Search
Search Results
-
Eight genotypes of tall fescue were selected from a plant population consisting of 9000 individuals. Selection was based on early maturity, self-sterility, seed and forage potential, and nutritive value. Parental ...
Citation Citation
- Title:
- A comparison of progeny testing methods and estimates of combining ability for seed yield and associated variables in tall fescue (Festuca arundinacea, Schreb.)
- Author:
- Echeverri, Silvio
Eight genotypes of tall fescue were selected from a plant population consisting of 9000 individuals. Selection was based on early maturity, self-sterility, seed and forage potential, and nutritive value. Parental clones, single cross, open pollinated, self pollinated and polycross seeds were organized into a randomized block experiment with five replications each of which contained 63 entries with ten plants to the entry. The distance between entry rows was three feet; plants within the entry were planted one foot apart. The experimental nursery was planted on October 11, 1961, at the Hyslop Agronomy Farm, Corvallis, Oregon. Alta and Kentucky 31 varieties were used as checks. Parchment bags were used to secure information on self sterility and seed production. Height, maturity, disease incidence, number of culms per plant, panicles not harvested and forage yield were also evaluated. Griffing's model 1, experimental method 4, was used to test both general and specific combining ability. Simple correlation coefficients were calculated to measure the association between characters and between progeny testing methods within characters. The height of F₁ progenies was positively correlated with the height of both the mean of the parents and the largest parent. Height was also correlated with the number of culms per plant and the stage of maturity before harvesting. The stage of maturity of the single crosses was positively correlated with the stage of maturity of both the mean of the parents and the largest parent. Either polycross or open pollination may be used for evaluating maturity with comparable results. Although there was an incidence of Helminthisporium dyctioides, it was not serious enough to affect markedly the general performance of the plants in the nursery. The number of culms per plant for single crosses was negatively correlated with the number of culms for both the mean of the parents and the smallest parent. The number of culms per plant was also correlated with seed yield. From parents to S₁ there was a reduction in the number per plant.
-
2. [Article] Heritabilities and associations of seed yield components and seed yield in tall fescue (Festuca arundinacea, Schreb.)
Estimates of heritability and associations of tiller number, seed number, seed weight, and seed yield were determined in two groups of tall fescue. Group E was early maturing and had eight parental clones, ...Citation Citation
- Title:
- Heritabilities and associations of seed yield components and seed yield in tall fescue (Festuca arundinacea, Schreb.)
- Author:
- Thomas, Johnny Ray
Estimates of heritability and associations of tiller number, seed number, seed weight, and seed yield were determined in two groups of tall fescue. Group E was early maturing and had eight parental clones, while group I was intermediate in maturity and had nine parental clones. The parents and five progeny groups were included in the study: first generation selfed progeny (S₁), open-pollinated progeny, polycross progeny, single-cross progeny, and selfed progeny from the single-crosses (F₂). Each entry consisted of ten plants in a nine foot row, with three feet between rows. The plants were established in the field in October 1962, in a randomized block design with four replications. Data were collected during the summers of 1963, 1964, and 1965. Group E plants had slightly higher seed yields than group I plants. This yield advantage was mainly the result of heavier seeds in group E. The parents and the open-pollinated, polycross, and single-cross progeny were generally highest in performance for all four characteristics, while the S₁ progeny were intermediate and the F₂ progeny were lowest in performance. The open-pollinated and polycross progeny were limited in their usefulness for evaluating breeding material since there were no significant differences among the open-pollinated progenies in 13 of 24 analyses of variance and there were no significant differences among the polycross progenies in 12 of 24 analyses of variance. Six methods (regression of the S₁ progeny on the parents, two times the regression of the open-pollinated progeny on the parents, two times the regression of the polycross progeny on the parents, regression of the single-cross progeny on the mid-parents, regression of the F₂ progeny on the mid-parents, and mean squares from the diallel analysis) of estimating heritability were compared for both populations. There were large differences in the heritability estimates among years and among methods. Two methods, 2([superscript b][subscript OP.P]) and [superscript b][subscript SX.MP'] generally resulted in higher heritability estimates than the other four methods. The "average heritability" (averaged across six methods and three years) for tiller number, seed number, seed weight, and seed yield were .187, .371, .506, and .193 respectively for group E and .567, .417, .622, and .374 for group I. The correlation and path-coefficient analysis for groups E and I indicated that tiller number had the largest direct effect in 1963, while seed number was most important in 1964 and both were equally important in 1965. Seed weight had relatively small effects all three years. All indirect effects were minor in both populations. The correlations and the direct and indirect effects for the parent, single-cross and F₂ generations were more variable and they were not always in agreement with the direct and indirect effects for the combined data of the parents and five progeny groups. There were large negative indirect effects in the individual generations which were not evident for the combined data of the parents and five progeny groups. Group E could be improved most by breeding for increased seed number while selection for yield would be most worthwhile in group I. However, maximum improvement could be made by combining the attributes of the two populations. The high seed weight of group E could be incorporated into group I and seed number of group E could be increased by crossing plants of group E with plants of group I.
