Crop Science 42:513-517 (2002)
© 2002 Crop Science Society of America
TURFGRASS SCIENCE
Patterns of Variation in Poa annua Populations as Revealed by Canonical Discriminant Analysis of Life History Traits
J. Scott McElroy,
Robert H. Walker and
Edzard van Santen*
Dep. of Agronomy and Soils, 202 Funchess Hall, Auburn Univ., Auburn, AL 36849-5412
* Corresponding author (evsanten{at}acesag.auburn.edu)
 |
ABSTRACT
|
|---|
Infestation of southern golf courses with weedy annual bluegrass can be a serious problem and chemical control is less than fully effective. Researchers have classified Poa annua L. into an annual var. annua and a perennial var. reptans. Field and laboratory studies were conducted to evaluate eight P. annua ecotypes: ‘Augusta 4’, ‘Augusta 8’, ‘Augusta 14’, ‘Augusta 17’, ‘Auburn’, ‘Birmingham’, ‘Columbia’, and ‘Purchased’, the latter established from seed grown in Oregon. Observations were taken on seedling and reproductive traits, regrowth potential of clones in early summer, and germination of freshly harvested seed to evaluate variability among the selected ecotypes. Only the Birmingham ecotype represented var. reptans, whereas the other seven represented var. annua. The Birmingham population had smaller flag leaves than the Augusta populations (21 vs. 26 mm), fewer panicles at the end of the study (46 vs. 128 panicles plant-1), and a higher germination percentage of freshly harvested seed (50 vs. 22%). The Birmingham population was the only one of the eight studied which had significant regrowth of clones harvested after completion of the annual life cycle in early June. Canonical discriminant analysis revealed that the cluster of Augusta ecotypes was fairly homogeneous and quite similar to the Purchased check population. The Auburn, Birmingham, and Columbia populations were quite distinct from one another as well as the Augusta–Purchased group. The similarity between the Augusta populations and the Purchased check suggests that the Augusta populations may have been established originally from a similar source of Oregon-grown material; however, similar turfgrass management practices could account for apparent similarities between the Augusta and Purchased populations.
|
INTRODUCTION
|
|---|
ANNUAL BLUEGRASS is a native of Europe and thought to have developed from a cross between Poa supina Schrad. and Poa infirma H.B.K. (Tutin, 1957). It is distributed throughout the temperate regions of the world (Beard et al., 1978). Defining characteristics of P. annua include a keel-shaped leaf blade, boat-shaped leaf apex, membranous ligule, and a five-nerved lemma with no webbing at the base (Beard et al., 1978). Biological and morphological characteristics of P. annua are thought to be highly variable because of various ecological pressures and turfgrass management regimes.
Annual bluegrass is a common weed in turfgrass. It is undesirable because it disrupts turfgrass uniformity, produces unsightly flowers at mowing heights as low as 6 mm, dies out quickly because of summer heat stress, and competes for growth essentials with the desired turfgrass (Hall and Carey, 1992; Bingham et al., 1969; Goss and Zook, 1971). Poa annua presents a problem in bermudagrass (Cynodon spp.) that is overseeded with cool-season grasses because of the lack of herbicides that will selectively control P. annua without harming the cool-season species (Coats and Krans, 1986; Johnson, 1983). In cool-season turfgrasses, P. annua can become so competitive as to dominate the desired grass eventually (Gaussoin and Branham, 1987; Johnson and Murphy, 1995).
Both annual and perennial varieties of P. annua are reported to exist. The perennial variety, P. annua var. reptans (Hauskn.) Timm, is prevalent in more intensively managed turfgrass areas, such as golf putting greens, while the annual variety, P. annua var. annua (L.) Timm, is prevalent in less intensively managed areas, such as golf roughs (Breuninger, 1993). Gibeault (1971) reported that var. annua has fewer leaves, nodes, secondary tillers, and adventitious roots. In addition, it reaches reproductive maturity quicker and produces a greater percentage of flowering tillers relative to var. reptans. Gibeault (1971) also noted that var. reptans did not exhibit post-harvest seed dormancy, while var. annua required a ripening period of up to 90 d. Germination of P. annua was thought to be favored by light (Beard et al., 1978); however, Itoh et al. (1997) reported that some P. annua selections germinated >70% with no light at 20°C.
