9512.net
甜梦文库
当前位置:首页 >> >>

Responses to Divergent Selection for Fiber Concentration at Two Disease Potentials in Smoot



Responses to Divergent Selection for Fiber Concentration at Two Disease Potentials in Smooth Bromegrass
L. X. Han, M. D. Casler,* and C. R. Grau ABSTRACT
Neutral detergent fiber (ND

F) concentration is related to intake potential of forages but might also be related to plant reaction to pathogens. Three smooth bromegrass (Bromus inermis Leyss.) populations from these cycles of uniparental mass selection for low NDF, one cycle of biparental mass selection for high NDF, and the base population were evaluated at high and low disease potentials at three locations for 2 yr and three harvests each year. High disease severity, measured as proportion of foliage diseased, was associated with high NDF and high lignin concentration (on a dry-matter basis and a cellwall basis). Nevertheless, disease potentials did not affect the relative performance of selection cycles on any traits measured. On average, NDF was reduced 3.5 g kg 1 DM per cycle (P 0.001) by selection for low NDF, and increased 11.1 g kg 1 DM in one cycle (P 0.001) by selection for high NDF. For the spring harvest, the response in NDF could partially be attributed to a change in heading index. A positive correlated response in acid detergent lignin was found in the spring (P 0.001) while a negative correlated response in cell wall lignin was detected in the summer and fall (P 0.001). Selection for NDF modified the composition of NDF, as cell wall hemicellulose changed in the opposite direction of NDF. Lower dry matter yield per plant appeared to be associated with lower NDF and higher inbreeding. Selection for low NDF, accompanied by slight selection pressure on disease resistance, resulted in a reduction of NDF without an increase in disease susceptibility.

eutral detergent fiber is a measure of cell wall concentration for forages (Goering and Van Soest, 1970). It is also an indicator of intake potential of feed for ruminant animals. When forage alone is fed to sheep or cattle, NDF is negatively correlated with dry matter intake (Osbourn et al., 1974; Reid et al., 1988). When forage is fed together with concentrates, NDF is negatively correlated with dry matter intake if the rumen capacity of the animal is not reached (Conrad et al., 1964; Mertens, 1987; Ruiz et al., 1995). Thus, it is desirable to reduce NDF of forages. Genetic modification of NDF in forages has been reported. In reed canarygrass (Phalaris arundinacea L.), one cycle of bidirectional selection for NDF and forage yield created the following groups: low NDF–low yield, low NDF–high yield, mean NDF–mean yield, high NDF–low yield, and high NDF–high yield (Surprenant et al., 1988). At both reproductive and vegetative stages, the two low-NDF groups, on average, had lower NDF and acid detergent fiber (ADF) than the two high-NDF
L.X. Han, Dep. 436, AP9A/2, Abbott Laboratories, 100 Abbott Park Road, Abbott Park, IL 60064-6124; M.D. Casler, Dep. of Agronomy, Univ. of Wisconsin-Madison, 1575 Linden Dr., Madison, WI 537061597; C.R. Grau, Dep. of Plant Pathology, Univ. of Wisconsin-Madison, 1630 Linden Dr., Madison, WI 53706-1598. Part of a dissertation submitted by the senior author for the partial fulfillment of the requirements of the Ph.D. degree at Univ. of Wisconsin-Madison. Received 23 Nov. 1999. *Corresponding author (mdcasler@facstaff.wisc.edu). Published in Crop Sci. 41:30–39 (2001).

N

groups, although the difference in NDF at reproductive stage was attributed to the difference in maturity. At reproductive stage, the low-NDF groups also had lower yield per plant than the high-NDF groups; at the vegetative stage, their difference in yield was not significant. In smooth bromegrass (Bromus inermis Leyss.), one cycle of divergent selection for NDF produced a lowNDF synthetic and a high-NDF synthetic (Carpenter and Casler, 1990). The low-NDF synthetic was lower in NDF and ADF than the high-NDF synthetic, but the two synthetics were not significantly different in sward dry matter yield or acid detergent lignin (ADL). Most recently, Casler (1999) reported on the evaluation of selection cycles from six recurrent selection methods for reducing NDF in smooth bromegrass. Three of these methods completed three cycles; the other three methods completed two cycles. The response to the selection was significant at the vegetative and/or reproductive stage for five of the six methods. The least effective method in that study was uniparental mass selection at heading stage using wet-laboratory analysis for NDF determination. The three selection cycles selected by that method as well as the base population of the selection were independently evaluated in this study with more replications and at more sites. Cell walls, especially when highly lignified, could serve as preexisting structural defenses against microorganisms which are incapable of producing the appropriate degradative enzymes (Ride, 1983). Therefore, it is conceivable that genetic modification of NDF, of which lignin is a part, could change the reaction of plants to pathogens. Water soluble carbohydrates (WSC) make up the largest share of noncell-wall carbohydrates. Breese and Davies (1970) reported that successful divergent selection for WSC in perennial ryegrass (Lolium perenne L.) for three generations resulted in higher susceptibility to crown rust infection (caused by Puccinia coronata Corda) for the families with high WSC. Webb et al. (1996) reported that plants selected for low ADL in alfalfa (Medicago sativa L.) showed greater susceptibility to alfalfa rust (caused by Uromyces striatus J. Schrot.) and spring blackstem (caused by Phoma med¨ icaginis Malbr. & Roum. in Roum.) in the field than those selected for high ADL, although a further experiment in the greenhouse found little or no relationship between ADL and the reaction of alfalfa to the rust pathogen. It was our concern that selection for low NDF might have increased susceptibility of smooth bromegrass to diseases. On the other hand, selection for disease resistance has been shown to alter NDF in some cases further
Abbreviations: ADF, acid detergent fiber; ADL, acid detergent lignin; CW, cell wall; NDF, neutral detergent fiber; NIRS, near-infrared reflectance spectroscopy; WSC, water soluble carbohydrate.

