| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Dep. of Agronomy, Univ. of Wisconsin-Madison, Madison, WI 53706-1597
* Corresponding author (mdcasler{at}facstaff.wisc.edu)
 |
ABSTRACT |
|---|
 TOPNOTES ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
Abbreviations: SEC, standard error of calibration • SEV, standard error of validation
|
INTRODUCTION |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
Cultivated forage-type timothy (P. pratense) is hexaploid with 2n = 6x = 42. There are numerous diploid and tetraploid species within Phleum, three of which are considered to be progenitors of P. pratense. The diploid P. alpinum (alpine timothy) is the most likely donor of the A genome of P. pratense, perhaps after tetraploidization to P. commutatum, while the diploid P. bertolonii (turf timothy) is the most likely donor of the B genome of P. pratense (Cai and Bullen, 1991; Joachimiak and Kula, 1997; Walton, 1983; Wilton and Klebesadel, 1973). These three species are all members of section Phleum Griseb., are geographically widespread, and are virtually indistinguishable morphologically, except for the generally larger plant size of P. pratense (Joachimiak and Kula, 1997). Phleum alpinum and P. commutation appear to be the only members of the genus that are indigenous to North America, with widespread occurrence in subalpine regions (Wilton and Klebesadel, 1973).
Timothy cultivars are well adapted to hay management practices based on relatively infrequent harvests. The frequent harvests (three or four per year) and persistent competition in a timothy-alfalfa (Medicago sativa L.) mixture reduces persistence of timothy cultivars (Casler and Walgenbach, 1990; Smith et al., 1973). Cultivars with earlier heading tend to be more persistent under frequent harvests, but earliness is not a warranty against low persistence (Casler and Walgenbach, 1990). Timothy reproduces vegetatively by swollen sections of basal culms, called corms. Individual corms are biennial, so long-term persistence of timothy plants is highly dependent on continual production of new corms (Childers and Hanson, 1985). Frequent harvesting of timothy can increase the number of dead corms per plant by up to 93% and decrease living corm mass by 29% (Peters, 1958).
Relatively few timothy cultivars developed or commercialized in North America were bred to withstand frequent defoliation. Timothy cultivars show considerable variation for persistence in mixture with alfalfa under hay management (Casler and Walgenbach, 1990) or under management-intensive rotational grazing in pure or mixed stands (Casler et al., 1998). Most timothy cultivars ranked low in persistence compared with cultivars of other species in both studies. Jönsson et al. (1992) concluded that hay-type timothy plants should be erect, tall, and early heading, with long, wide leaf blades. Conversely, timothy plants that are most persistent under long-term frequent defoliation are relatively prostrate and late heading (Van Dijk, 1955). As suggested by these conclusions, cultivar x management interactions are common for forage yield of timothy cultivars (Surprenant et al., 1993).
The agricultural importance of timothy in the Nordic countries has led to many studies of genetic variation and adaptation. A considerable amount of genetic variation within timothy is adaptive in nature, resulting from long-term natural selection under localized stressful conditions (Cenci, 1980; Helgadóttir, 1989; Rognli, 1988; Yumoto et al., 1984). The USDA-NPGS collection of timothy accessions contains over 650 accessions from 30 countries, including both cultivated and natural accessions originating from a considerable range of habitats and climates. While this collection likely contains genetic variability for traits that would confer tolerance to frequent defoliation, it has never been systematically evaluated for agronomic traits. A subset of 75 Japanese accessions was evaluated under hay management at five locations in the northeastern USA, illustrating considerable variation for a range of agronomic traits (Bryan et al., 1988). The objectives of this study were to quantify genotypic variation for agronomic traits within the USDA-NPGS collection of timothy accessions, including an assessment of persistence under frequent defoliation, and to relate that variation to geographic source of the accessions.
|
MATERIALS AND METHODS |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
|
Plots were clipped four times in 1993 and six times each in 1994 and 1995 to simulate the defoliation frequency of a management-intensive grazing system. Plots were clipped to a 10-cm stubble height when the tallest timothy plants were approximately 25 cm tall. Prior to each clipping, forage yield was rated on a visual scale of 0 = missing plant to 20 = the greatest apparent biomass at that point in time. Following the forage yield rating, 40 plants (two from each nonzero numerical rating category) were harvested from the non-test plots at a 1-cm stubble height. These plants were dried at 60°C and weighed. Following the last forage yield rating in October 1995, each plant was coded as living or dead. Plant survival percentage was computed on an individual plot basis.