-
3. [Article] Heterosis and inbreeding in the progeney of genetically diverse parental clones of Festuca arundinacea Schreb
The response of maturity, vigor, plant spread, plant height, three forage harvest yields, and total forage yield to heterosis and inbreeding was studied in tall fescue. Parent clones selected for diverse ...Citation Citation
- Title:
- Heterosis and inbreeding in the progeney of genetically diverse parental clones of Festuca arundinacea Schreb
- Author:
- Johnson, Blaine Earl
The response of maturity, vigor, plant spread, plant height, three forage harvest yields, and total forage yield to heterosis and inbreeding was studied in tall fescue. Parent clones selected for diverse anthesis date, origin, and morphology were separated into two groups based on anthesis date, those having early maturity and those having late maturity. Parents, single-cross (SX), and first generation selfed (S₁) progeny were field evaluated. Parents, S₁, and SX progeny were space planted on a . 914 x 1.219 meter basis at the Hyslop Agronomy Field Laboratory, Corvallis, Oregon, in September, 1969. A randomized block design having four blocks and 14 plant rows as entries was used. Data were collected during the spring, summer, and fall of 1972. Single-cross progeny had a mean performance better than their midparents ' mean performance for all characteristics, with all differences significant except for plant height and third harvest forage yield. Single-cross progeny averaged 7. 03, 13. 18, 13. 50, 3. 89, 42. 93, 51. 72, 15. 14, and 37. 12 percent better than their midparents for maturity rating, vigor rating, plant spread, plant height, first, second, and third harvest forage yield, and forage yield, respectively. There was a greater frequency of individual single-crosses that exhibited significant heterosis in the early x late group than in either the early x early or late x late groups. In addition the average heterotic response of the early x late group was consistently greater than that of either of the other two groups. Thus it appears significant heterosis is more likely to occur, and is likely to be of a greater magnitude, in the progeny of parents having maturity differences, and presumably, greater genetic differences. Significant variation occurred among the single-cross maturity group means for all characteristics with the early x early single crosses most often being the better performers. Consequently the additional heterosis observed in the early x late group was usually insufficient to bring performance of this group up to that of the higher performing but less heterotic early x early single-crosses. This suggests heterosis may be of little practical importance. Inbreeding depression of S₁ progeny was significant for all characteristics. The inbreeding depression was of the greatest magnitude in the progeny of the early parents. Performance of early parents and early S₁ progeny was generally better than that of their late counterparts. Heritability estimates were high, particularly those obtained by regressing single-cross progeny on midparents, although two did not differ significantly from zero. The coefficients of determination for the same associations were likewise high, with over 70 percent of progeny variation being explained by linear association with parents for most characteristics. As superior progeny came from superior parents, this high degree of association suggests clonal evaluation may be an effective screening method in tall fescue. There was strong association among the characteristics of maturity rating, plant spread, plant height, first and third forage yield, and total forage yield, with the lowest of these coefficients of determination, R² = .6031, occurring between maturity rating and plant spread. Vigor rating, second harvest yield, seedling vigor index, and panicle number showed little association among themselves or among the other characteristics, with the highest association, R² = . 3599, occurring between panicle number and third harvest yield. Seed yield was moderately associated with forage characteristics, with R² values of about .5000, but showed no association with panicle number.
-
Response to heterosis for plant height, anthesis date, panicle number, seed yield, and fall vigor rating was determined in a tall. fescue group selected for diverse morphology, origin, and anthesis date. ...
Citation Citation
- Title:
- An examination of seedling vigor and the effects of genetic diversity on response to heterosis in tall fescue (Festuca arundinacea Schreb.)