30

HAN ET AL.: FIBER CONCENTRATION AND LEAF DISEASE IN SMOOTH BROMEGRASS

31

raising the concern that NDF and disease resistance could be genetically correlated. In intermediate wheatgrass [Thinopyrum intermedium (Host) Barkworth and D.R. Dewey], strains resistant to foliar diseases had lower NDF than those susceptible when diseases were present (Karn et al., 1989). In orchardgrass (Dactylis glomerata L.), three cycles of selection for resistance to purple leaf spot (caused by Stagonospora arenaria Sacc.) did not change NDF, even though the third cycle was later in maturity than the original cycle (Oberheim et al., 1987). In alfalfa, the cultivar Saranac, which was susceptible to anthracnose (caused by Colletotrichum trifolii Bain), and the cultivar Saranac AR, which was selected from Saranac for resistance to anthracnose, were compared for NDF in the greenhouse with and without inoculation with Colletotrichum trifolii (Lenssen et al., 1991). Saranac AR was lower than or similar to Saranac in NDF in stems or leaves. In addition, the difference between the two cultivars was not consistent in the two treatments or in different plant tissues. The relationship between cell wall traits and disease reaction goes beyond genetic correlations. Forage plants with leaf diseases could have different NDF, ADL, and yield than healthy plants of the same genotypes. In a greenhouse study on intermediate wheatgrass, plants infected with Cochliobolus sativus (Ito and Kurib.) Drechsl. ex Dastur had higher NDF than healthy plants of the same strains (Karn and Krupinsky, 1983). In two greenhouse trials on orchardgrass, leaves of inoculated plants had higher ADL (permanganate method) than the leaves of the uninoculated plants (Sherwood and Berg, 1991). In a field study on alfalfa, crude fiber of fungicide-protected plots was similar to that of unprotected plots while dry matter yield of fungicide-protected plots was higher than or similar to that of unprotected plots depending on which fungicide was used (Summers and McClellan, 1975). In a field study on pearl millet [Pennisetum americanum (L.) Leeke] and its rust (Puccinia substriata Ellis and Barth. var. indica Ramachar and Cummins), fungicide-protected and urediniospore-inoculated plots had similar dry matter yield for two resistant cultivars while fungicide-protected plots yielded better than inoculated plots for two susceptible cultivars (Wilson et al., 1991). In a greenhouse study on red clover (Trifolium pratense L.), plants inoculated with Stemphylium sarciniforme (Cav.) Wiltsh. had 20 to 30% lower dry matter yield than uninoculated plants (Berg and Leath, 1996). The objectives of this study were (i) to evaluate divergent selection for NDF in smooth bromegrass for direct response in NDF and correlated responses in disease resistance and other traits, (ii) to determine whether these responses change at different disease potentials, and (iii) to ascertain how plants at different disease potentials differ in cell wall traits and dry matter yield. MATERIALS AND METHODS Divergent Selection
Divergent selection for NDF in smooth bromegrass was conducted during 1985–1991 at the Arlington Agricultural Research Station, Arlington, WI (Plano silt loam [fine-silty,

mixed, mesic Typic Argiudolls]). The base population of the selection, WB-RP1, is a Syn 2 generation of 129 clones of diverse sources (Casler, 1992). Three cycles of uniparental mass selection were completed in the selection for low NDF. One cycle of biparental mass selection was completed in the selection for high NDF. In each cycle, 400 to 420 plants were planted at 1.0-m spacing for the population, and 350 of them were sampled for forage at fully headed growth stage in the spring and analyzed for NDF by the wet-lab procedure of Robertson and Van Soest (1981). The plants that were not sampled were either heavily diseased or ignored arbitrarily. The diseases in the field were mostly brown leafspot caused by Pyrenophora bromi (Died.) Drechs. Five to 10 tillers were taken from each plant and were bulked to represent the plant. In the selection for low NDF, 35 plants were selected in each cycle on the basis of t-score of the grid system (Casler, 1992). Equal weight of seeds from each selected plant was bulked to plant the next cycle and used for the evaluation of selection responses. In the selection for high NDF, 10 plants were selected on the basis of t-score; and the selected plants were transplanted to a polycross block in three replications. Equal weight of seeds from each plant in the polycross block was bulked for the evaluation. We will designate the base population as C0, the first, second, and third cycle from the selection for low NDF as C-1, C-2, and C-3, respectively, and the cycle from the selection for high NDF as C 1. The cycles C0, C-1, C-2, and C-3 were evaluated previously for NDF in a methodology study by Casler (1999).

Evaluation in the Field
The five cycles, C-3 to C 1, were evaluated at three locations: Arlington Agricultural Research Station, Arlington, WI; West Madison Agricultural Research Station, Verona, WI (both Plano silt loam [fine-silty, mixed, mesic Typic Argiudolls]); and Agrecol Corp., Sun Prairie, WI (Houghton muck [Euic, mesic Typic Medisaprists]). A split-plot design was used at each location with disease potential (high or low) as the whole plot factor and selection cycle as the subplot factor. The whole plot factor was arranged in four randomized complete blocks at each location, but some plots were lost because of flooding at Sun Prairie, resulting in incomplete data at that location. Each subplot had two rows of 10 plants each. Centers of two adjacent plants were spaced at 0.91 m. All selection cycles were seeded in June of 1994 in the greenhouse, and the seedlings were transplanted to the field in August of 1994. To create different disease potentials, we selected Cochliobolus sativus, a cause of leaf blight on smooth bromegrass, as a pathogen. The isolates of C. sativus used for inoculation were isolated from the diseased leaves of smooth bromegrass at the Arlington Agricultural Research Station in 1993 and were cultured on potato dextrose agar to produce conidia. Before transplanting, plants at the high disease potential were sprayed using a hand sprayer until runoff with a conidial suspension of C. sativus (5.7 104 conidia per mL sterilized distilled water, with two drops of Tween 20 added per 100 mL). Then the inoculated plants were kept in a humidity chamber for 24 h before being transplanted 4 to 8 d later. At the time of transplanting, all inoculated plants expressed moderate to severe disease symptoms. Additional inoculum was not delivered to plants once they were established in the field. The plants at the low disease potential were sprayed with sterilized distilled water (two drops of Tween 20 added per 100 mL) in place of conidial suspension before being placed in a humidity chamber and transplanted to the field. In addition, the plants at the low disease potential were sprayed with 0.9 to 2.2 kg a.i. ha 1 mancozeb (zinc ion and manganese ethylene-bisdithiocarbamate) one to two times for

