Published in Crop Sci. 44:443-448 (2004).
© 2004 Crop Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA
CROP BREEDING, GENETICS & CYTOLOGY
Genetic Variability and Trait Relationships in Switchgrass
Modan K. Das*,a,
Roger G. Fuentesb and
Charles M. Taliaferroa
a Dep. of Plant & Soil Sciences, Oklahoma State Univ., Stillwater, OK 74078
b Dep. of Agronomy & Plant Genetics, Univ. of Minnesota, 411 Borlaug Hall, St. Paul, MN 55108
* Corresponding author (dmodan{at}mail.pss.okstate.edu).
 |
ABSTRACT
|
|---|
Greater knowledge of the magnitude of genetic variability for biomass yield and yield components in switchgrass, Panicum virgatum L., and relationships among the biomass yield component traits would facilitate the breeding improvement of the species. Accordingly, we conducted two replicated experiments to assess genetic variation for biomass yield and yield components and quantify relationships among those traits in different switchgrass populations. In Exp. 1, 228 half-sib families from SU C3 and 261 from NU C3 populations were evaluated at Perkins, OK, while 278 half-sib families from the SL C0 population were evaluated at Stillwater, OK. Exp. 2 comprised 11 lowland switchgrass populations tested at Chickasha and Perkins, OK, in 1998. Substantial differences (P < 0.01) in biomass yield per plant existed among half-sib families within the respective SU C3, NU C3, and SL C0 populations. Estimates of genetic variance for biomass yield were significant for each population. However, significant family x year and family x block variance estimates indicated substantial environmental influence in each of the populations. In Exp. 2, significant (P 
0.05) variation existed among the 11 populations over locations for biomass yield per plant, tiller number per plant, tiller length, leaf blade length, and leaf blade width. Phenotypic correlation between biomass yield and tiller number per plant was positive (r = 0.68* at Chickasha and r = 0.60* at Perkins). Path coefficient analyses revealed that number of tillers per plant had the highest positive direct effect on biomass yield at both locations (0.74 at Chickasha and 0.66 at Perkins). Adequate genetic variability was present within the switchgrass populations to allow breeding improvement of biomass yield. Selection for increased number of tillers per plant would be the most effective means of indirectly increasing biomass yield.
|
INTRODUCTION
|
|---|
SWITCHGRASS is a warm-season perennial grass native to most of the continental USA except the Pacific Northwest (Talbert et al., 1983). It is one of the major component species of the tall- and mixed-grass prairies of North America. Switchgrass is used as a component in seeded native grass mixtures and is grown in monoculture for hay, grazing, and erosion control (Moser and Vogel, 1995). Switchgrass has received substantial research interest over the past decade because of its selection by the U.S. Department of Energy as a model herbaceous species to develop as a bioenergy feedstock crop (McLaughlin, 1993; Hohenstein and Wright, 1994; McLaughlin and Walsh, 1998). It was selected because of its high biomass production potential on marginal land with relatively low inputs and possession of other environmentally friendly traits.
Switchgrass constitutes a polyploid series with reported chromosome numbers ranging from 2n = 2x = 18 to 2n = 12x = 108 (Nielson, 1944; McMillan and Weiler, 1959; Henry and Taylor, 1989). Switchgrass is classified into upland and lowland ecotypes based on morphology and habitat preference (Porter, 1966). All confirmed lowland ecotypes have been tetraploids and most upland ecotypes are octoploids (Hopkins et al., 1996).
Information on the variability and associations of biomass yield and yield components in switchgrass is limited. Talbert et al. (1983) reported narrow-sense heritability estimates of 0.25 and 0.59 based on individual half-sib progeny and half-sib progeny means, respectively, for plant dry weight in lowland switchgrass populations. Eberhart and Newell (1959) estimated broad-sense heritability as 0.78 for plant yield in an upland switchgrass population comprising strains endemic to Nebraska. Newell and Eberhart (1961) reported heritability estimates for upland switchgrasses from Nebraska and northern Kansas separated into "small blue-green," "medium-tall blue-green," and "tall-green" plant populations. Their estimates of narrow-sense heritability for total plant yield were 0.18, 0.52, and 0.05 for the small blue-green, medium-tall blue-green, and tall-green types, respectively.