Some researchers are skeptical that separate annual and perennial P. annua varieties exist. Lush (1989), determined that while there are different growth characteristics between var. reptans and var. annua, var. reptans can give the appearance of a perennial by producing successive generations which overlap in time. Gibeault's (1971) findings of the lack of seed dormancy in var. reptans are consistent with the "successive overlapping generations" theory. Johnson et al. (1993) pointed out that P. annua is represented by a continuum of types ranging from strictly annual to extreme perennial. The latter with a distinct juvenile phase, vernalization temperatures of 
8°C, a 10- to 12-d vernalization period, and a single spring-time reproductive period (Johnson and White, 1997a) would represent true var. reptans.
Throughout this paper, we will refer to these populations of P. annua as ecotypes. The populations selected exhibit ecotypic variation ("genetically-based variation within a species that is related to habitat") (Sosebee and Wester, 1995). Others have referred to selected populations of P. annua as "biotypes" (Yelverton et al., 1999; McCarty, 1998; Gaul and Christians, 1988). Biotype refers to plants of a given population that have the same genotype, thus have reproduced either vegetatively or by apomixis (Sosebee and Wester, 1995). An ecotype, however, can be made up of many biotypes that have selected a certain habitat because of environmental factors (Sosebee and Wester, 1995). While the term biotype could possibly be applied to a single population of var. reptans, it is not applicable when referring to var. annua.
The objectives of our research were to characterize and evaluate the variation among eight populations of P. annua, collected from golf courses in Alabama, Georgia, and South Carolina by comparing them with a commercial population of plants grown from purchased seed.
|
MATERIALS AND METHODS
|
|---|
Three experiments were conducted during 1999 and 2000 at the Auburn University Turfgrass Research Unit and the Auburn University environmental growth chambers, Auburn, AL. Eight P. annua ecotypes were obtained in 1997 and 1998 (Table 1). Twenty individual plants of each ecotype were collected in an approximately 93-m2 (1000-ft2) area at each location. All collected plants were maintained in isolation blocks in a greenhouse environment and the seed was collected for future tests. Seed collected from greenhouse grown plants were allowed to ripen for about 2 mo at 20 ± 2°C at 60% relative humidity to overcome possible dormancy and then stored at 4°C and 30% relative humidity until needed.
View this table:
[in this window]
[in a new window]
|
Table 1. Description of eight Poa annua L. genotypes evaluated for life history traits during the 1999/2000 growing season at the Auburn University Turfgrass Research Unit.
|
|
Biological Characteristics
This field experiment was conducted to evaluate differences in biological characteristics among ecotypes. Eighty 15-cm-diam plastic pots were filled within the top 5 cm with a growth medium containing 700 g kg-1 sand and 300 g kg-1 reed sedge peat (Dakota Reed-Sedge Peat, Pioneer Peat, Inc., Box 1798, Grand Forks, ND). Pots were spaced 0.8 m apart in eight rows of 10, and buried in the soil leaving the top rim of the pot 5 cm above soil surface. Twenty seeds of a given ecotype were planted per pot on 15 Sept. 1999. Each ecotype was replicated 10 times in a completely randomized design. Thus a total of 200 seeds per ecotype were evaluated. Pots were covered with a tobacco-seedbed cloth that allowed sunlight and water to pass through, but prevented dispersal of seed from pots because of water splashing. Pots were watered as needed to maintain adequate soil moisture. The seedbed cloth was removed 2 wk after seeding on 27 Sept. 1999. Pots were fertilized biweekly beginning 11 Nov. 1999 with 23-8-14 fertilizer applied with a drop spreader at a rate of 1.2 g N m-2. We collected data on 11 response variables related to seedling, vegetative, and reproductive phases of plant development (Table 2).