32

CROP SCIENCE, VOL. 41, JANUARY–FEBRUARY 2001

each growth-harvest cycle to reduce disease occurrence. The plants were fertilized with 100 kg N ha 1 2 wk after being transplanted. Yield data were collected in June, August, and October of 1995 and 1996. Forages were harvested by subplot with a small-plot harvester, and fresh forage yield was measured for each subplot. The plots at Sun Prairie were not harvested at the third harvest in 1996 due to inadvertent plot damage. One or two handfuls of forage were taken from the harvester for each subplot and air-dried at 55 C to provide the dry matter percentage of the fresh forage. Dry matter yield for each subplot was calculated from the fresh forage yield and the dry matter percentage of the sample from the harvester, and then divided by the number of live plants in the subplot to convert to dry matter yield per plant for a subplot. Plants were fertilized at the rate of 56 kg N ha 1 in the spring and after spring and summer harvests. Weeds were controlled by a combination of hand weeding and application of 1.12 kg a.i. ha 1 alachlor [2-chloro-N-(2,6-diethylphenyl)-N-(methoxymethyl) acetamide] and 0.56 kg a.i. bromoxynil [3,5-dibromo-4-hydroxybenzonitrile] in the spring. Disease severity was assessed for the second and third harvests in 1995 and all three harvests in 1996. One to two weeks before harvesting forage for yield, each plant was scored visually for the percentage of the foliage diseased on the scale of 0 to 10: 0 less than 1% of all leaf tissues diseased, 1 1 to 10%, 2 11 to 20%, 3 21 to 30%, 4 31 to 40%, 5 41 to 50%, 6 51 to 60%, 7 61 to 70%, 8 71 to 80%, 9 81 to 90%, and 10 91 to 100%. The average of all plants in a subplot was used to represent the disease severity of that subplot. Further, the average disease severity of a subplot was transformed to a 0 to 2 log scale for statistical analysis by the formula y log10(10x 1) to correct for unequal variances on the original scale, where x is a value on the original scale and y is the transformed value on the log scale. No attempt was made to distinguish different diseases in disease assessment, although three diseases were predominant: leaf blight caused by C. sativus, brown leafspot caused by P. bromi, and rust caused by Puccinia coronata Corda (Delgado, 1998). Plants of all selection cycles at both disease potentials at all three locations survived well throughout the experiment, thus no survival data were taken. Frequency of heading prior to harvest was scored as 0 to 2 according to the number of tillers with fully emerged panicles: if none, score 0; if 1 to 5, score 1; if more than 5, score 2. The scores of all plants for a subplot were averaged to give a variable, which we call heading index. It was taken for the second and third harvests in 1995 and all three harvests in 1996. Heading index was a measure of the number of emerged heads at the time of assessment for a subplot, combining variation in relative maturity with variation in number of tillers per plant.

tion samples, all on a dry-matter basis, were determined by sequential analysis on 0.5-g samples in duplicate (Robertson and Van Soest, 1981). To improve the precision of ADL, all samples were analyzed for ADF and ADL again on 0.5-g samples in duplicate, skipping the NDF step. Acid detergent fiber determined without the NDF step was, on average, 12 g kg 1 DM higher than ADF determined with the NDF step. However, the correlation coefficients between the four ADF determinations were all very high, in the range of 0.97 to 0.99. On the other hand, acid detergent lignin determined without the NDF step was, on average, 5.6 g kg 1 DM higher than ADL determined with the NDF step, but the correlation coefficients between the four ADL determinations were relatively low, in the range of 0.35 to 0.51, suggesting the need for multiple measurements. The averages of all determinations for each trait were used to develop the prediction equations for NDF, ADF, and ADL by the NIRS software of Infrasoft International (1995). The R2 for the prediction equations were 0.99, 0.99, and 0.73 for NDF, ADF, and ADL, respectively. The standard errors of cross-validation for NDF, ADF, and ADL were 6.51, 3.94, and 2.03 g kg 1 DM, respectively. Neutral detergent fiber, ADF, and ADL of all samples were predicted by the prediction equations. Hemicellulose and cellulose were calculated as NDF ADF and ADF ADL, respectively. Further, hemicellulose, cellulose, and lignin were divided by NDF to give cell wall hemicellulose (CW-hemicellulose), cell wall cellulose (CW-cellulose), and cell wall lignin (CW-lignin), respectively. Morrison (1980) reported that the Van Soest detergent system underestimates the hemicellulose and overestimates the cellulose. For this reason, CW-hemicellulose and CW-cellulose were used only for the comparison of samples. Such a comparison is valid under the assumption that the bias of the estimates on hemicellulose and cellulose is consistent across all samples in a block, which is likely to be true for comparisons made within one species at a single stage of maturity.

Statistical Analyses
The experiment was a split-plot design replicated across three locations. For dry matter yield and all cell wall traits, each subplot had repeated measurements at three harvests in each of two consecutive years. Each of these traits was fitted to the following linear model:

Yijklmn

m

Cell Wall Traits Determination
A forage sample was taken for each subplot by bulking one to two tillers from each plant in the subplot. For the spring harvest, this was done when most plants were at fully headed growth stage. For the summer and fall harvests, it was done when at least 20% of the plants were heading. The samples were air-dried at 55 C, ground through a 1-mm screen in a Wiley-type mill, and reground through a 1-mm screen in a cyclone mill. All 600 samples were scanned by a NIRS (nearinfrared reflectance spectroscopy) monochromator (Model 6500, NIRSystems/Perstorp Corp., Silver Spring, MD). A calibration set of 103 samples was selected by means of the NIRS software for the monochromator (Infrasoft International, 1995). Neutral detergent fiber, ADF, and ADL of the calibra-

Li Bij dk (Ld)ik Wijk cl (Lc)il (dc)kl (Ldc)ikl Eijkl ym (Ly)im (dy)km (Ldy)ikm (cy)lm (dcy)klm hn (Lh)in (dh)kn (Ldh)ikn (ch)ln (dch)kln (yh)mn (Lyh)imn (dyh)kmn (cyh)lmn [1] (dcyh)klmn Rijklmn

where Yijklmn is the response variable; m is the intercept, Li is the effect of location, Bij is the effect of block nested within location, dk is the effect of disease potential, cl is the effect of cycle, ym is the effect of year, hn is the effect of harvest (spring, summer, or fall); Wijk is the whole-plot error, Eijkl is the subplot error, Rijklmn is the error for repeated measurements within subplots; and the rest are interaction effects between factors for which the main effects have been previously defined. Uppercase and lowercase letters represent random and fixed effects, respectively. Interactions of fixed and random effects were treated as random. The residual maximum likelihood method was employed to fit the model for all traits by PROC MIXED of SAS (SAS Institute Inc., 1997). All interactions

HAN ET AL.: FIBER CONCENTRATION AND LEAF DISEASE IN SMOOTH BROMEGRASS

33

of fixed effects were included in the model. Some interactions involving random effects were excluded from the model because they had little or no effect on residual log likelihood. Several covariance structures for Rijklmn were entertained before the variance components structure was chosen on the basis of convergence and model fitting criteria. For disease severity and heading index, there were no measurements for the spring of 1995. Thus, repeated measurements for each subplot did not have a balanced combination of harvests and years. To avoid an unbalanced model, repeated measurements in the two harvests in 1995 and the three harvests in 1996 were treated as levels of one factor termed assessment, instead of as combinations of years and harvests. The linear model is

