (fine-loamy, mixed Aquic Glossoboralf); 44°39' N, 90°08' W]. The 49 populations were arranged in a randomized complete block with six replicates at each location. Plots consisted of 10 seedlings, two rows of five, spaced 0.3 m apart. Adjacent plots were 0.9 m apart. Weeds were controlled by application of 1.12 kg ha–1 alachlor [2-chloro-N-2,6-diethylphenyl)-N-(methoxymethyl)-acetamide] with 0.56 kg ha–1 bromoxynil [3,5-dibromo-4-hydroxybenzonitrile] and 0.07 kg ha–1 imazethapyr {( ± )-2-[4,5-dihydro-4-methyl-4-(1-methylethyl)-5-oxo-1H-imidazol-2-yl]-5-ethyl-3-pyridine-carboxylic acid}.

Plots were fertilized before spring growth in 2000 and 2001 with 112 kg N ha–1. Herbicide was applied before spring growth as previously described. Plots were harvested in late August 2000 and 2001 with a flair harvester to determine biomass yield per plot. Relative maturity and plant height were determined on each plant before harvesting. Relative maturity was determined using the 0-to-8 scale of Casler (1988), where 0 = vegetative and 8 = postanthesis. Plant height was measured from the soil surface to the tip of the highest panicle. A whole-plant tissue sample was clipped from five plants per plot before harvesting. Samples were dried at 60°C for 5 d and used to compute the concentration of dry matter in plant tissue. Survival was determined immediately after harvest by counting the surviving plants in each plot. Biomass yield per plot was adjusted to a dry matter basis.

Dry-matter samples were ground through a 1-mm screen of a Wiley-type mill and scanned on a near-infrared reflectance spectrophotometer (NIRS). A calibration set of 75 samples was chosen by cluster analysis of the reflectance data (Shenk and Westerhaus, 1991). Calibration samples were sequentially analyzed for neutral detergent fiber (NDF), acid detergent fiber (ADF), and acid detergent lignin (ADL), and ash with the ANKOM Fiber Analyzer (ANKOM Technology Corporation, Fairport, NY) and the procedures described by Vogel et al. (1999). Values of NDF, ADL, and ash were predicted for all samples using a single calibration equation per variable, respectively: SEP (standard error of prediction) = 11.6, 6.2, and 8.8 g kg–1; R2 = 0.83, 0.80, and 0.73. Following calibration and prediction, holocellulose (cellulose + hemicellulose {approx} total structural sugars) was estimated as the difference between NDF and ADL, and lignin was expressed on an NDF basis.

Data were analyzed by analyses of variance assuming populations and replicates to be random effects and years and locations to be fixed effects. Broad-sense heritability was computed as

Formula
where variance components were defined by their subscripts (P = populations, L = locations, R = replicates, Y = years, and e = error) and were estimated by equating mean squares to their expectations (Gaylor et al., 1970). Population x location interactions were described in terms of the rank correlation between population means at the two locations and the phenotypic variance at each location (Muir et al., 1992). Sums of squares for populations was partitioned into hardiness zones (3 df), ecoregions (1 df), and populations within regions (33 df), on the basis of the classification in Table 1.

The 49 population means for all nine variables were subjected to principal components analysis. Principal components were used as input variables in a cluster analysis, using the unweighted pair-group method of averages (UPGMA). Principal components were used in the cluster analysis because the large collinearity among several of the phenotypic variables would result in excessive weighting to the correlated variables. Six groups were identified from the cluster analysis and described on the basis of group means and standard deviations. Because of the presence of population x location interactions, principal components analysis, and cluster analysis were applied separately to data from each location. Because the results and conclusions from the separate principal components and cluster analyses were similar for the two locations, a single set of analyses, on the basis of population means over two locations, was used.

Spatial variation for phenotypic variables measured on the 38 remnant prairie populations was investigated by three methods. First, linear and multiple linear regression were computed for variable means on latitude and longitude of the collection site.

Second, seven distance classes were created based on pairwise geographic distances between remnant prairie sites: 0 to 5, 6 to 50, 51 to 100, 101 to 200, 201 to 400, 401 to 800, and 801 to 1600 km. Moran's I, a spatial autocorrelation coefficient, was computed for population means of each variable within each distance class (Sokal and Oden, 1978). A permutation test was conducted with 999 randomizations of the vector of population means (Smouse and Peakall, 1999).