Correlation coefficients measure the relationships among traits. Correlations provide only limited information because they disregard complex interrelationships among traits. Accordingly, they must be used with caution in making decisions regarding indirect selection criteria (Kang, 1994; Board et al., 1997). A path coefficient is a standardized, partial regression coefficient that measures the direct influence of one trait on another trait and permits the separation of a correlation coefficient into components of direct and indirect effects for a set of a priori cause-and-effect interrelationships (Dewey and Lu, 1959). Several studies (Dewey and Lu, 1959; Sidwell et al., 1976; Kang et al., 1983; Board et al., 1997) have demonstrated that partitioning correlation coefficients into direct and indirect effects provides more useful information. Newell and Eberhart (1961) reported correlation among several characters in switchgrass. For the small, blue-green switchgrass population, they calculated a positive and significant phenotypic correlation (r = 0.45) between height of leaves and total plant yield and a nonsignificant phenotypic correlation between plant height and total plant yield. For the medium-tall blue-green population, the phenotypic correlations of total plant yield with height of leaves and plant height were positive and significant. Talbert et al. (1983) reported a significant positive phenotypic correlation between plant dry weight and plant height and a significant negative phenotypic correlation between plant dry weight and maturity. Redfearn et al. (1997) studied associations among several morphological traits and forage yield in switchgrass. They reported that forage yields of switchgrass populations were affected primarily by tiller growth and development and the associated morphological modifications occurring in the leaf blades, leaf sheaths, and stems. However, no information exists on the relative importance of direct and indirect effects of biomass yield components on biomass yield in switchgrass. The objectives of this study were to (i) assess genetic variability of biomass yield and yield components, (ii) determine phenotypic correlation coefficients (subsequently referred to simply as correlation) among biomass yield and several yield components, and (iii) partition the correlation through path coefficient analysis to assess the relative importance of direct and indirect effects of the yield components on biomass yield in switchgrass.
|
MATERIALS AND METHODS
|
|---|
Experiment I
Plant materials used in this experiment comprised 228, 261, and 278 half-sib families derived from the switchgrass breeding populations SU C3, NU C3, and SL C0, respectively. The populations were used for phenotypic recurrent selection for increased biomass yield. The original SU (southern upland) population was synthesized from ‘Caddo’ and ‘Blackwell’. The original NU (northern upland) population was synthesized from ‘Nebraska 28’, ‘Pathfinder’, and ‘Cave-in-Rock’. The original SL (southern lowland) population was synthesized from ‘Alamo’ and ‘PMT-279’. Greenhouse grown seedlings of the SU C3 and NU C3 half-sib families were transplanted in separate field tests at the Agronomy Research Station, Perkins, OK, in summer 1997. The experimental design for both tests was a randomized complete block with two replications. Each plot contained two progeny plants from a half-sib family. Half-sib families from the SL C0 population were planted at the Agronomy Research Station, Stillwater, OK, in summer 1997. The experimental design was the same as that of the SU C3 and NU C3 populations, except that individual plots contained three half-sib progeny plants. Individual plants in each of the tests were equally spaced 107 cm apart. The soil at Perkins is classified as a member of the fine-loamy, mixed, active, thermic Udic Argiustolls, while that at Stillwater belongs to the fine, mixed, superactive, thermic Udertic Paleustolls. All tests received 90 kg N ha–1 yr–1 applied in early spring. Phosphorous and potash were incorporated before planting at rates indicated by soil tests to meet sufficiency levels for a yield goal of 18 Mg ha–1. Biomass yields of individual plants were obtained in fall of 1998 and 1999. Data were analyzed as a split-plot in time by ANOVA (SAS Institute, 1999). All effects were considered to be random effects in the analyses of variance. Variance components were estimated from the expected mean squares (Knapp and Bridges, 1987) and standard errors of the variance components were estimated after Anderson and Bancroft (1952).