Yijklm

m

Bij dk (Ld)ik Wijk cl (Lc)il (dc)kl (Ldc)ikl Eijkl am (La)im (da)km (Lda)ikm (ca)lm (dca)klm Rijklm [2]

Li

where Yijklm is the response variable; m, Li, Bij, dk, cl, Wijk, and Eijkl are defined as for Model [1]; am is the effect of assessment; Rijklm is the error for repeated measurements at different assessments; and the rest are interaction effects. Model fitting approach was similar to that for Model [1]. For the calculation of genetic gain per cycle averaged over all cycles (C-3 to C 1), we assigned C 1 a cycle number 2.6 (rounded up from 2.595) to account for the fact that C 1 was selected by a higher selection intensity than C-1, C-2, and C-3 and was selected by biparental mass selection instead of uniparental mass selection. Thus the cycle numbers for C-3 to C 1 were -3, -2, -1, 0, and 2.6, as if C 1 had been selected by the scheme that was used in selection for low NDF. These cycle numbers were derived under the assumption of randommating equilibrium at each cycle and constant phenotypic and additive variances ( 2 and 2 ) throughout the recurrent selecP A tion. The derivation was in six steps: (i) calculate the selection intensity (i ) from the proportion of plants selected for each cycle (Burrows, 1972); (ii) compute the expected genetic gain ( g) over the previous cycle for cycles C-1, C-2, and C-3 using formula g (1? 2) i 2 / P, or for cycle C 1, using g i 2 / A A P (note that these formulas are the same for both diploid [Empig et al., 1981] and autotetraploid [Rowe and Hill, 1984]); (iii) set the genotypic mean of C0 at an arbitrary value, say 0; (iv) add cumulatively the expected genetic gains to obtain

the expected genotypic mean of each cycle; (v) divide the expected genotypic mean of each cycle by 2 / P to cancel A 2 A/ P from all terms; (vi) multiply the expected genotypic mean of each cycle by a common constant so that cycle C-3 has a value of -3. The assumption underlying the above derivation implied that the response to the recurrent selection was linear. A linear contrast was constructed from these cycle numbers to test the linear response to the recurrent selection. The contrast coefficients were the cycle numbers subtracted by their average. Inbreeding coefficients were calculated for all cycles assuming that the inheritance was disomic, that every plant in a population had the same chance to be pollinated by any plant in the population, and that the plants of C0 had an inbreeding coefficient of zero and were unrelated to each other (Han and Casler, 1999). The inbreeding coefficients calculated under the assumptions are (from C-3 to C 1): 0.0107, 0.0061, 0.0015, 0.0000, 0.0500. The pattern of inheritance in smooth bromegrass can be disomic or tetrasomic (Armstrong, 1981). If the inheritance were tetrasomic, we suspect that the inbreeding coefficient would be lower for all selected cycles. A linear contrast was constructed from the inbreeding coefficients by subtracting their average from each of them. This contrast tests whether cycles are linear for a trait with respect to their inbreeding coefficients.

RESULTS AND DISCUSSION Differences between the Two Disease Potentials
The differences between the high and low disease potentials in NDF, ADF, ADL, CW-hemicellulose, CW-cellulose, and CW-lignin depended on harvests and/or year and harvest combinations, as suggested by the significant (P 0.05) potential harvest interactions and/or potential year harvest interactions (Table 1). Similarly, the differences between the high and low disease potentials in disease severity and heading index depended on year and harvest combinations, or assessments, as suggested by the significant (P 0.05) potential assessment interactions (Table 2). Of the five assessments for disease severity, only the fall of 1995 and the fall of 1996 differed (P 0.05) in disease severity between the high and low disease

Table 1. P-values of F-tests for all fixed effects in an experiment on disease potentials (high vs. low) and cycles from divergent selection for neutral detergent fiber (NDF) in smooth bromegrass. Traits presented in this table are NDF, acid detergent fiber (ADF) and acid detergent lignin (ADL) on a dry-matter basis; hemicellulose, cellulose and lignin on a cell-wall (NDF) basis; and dry matter yield.
Traits on a dry-matter basis Source of variation? Potential Cycle Potential Cycle Year Potential Year Cycle Year Potential Cycle Year Harvest Potential Harvest Cycle Harvest Potential Cycle Harvest Year Harvest Potential Year Harvest Cycle Year Harvest Potential Cycle Year Harvest df 1 4 4 1 1 4 4 2 2 8 8 2 2 8 8 NDF 0.546 0.001*** 0.879 0.390 0.690 0.044* 0.960 0.001*** 0.003** 0.001*** 0.254 0.033* 0.004** 0.647 0.966 ADF 0.976 0.001*** 0.753 0.720 0.793 0.109 0.954 0.002** 0.009** 0.001*** 0.476 0.061 0.042* 0.760 0.977 ADL 0.102 0.001*** 0.837 0.007** 0.127 0.097 0.999 0.020* 0.008** 0.001*** 0.506 0.048* 0.024* 0.930 0.905 Traits on a cell-wall basis Hemicellulose 0.368 0.001*** 0.484 0.350 0.831 0.460 0.951 0.014* 0.001*** 0.001*** 0.670 0.410 0.004** 0.475 0.983 Cellulose 0.178 0.001*** 0.490 0.338 0.906 0.693 0.931 0.012* 0.001*** 0.001*** 0.624 0.374 0.005** 0.438 0.994 Lignin 0.030* 0.001*** 0.526 0.001** 0.277 0.140 0.941 0.009** 0.011* 0.001*** 0.901 0.062 0.462 0.875 0.695 Dry matter yield 0.461 0.006** 0.277 0.071 0.351 0.811 0.864 0.081 0.043* 0.001*** 0.950 0.040* 0.971 0.628 0.512

* P 0.05; **P 0.01; ***P 0.001. ? The experiment was conducted at three Wisconsin locations with four replicates at each location. Data were collected for two years (1995 and 1996) and three harvests (spring, summer and fall) each year.