Each set of 40 harvested plants was divided into two sets of 20, each set including one plant of each size class. A linear regression calibration was computed between forage yield ratings and dry matter yield of the 20 harvested plants for each of the 32 sets. Paired sets for each of the 16 cuttings were used to validate the calibration developed in the alternate set of the pair. Regressions were judged adequate for predicting forage yield from visual ratings, with calibration r2 ranging from 0.64 to 0.92 and validation r2 ranging from 0.62 to 0.87. On the basis of these results, a single linear regression was computed from the 40 harvested plants of each cutting and used to predict forage yield of that cutting for all plants. Forage yield estimates were summed over six cuttings in 1994 and 1995 prior to any statistical analysis. Previous work has shown that visual ratings of forage yield made by an experienced researcher had high genetic correlations (r = 0.93 to 0.97) with forage yield of meadow fescue, Festuca pratensis Huds. (Aastveit and Aastveit, 1989). Fertilization with P and K was done according to recommendations derived from soil tests. Nitrogen fertilizer was not applied to this experiment.
Following the visual forage yield rating in August of 1994 and 1995, a bulk forage sample was harvested from each 10-plant row at a 10-cm stubble height. Ten accessions had plants that did not exceed a 10-cm height. For these plants, forage samples were clipped at a stubble height of approximately half their maximum leaf height. Bulk samples consisted of approximately 50 g fresh leaf tissue per plant. Samples were dried and ground through a 1-mm screen of a Wiley-type mill and reground through a 1-mm screen of a cyclone mill. Near-infrared reflectance spectra (NIRS) were obtained on each sample with a Pacific Scientific 6500 scanning monochromator (Pacific Scientific, Rockford, IL). A subset of 120 calibration samples were chosen by the NIRS software using a cluster analysis of reflectance spectra (Shenk and Westerhaus, 1991). These samples were analyzed for neutral detergent fiber (NDF) concentration using the procedure of Van Soest et al. (1991) with the exceptions that sodium sulfite and 
-amylase were excluded. Predicted NDF values were generated from a single calibration equation (SEC = 7.9 g kg-1, R2cal = 0.92, SEV = 10.8 g kg-1, R2val = 0.86).
Non-Overseeded Plots
The experiment was repeated, with the same design and plot size, adjacent to the overseeded experiment. The second experiment was also established in May 1993, but was not overseeded with white clover. Annual weeds were controlled by pre-emergence herbicide applications in 1993 and 1994, as described by Falkner and Casler (1998). Kentucky bluegrass (Poa pratensis L.) and annual bluegrass (P. annua L.) provided sufficient ground cover after 1994 that additional weed control was unnecessary. Plants were harvested twice in 1993 and three times in 1994 without data collection. Fertilization with P and K was done according to recommendations derived from soil tests.
In April 1995 and 1996, all plants were fertilized with 90 kg N ha-1. Plants were allowed to grow into a reproductive mode, during which time the following measurements were made on all living plants. Heading date was recorded as the day-of-year when the fifth panicle was fully emerged from the boot. For plants that clearly had fewer than five floral primordia, heading date was determined as the earliest day that the majority of panicles were emerged. Plants that never produced panicles were coded as missing values.