View this table:
[in this window]
[in a new window]
|
Table 2. List, units of measurement, and description of characters scored for the evaluation of eight Poa annua ecotypes at the Auburn University Turfgrass Research Unit.
|
|
Regrowth Ability
Field-grown plants, which had been used to evaluate biological characteristics, were also used to evaluate plant regrowth ability. All pots were removed from the field on 12 June 2000. Brown plant material, assumed to be either dead or dormant, was removed from pots. Plants were then divided into single rooted shoot sections. Excess growth was removed to leave only 2 cm of shoot material. Plant tissue sections were washed with tap water to remove soil or seed attached to roots. Cone-shaped containers were filled to capacity with a mixture of 700 g kg-1 sterilized sand and 300 g kg-1 reed sedge peat. Plant material sections were planted into individual cones to a depth that covered the crown and roots. Plants were watered every 48 h to maintain adequate soil moisture. Each pot within each ecotype containing a living plant at the completion of evaluation of biological characteristics (15 Apr. 2000) was replicated 10 times in a completely randomized design. Growth chambers were utilized to simulate a day/night regime of 19/10°C and 8/16 h. The percentage of plants with green tissue was evaluated after 3 wk in the environmental chamber.
Seed Dormancy
The field-grown plants, which had been used to evaluate biological characteristics (previously described), were also used to evaluate seed dormancy. To obtain seed of a similar maturity, nylon netting was placed on the soil surface of each pot and secured with metal pins. This allowed for passage of water to the soil surface but captured dehiscent seeds. Netting was placed in pots on 31 March 2000 and removed with collected seed on 28 April 2000. Evaluation of seed dormancy utilized acrylamide copolymer gel and tissue culture. Growth medium consisted of 4 mg Terra-Sorb GB acrylamide copolymer (Industrial Services International, Inc., Bradenton, FL) dissolved in 1 mL tap water. Twelve-cell tissue culture trays (Fisher Scientific, Pittsburgh, PA) were filled to capacity with this growth medium. Seeds collected from greenhouse plants were washed with 10% (v/v) sodium hypochlorite solution, thoroughly rinsed with tap water and allowed to air dry for 30 min before placing in growth medium. All experiments were conducted in environmental chambers. Photosynthetic flux density during the light period was 380 µmol m-2 s-1, which came from a mixture of fluorescent and incandescent lamps. Five seed from each plant were added to each cell. Germination was evaluated at a single photoperiod (8/16 day/night h) and a single temperature regime (19/10 day/night °C). A completely randomized design was used. The percentage of germinated seeds was determined after 3 wk.
Data Analysis
Response data were subjected to ANOVA by mixed model methodology (Littell et al., 1996). Because not all plants survived the experiment, standard errors may differ among populations to account for the number of observations contributing to each mean. Since some of the traits were highly correlated, e.g., SDL_10, SDL_11 (see Table 2), we used canonical discriminant analysis (Hair et al., 1987) to separate the eight populations on the basis of a linear combination of the original 13 response variables. With this multivariate statistical technique, a priori known groups, e.g., populations, are compared for a set of interdependent quantitative traits. The main difference to traditional techniques such as ANOVA and regression, as pointed out by Vaylay and van Santen (2002), is that the dependent variable is categorical and the independent variable is metric, i.e., contains the observed values. Thus in this study, the eight populations were treated as dependent variables, whereas the response variables (SDL_1, SDL_10, etc.) were independent variables in the canonical discriminant analysis. The group centroid is the mean of the canonical discriminant function. Mahalanobis squared distance is the differences between two group centroids and is calculated as:
 |
where D2 = Mahalanobis squared distance, 
1 and 2 are the estimated mean vectors in the respective groups, and S-1 is the inverse of the pooled sample variance-covariance matrix (Dillon and Goldstein, 1984).
|
RESULTS
|
|---|
Biological Characteristics
Significant differences among populations were observed for all traits (Table 3). The Birmingham (BHAM) population in particular had responses that were significantly lower than other populations for days to emergence of the first seedling (SDL_1) and number of panicles at the conclusion of the study (PAN_END). The Columbia (COL) population had significantly more PAN_END than the other populations. Inspection of the data revealed that the extremes were represented by the Auburn (AUB) and BHAM populations. The AUB population germinated 17 d after planting, first produced panicles 6 wk after emergence, developed a 32.3 mm flagleaf, and produced >100 panicles per plant 7 mo after planting (Table 3). In contrast, BHAM germinated 7 d after planting, first produced panicles 10 wk after emergence, developed a 21.3-mm flagleaf, and produced 45 panicles per plant 7 mo after planting.