34

CROP SCIENCE, VOL. 41, JANUARY–FEBRUARY 2001

Table 2. P-values of F-tests for all fixed effects in an experiment on disease potentials (high vs. low) and cycles from divergent selection for neutral detergent fiber in smooth bromegrass. Traits presented are disease severity and heading index.
Source of variation? Potential Cycle Potential Cycle Assessment Potential Assessment Cycle Assessment Potential Cycle Assessment df 1 4 4 4 4 16 16 Disease severity 0.001*** 0.088 0.931 0.016* 0.001*** 0.001*** 0.995 Heading index 0.070 0.001*** 0.932 0.005** 0.001*** 0.003** 0.632

* P 0.05; **P 0.01; ***P 0.001. ? The experiment was conducted at three Wisconsin locations with four replicates at each location. Data were collected for five assessments (two harvests in 1995 and three harvests in 1996).

potentials (Table 3). In both cases, the high disease potential was higher than the low disease potential in disease severity. For the fall of 1995, the high vs. low disease potential was 1.50 vs. 1.20 (P 0.001) in disease severity on the log scale. For the fall of 1996, 1.34 vs. 1.18 (P 0.01). Notably, the differences between the two disease potentials for different year and harvest combinations in NDF, ADL, and CW-lignin paralleled those in disease severity (Table 3), i.e., for year and harvest combinations for which we observed a significant difference in disease severity between the high and low disease poten-

tials (fall 1995 and fall 1996), we also observed significant differences in NDF, ADL, and CW-lignin between the two disease potentials, with the high disease potential being higher in these traits than the low disease potential (P 0.01, Table 3). Even though the high disease potential was higher in heading index than the low disease potential for the fall of 1995 and the fall of 1996 (P 0.05, Table 3), it does not appear that this was the reason for the differences between the two disease potentials in NDF, ADL, and CW-lignin, for differences of a similar scale between the two disease potentials in heading index in the spring of 1996 and the summer of 1996 were not associated with differences in NDF, ADL, or CW-lignin (Table 3). There are two possible explanations for the result that high disease severity was associated with high ADL and high CW-lignin. First, our measurements of acid detergent lignin and CW-lignin included chitin from fungal cell walls, and there was a larger amount of pathogenic fungi in the plants at the higher disease severity. Second, the diseased tissues had additional lignification in the cell walls. This implies that lignification was involved in plant response to the disease, but not necessarily in plant resistance response to the disease. The result on NDF disagrees with Summers and McClellan (1975), who found that fungicide-protected alfalfa plots were similar to unprotected plots for crude fiber. The association between high disease severity and high NDF, ADL,

Table 3. Means of low and high disease potentials (LowDP vs. HighDP) for nine traits at each year and harvest combination in an experiment on disease potentials and cycles from divergent selection for neutral detergent fiber (NDF) in smooth bromegrass.?
NDF Year 1995 1995 1995 1996 1996 1996 Year 1995 1995 1995 1996 1996 1996 Year 1995 1995 1995 1996 1996 1996 Harvest Spring Summer Fall Spring Summer Fall Harvest Spring Summer Fall Spring Summer Fall Harvest Spring Summer Fall Spring Summer Fall LowDP 590* 564 453 612 558 413 Hemicellulose LowDP 465 499 514 465 482 516 HighDP LowDP g kg
1

ADF HighDP 572** 567 474** 605 556 431** LowDP g kg 319 292 230 332 295 207
1

ADL HighDP LowDP 30.9 34.9 28.4 30.6 30.8 24.4 Lignin LowDP 52.3 61.9 62.7 50.0 55.3 59.2 HighDP HighDP 29.3** 35.5 30.8** 30.3 31.3 26.7***

DM 310* 295 235 328 294 213 Cellulose HighDP

465 497 523* 466 483 519 Dry matter yield HighDP g plant
1

488 456 443 493 473 440

NDF 490 457 430** 492 471 433* Disease severity? HighDP – 1.09 1.50*** 1.12 1.23 1.34**

51.2 62.6 65.1** 50.0 56.3 61.9*** Heading index? HighDP – 1.35 1.02* 1.39* 0.80* 0.97*

LowDP 132 186 135 279 282 153

LowDP – 1.13 1.20 1.12 1.18 1.18

LowDP – 1.30 0.93 1.50 0.68 0.87

105 175 118 272 290 154

* P 0.05; **P 0.01; ***P 0.001. From comparison of LowDP with HighDP. ? The experiment was conducted at three Wisconsin locations with four replicates at each location. Data were collected for two years (1995 and 1996) and three harvests (spring, summer and fall) each year for NDF, acid detergent fiber (ADF) and acid detergent lignin (ADL) on a dry-matter basis; hemicellulose, cellulose and lignin on a cell-wall (NDF) basis; and dry matter yield; and for two harvests (summer and fall) in 1995 and three harvests in 1996 for disease severity and heading index. ? Measured as scores. The score for disease severity for a plant is 0–2 after log transformation of the scale of 0–10 where 0 no diseased leaf tissues and 10 leaf tissues completely diseased. The score for heading index for a plant is 0–2 according to the number of fully emerged panicles prior to harvest: if none, score 0; if 1 to 5, score 1; if more than 5, score 2.

HAN ET AL.: FIBER CONCENTRATION AND LEAF DISEASE IN SMOOTH BROMEGRASS

35

and CW-lignin suggests that diseases could reduce forage nutritional quality. None of the significant differences in disease severity between the two disease potentials was associated with a significant difference in dry matter yield (Table 3). This result partially agrees with Summers and McClellan (1975), who found that plots protected by one fungicide yielded more than the unprotected plots while plots protected by another two fungicides were similar in yield to the unprotected plots. The result disagrees with Wilson et al. (1991), who found in pearl millet that fungicide-protected plants had higher dry matter yield if the cultivars were susceptible to the diseases. The interaction between disease potential and selection cycle was not significant for any trait measured, indicating that the relative differences among cycles in these traits were not affected by disease potentials, i.e., the responses in these traits to divergent selection for NDF were not affected by disease severities.

Direct Response to Selection for NDF
Selection for NDF was effective in both high- and low-NDF directions, as indicated by the overall linear response to selection (P 0.001), the linear response in the low-NDF direction (P 0.001), and the difference (P 0.001) between the high-NDF cycle (C 1) and the base population (C0) (Table 4). Averaged over the two disease potentials and all harvests in 2 yr, the third cycle from selection for low NDF (C-3) was 1.9% lower than the base population C0 (525.3 vs. 535.5 g kg 1 DM, P 0.001), and the first cycle from the selection for high NDF (C 1) was 2.1% higher than the base population C0 (546.6 vs. 535.5 g kg 1 DM, P 0.001). Even though cycle year interaction was significant (P 0.044), the relative NDF levels of cycles were fairly consistent over the 2 yr. Cycle harvest interaction, however, was statistically significant (P 0.001) as well as large in magnitude.