Length and maximum width of the first leaf blade below the flag leaf and the total number of leaves were recorded for one tiller per plant in late July, after most plants had reached anthesis. Height to the top of the tallest panicle was recorded in late July. The number of panicles per plant was recorded in early August. Ten random panicles (or fewer, as necessary) were harvested from each plant prior to seed shattering. Total length and maximum width of each panicle were measured during the following winter. Seeds of each 10-panicle sample were threshed and cleaned by one person to maximize uniformity of technique. Seed weight per panicle was determined for each plant. One thousand seeds were counted from each 10-panicle sample and weighed to determine 1000-seed weight. All plants were clipped in late August 1995, following the 1995 seed harvest. Germinating timothy seeds did not interfere with 1996 data collection because of the excellent bluegrass ground cover.
Statistical Analysis
All variables were analyzed by a random effects model analysis of variance by means of generalized least squares to handle missing and unbalanced data (Searle, 1971). The initial ANOVA partitioned accessions into sources of variation for species and accessions within species. A second ANOVA and all additional analyses were based on accessions classified as P. pratense (Table 1).
In the second ANOVA, sums of squares for P. pratense accessions were divided into three sources of variation related to the source of the accession: region, country(region), and accession(country). Ten geographic regions were defined as JPN = Japan, RUS = Russia (mostly northwestern), NZL = New Zealand, NAM = North America (Canada and USA), SWE = Southwestern Europe (France, Italy, and Spain), NWE = Northwestern Europe (Belgium, Great Britain, and The Netherlands), SCA = Scandinavia (Denmark and Sweden) plus Finland, NEE = Northeastern Europe (The Czech Republic, Germany, Poland, and Slovakia), SEE = Southeastern Europe (Bulgaria, Hungary, Greece, Romania, and the former Yugoslavia), and SWA = Southwestern Asia (Afghanistan, Azerbaijan, Iran, Iraq, Kazakhstan, Turkey, and Uzbekistan). All factors (years, replicates, accessions, countries, and regions) were assumed to have random effects. Separate from the above region/country/accession partition, a single degree of freedom was partitioned from the sum of squares for accessions to compare the mean of cultivated accessions to the mean of natural accessions.
The 13 variables were standardized and organized into 13 principal components. The principal component scores were subjected to cluster analysis using nearest centroid sorting (Anderberg, 1973; Milligan, 1980). The number of discrete non-hierarchical clusters was arbitrarily determined as that which divided the total sum of squares into 50% among clusters and 50% pooled within clusters. While it would be more desirable to work with a group of clusters that described more of the phenotypic variation, such as 70%, this would have required a minimum of 35 clusters, reducing the effectiveness of presentation and discussion, and placing additional limitations on the structure of a core collection.
|
RESULTS AND DISCUSSION |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
Accession x year interactions were significant (P < 0.05) for 9 of 12 variables (excluding plant height, forage yield, and NDF; also excluding plant survival which was scored in only one year). Variation among accessions was significant at P < 0.01 for all 13 variables. Genotypic correlation coefficients between years were generally high (r 
0.8 for most variables), indicating that accession differences and rankings were generally similar between years. Furthermore, the accession variance component always exceeded the accession x year variance component. Means over years and replicates were used in all tables and discussions to follow.
Minor Species
Analyses of variance demonstrated variation among Phleum species means for eight of the 13 variables. There were no differences among species in plant height, leaf number per tiller, panicle length or width, or forage yield. Only those variables that varied among species are presented for the minor species. Because of the immense number of P. pratense accessions and the large amount of phenotypic variation among these accessions, the range among P. pratense accessions was sufficiently large that it encompassed most of the variability among the minor species' accessions (Table 2). Nevertheless, there were numerous significant deviations of minor species' means from the mean of the P. pratense accessions.
|
P. arenarium
The five accessions of P. arenarium, all from Turkey, were uniformly 4 d earlier and had 11% shorter leaves than the average P. pratense accession (Table 2). These accessions varied for the other six variables presented in Table 2. Only one of the five P. arenarium accessions was moderately adapted to southern Wisconsin (40% plant survival; high panicle- and 1000-seed weights), but it had extremely high NDF concentration, limiting its usefulness as a forage crop. The other four P. arenarium accessions fell into narrow- or wide-leaf-blade classes, but had very low plant survival. The high panicle- and 1000-seed weights of some accessions indicated that some individual plants were well adapted to this environment, suggesting that selection for improved adaptation might be successful. Nevertheless, the extremely high NDF concentration of all five accessions, relative to the average P. pratense accession, would severely limit the value of P. arenarium accessions for forage.