Third, Euclidean phenotypic distance values were computed among the 38 remnant prairie populations by the formula

Formula
where Mik and Mjk are means for populations i and j, respectively, and variable k; summation was across nine variables (k = 1,...,9). Population means were standardized to µ = 0 and {sigma} = 1 before computation of PD. Phenotypic distances were converted to normalized phenotypic distances (NPD) by dividing each value by the mean phenotypic distance. The NPD was adapted from Smouse and Peakall (1999). The Mantel test of matrix correlation between the NPD matrix and the geographic distance matrix was computed (Smouse et al., 1986). A permutation test was conducted with 999 randomizations of the geographic distance matrix (Smouse et al., 1986).


   RESULTS AND DISCUSSION
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
CONCLUSIONS
 REFERENCES
 
Genotypic variability among populations was significant and broad-sense heritability was moderate to high for all nine variables (Table 2). The population x location interaction was significant for five of the nine variables, but made a significant contribution to the variance of a population mean only for two variables, lodging and survival. For lodging, this interaction was largely due to an almost four-fold difference in phenotypic variance between the two locations. For survival, this interaction was largely due to some significant changes in rank values between locations. Three populations in particular had high survival at Arlington and extremely low survival at Marshfield, contributing to this interaction (data not shown). Population x year and population x location x year interactions were significant only for one and four variables, respectively, and were always relatively small, not involving significant changes in population rankings.


View this table:
[in this window]
[in a new window]

  		Table 2. Analysis of variance results for 49 switchgrass populations evaluated for 2 yr at two locations (Arlington and Marshfield, WI).

 
Because of the population x location interactions for lodging and survival, all subsequent analyses were conducted separately for Arlington and Marshfield data. Because the results and conclusions from these separate analyses were similar or nearly identical across locations, these interactions were ignored for all remaining data presentations. Means over locations and years were utilized for all subsequent analyses and presentations.

The range among prairie-remnant population means was greater than the range among cultivar means for each of the nine variables (Table 3). Generally, the minimum and maximum cultivar means were within the range of the prairie-remnant population means. Cultivars averaged 13.8% higher biomass yield, 15.0% higher survival, 4.6% less dry matter, 0.4% more holocellulose, 3.5% less lignin, and 1.4% less ash than prairie-remnant populations. The higher biomass yield and survival for cultivars likely reflects the effects of selection and breeding. Three types of cultivars are represented in this group. Blackwell, Cave-in-Rock, Shelter, and Summer are ecotypes, direct seed increases of collections from prairie remnants. While ecotypes have not undergone breeding, they have been selected for vigor and other agronomic traits from among other populations collected from prairie-remnant sites, following evaluation in a common nursery. In this study, 12 of the 38 prairie-remnant populations had both mean biomass yield and survival lower than the lowest of the cultivars. Populations such as these would probably not be elevated to cultivar status after an evaluation in a common nursery. NE-HZ4 and Sunburst are ecopopulations, strain crosses of plants from several ecotypes. As with ecotypes, they represent some selection among ecotypes, but no breeding. Finally, the other five cultivars are products of the USDA-ARS breeding program at Lincoln, NE, where there has been considerable emphasis placed on increasing biomass yield and survival.


View this table:
[in this window]
[in a new window]

  		Table 3. Minimum, maximum, and overall means for 38 prairie-remnant switchgrass populations and 11 switchgrass cultivars evaluated for 2 yr at two locations (Arlington and Marshfield, WI).