Experiment II
Plant materials consisted of 11 lowland switchgrass populations (Alamo, SL C0, SL C1, SL C2, SL92-1, SL94-1, ‘Kanlow’, NL C1, NL C2, NL92-1, and NL94-1). Alamo and Kanlow are commercial cultivars. The origin of the SL C0 population is as described under Exp. I. The SL C1 and SL C2 populations resulted from phenotypic recurrent selection within the SL population. The SL 92-1 and 94-1 populations are 14- and 8-parent clone experimental synthetic cultivars, respectively. The NL C1 and C2 populations resulted from phenotypic recurrent selection within Kanlow. The NL 92-1 and 94-1 populations are 23- and 9-parent clone experimental synthetic cultivars, respectively. Greenhouse grown seedlings of the 11 populations were transplanted to a field nursery on the South Central Research Station, Chickasha, OK, on 23 May 1996. The experimental design was a randomized complete block with four replications. Each plot comprised four rows of 12 plants row –1. Plants within plots were spaced on 61 cm centers. An identical planting was made on the Agronomy Research Station, Perkins, OK, on 30 May 1996. The soil at Chickasha is classified as a member of the course-silty, mixed, superactive, thermic Pachic Haplustolls and that at Perkins as previously stated. Rate of fertilizer application was the same as Exp. I. Biomass yields were first measured from these nurseries in fall of 1997. For this study, data for biomass yield and yield components were collected during summer and fall of 1998. The yield components studied were (i) tiller number per plant, (ii) tiller length (from stem base to the ligule of the most recent fully expanded leaf), (iii) stem width at the base, (iv) node number per tiller, (v) internode length (1st fully developed internode below the shoot apex), (vi) leaf blade length (one random leaf blade per tiller), and (vii) leaf blade width (middle of one random leaf blade per tiller). All data on the components except tiller number per plant were recorded from five tillers of each of ten randomly selected plants from the middle two rows of each plot. The five tillers were selected randomly and cut from the base after the plants had flowered. Selected tillers were brought to the laboratory and data were recorded about a month later. The 10 plants selected from each plot were hand harvested when inflorescences reached physiological maturity. The number of tillers per plant was counted and the air-dried weights were recorded. A 100- to 200-gm biomass sample was taken from each harvested plant and oven dried to calculate dry biomass yield per plant. Plot mean value was determined for all traits and these mean values were used in ANOVA to test population differences and the significance of population x location interaction effects. Population effect was considered fixed and all other effects were considered to be random effects in the analyses of variance. Separate ANOVAs were conducted for each location for traits with significant population x location interactions. Comparisons of population means were conducted by Duncan's Multiple Range Test (DMRT) for the traits that had significant variations among the populations. DMRT was conducted by SAS (SAS Institute, 1999). Correlation and path coefficient analyses were done by standard methods (Dewey and Lu, 1959; Kang, 1994). The causal relationships for the path coefficient analysis involved the seven biomass yield components as predictor (cause) variables and biomass yield as the response (effect) variable (Fig. 1)
.
|
RESULTS AND DISCUSSION
|
|---|
Experiment I
Analyses of variance indicated significant differences (P < 0.01) in biomass yield among half-sib progeny families from the SU C3, NU C3, and SL C0 populations (data not presented). Mean biomass yield per plant of the half-sib families from the SU C3 and NU C3 populations were 0.73 ± 0.14 and 0.94 ± 0.15 kg, respectively. Biomass yield of the half-sib families ranged from 0.37 kg to 1.40 kg and 0.21 kg to 1.64 kg plant–1 for the SU C3 and NU C3 populations, respectively. The SL C0 half-sib family mean biomass yield per plant was 1.14 ± 0.13 kg and the range was 0.77 to 1.68 kg. These results confirm substantial differences for biomass yield among the half-sib families in each of the three switchgrass populations. Estimates of the variance components for biomass yield of these three populations are presented in Table 1. For all three populations, estimates of each of the family (or genetic) variance 
, family x year variance 
, and family x block 
variance were significant based on the significance test given by Stuthman and Stucker (1975). The magnitude of 
2f was higher than that of the 2fxy but lower than 2fxb variance for all three populations. These results suggest substantial genetic variability among the half-sib families for biomass yield, but also clearly indicate that environmental variability was important and of a magnitude to warrant testing through space and time. These results are in congruence with those of Eberhart and Newell (1959), Newell and Eberhart (1961), and Talbert et al. (1983) based on the magnitudes of their reported heritability estimates.