- Author:
- Moutray, Jim B., 1939-
Response to heterosis for plant height, anthesis date, panicle number, seed yield, and fall vigor rating was determined in a tall. fescue group selected for diverse morphology, origin, and anthesis date. Parents, single-cross progeny (SX), and first generation selfed progeny (S₁) were included in the study. Each experimental plot consisted of 14 plants spaced three feet apart in a row with four feet between rows. Plants were established in the field in September, 1969, in a randomized block design with four replications. Data were collected during the summer and fall of 1970. Single-cross progeny were superior to parents with the exception of seed yield, with parents above S₁ progeny except in plant height. Single-crosses averaged 15. 36, 2. 60, 28.21, 24. 79, and 23. 04 percent above the mid-parent for plant height, early anthesis date, panicle number, seed yield, and fall vigor rating respectively. Single-crosses averaged 5.85 and 10.27 percent above the high-parent for panicle number and fall vigor rating respectively. Crosses between maturity groups resulted in the greatest heterosis above the mid-parent for all characteristics and above the high-parent for panicle number (117.50 percent of the high-parent). Early x early single-crosses were tallest (106. 49 percent of the high-parent) and had the most vegetative vigor (115. 36 percent of the high-parent). Indications are that crosses between parents of diverse morphology and origin result in a greater expression of heterosis than crosses among similar parents. Diverse anthesis date appears to increase response to heterosis. Vigor and associations among vigor characteristics were determined from single-cross and S₁ seed and seedlings. Respiration, germination, root and shoot growth, rate of growth, and unit growth characteristics were measured using a Gilson differential respirometer and a seed germinator. For most seed and seedling vigor characteristics S₁ 's and SX's were very similar. Early S₁ and early x early SX groups were consistently more vigorous than late S₁ and late x late SX groups with the early x late SX group intermediate between early x early and late x late SX groups for all seedling vigor characteristics. Root length and vigor index were two of the better indicators of early seedling vigor, with vigor index favored because it takes less time to measure and requires less space. Selection on the basis of vigor index should be an effective laboratory method of screening for higher seedling vigor in tall fescue. There was a significant association between most seed and seedling vigor characteristics in SX's and S₁'s, and most seed and seedling vigor characteristics were associated with from one to three mature plant characteristics, most often fall vigor rating. Three methods of selection for high and low seedling vigor (seed weight, head selection on shoot length, and emergence from deep seeding) resulted in small changes in seed weight and vigor index of progeny. Clones from a population with low seed weight and seedling vigor were self-pollinated and topcrossed to a source with high seed weight and seedling vigor. Positive associations between topcrosses and S₁'s for seed weight and vigor index suggests a lack of appreciable heterosis for these characteristics. The association of forage and seed yield was studied in duplicate populations of parents and five progeny groups. These were: first generation selfed (S₁), open-pollinated OP), polycross (PX), single-cross (SX), and selfed single-cross (F ). Nurseries were planted in a randomized block design with four replications, with harvests made during the spring and summer of 1969. There were significant associations of forage and seed yield in four progeny groups with no association in parents and polycross progeny. The highest correlation occurred in the S₁ progeny (r .794). High forage and seed yields in certain OP, PX, and SX progenies indicate the possibility of breeding for both characteristics simultaneously.
-
5. [Article] The heritability of agronomic characteristics in tall fescue (Festuca arundinacea Schreb.)
Three genotypes of tall fescue (Festuca arundinacea Schreb. cultivar Fawn and four genotypes from the cultivar Fortune were used as parents to form 15 of a possible 21 single crosses to study the heritability ...Citation Citation
- Title:
- The heritability of agronomic characteristics in tall fescue (Festuca arundinacea Schreb.)