 
Linear regressions of the 38 prairie-remnant population means on latitude or longitude were significant for five of nine variables, accounting for 13 to 29% of the variability among population means (Table 4). Partitioning of the population sums of squares in analyses of variance give similar results (data not shown). Substantial sums of squares could be attributed to hardiness zones (15–21%) or ecoregions (4–10%) only for three variables: maturity, holocellulose, and ash. Prairie-remnant collections from the northern parts of this region had greater lodging, were earlier in maturity, and had greater concentrations of holocellulose, lignin, and ash (Table 4). The effect of latitude on maturity of prairie-remnant switchgrass collections has been documented on a larger landscape scale (Casler et al., 2004). These results show that this effect is more-or-less continuous, operating across relatively small changes in latitude (40–47°N latitude). There was one obvious outlier in the regression of mean maturity on latitude (Fig. 1) , population 36, collected near Westport, WI (site WP). Removal of this population from the sample increased the R2 from 0.15 to 0.24 and decreased the P value from 0.0154 to 0.0022. This population had a mean maturity score of 1.6, indicating most panicles barely in the boot stage. The unusually late maturity and the blue-green stem coloration of the plants collected from this prairie-remnant site are suggestive of the lowland phenotype, which is typically 2 to 4 wk later in heading than the upland phenotype (Casler et al., 2004; McMillan, 1965). This population may represent an extreme northern population of the lowland phenotype, a hybrid between upland and lowland populations, or a recent human-facilitated introduction of a lowland population into a prairie remnant that was formerly populated by the upland phenotype. Some lowland populations, such as the cultivar Kanlow, have sufficient cold tolerance that a few plants can survive in southern Wisconsin (Casler et al., 2004).


View this table:
[in this window]
[in a new window]

  		Table 4. Linear or multiple linear regression equations for the regressions of 38 switchgrass population means on latitude and/or longitude of prairie-remnant collection sites.

 

Figure 1
View larger version (17K):
[in this window]
[in a new window]

  		Fig. 1. Linear regression of mean maturity vs. latitude of the collection site for 38 switchgrass populations collected from prairie-remnant sites. The regression equation is shown in Table 4.

 
The effect of longitude was always smaller than the effect of latitude, although it was significant for four variables (Table 4). Lodging tended to be higher for the western collections, while holocellulose, lignin, and ash tended to be higher for the eastern collections. These results suggested that latitude-related factors, such as daylength, cold tolerance, and heat tolerance, have an influence on phenotypic variability among switchgrass populations at prairie-remnant sites in the northern USA. Longitude had a small effect on phenotypic variability, largely related to differences between ecoregions (Bailey, 1998). The Prairie Parkland ecoregion is subhumid with precipitation that is almost entirely lost by evapotranspiration. Soils are largely Mollisols, which have a high-humus surface horizon and relatively high pH. The Eastern Broadleaf Forest ecoregion has a humid continental climate in which precipitation exceeds evapotranspiration. Soils are largely Inceptisols, Untisols, and Alfisols with relatively high organic matter, moderately leached with a distinct light-colored leached horizon, and lower in pH than Mollisols. Climatic and/or soil differences between these two ecoregions appears to be partly responsible for some phenotypic variation among prairie-remnant populations of switchgrass.

Autocorrelation analyses (Moran's I) were not significant for any variable, except for a slight spatial trend observed for lodging (P = 0.036). Populations that originated from 51 to 100 km apart tended to be positively correlated with each other (r = 0.31, P = 0.019), while populations that originated greater than 800 km apart tended to be negatively correlated with each other (r = –0.26, P = 0.012). With this slight exception, these results indicated that individual variables measured on switchgrass populations from nearby prairie remnants were not correlated with each other. These results supported the regressions of variable means on latitude and longitude, indicating that most of the phenotypic variability in this collection is unrelated to latitude, longitude, or geographic distance between collection sites.

Finally, normalized phenotypic distances, taking into account all nine phenotypic variables, were not correlated with geographic distances of paired prairie-remnant sites (r = –0.03; Fig. 2) . The majority of population pairs had geographic distances less than 500 km and phenotypic distances less than 1.5. Surprisingly, pairs of populations that had the greatest geographic distances generally had relatively low phenotypic distances, while pairs of populations that had the greatest phenotypic distances were typically less than 500 km distant from each other. There were only seven population pairs that had a geographic distance greater than 800 km and an NPD >2.0. Populations 4 and 5 (both from site HW in central Indiana) and Population 27 (from site BT in western New York) were represented in six of these seven pairs.


Figure 2
View larger version (32K):
[in this window]
[in a new window]

  		Fig. 2. Scatterplot of normalized phenotypic distance vs. geographic distance for 722 paired populations of switchgrass.

 
Six of the prairie-remnant collection sites were represented by two populations that were collected less than 5 km apart. Paired populations within each of these six sites differed for three to eight of the nine variables (Table 5). Within-site variability was significant for all nine variables, ranging from 19 to 43% of the total variability among the 12 population means. Some of the individual differences between paired populations (Table 5) represented more than half the range among all population means (Table 3). Slope, aspect, and habitat were the factors that generally formed the basis for multiple collections within a site. However, there was no single factor that seemed to account for phenotypic differences between collections within a site. Numerous additional examples could be found of phenotypically divergent populations originating in close proximity to each other, resulting in the general lack of autocorrelation or spatial variation at the landscape level.