View this table:
[in this window]
[in a new window]
|
Table 1. Estimates of variance components and their associated standard errors for biomass yield of three switchgrass populations.
|
|
Experiment II
Analyses of variance for biomass yield and seven yield components are presented in Table 2. The yield components stem width, internode length, and leaf blade length had significant (P 0.05) population x location interactions and data for these three traits were analyzed separately for each location. Significant (P 0.05) variation among the populations were observed for biomass yield per plant, tiller number per plant, tiller length, and leaf blade width for the across-location analyses (Table 2). Among population variation for stem width and leaf blade length was significant (P 0.05) at both locations (data not presented). However, variation among populations for internode length was significant (P 0.01) only at Perkins (data not presented). Means for biomass yield per plant, tiller number per plant, tiller length, leaf blade length, and node number per tiller are given in Table 3. The SL94-1 population had significantly higher biomass yield per plant as compared to SL C2 and the five NL populations including Kanlow. SL94-1 population had significantly higher tiller number per plant than all other populations. In general the NL populations had longer tillers than the SL populations. The SL C0 population had significantly wider leaf blades than any other populations. Means for stem width, leaf blade length, and internode length for each location are presented in Table 4. Alamo had significantly wider stem as compared with NL94-1 and all SL populations at Chickasha; however, at Perkins, SL C1 population had significantly wider stem than Alamo and several other populations (Table 4). Alamo also had longer leaf blades only at Chickasha. At Perkins, NL92-1 had the longest internode among the populations. While variation among the populations varied considerably for different traits, they indicated that scope of selection exists for all traits except node number per tiller and internode length.
View this table:
[in this window]
[in a new window]
|
Table 2. Analyses of variance for biomass yield (kg plant–1) (BY), tiller number per plant (TN), tiller length (cm) (TL), stem width (mm) (SW), node number per tiller (NN), internode length (cm) (IL), leaf blade length (cm) (LBL), and leaf blade width (mm) (LBL) of 11 switchgrass populations grown at Chickasha and Perkins, OK, in 1998.
|
|
View this table:
[in this window]
[in a new window]
|
Table 3. Means for biomass yield, number of tillers, tiller length, leaf blade width, and number of nodes per tiller of 11 switchgrass populations grown at Chickasha and Perkins, OK, in 1998.
|
|
View this table:
[in this window]
[in a new window]
|
Table 4. Means for stem width, leaf blade length, and internode length of 11 switchgrass populations grown at Chickasha and Perkins, OK, in 1998.
|
|
Correlation for biomass yield per plant and yield components from the Chickasha and Perkins experiments are given in Table 5. Correlations between biomass yield per plant and tiller number per plant were positive and significant at both locations. Among the yield components, significant positive correlations between tiller length and node number per tiller were also found at both locations. This indicated that as tiller length increased node number per tiller also increased. Positive and significant correlation between leaf blade length and leaf blade width was found only at Perkins. Correlation between node number per tiller and stem width was positive and significant only at Chickasha. The later two correlations cannot be generalized since they were not similar across locations. Redfearn et al. (1997) reported that forage yields of switchgrass populations were affected primarily by tiller growth and development and the associated morphological modifications occurring in the leaf blades, leaf sheaths, and stems.