- Author:
- Subhanij, Thiravira
Three genotypes of tall fescue (Festuca arundinacea Schreb. cultivar Fawn and four genotypes from the cultivar Fortune were used as parents to form 15 of a possible 21 single crosses to study the heritability of certain agronomic characteristics. The Fawn variety is considered to be tall in growth habit, vigorous and early in anthesis. The Fortune variety represents germ plasm with a shorter growth habit, less vigor and later anthesis. Crossing was accomplished by mutual pollination in the greenhouse during February 1972. A greenhouse experiment was started in June 1972 using the 15 single crosses in a randomized block with three replications. Plants were supplied with one half strength of a modified Hougland's solution twice weekly until August 3 followed with an application of the same nutrient solution every two weeks until they were taken to the field for establishment. Greenhouse data were collected on July 7, August 4 and September 11. These data indicated that the variation among single crosses for tiller numbers and dry weight were greater as the plant grew older. Nutrient stress affected dry matter yield by reducing the variation and mean performance of the single crosses. Differences among single crosses for shoot length, tiller number, and plant spread were still quite evident under the stress conditions. Shoot length showed significant general combining ability (GCA) with non significant specific combining ability effects (SCA) on July 7 and September 11. GCA effects for tiller numbers were evident on August 4 and September 11. Both GCA and SCA effects were significant for plant spread on September ii. GCA for dry weight were significant only on July 7. Based on the diallel analysis, additive gene action was primarily responsible for shoot length, tiller numbers and dry weight. Both additive and non additive gene action influenced plant spread. The field establishment included the 15 single crosses from greenhouse experiments and the seven parent clones, established in a randomized block design with three replications. From the diallel analysis, it was observed that the GCA:SCA mean square for the characters measured were: plant height (19:1), plant spread (2. 1:1), anthesis date (3.2:1), panicle number (1.4:1), panicle length(l2:1), number of primary pedicels per panicle (7.4:1), five panicle seed weight (0. 2:1), seed yield per plant (0. 4:1) and 100 seed weight (15.1:1). This indicated that additive gene action was of major importance in the expression of plant height, number of primary pedicels and 100 seed weight. Non additive gene action was contributed substantially more to the expression of five panicles seed weight and seed yield. Both additive and non additive gene action contributed to plant spread and anthesis. Inferences about gene action for panicle numbers and panicle length were not as evident. This is because of the low GCA:SCA ratio, a barely significant GCA for panicle number, and no significant GCA or SCA for panicle length. Single cross progeny were superior to mid parent for plant height (9.27 percent), plant spread (20. 33 percent), anthesis (20.90 percent, which suggested earlier anthesis than mid parent), panicle number (36. 69 percent), and panicle length (11. 44 percent). No apparent superiority of single cross progeny over their corresponding mid parent for number of primary pedicels and 100 seed weight was observed. Of all single crosses, only Fawn x Fortune crosses were inferior to mid parents for five panicle seed weight (65.93 percent) and seed yield (60. 22 percent). The Fawn x Fortune performance for both of these characters (0.4062 and 1.23 g per plant, respectively)was low, The diverse parentage of these crosses may have resulted in irregular meiotic behavior in the F₁ which in turn may have caused inviable gametes to be formed. This sterility would result in the low seed yields observed. This is a problem that should be cytologically investigated. Crosses between diverse germ plasm (Fawn x Fortune crosses) did result, however, in heterosis for all characteristics except number of primary pedicels and 100 seed weight. But Fortune x Fortune crosses, for certain characters such as plant height, plant spread, panicle numbers, panicle length,exhibited even more heterosis. No heterosis was observed in the Fawn x Fawn crosses. Medium to high heritability estimates were observed for 100 seed weight, plant spread, anthesis date, panicle number and panicle length, while medium to low heritability were observed for panicle length. Low heritable value was evident for plant height, number of primary pedicels, five panicle seed weight and seed yield. Heritability estimates were based on parent progeny relationships involving all single crosses and mid parents. The relationships among various plant characteristics and seed yield was studied. Five panicle seed weight and panicle numbers were significantly correlated with seed yield. However there was a non-significant association of panicle length with seed yield and number of primary pedicels with seed yield.These relationships suggest that higher yielding genotypes may be identified by observing only few panicles per plants. The panicle characteristics examined in this study (other than seed weight) were of little value in identifying high seed yield genotypes, but panicle seed weight was positively related to plant yield.
-
6. [Article] Evaluation of methods of progeny testing in tall fescue seed production (Festuca arundinacea, Schreb.)
Seventeen parental clones were selected, from 9,000 plants, on the basis of self-sterility, seed and forage yield potential using phenotypic ratings, crude protein and chromogen content. The parental clones ...Citation Citation
- Title:
- Evaluation of methods of progeny testing in tall fescue seed production (Festuca arundinacea, Schreb.)