View this table:
[in this window]
[in a new window]
|
Table 3. Least squares means and associated standard errors (in parenthesis) for 13 response variables of eight Poa annua L. genotypes evaluated during the 1999–2000 growing season at the Auburn University Turfgrass Research Unit.
|
|
Regrowth Ability
Not all plants were completely void of green tissue when the experiment began on 12 June 2000. Four populations had varying percentages of green tissue remaining: BHAM (44%), Augusta 4 (AUG04, 9%), Augusta 17 (AUG17, 4%), and PUR (2%). The remaining populations had no scorable green leaf tissue at that time. Only three populations produced regrowth: AUG14 (2 of 9 plants established from survivors of the biological characteristics experiment), BHAM (6 of 8) and PUR (1 of 7). Neither AUG14 nor PUR had plants exceeding 5% regrowth. However, BHAM plants regrew as much as 100% with an average of 56.3% regrowth (Table 3). Only plants that had green tissue present at the initiation of the study demonstrated vegetative regrowth ability.
Seed Dormancy
Significant differences in seed dormancy existed among ecotypes (P 0.05). Two populations (AUG14 and COL) germinated but were not significantly different from zero (Table 3). Only plants from BHAM, AUG08, and AUG17 had greater than 25% germinated seed. The BHAM population had the highest germination percentage (50%).
Discriminant Analysis
The first three discriminate functions accounted for 81.7% of the total observed variance among the 13 response variables (Table 4); only functions with an Eigenvalue >1.0 were selected. Discriminant analysis was successful in grouping populations (Fig. 1)
. The standardized coefficients (loadings) for the functions (Table 4) reflect the correlation between the original variable and the function and can be thought of as the proportion of the variance that is shared (Hair et al., 1987). The first canonical function separated the BHAM population from the other populations. It was dominated by large loadings for SDL_11, PAN_END, and REGROWTH. These are the very traits, which distinguished the BHAM population in Table 3. The second discriminant function separated AUB from the remaining populations. This function had large loadings for PAN_1, PAN_2wk, and SDL_10. Again, these are the variables for which AUB differed from most of the other populations. The cluster of the four populations from Augusta National (AUG04, AUG08, AUG14, AUG17) were not significantly (P = 0.05) different from one another on the basis of Mahalanobis distances (Table 5).
View this table:
[in this window]
[in a new window]
|
Table 4. The standardized coefficients for the first three canonical discriminant functions, the Eigenvalue, cumulative percent of the total variance accounted for, and the canonical correlation.
|
|
View larger version (20K):
[in this window]
[in a new window]
|
Fig. 1. Scatterplot of the centroid values (means of discriminant functions) separating eight Poa annua ecotypes evaluated at the at the Auburn University Turfgrass Research Unit.
|
|
View this table:
[in this window]
[in a new window]
|
Table 5. Mahalanobis distance (upper triangle) and associated probability levels (lower triangle) from the canonical discriminant analysis comparing eight Poa annua populations using 13 independent variables.
|
|
|
DISCUSSION
|
|---|
The most surprising result of this study was that the purchased check (PUR) was much closer to the populations obtained from Augusta National (AUG04, AUG08, AUG14, AUG17) than it was to the other populations (Fig. 1). These populations had no regrowth ability following the completion of the life cycle of an annual (Table 3). They also tended to have a large number of spikelets and florets per panicle, which would underscore the reliance of these populations on seed dissemination. Thus, these populations appear to represent var. annua as defined by Gibeault (1971).