View this table:
[in this window]
[in a new window]
|
Table 5. Phenotypic correlation coefficients among biomass yield and yield components of 11 switchgrass populations grown at Chickasha (above diagonal) and Perkins (below diagonal), OK, in 1998.
|
|
Correlations were partitioned into direct and indirect effects through path coefficient analysis. Results of the path coefficient analyses are given in Tables 6 and 7 for Chickasha and Perkins, respectively. Tiller number per plant had the greatest positive direct effect on biomass yield at both locations. The magnitudes of these direct effects were similar to the magnitudes of correlations between biomass yield and tiller number per plant indicating that the correlations explain the true relationship and selection for more tillers per plant would indirectly select for higher biomass yield. At Chickasha, leaf blade length had the second highest positive direct effect on biomass yield, while at Perkins stem width had the second highest positive direct effect on biomass yield. Node number per tiller had the third highest positive direct effect on biomass yield for both locations. However, node number per tiller had negative indirect effect on biomass yield via tiller number. Node number per tiller also had positive correlation with tiller length and tiller length had large negative indirect effect on biomass yield via tiller number. The large negative indirect effect of tiller length via tiller number indicated that tiller length affects tiller number. These results indicated that node number per tiller could not be used as an indirect selection criterion for biomass yield. Leaf blade length had high positive indirect effect on biomass yield via tiller number at both locations. A possible explanation of this is that plants with longer leaves had more photosynthate that contributed to higher number of tiller development. That leaf area plays a role in higher biomass yield is also reflected by the considerable amount of direct effects (fourth highest in ranking) of leaf blade width at both locations. In phenotypic path coefficient analysis, a large residual effect usually indicates that there are traits other than those included in pathways that contribute to the dependent variable (Wang et al., 1999). Path coefficient analyses in our study did not account for all the variation in biomass yield as indicated by the magnitude of residual effects (0.59 at Chickasha and 0.53 at Perkins), which pointed out that there are traits in addition to the seven traits we have included in the path analysis that contributed to biomass yield.
View this table:
[in this window]
[in a new window]
|
Table 6. Phenotypic path coefficients showing direct and indirect effects of yield components on biomass yield of 11 switchgrass populations grown at Chickasha, OK, in 1998.
|
|
View this table:
[in this window]
[in a new window]
|
Table 7. Phenotypic path coefficients showing direct and indirect effects of yield components on biomass yield of 11 switchgrass populations grown at Perkins, OK, in 1998.
|
|
Collectively, the results from the two experiments indicate the presence of adequate genetic variability within the populations studied for breeding improvement of biomass yield. The genetic variation observed among the populations in Exp. 2 suggests that positive response to direct selection is possible for all yield components studied with the possible exception of node number per tiller and internode length. Results from both the correlation and path coefficient analyses indicate that selection for increased tiller number per plant would be the best indirect selection trait for increasing biomass yield under spaced-plant conditions. Whether this would hold for sward-type stands has not been determined.
|
ACKNOWLEDGMENTS
|
|---|
We thank Dr. Manjit Kang, Professor, Quantitative Genetics, Louisiana State University for technical assistance. Research support was provided by the Biofuels Feedstock Development Program, U.S. Department of Energy, Oak Ridge National Laboratory, Oak Ridge, TN.
|
NOTES
|
|---|
Research supported by the Department of Energy Biofuels Feedstock Development Program.
Received for publication October 1, 2002.
|
REFERENCES
|
|---|
- Anderson, R.L., and T.A. Bancroft. 1952. Statistical theory in research. McGraw-Hill Book Co., New York.
- Board, J.E., M.S. Kang, and B.G. Harville. 1997. Path analyses identify indirect selection criteria for yield of late-planted soybean. Crop Sci. 37:879–884.
- Dewey, D.R., and K.H. Lu. 1959. A correlation and path coefficient analysis of components of crested wheatgrass seed production. Agron. J. 51:515–518.[Abstract/Free Full Text]
- Eberhart, S.A., and L.C. Newell. 1959. Variation in domestic collection of switchgrass. Agron. J. 51:613–616.[Abstract/Free Full Text]
- Henry, D.S., and T.H. Taylor. 1989. Registration of KY 1625 switchgrass germplasm. Crop Sci. 29:1096.[Free Full Text]
- Hohenstein, W.G., and L.Y. Wright. 1994. Biomass energy production in the United States: An overview. Biomass Bioenergy 6(3):161–173.
- Hopkins, A.A., C.M. Taliaferro, C.D. Murphy, and D. Christian. 1996. Chromosome number and nuclear DNA content of several switchgrass populations. Crop Sci. 36:1192–1195.[Abstract/Free Full Text]
- Kang, M.S. 1994. Applied quantitative genetics. M.S. Kang Publishers, Baton Rouge, LA.