View this table:
[in this window]
[in a new window]

  		Table 5. Means of paired switchgrass populations collected from different areas of six prairie-remnant sites, based on evaluations over 2 yr at two locations (Arlington and Marshfield, WI).

 
Cluster analysis, based on the nine principal components, resulted in six cluster groups that were separated from each other by normalized distances of 0.93 or greater (Fig. 3) . The Great Plains group contained all but one of the cultivars developed in the Great Plains region, plus 14 prairie-remnant populations (seven of the nine Minnesota populations, six Wisconsin populations, and one Indiana population). The four populations from north-central Minnesota originated in USDA Hardiness Zone (HZ) 3; the Indiana population and the two populations in southeastern Wisconsin originated in HZ5; the remaining populations in this group originated in HZ4 (Cathey, 1990). Three of the Minnesota populations in this group, from the southern part of the state, originated in the Prairie Parkland Ecoregion (Bailey, 1998) which includes much of the central Great Plains of the USA. Thus, there is a strong phenotypic and ecological connection between the cultivars and prairie-remnant populations in the Great Plains cluster group.


Figure 3
View larger version (28K):
[in this window]
[in a new window]

  		Fig. 3. Cluster dendrogram of 49 switchgrass populations, identified by their state of origin, clustered by the UPGMA clustering method on the basis of principal components of nine phenotypic traits. Switchgrass cultivar population numbers were as follows: 70 = Blackwell, 71 = Cave-in-Rock, 72 = Pathfinder, 73 = Shawnee, 74 = Shelter, 75 = Summer, 76 = Sunburst, 77 = Trailblazer, 78 = NE-HZ4-Syn1, 79 = NEearly-HYC3-HDC2, and 80 = NE28-HYC3-HDC2. Cluster groups are identified by vertical lines to the left of population numbers and by names to the right of cluster nodes.

 
The prairie-remnant populations in the Great Plains group were strongly representative of the western portion of the geographic region sampled for this study (Fig. 4) . Many populations within this group had small pairwise distances (Fig. 3), including two populations sampled from Wisconsin site RR (52 and 53) and several pairs of cultivars (NE-HZ4 and NEearly-HYC3-HDC2, Pathfinder and Shawnee, Blackwell, and NE28-HYC3-HDC2). One of the Minnesota populations (22 from site SP) was phenotypically similar to Trailblazer.


Figure 4
View larger version (16K):
[in this window]
[in a new window]

  		Fig. 4. Albers equal-area projection of a portion of the north-central and northeastern USA, showing the location of 38 switchgrass collections made from remnant prairie sites in 1997 and 1998. Each one-letter code identifies the cluster to which each population was assigned in Fig. 3: G = Great Plains, S = HZ4,5-S, N = HZ4,5-N, K = Koro Prairie, H = Howard, and L = Lowland. The site to the east of the Michigan Lower Peninsula is on Hansen's Island in Lake St. Clair. The stars indicate the two evaluation sites, Arlington and Marshfield, WI.

 
The HZ4,5-S cluster group contained the remaining three cultivars plus 12 prairie-remnant populations ranging from Wisconsin to Ohio (Fig. 3 and 4). The HZ4,5-N cluster group contained eight prairie-remnant populations ranging from southeastern Minnesota to western New York. All but two of these populations originated in HZ4 or HZ5, one exception from each group in eastern Michigan (population 13 from site SC and population 14 from site HI). There was a slight, but significant, difference in mean latitude of the collection sites represented in these two groups (41°57' vs. 43°4'; P = 0.05 by t test), hence the name of the two groups. The cultivars in HZ4,5-S originated from HZ5 or the northern portion of HZ6.

The last three cluster groups represented the most phenotypically distinct of the prairie-remnant populations (Fig. 3). The Koro Prairie collection (population 51) was the most unique, followed by the Westport collection (Population 36), and the two collections from Howard Twp. (Populations 4 and 5).