The linear response in NDF to selection for NDF was more noticeable in the spring than in the summer or the fall (Fig. 1 A), even though the linear response in NDF was significant for all three seasons (Table 4). In the spring, C-3 was 3.3% lower than C0 (580.3 vs. 600.0 g 0.001), and C 1 was 2.3% higher than kg 1 DM, P 0.001). The larger C0 (613.8 vs. 600.0 g kg 1 DM, P linear response in the spring could be due to three reasons: (i) the selection for NDF was based on forage samples in the spring, thus response in the spring was a direct response while response in the summer or the fall was a correlated response; (ii) genetic correlation between NDF of the spring harvest and NDF of the summer or fall harvest was less than 1; or (iii) there might be more measurable genetic variation in NDF in the spring than in the summer or the fall. Previously, Surprenant et al. (1988) in reed canarygrass and Carpenter and Casler (1990) in smooth bromegrass reported the success of divergent selection for NDF; however, both studies went through only one cycle of selection in the low-NDF direction. Casler (1999) reported that five out of six methods reduced NDF in smooth bromegrass. Uniparental mass selection for low NDF at heading stage, with NDF determined by wet-laboratory method, was the selection method which did not produce a significant response in that study. The low-NDF selection cycles developed by this selection method were evaluated in this study with many more replications, therefore much greater power. We demonstrated that this method was also successful in reducing NDF. Thus, animal intake potential of forages can be improved through a number of recurrent selection methods (Casler, 1999).

Correlated Responses in Other Cell Wall Traits
The correlated response in ADF to selection for NDF almost mirrored the response in NDF (Table 4; Fig. 1 A,

Table 4. Change over selection cycles (C 3, C 2, C 1, C0, C 1) for neutral detergent fiber (NDF), acid detergent fiber (ADF) and acid detergent lignin (ADL) on a dry-matter basis; hemicellulose, cellulose and acid detergent lignin on a cell-wall (NDF) basis; and dry matter yield. The significance indicators accompanying the values are from F-tests of the corresponding contrasts testing linear responses to selection for NDF (all but last row) or to inbreeding (last row).
Measurements of selection response? Change per cycle? All cycles Low-NDF direction High-NDF direction All cycles—spring All cycles—summer All cycles—fall Gain per 0.01 increase in inbreeding coefficient All cycles§ Traits on a dry-matter basis NDF ADF g kg 3.98*** 3.46*** 11.10*** 5.94*** 3.61*** 2.39***
1

Traits on a cell-wall basis Hemicellulose Cellulose g kg
1

ADL

Lignin

Dry matter yield g plant
1

DM 0.125*** 0.217*** 0.033 0.356*** 0.038 0.020 1.90*** 2.68*** 2.75* 2.70*** 2.06*** 0.95**

NDF 0.194*** 0.045 1.256*** 0.105 0.308*** 0.380***

3.00*** 3.22*** 6.76*** 4.66*** 2.84*** 1.49***

1.82*** 2.62*** 2.65* 2.37*** 2.00*** 1.10***

0.98 6.87*** 14.08* 4.00** 3.51** 2.44

3.26***

2.18***

0.056

1.11***

1.04***

0.250***

1.81

* P 0.05; **P 0.01; ***P 0.001. ? Results were from an evaluation of smooth bromegrass cycles from divergent selection for NDF at two disease potentials (high vs. low). The experiment was conducted at three Wisconsin locations with four replicates at each location. Data were collected for two years (1995 and 1996) and three harvests (spring, summer and fall) each year. ? All cycles and low-NDF direction: response is measured as the linear regression of cycle mean on cycle number ( 3, 2, 1, 0, 2.6). The high-NDF cycle (C 1) is given a cycle number 2.6 so that it has the same amount of expected genetic gain per cycle as the low-NDF cycles. High-NDF direction: response is measured as the mean of C 1 minus the mean of base population C0. § The inbreeding coefficient is assumed to be 0 for the base population C0, and estimated to be 0.0107, 0.0061, 0.0015, and 0.0500 for cycles C 3, C 2, C 1, and C 1, respectively.

36

CROP SCIENCE, VOL. 41, JANUARY–FEBRUARY 2001

B). On average, C-3 was 3.4% lower than C0 in ADF 0.001) and C 1 was (272.2 vs. 281.9 g kg 1 DM, P 2.4% higher than C0 in ADF (288.7 vs. 281.9 g kg 1 DM, P 0.001). The relative ADF levels of cycles were consistent over the 2 yr, as indicated by the nonsignificant cycle year interaction. Similar to the response in NDF, the response in ADF was more appreciable in the spring than in the summer or the fall, but linear response to selection was significant for all three seasons (Table 4). The correlated response in ADL was only evident in the spring harvest (Table 4; Fig. 1 C): C-3 was 5.5% lower than C0 (29.31 vs. 31.03 g kg 1 DM, P 0.001) although C 1 was similar to C0 (31.29 vs. 31.03 g kg 1 DM, P 0.05). The linear response in ADL was not

significant for the summer or the fall harvests, possibly because selection for NDF was based on forage samples in the spring. The correlated response in CW-hemicellulose was in the opposite direction of NDF (Table 4, Fig. 2 A) while the correlated response in CW-cellulose was in the same direction of NDF (Table 4, Fig. 2 B). In selection for low NDF, CW-hemicellulose increased 2.7 g kg 1 NDF per cycle (P 0.01) while CW-cellulose decreased by 2.6 g kg 1 NDF per cycle (P 0.01). In selection for high NDF, cycle C 1 was 2.8 g kg 1 NDF lower (P 0.05) in CW-hemicellulose and 2.7 g kg 1 NDF higher (P 0.05) in CW-cellulose than the base population C0. The pattern of responses in CW-hemicellulose and CW-cellulose to selection for NDF was consistent across all harvests (Table 4, Fig. 2 A, B). For CW-lignin, there was no correlated response in the spring harvest (Table 4). In the summer and fall harvests, CW-lignin responded in the opposite direction of NDF, increasing 0.01), 0.3 and 0.4 g kg 1 NDF per cycle (both P respectively, as NDF was reduced (Fig. 2 C). The parallel responses in ADF and NDF to selection for NDF were reported by Surprenant et al. (1988) and Carpenter and Casler (1990). The latter also reported a nonsignificant difference in ADL between synthetics selected for low and high NDF. The results from these previous studies and from this study suggest that the genetic correlation between NDF and ADF is high, which is probably a reflection of the fact that ADF constitutes a large portion of NDF, and that the genetic correlation between NDF and ADL is possibly low. There were no literature reports, to our knowledge, on the correlated responses in cell wall composition due to selection for NDF, as evaluation of populations from selection for NDF generally did not include cell-wall based traits (Surprenant et al., 1988; Carpenter and Casler, 1990; Buendgen et al., 1990; Casler, 1999). This pattern of a reduction in NDF leading to an increase in CW-hemicellulose and a decrease in CW-cellulose suggests that in the base population of this experiment (WB-RP1), NDF concentration is negatively correlated with CW-hemicellulose and positively correlated with CW-cellulose. Even though this pattern may or may not hold for all populations in all forage species, the composition of NDF should be investigated following selection for NDF concentration, because changes in NDF composition may affect NDF digestibility.