- Kang, M.S., J.D. Miller, and P.Y.P. Tai. 1983. Genetic and phenotypic path analyses and heritability in sugarcane. Crop Sci. 23:643–647.[Abstract/Free Full Text]
- Knapp, S.J., and W.C. Bridges, Jr. 1987. Confidence interval estimators for heritability for several mating and experimental designs. Theor. Appl. Genet. 73:759–763.
- McLaughlin, S.B. 1993. New switchgrass Biofuels research program for the southeast. p. 111–115. In Proc. Annual Automative Technol. Dev. Contractors Coordinating Meeting. Nov. 2–5, 1992, Dearborn, MI.
- McLaughlin, S.B., and M.E. Walsh. 1998. Evaluating environmental consequences of producing herbaceous crops for bioenergy. Biomass and Bioenergy 14:317–324. 1993.
- McMillan, C., and J. Weiler. 1959. Cytogeography of Panicum virgatum in central North America. Am. J. Bot. 78:183–188.
- Moser, L.E., and K.P. Vogel. 1995. Switchgrass, big bluestem and indiangrass. p. 409–420. In R.F Barnes et al. (ed.) Forages Vol. 1: An introduction to grassland agriculture. 5th ed. Iowa St. Univ. Press, Ames, IA.
- Nielson, E.L. 1944. Analysis of variation in Panicum virgatum. J. Agric. Res. 69:327–353.
- Newell, L.C., and S.A. Eberhart. 1961. Clone and progeny evaluation in the improvement of switchgrass, Panicum virgatum L. Crop Sci. 1:117–121.[Free Full Text]
- Porter, C.L., Jr. 1966. An analysis of variation between upland and lowland switchgrass, Panicum virgatum L., in central Oklahoma. Ecology 47:980–992.[ISI]
- Redfearn, D.D., K.J. Moore, K.P. Vogel, S.S. Waller, and R.B. Mitchell. 1997. Canopy architecture and morphology of switchgrass populations differing in forage yield. Agron. J. 89:262–269.[Abstract/Free Full Text]
- SAS Institute Inc. 1999. SAS online doc, Version 8. SAS Institute Inc. Cary, NC
- Sidwell, R.J., E.L. Smith, and R.W. McNew. 1976. Inheritance and interrelationships of grain yield and selected yield-related traits in hard red winter wheat cross. Crop Sci. 16:650–654.[Abstract/Free Full Text]
- Stuthman, D.D., and R.E. Stucker. 1975. Combining ability analysis of near-homozygous lines derived from a 12-parent diallel cross in oats. Crop Sci. 15:800–803.[Abstract/Free Full Text]
- Talbert, L.E., D.H. Timothy, J.C. Burns, J.O. Rawlings, and R.H. Moll. 1983. Estimates of genetic parameters in switchgrass. Crop Sci. 23:725–728.[Abstract/Free Full Text]
- Wang, G., M.S. Kang, and O. Moreno. 1999. Genetic analyses of grain-filling rate and duration in maize. Field Crops Res. 61:211–222.
This article has been cited by other articles:
|
|
|
|
 
A. Boe and D. L. Beck
Yield Components of Biomass in Switchgrass
Crop Sci.,
July 1, 2008;
48(4):
1306 - 1311.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
Y. Q. Wu, C. M. Taliaferro, D. L. Martin, J. A. Anderson, and M. P. Anderson
Genetic Variability and Relationships for Adaptive, Morphological, and Biomass Traits in Chinese Bermudagrass Accessions
Crop Sci.,
September 1, 2007;
47(5):
1985 - 1994.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
A. Boe and D. K. Lee
Genetic Variation for Biomass Production in Prairie Cordgrass and Switchgrass
Crop Sci.,
May 31, 2007;
47(3):
929 - 934.
[Abstract]
[Full Text]
[PDF]
|
|
|
TOP
ABSTRACT
INTRODUCTION
) represent direct path (P) and two-directional arrows (
) represent correlations (r).