Phenotypic differences among the six cluster groups are illustrated in Fig. 5 , using the first three principal components, which accounted for 81% of the phenotypic variability. The first component described high dry matter, earliness, and high lignin and ash, accounting for 37% of the variability. The second component described high biomass yield and survival and tall plants, accounting for 26% of the variability. The third component described high lodging and holocellulose, accounting for 18% of the variability.


Figure 5
View larger version (15K):
[in this window]
[in a new window]

  		Fig. 5. Scatterplot of the first two principal components (PRIN1 and PRIN2) for 49 switchgrass populations identified by the six cluster groups in Fig. 3 and Table 6. Upper graph includes populations that had a negative value of the third principal component (PRIN3) and lower graph includes populations that had a positive value of PRIN3.

 

View this table:
[in this window]
[in a new window]

  		Table 6. Statistics associated with six cluster groups of 49 switchgrass populations based on principal components of nine variables measured for 2 yr at two locations (Arlington and Marshfield, WI).

 
Members of the Great Plains cluster group were characterized by low values of PRIN1 and mostly high values of PRIN3 (Fig. 5). These populations had relatively low dry matter, lignin, and ash; relatively high lodging and holocellulose; and were relatively late in maturity (Table 6). The cultivars within this group were phenotypically indistinguishable from the prairie-remnant populations for any phenotypic variable or on the basis of the principal components. Collectively, these populations represent potentially valuable germplasm for population improvement and breeding of switchgrass as a bioenergy feedstock. Their relatively high lodging potential was the only negative trait for bioenergy feedstock production.

Members of the HZ4,5-S cluster group were characterized by high values of PRIN2 and mostly low values of PRIN3 (Fig. 5). These populations had relatively high biomass yield and survival, low lodging and holocellulose, and were relatively tall (Table 6). While low lodging potential would be an advantage of these populations, the relatively low holocellulose concentrations would limit their value in breeding switchgrass for bioenergy feedstock production. Holocellulose represents the majority of the fermentable sugars in the cell wall, which comprise the major source of energy in dry-cured switchgrass biomass. A positive phenotypic correlation coefficient between lodging and holocellulose (r = 0.57, P < 0.01) suggested that lodging may be a possible consequence of high holocellulose concentration. Lodging was negatively correlated with lignin concentration (r = –0.41, P < 0.01), suggesting that high lignin may also be a factor contributing to lodging resistance. A high lignin concentration in switchgrass biomass would be detrimental for fermentation, because lignin cannot be broken down during fermentation (Jung and Deetz, 1993). However, a previous study of switchgrass germplasm did not reveal a relationship between lignin and lodging (Casler et al., 2004). Furthermore, the literature suggests that low-lignin grasses may have greater stem flexibility, resulting in greater resistance to bending stress and greater lodging resistance (Casler, 2001).

Members of the HZ4,5-N cluster group were characterized by high values of PRIN1, low values of PRIN2, and low values of PRIN3 (Fig. 5). These populations had relatively high dry matter and lignin; low biomass yield, survival, lodging, and holocellulose; and were relatively short and early in maturity (Table 6). These populations were distinguished from those in the HZ4,5-S group, largely by their reduced survival, biomass, and plant height. Because of these traits, these populations have relatively little value in a breeding program designed to improve the bioenergy feedstock production of switchgrass. However, these populations should be included in germplasm pools created to represent the range of phenotypic variability within this region.

Two populations of the HZ4,5-N cluster group were members of two paired-population collections from individual collection sites (Table 5). Population 23 (site SP in Minnesota) was highly divergent from Population 24, which was a member of the Great Plains cluster group. Population 28 (site YC in Ohio) was highly divergent from Population 29 (site SA in Ohio), which was a member of the HZ4,5-S cluster group. The assignment of paired populations from these sites to different cluster groups provided a measure of the potential for phenotypic differentiation between collections from nearby sites.

The collection from Koro Prairie (population 51 from site KP in Wisconsin) ranked 49th for biomass yield and plant height, 47th for survival, and 2nd in ash concentration among the 49 populations. This was the most extreme phenotype among the 49 populations in the study, with the fifth highest value of PRIN1, the 4th lowest value of PRIN2, and the highest value of PRIN3 (Fig. 4 and 5; Table 6). This collection derived from what appeared to be a single plant, so inbreeding is a potential explanation for the relatively poor survival and vigor of this collection.