- Author:
- Thomas, Johnny Ray
Seventeen parental clones were selected, from 9,000 plants, on the basis of self-sterility, seed and forage yield potential using phenotypic ratings, crude protein and chromogen content. The parental clones were divided into two populations based on date of maturity. Group I was intermediate in maturity and had nine parental clones, while group E was early maturing and had eight parental clones. Six progeny testing methods (vegetative cuttings of parental clones, first generation selfed-seed (S₁ progeny), open-pollinated progeny, polycross progeny, seed from all possible single crosses, and selfed seed from the single-crosses or F₂ progeny) were compared using simple correlation coefficients between the progeny test and by comparing the rankings of the clones on the basis of each progeny test. The actual polycross performance was compared with the theoretical polycross performance or average of all possible single crosses, to see if equal and random fertilization occurred in the actual polycross. The results from the comparison of the six progeny testing methods indicated the vegetative cuttings of the parental clones and the single-cross progeny tests are the two most effective progeny testing methods when considering all four traits and both populations of plants. The open pollinated and polycross progeny tests were generally undesirable, because frequently there was no significant difference among their progeny. The S₁ progeny test was generally not desirable in group I; however, it appeared to be the most satisfactory test for seed yield when both populations of plants were considered. The results illustrate why there are inconsistencies in the literature concerning the effectiveness of the progeny testing methods. The effectiveness of the six progeny tests varied, to some extent, from trait to trait and for the two populations. For all four traits there was much less agreement among the six progeny tests in group E than in group I. In all four traits the F₂ generation showed a great inbreeding depression. The average of the single crosses was lower than the parents for tiller number, plant height, and 100 seed weight, but equalled the parents in yield per plant. The S₁ progeny were usually approximately equal to the average of the single cross progeny. The open-pollinated and polycross progeny were generally the highest for all four traits. The results from the comparison of the actual polycross with the theoretical polycross indicated that equal and random fertilization did not occur in the actual polycross. The more desirable clones appeared to contribute the majority of the pollen, since the actual polycross performance was usually greater than the theoretical polycross performance.
-
All possible single crosses from three different maturity groups of 27 selected parental genotypes of tall fescue were established in a randomized block experiment to be evaluated in 1962 for five agronomic ...
Citation Citation
- Title:
- Single cross progeny evaluation in tall fescue, Festuca arundinacea, Schreb
- Author:
- Cáceres, Oscar
All possible single crosses from three different maturity groups of 27 selected parental genotypes of tall fescue were established in a randomized block experiment to be evaluated in 1962 for five agronomic characters: height, width and maturity per plant; self -seed yield per plant; and forage yield per plant. For each character, general and specific combining ability effects, relationship among these characters and heritability estimates based on mean square expectations were obtained. Results of this study indicate that genetic diversity present in the original parental population resulted in significant differences among maturity groups for the characters measured, with the exception of self-seed yield. The variability within each group was small and significant only for maturity and self-seed yield in the intermediate group and for height and maturity in the late group, indicating that the progenies were quite uniform for the other characters. Significant correlation coefficients were noted between forage yield with height and width of the plant. This significant association was observed in all three maturity groups. Estimates of heritability in narrow sense for maturity and self-seed yield in the intermediate maturity group were 0.19 and 0.46, respectively. With the preponderance of additive gene effects for self-seed yield, the value of breeding material for this character may be indicated early in a breeding program.
-
8. [Article] The breeding behavior of forage yield in two populations of tall fescue, (Festuca arundinacea, Schreb.)
The single-cross and most F₂ progenies of 19 genotypes of tall fescue (Festuca arundinacea, Schreb. ) were used to study the breeding behavior of forage yield. Parental genotypes were selected for high ...Citation Citation
- Title:
- The breeding behavior of forage yield in two populations of tall fescue, (Festuca arundinacea, Schreb.)