P. bertolonii
The five accessions of P. bertolonii were uniformly 2 d earlier than the average P. pratense accession (Table 2). The relatively long and wide leaf blades of these five accessions, similar to the average P. pratense accession, indicated that they are forage phenotypes, rather than turf phenotypes of this species. Only one P. bertolonii accession had high plant survival, the Turkish accession. Three of the other four accessions, which were all from Spain, demonstrated some potential for adaptation (high panicle seed weight) or agronomic value (low NDF).
P. boissieri
The two P. boissieri accessions averaged 5 d earlier, 28% higher 1000-seed weight, 69% lower plant survival, and 4% higher NDF concentration than the average P. pratense accession (Table 2). The Turkish accession had 64% more panicles per plant, 28% higher 1000-seed weight, and 75% lower plant survival than the Azerbaijani accession. Because of their low survival and high NDF concentration, neither P. boissieri accession appeared to offer much agronomic potential.
P. montanum
The three P. montanum accessions averaged 5 d earlier and 79% lower plant survival than the average P. pratense accession (Table 2). Despite their low survival percentages, these three accessions were highly variable for leaf blade length, panicle number per plant, panicle seed weight, 1000-seed weight, and NDF concentration, expressing a range of variation equivalent to that among the 482 P. pratense accessions. The Azerbaijani accession had extremely short, but wide, leaf blades and low NDF concentration. The Afghani accession had relatively few panicles per plant, but extremely high panicle seed weight. The Turkish accession contrasted with more panicles per plant than any other accession, but extremely low panicle seed weight. Both the Afghani and Turkish accessions had extremely high NDF concentration, limiting their agronomic value.
P. phleoides
Of the minor species, P. phleoides was the most similar to P. pratense, but also the most variable, most likely due to the larger number of accessions (Table 2). The greatest differences between P. phleoides and P. pratense accessions were the lower panicle seed weight, 1000-seed weight, and plant survival and the higher NDF concentration of most P. phleoides accessions. Without intensive selection for increased adaptation, the only accession that appears to have agronomic value is PI 251391 (Afghanistan) which had moderately high panicle seed weight, 1000-seed weight, and plant survival, combined with NDF concentration similar to the average P. pratense accession.
P. commutatum and P. paniculatum
The P. commutatum accession (Argentina) had 39% lower plant survival than the average P. pratense accession and the P. paniculatum accession (Azerbaijan) was 6 d earlier than the average P. pratense accession. Otherwise these two accession were phenotypically similar to the average P. pratense accession.
Minor Species Summary
These minor species of Phleum appear to offer as much variation as present within P. pratense, limited primarily by the number of accessions present in the collection. The low NDF concentration, moderate plant survival percentages, and relatively high panicle- or 1000-seed weights indicate some potential for agronomic adaptation to environments similar to southern Wisconsin. The lack of variation among species for forage yield was notable, given previous reports that diploid and tetraploid Phleum species are lower in vigor and/or forage yield than hexaploid P. pratense (Caradus, 1978; Joachimiak and Kula, 1997). The distribution of forage yield means for P. pratense (minimum = 37, maximum = 264, mean = 98, and median = 95 g plant-1) was nearly identical to that for the minor species (minimum = 43, maximum = 254, mean = 107, and median = 100 g plant-1), suggesting that some of these species may have utility for livestock agriculture. Utilization of any of these minor species will likely require additional collection to broaden the genetic base and intensive selection for adaptation both within and among accessions.