It is not hard to envision why the four AUG populations should be similar, considering that they came from a single golf course, but what is responsible for their similarity to PUR? One management practice that distinguishes Augusta National from the other golf courses in this study is the intensity of overseeding with cool season species in late summer to provide a green playing surface for the winter months. Augusta National has bermuda (Cynodon dactylon L. x C. transvaalensis) fairways, which are regularly and completely overseeded (Coleman Ward, personal communication, 2000) with either rough bluegrass (Poa trivialis L.) or perennial ryegrass (Lolium perenne L.). Columbia country club, on the other hand, uses only perennial ryegrass to overseed tee boxes only, yet P. annua constitutes a weed problem on this entire course. The seed for both overseeding species is produced in the Pacific Northwest, primarily Oregon. The 1999 production estimate for Oregon-grown P. trivialis was 2104 Mg (http://www.css.orst.edu/seed-ext/FnF.html; verified October 31, 2001). Oregon seed certification rules (http://www.oscs.orst.edu/standards/bluegrass.pdf; verified October 31, 2001) prohibit the presence of P. annua in Foundation and Registered classes. Certified seed, however, may contain up to 3 g kg-1 (0.3%) of weed seed. The seed size differential between P. annua and P. trivialis is much closer than between P. annua and L. perenne and contamination would therefore be a greater problem. Thus, it is possible, may be even likely, that the Augusta populations were originally established from P. annua "imported" from Oregon. However, because of the extensive prevalence of Poa annua throughout the world, similar selective pressures could have produced the similarities seen in the AUG and PUR ecotypes.
The AUB, BHAM, and COL populations were clearly different from the remaining five (Fig. 1, Table 5). They represent native populations, which have adapted to their particular habitat, even though they may have originally been imported from other places. Primarily the second canonical variate, which is dominated by the traits PAN_1, and PAN_2wk distinguished AUB and COL. AUB developed panicles faster but had significantly fewer panicles than COL (Table 3). The BHAM population had the fewest panicles (PAN_END) of all populations. Gibeault (1971) states that var. reptans produces fewer flowering tillers than var. annua. Thus, BHAM may be a representative of var. reptans, while all other ecotypes resembled var. annua more closely. It is interesting to note that AUB and BHAM produced panicles at 51 and 93 d, respectively, after the first occurrence of night temperatures below 8°C. Variety reptans has been reported to require vernalization at constant temperatures of 4 to 8°C for 8 to 12 wk prior to the initiation of panicle production, while var. annua does not respond to vernalization (Johnson and White, 1997b). Although constant temperatures below 8°C were not encountered in this experiment, the delay in panicle production by BHAM until the onset of cooler temperatures, further supports the claim that BHAM be classified as var. reptans.
Furthermore, Gibeault (1971) concluded that var. reptans does not have seed dormancy and seeds can germinate shortly after senescing from the panicle, while var. annua has a 2- to 3-mo dormancy period before germination can occur. Our results indicate that this may be a matter of degree but not an absolute distinguishing characteristic. Considering that all ecotypes were highly variable in germination, it could be concluded that no ecotypes were definitely consistent with the description set forth by Gibeault (1971). However, while this does not refute conclusions set forth by Gibeault (1971), it does demonstrate the large amount of variation that is present within the ecotypes collected. Variation within the species of P. annua has also been demonstrated with regard to photoperiod and temperature requirements for flowering (Johnson and White, 1997a,b).
|
NOTES
|
|---|
Salary support provided in part by state and federal funds appropriated to the Alabama Agric. Exp. Stn.
Received for publication February 1, 2001.
|
REFERENCES
|
|---|
- Beard, J.B., P.E. Rieke, A.J. Turgeon, and J.M. Vargas, Jr. 1978. Annual bluegrass (Poa annua L.): Description, adaptation, culture and control. Res. Rep. 352. Michigan State Univ. Agric. Exp. Stn.
- Bingham, S.W., R.E. Schmidt, and C.K. Curry. 1969. Annual bluegrass control in overseeded bermudagrass putting green turf. Agron. J. 61:908–911.[Abstract/Free Full Text]
- Breuninger, J. 1993. Poa annua control in bentgrass greens. Golf Course Management 61(8):68–73.