The two collections from Howard Twp. (Populations 4 and 5 from site HW in Indiana) had extremely low values of all three principal components (Fig. 5). This was caused by relatively low values of all nine phenotypic variables, although there was some phenotypic variability between the two populations (Tables 5 and 6). Their relatively late maturity was probably the most notable unique feature of these two populations.

Finally, Population 36 (site WP in Wisconsin) was labeled as the Lowland cluster group, because of its similarity to the typical phenotype of the lowland cytotype of switchgrass (Casler et al., 2004). This population had unusually late maturity and the lowest dry matter of the 49 populations (282 g kg–1, compared with the next lowest population with 332 g kg–1). Population 36 was the 3rd tallest population, another indication of a possible lowland cytotype. Population 36 had the lowest values of both PRIN1 and PRIN3 (Fig. 5). Lowland cytotypes of switchgrass are not known north of 38°N latitude (Hultquist et al., 1997). The population collected from site WP may be a relatively recent human introduction of a lowland cytotype or an extremely rare naturally occurring lowland cytotype from 43°N latitude.


   CONCLUSIONS
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
REFERENCES
 
Switchgrass is an allogamous species, resulting in highly heterogeneous and variable populations. For this collection of switchgrass populations, 66% of the genotypic variability occurs within populations (Stendal et al., 2003). In this regard, switchgrass is fairly typical of polyploid, allogamous grasses. Within-population variability creates the potential for natural selection to occur in response to local environmental conditions, such as soil or slope characteristics, climate, daylength, or habitat. Differentiation among sites creates differential selection pressures, diverging populations from each other over hundreds or thousands of years. Site-to-site differentiation may occur over vast geographic distances, caused by differential climate, daylength, or habitat, or over distances of a few meters, caused by soil or slope characteristics.

Phenotypic differentiation among switchgrass populations occurred at regional, landscape, and neighborhood levels in this study. Because so much of the phenotypic variability occurred at landscape and neighborhood levels, there were no autocorrelation or spatial patterns to the variability. Large-scale regional factors were clearly important in distinguishing a group of populations, largely from the western portion of the sampled region. This group of populations originated largely in the HZ4 section of the Prairie Parkland ecoregion or in the HZ3 or HZ4 sections of the Eastern Broadleaf Forest ecoregion. These classifications, combined with the phenotypic similarity of these populations to cultivars from the Great Plains (HZ4 to HZ6 of the Prairie Parkland ecoregion), suggested that daylength and climate were likely factors driving natural selection at many of these prairie-remnant sites. However, these factors could only account for small amounts of phenotypic variability and many of the other populations could not be distinguished on the basis of large-scale latitude, climatic, or habitat factors. Differential soil or slope characteristics or the possibility of human disturbance have created small-scale phenotypic differentiation within hardiness zones, ecoregions, and individual prairie-remnant sites. Human disturbance cannot be ruled out for some sites. Verification of putative prairie-remnant sites is often achieved without written documentation, largely based on oral documentation and institutional or personal memories. Gene migration can occur via pollen or seed, potentially creating phenotypic differentiation without natural selection and unrelated to any climatic or edaphic factors.

Switchgrass is a highly adaptable species with an adaption zone that includes much of eastern North America. Daylength controls the adaptation zone of individual plants or populations of switchgrass, such that most populations cannot be moved north or south more than one hardiness zone without adversely affecting vigor, survival, or flowering (Casler et al., 2004). These results build on this conclusion, suggesting that there may be a longitudinal component to the large-scale adaptive pattern of switchgrass, associated with the Prairie Parkland vs. Eastern Broadleaf Forest ecoregions of the central and eastern USA. More research will be required to verify and refine this conclusion and to determine why some natural populations of switchgrass do not conform to this pattern.

Finally, most populations of switchgrass have an adaptation range that extends beyond their point of origin, at least within their hardiness zone, possibly including one zone north and south of their zone-of-origin, and within their ecoregion of origin (Casler et al., 2004; Casler and Boe, 2003; Hopkins et al., 1995). In some cases, this can include a relatively large geographic region, e.g., HZ5 of the Eastern Broadleaf Forest ecoregion, which includes parts of Wisconsin, Illinois, Indiana, Michigan, Ohio, Pennsylvania, and New York. The observed lack of spatial patterns within ecoregions or hardiness zones suggests that switchgrass populations within combined ecoregions and hardiness zones may be considered as regional populations. Some phenotypic differentiation is present among populations within each regional population, but members of a regional population share some basic genotypic information and phenotypic traits defined, in part, by their adaptation zone. Switchgrass populations within a combined ecoregion and hardiness zone can be utilized for conservation and restoration projects without undue fear of contaminating, diluting, or swamping the local switchgrass gene pool.