Correlated Response in Dry Matter Yield and Inbreeding Depression
Linear response in dry matter yield to selection for NDF was significant (P 0.001) for the low-NDF direction when averaged over all harvests in the 2 yr (Table 4). Dry matter yield was lower for cycles with lower NDF, dropping from 200 g plant 1 in C0 to 180 g plant 1 in C-3 (P 0.001); however, if C 1 is included, the positive linear response in yield to selection for NDF cannot be detected. From Fig. 1 D, it is obvious that C 1 did not follow the trend we found in the low-NDF direction. We have already described that the linear response in NDF extended to both the low-NDF and the high-NDF directions. If we recall that C 1 had the

Fig. 1. Means of smooth bromegrass selection cycles in spring, summer and fall harvests averaged over 2 yr. The traits are neutral detergent fiber (NDF), acid detergent fiber (ADF), and acid detergent lignin (ADL) on a dry-matter basis and dry matter yield. The horizontal axes (cycle) are scaled by the expected genetic gains of recurrent selection for NDF. The least significant difference (LSD) applies to comparisons between cycles within harvests.

HAN ET AL.: FIBER CONCENTRATION AND LEAF DISEASE IN SMOOTH BROMEGRASS

37

highest inbreeding coefficient of all cycles, the most reasonable explanation to low yield in C 1 is inbreeding depression. However, the contrast testing the linear trend of cycles with respect to inbreeding coefficients was not significant (P 0.072) even though the overall test for differences among cycles was significant (P 0.006), suggesting that inbreeding alone cannot explain all the variation in yield of cycles. In the low-NDF direction, a likely explanation is that inbreeding depression and genetic correlation between NDF and yield acted synergistically to cause the yield to go down. In the high-NDF direction, the explanation is that inbreeding depression and genetic correlation between NDF and yield were counteractive. In reed canarygrass, Surprenant et al. (1988) found that the low-NDF groups had lower yield per plant than the high-NDF groups in the spring, but did not differ from the high-NDF groups in the summer or the fall. In smooth bromegrass, Carpenter and Casler (1990) did not find a significant difference in sward dry matter yield between low- and high-NDF synthetics. Both Surprenant et al. (1988) and Carpenter and Casler (1990) compared low- and high-NDF populations at similar inbreeding levels, thus their results were not affected by inbreeding. From this study, it appears that both inbreeding and genetic correlation between NDF and dry matter yield contributed to the differences among cycles from recurrent selection for NDF, the trend being that either reduced NDF or increased inbreeding could cause reduced yield.

Correlated Response in Disease Resistance
There was no evidence that selection for low NDF modified disease reaction, as the linear response in the

low-NDF direction was nonsignificant (Table 5). The disease severities of C-3 and C0 were not significantly different (Fig. 3 A, B) for any of the five assessments. Plants of C 1 expressed severe symptoms in the summer and the fall of 1995, causing the contrast testing overall linear response to selection for NDF to be significant for those two harvests in 1995 (P 0.05 and P 0.001, respectively). We had suspected that low CWlignin might be responsible for the high disease severity of C 1 in the summer and fall harvests of 1995 because C 1 had significantly lower (P 0.05) CW-lignin than all other cycles for those two assessments (data not shown). Yet C 1 also had significantly lower (P 0.05) CW-lignin than all other cycles for the summer and fall harvests of 1996 (data not shown), for which we did not observe a significantly higher disease severity on C 1. Interestingly, C 1 happened to be the cycle with the highest inbreeding coefficient. Breese and Davies (1970) found higher susceptibility to crown rust infection for the families with higher water soluble carbohydrate (WSC) in perennial ryegrass. If WSC can be interpreted roughly as the complement of NDF, then the results from the two harvests of 1995 are just the opposite of Breese and Davies (1970). The results for the two harvests of 1995 are in agreement with Karn et al. (1989), who found that intermediate wheatgrass strains resistant to foliar diseases had lower NDF than susceptible strains when diseases were present. The results for the three harvests in 1996 supported Oberheim et al. (1987), who reported in orchardgrass that three cycles of selection for resistance to purple leaf spot did not change NDF, although very little purple leaf spot occurred in their evaluation. Overall, there was not strong evidence from this study to suggest that

Fig. 2. Means of smooth bromegrass selection cycles in spring, summer and fall harvests averaged over 2 yr. The traits are hemicellulose, cellulose, and lignin on the basis of cell wall or neutral detergent fiber (NDF). The horizontal axes (cycle) are scaled by the expected genetic gains of recurrent selection for NDF. The least significant difference (LSD) applies to comparisons between cycles within harvests.

38

CROP SCIENCE, VOL. 41, JANUARY–FEBRUARY 2001

Table 5. Change over selection cycles (C 3, C 2, C 1, C0, C 1) for disease severity and heading index. The significance indicators accompanying the values are from F-tests of the corresponding contrasts testing linear responses to selection for neutral detergent fiber, NDF (all but last row) or to inbreeding (last row).
Measurements of selection response? Change per cycle§ All cycles Low-NDF direction High-NDF direction All cycles—summer ’95 All cycles—fall ’95 All cycles—spring ’96 All cycles—summer ’96 All cycles—fall ’96 Gain per 0.01 increase in inbreeding coefficient for: All cycles? Disease severity? 0.0047* 0.0002 0.0347* 0.0110** 0.0197*** 0.0029 0.0062 0.0020 Heading index? 0.0289*** 0.0083 0.1270** 0.0160 0.0301** 0.0469*** 0.0188 0.0325**

0.0055*

0.0390***

* P 0.05; **P 0.01; ***P 0.001. ? Results were from an evaluation of smooth bromegrass cycles from divergent selection for NDF at two disease potentials (high vs. low). The experiment was conducted at three Wisconsin locations with four replicates at each location. Data were collected for two harvests (summer and fall) in 1995 and three harvests (spring, summer and fall) in 1996. ? Units are change in score per cycle or per 0.01 increase in inbreeding coefficient. The score for disease severity for a plant is 0–2 after log transformation of the scale of 0–10 where 0 no diseased leaf tissues and 10 leaf tissues completely diseased. The score for heading index for a plant is 0–2 according to the number of fully emerged panicles prior to harvest: if none, score 0; if 1 to 5, score 1; if more than 5, score 2. § All cycles and low-NDF direction: response is measured as the linear regression of cycle mean on cycle number ( 3, 2, 1, 0, 2.6). The high-NDF cycle (C 1) is given a cycle number 2.6 so that it has the same amount of expected genetic gain per cycle as the low-NDF cycles. High-NDF direction: response is measured as the mean of C 1 minus the mean of base population C0. ? The inbreeding coefficient is assumed to be 0 for the base population C0, and estimated to be 0.0107, 0.0061, 0.0015, and 0.0500 for cycles C 3, C 2, C 1, and C 1, respectively.

low NDF and disease susceptibility are necessarily correlated.