- Coats, G.E., and J.V. Krans. 1986. Evaluation of ethofumesate for annual bluegrass (Poa annua L.) and turfgrass tolerance. Weed Sci. 34:930–935.
- Dillon, W.R., and M. Goldstein. 1984. Multivariate analysis-methods and applications. John Wiley and Sons, New York.
- Gaul, M.C., and N.E. Christians. 1988. Selective control of annual bluegrass in cool-season turfs with fenarimol and chlorsulfuron. Agron. J. 80:120–125.[Abstract/Free Full Text]
- Gaussoin, R.E., and B.E. Branham. 1987. Annual bluegrass and creeping bentgrass germination response to flurprimidol. HortScience 22:441–442.
- Gibeault, V.A. 1971. Perenniality in Poa annua L. Ph.D Dissertation (Diss. Abstr. AAT 7115643). Oregon State University, Corvallis, OR.
- Goss, R.L., and F. Zook. 1971. New approach for Poa annua control. The Golf Superintendent. 39(1):46–48.
- Hair, J.F., Jr., R.E. Anderson, and R.L. Tatham. 1987. Multivariate data analysis. Macmillan Publishing Company, New York.
- Hall, J.C., and C.K. Carey. 1992. Control of annual bluegrass (Poa annua) in Kentucky bluegrass (Poa pratensis) turf with linuron. Weed Technol. 6:852–857.
- Itoh, M., H. Kobayashi, and K. Ueki. 1997. Variation in seed germination and dormancy of Poa annua L. in golf course. Grassland Sci. 42(4):299–306.
- Johnson, B.J. 1983. Response to ethofumesate of annual bluegrass (Poa annua) and overseeded bermudagrass (Cynodon dactylon). Weed Sci. 31:385–390.
- Johnson, B.J., and T.R. Murphy. 1995. Effect of paclobutrazol and flurprimidol on suppression of Poa annua spp. reptans in creeping bentgrass (Agrostis stolonifera) greens. Weed Technol. 9:182–186.
- Johnson, P.G., B.A. Ruemmele, P. Velhuth, D.B. White, and P.D. Ascher. 1993. An overview of Poa annua L. reproductive biology. p 798–804. In R.N. Carrow et al. (ed.) International Turfgrass Society Research Journal, Vol. 7. Intertec Publishing Corp., Overland Park, KS.
- Johnson, P.G., and D.B. White. 1997a. Flowering responses of selected annual bluegrass genotypes under different photoperiod and cold treatments. Crop Sci. 37:1543–1547.[Abstract/Free Full Text]
- Johnson, P.G., and D.B. White. 1997b. Vernalization requirements among selected genotypes of annual bluegrass (Poa annua L.). Crop Sci. 37:1538–1542.[Abstract/Free Full Text]
- Littell, R.C., G. Milliken, W.W. Stroup, and R.D. Wolfinger. 1996. SAS system for mixed models. SAS Inst. Inc., Cary, NC.
- Lush, W.M. 1989. Adaptation and differentiation of golf course populations of annual bluegrass (Poa annua L.). Weed Sci. 37:54–59.
- McCarty, L.B. 1998. Annual bluegrass control in overseeded turfgrasses. Proc. South. Weed Sci. Soc. 51:88–90.
- Sosebee, R.E., and D.B. Wester. 1995. Genetic variation in plants: environmental and biotic effects. p. 1–21. In D.J. Bedunah and R.E. Sosebee (ed.) Wildland plants: Physiological ecology and developmental morphology. Society for Range Management, Denver, CO.
- Tutin, T.G. 1957. A contribution to the experimental taxonomy of Poa annua L. Watsonia. 4:1–10.
- Vaylay, R., and E. van Santen. 2002. Application of canonical discriminant analysis for the assessment of genetic variation in tall fescue. Crop Science 42:534–539 (this issue).[Abstract/Free Full Text]
- Yelverton, F.H., J. Isgrigg III, and J. Hinton. 1999. New approaches to management of annual bluegrass in bentgrass putting greens. Proc. South. Weed Sci. Soc. 52:72.
TOP
ABSTRACT
INTRODUCTION