   ACKNOWLEDGMENTS
 
I thank the following individuals for collecting switchgrass seed from prairie-remnant sites and for providing provenance information on local sites: Valerie Berglund-Garcia, Bob Berkemeier, Barry Bortner, Steve Breaker, William Bronder, Mitch Cattrell, Rebecca Cifaldi, Don Cree, Dave Dortman, Rick Grooters, Russ Haas, Bob Hanson, Glenn Hartman, Kim Herman, Lee Johnson, Doug Keene, Bruce Knapp, Paul Labus, Nicole McClain, Dan McGuckin, Chris Newell, Julius Pigott, Roger Powell, Mary Jane Reetz, Dennis Reimers, Ivan Reinke, David Stanley, Steve Smith, Ed Stuff, Walter Summers, David Walter, and Greg Wheeler. I also thank Dave Burgdorf and Phil Koch, USDA-NRCS, Rose Lake Plant Materials Center, for organizing the aforementioned individuals to make the collections utilized in this study. I also thank Andy Beal for his tireless and dedicated efforts to collect seed from numerous prairie remnants in Wisconsin and Mark Martin, Wisconsin Dep. of Natural Resources, for assistance in obtaining permission to collect switchgrass seed throughout Wisconsin. I thank Doug Foy, Susan Selman, and Mary Becker for assistance with laboratory data analysis. I thank Ken Vogel for advice, encouragement, and many productive discussions about switchgrass biology and genetics.


   NOTES
TOP
 NOTES
ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
This research was funded in part by Specific Cooperative Agreement 58-5440-7-123 between the USDA-ARS and the University of Wisconsin-Madison, which was a component of the U.S. Department of Energy, Oak Ridge National Laboratory and USDA-ARS Interagency Agreement under contract DE-A105-900R21954.

Received for publication July 23, 2003.


   REFERENCES
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 


Related articles in Crop Science:

THIS ISSUE IN CROP SCIENCE

Crop Science 2005 45: xi. [Full Text]  



This article has been cited by other articles:


Home page
 Crop Sci.Home page
M. D. Casler and E. C. Brummer
Theoretical Expected Genetic Gains for Among-and-Within-Family Selection Methods in Perennial Forage Crops
Crop Sci., May 1, 2008; 48(3): 890 - 902.
[Abstract] [Full Text] [PDF]

Home page
 Crop Sci.Home page
M. D. Casler, K. P. Vogel, C. M. Taliaferro, N. J. Ehlke, J. D. Berdahl, E. C. Brummer, R. L. Kallenbach, C. P. West, and R. B. Mitchell
Latitudinal and Longitudinal Adaptation of Switchgrass Populations
Crop Sci., November 7, 2007; 47(6): 2249 - 2260.
[Abstract] [Full Text] [PDF]

Home page
 Crop Sci.Home page
M. D. Casler, C. A. Stendal, L. Kapich, and K. P. Vogel
Genetic Diversity, Plant Adaptation Regions, and Gene Pools for Switchgrass
Crop Sci., November 7, 2007; 47(6): 2261 - 2273.
[Abstract] [Full Text] [PDF]

Home page
 Crop Sci.Home page
M.D. Casler, K.P. Vogel, and A.C. Beal
Registration of WS4U and WS8U Switchgrass Germplasms
Crop Sci., February 24, 2006; 46(2): 998 - 999.
[Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Related articles in Crop Science
Right arrow Similar articles in this journal
Right arrow Similar articles in Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Web of Science (10)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Casler, M. D.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Casler, M. D.
Agricola
Right arrow Articles by Casler, M. D.
Related Collections
Right arrow Other Forage Crops
Right arrow Crop Genetics

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
The SCI Journals Agronomy Journal Vadose Zone Journal
Journal of Natural Resources
and Life Sciences Education		 Soil Science Society of America Journal
Journal of Plant Registrations Journal of
Environmental Quality		 The Plant Genome