Major Species: Phleum pratense
Cultivated vs. Natural Populations
Differences between cultivated and natural populations represent the accumulated and combined effects of over 100 yr of timothy breeding in numerous temperate regions of the world (Casler et al., 1996). Cultivated and natural accessions did not differ in mean plant height, heading date, or number of panicles per plant (Table 3). Cultivated accessions averaged 2% longer leaf blades, 2% wider leaf blades, and 2% fewer leaves per stem than natural accessions. Cultivated accessions averaged 5% longer panicles, 5% narrower panicles, 4% greater panicle seed weight, 4% lower 1000-seed weight, and 10% more seeds per panicle than natural accessions. Cultivated accessions also averaged 10% higher forage yield, but 6% lower plant survival than natural accessions.
|
21 yr) grazing pressure was necessary to bring about measurable adaptive responses (Van Dijk, 1955). Thus, breeding programs that rely on infrequent mechanical harvesting are unlikely to result in cultivars that are adapted to frequent and/or severe defoliation.
Variation among Region and Country Germplasm Sources
Phleum pratense accessions demonstrated considerable phenotypic variation among region and country sources (Table 4). Variation among country means within regions was significant (P < 0.05) for five of 13 variables and variation among region means was significant for the other eight variables (Table 4). The sum of squares attributable to regions always exceeded the percentage of degrees of freedom for regions (1.9%), an indication of phenotypic differentiation among regions. The sum of squares attributable to countries within regions exceeded the percentage of degrees of freedom allocated to countries (2.7%) for nine of 13 variables. Variance components indicated that regions and accessions within countries were the most important sources of variation. The variance component for these two sources of variation were largest, among the three sources, for 5 of 13 variables each. For countries within regions, this variance component was largest for only 3 of 13 variables and there were three zero estimates.
|
|
|
|
The cluster analysis identified some striking phenotypic differences among groups of accessions (Table 8). Cluster P was perhaps the most notable, with 44 accessions that had extremely short and wide panicles. Cluster K consisted of accessions that were tall and early heading with high panicle- and 1000-seed weights. Clusters H and T consisted of accessions that were short and late heading with wide leaf blades and low 1000-seed weight. Clusters I, M, and O consisted of accessions with low panicle- and 1000-seed weights. Clusters L, M, Q, and T consisted of accessions with low NDF concentration. However, Clusters M, Q, and T had problems that might limit the use of this low-NDF germplasm, such as extremely low seed weights or forage yield. Cluster L may be a valuable source of germplasm for low NDF, because its accessions had no such obvious agronomic limitations.
|
The results of this study provide a mechanism to select a core subset of accessions from the USDA-NPGS timothy collection. The core subset for Phleum should consist of representatives of each species somewhat proportional to the number of accessions of each species. For P. pratense, a core subset can be described based on the cluster analysis, with one to five accessions per cluster, proportional to the number of accessions in each cluster. Such a sampling scheme would produce a core subset of approximately 46 accessions. The greatest utility of this core subset will be for locations similar to southern Wisconsin, for which phenotypic expression and variability should be similar. For extremely different locations, this core subset may represent a more-or-less random sample of the accessions. Phenotypic data and cluster numbers for each accession evaluated in this experiment are available on the internet through the GRIN homepage.
|
ACKNOWLEDGMENTS |
|---|
|
NOTES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
Received for publication October 26, 2000.
|
REFERENCES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONREFERENCES |
|---|
Related articles in Crop Science:
This article has been cited by other articles:
|
|
 |
|
  M. D. Casler and M. Diaby Positive Genetic Correlation between Forage Yield and Fiber of Smooth Bromegrass Crop Sci., November 24, 2008; 48(6): 2153 - 2158. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. D. Casler Agronomic Performance of Timothy Germplasm from Forage and Turf Sods under Two Harvest Managements Crop Sci., August 26, 2005; 45(5): 1990 - 1996. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| 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 | |||
29, right margin) or accession distribution among clusters (