- Author:
- Matheson, Keith Isom
The single-cross and most F₂ progenies of 19 genotypes of tall fescue (Festuca arundinacea, Schreb. ) were used to study the breeding behavior of forage yield. Parental genotypes were selected for high self-sterility, high chromogen and high crude protein. These were studied in two groups of nine intermediate and ten late-flowering genotypes. Progenies were grown on the Hyslop Agronomy Farm, Corvallis, Oregon as plants spaced one foot apart with three feet between rows. Ten plants of each cross for each generation were arranged in a randomized block design with four replications. Variables measured were plant width, natural plant height, plant spread, leaf length, total plant height of regrowth tissue, plant density and forage yield. The collected data were converted to plant means. Broad- and narrow-sense estimates of heritability were obtained from expected mean squares according to Griffing's analysis (random model, method four) and compared with estimates of heritability obtained by F₂-single-cross regression. General and specific combining ability effects were observed and the single-cross-F₂ inbreeding depressions were studied. The F₂ families were obtained by the self-pollination of five single-cross plants. General combining ability mean squares averaged from two to nine times greater than the mean squares for specific combining ability, with the largest difference being 54.7 times greater. The two populations responded similarly for the variables measured except for total plant height and plant spread. Greater progress from selection could be expected in the late- maturing group with the exception of these two traits. In most cases, single-cross progenies performed better than their corresponding F₂ progenies. The average of all single-crosses differed from the average of all F₂ progenies in 12 of 20 analyses in 1963, but in only 2 of 12 in 1964. The inbreeding characteristic was most pronounced for forage yield. High-performing single-crosses for the traits studied tended to produce high-performing F₂ progenies. Little or no heterosis was expressed and the expression of inbreeding depression appeared to be influenced greatly by the environment. Heritability estimates indicated that most traits could be studied more effectively after the second year of plant establishment. The environmental influence appeared to be less pronounced at that time. Where the coefficient of determination was high between the single-cross and F₂ generations, traits were less influenced by the environment and the disparity between the estimates of heritability was less. When used on cross-pollinated polyploid grasses, the method of diallel analysis proposed by Griffing does not appear to remove all environmental influence and thus overestimates heritability. Forage yield in the two populations of tall fescue studied was governed principally by additive gene action. Heterosis expressed was mostly in crosses involving low-performing parents.
-
Tall fescue often results in poor liveweight gains. This may be accounted for by variations in perloline content, nitrogen content, or digestibility of the plant. Two groups of tall fescue plants were ...
Citation Citation
- Title:
- Heritability for perloline, nitrogen, and digestibility characteristics in tall fescue (Festuca arundinacea, Schreb.) single-crosses grown in two locations
- Author:
- Watson, Clarence Ellis
Tall fescue often results in poor liveweight gains. This may be accounted for by variations in perloline content, nitrogen content, or digestibility of the plant. Two groups of tall fescue plants were examined for each of these traits. These groups consisted of (a) fifteen single-crosses of a six parent diallel and (b) seven cultivars of tall fescue. The diallel cross was planted at Corvallis, Oregon and Columbia, Missouri. Plots were harvested and sampled on April 8 and September 22, 1975 at Corvallis; Oregon and on October 16, 1975 at Columbia, Missouri. The two cuttings at Corvallis, Oregon and the fall cuttings at Corvallis and Columbia were analyzed as multiple environments. The cultivars were planted at Corvallis, Oregon and harvested April 10, July 11, and September 22, 1975. All samples were analyzed for perloline and nitrogen content. Samples from the fall harvests at Corvallis, Oregon and Columbia, Missouri were analyzed for in vitro digestibility. There were significant differences among crosses for perloline content. Significant general and specific combining ability was found for perloline content. Broad-sense heritabilities (H[subscript B. S.] = .21-.94) tended to be larger than narrow-sense heritabilities (H[subscript N.S.] = .15-.53). There appeared to be a high degree of dominance for low perloline. Significant differences existed among the cultivars for perloline content. Dates of harvest were a significant source of variation for perloline content, with perloline increasing steadily over the season. Significant genotype x environment interactions were found, suggesting that lines should be tested over locations and years. It should be possible to select for low perloline lines with a program of recurrent selection using a high perloline tester. Significant differences were found among crosses for nitrogen content. General combining ability was significant in all cases. Narrow-sense heritabilities (H[subscript N.S.] = .50-.72) were almost as large as the broad-sense heritabilities (H[subscript B. S.]= .55-.89), suggesting primarily additive gene action for nitrogen content. No differences were found among the cultivars for nitrogen content, but cuttings were a significant source of variation. Differences among cuttings as well as the cutting x cultivar interaction appeared to be due to soil fertility levels and disease. Selection for nitrogen content should be possible with recurrent selection. Lines should be tested over environments. There were differences in in vitro digestibility among crosses for the September harvest at Corvallis, Oregon, but not for the October harvest at Columbia, Missouri. General combining ability was a significant source of variation. The broad-sense heritability (H[subscript B. S.] = .74) was high. The narrowsense heritability (H[subscript N.S.] = .18) was very low, however, indicating that in vitro digestibility was controlled primarily by nonadditive gene action. Digestibilities were significantly higher at Corvallis, Oregon than at Columbia, Missouri. No genotype x environment interaction was found. Selection for high in vitro digestibility should be possible with a program of recurrent selection for specific combining ability.