Correlated Response in Heading Index
Heading index of cycles for the spring harvest of 1996 (Fig. 3 D) suggested that selection for low NDF resulted in low heading index. The contrast testing linear response in heading index to selection for NDF indicated the same (Table 5, P 0.001). Surprenant et al. (1988) attributed the change in NDF at reproductive stage (spring harvest) to change in growth stage while Carpenter and Casler (1990) found no variation for maturity in their base population of selection. The trend of heading index for the summer and fall harvests was quite different from that for the spring harvest. It appears that the association between lower NDF and lower heading index held for C 1, C0, and C-1 while the higher inbreeding coefficient of C-2 and C-3 was associated with more heads (Fig. 3 C, D). Smooth bromegrass has little genetic variation for timing of reproductive maturity in the spring, probably due to its photoperiod sensitivity. Carpenter and Casler (1990) found no variation for maturity in their base population of selection. However, the result of this study, which used a different base population from Carpenter and Casler (1990), suggests that there is some

Fig. 3. Means of smooth bromegrass selection cycles for each assessment of disease severity and heading index. The scale for disease severity is 0 to 2 after log transformation of the scale of 0 to 10 where 0 no diseased leaf tissues and 10 leaf tissues completely diseased. The scale for heading index is 0 to 2 according to the number of fully emerged panicles prior to harvest: if none, score 0; if 1 to 5, score 1; if more than 5, score 2. The horizontal axes (cycle) are scaled by the expected genetic gains of recurrent selection for NDF. The least significant difference (LSD) applies to comparisons between cycles within harvests.

variation in either growth stage or number of heads per se in smooth bromegrass, and selection for low NDF could lead to either late maturity or reduced number of heads for the spring harvest. Given that forage plants at a more advanced growth stage are usually higher in NDF, a change in growth stage possibly contributed to the change in NDF among selection cycles. We would not attribute all the change in NDF in the spring harvest to change in growth stage because the “match” between the curve of heading index over cycles and the curve (line) of NDF over cycles does not warrant it. In reed canarygrass, Surprenant et al. (1988) did attribute the change in NDF for spring harvest to a change in growth stage. Further study is needed to determine whether

HAN ET AL.: FIBER CONCENTRATION AND LEAF DISEASE IN SMOOTH BROMEGRASS

39

the direct response in spring harvest to recurrent selection for NDF is partially due to change in growth stage.

CONCLUSIONS
The responses to selection for NDF were not different between the high and low disease potentials in this experiment although plants at the high disease potential expressed more severe symptoms and had slightly higher CW-lignin than those at the low disease potential. Recurrent selection was effective in modifying NDF of smooth bromegrass, both in the low-NDF direction and in the high-NDF direction. Correlated response in ADF to selection for NDF usually mimicked the direct response in NDF because of the high correlation between these two traits. Correlated response in ADL to selection for NDF was present in the spring and was in the same direction as the response in NDF while CW-lignin did not show differences among selection cycles in the spring. Selection for NDF concentration changed the composition of NDF as CW-cellulose responded in the same direction of NDF and CW-hemicellulose changed in the opposite direction. Reduction in dry matter yield in the process of recurrent selection for NDF was associated with lower NDF and higher inbreeding level. Selection for low NDF, accompanied by elimination of up to 1/7 of the plants because of severe disease symptoms, did not appear to affect disease resistance in the range of NDF observed in this experiment.
ACKNOWLEDGMENTS We thank Agrecol Corp., Sun Prairie, WI, for providing land and management for the plots in Sun Prairie. We also thank Drs. J.G. Coors, R.R. Smith, and S.M. Kaeppler for comments and suggestions.

REFERENCES
Armstrong, K.C. 1981. The evolution of Bromus inermis and related species of Bromus sect. Pnigma. Bot. Jahrb. Syst. 102:427–443. Berg, C.C., and K.T. Leath. 1996. Responses of red clover cultivars to Stemphylium leaf spot. Crop Sci. 36:71–73. Breese, E.L., and W.E. Davies. 1970. Selection for factors affecting nutritive value. p. 33–37. In Jubilee Report Welsh Plant Breeding Stn. 1919–1969. Welsh Plant Breeding Stn., Plas Gogerddan near Aberystwyth, Wales. Buendgen, M.R., J.G. Coors, A.W. Grombacher, and W.A. Russell. 1990. European corn borer resistance and cell wall composition of three maize populations. Crop Sci. 30:505–510. Burrows, P.M. 1972. Expected selection differentials for directional selection. Biometrics 28:1091–1100. Carpenter, J.A., and M.D. Casler. 1990. Divergent phenotypic selection response in smooth bromegrass for forage yield and nutritive value. Crop Sci. 30:17–22. Casler, M.D. 1992. Usefulness of the grid system in phenotypic selection for smooth bromegrass fiber concentration. Euphytica 63: 239–243. Casler, M.D. 1999. Phenotypic recurrent selection methodology for reducing fiber concentration in smooth bromegrass. Crop Sci. 39:381–390. Conrad, H.R., A.D. Pratt, and J.W. Hibbs. 1964. Regulation of feed intake in dairy cows. I. change in importance of physical and physiological factors with increasing digestibility. J. Dairy Sci. 47:54–62. Delgado, N. 1998. Relationships between cell wall components and fungal disease resistance, host range of a Puccinia coronata strain,

and inheritance of resistance to P. coronata in smooth bromegrass. Ph.D. Dissertation (Diss abstr. 9910417), University of WisconsinMadison, Madison, WI. Empig, L.T., C.O. Gardner, and W.A. Compton. 1981. Theoretical gains for different population improvement procedures. University of Nebraska College of Agriculture, The Agricultural Experiment Station. MP 26, revised. Goering, H.K., and P.J. Van Soest. 1970. Forage fiber analysis (apparatus, reagents, procedures, and some applications). USDA Agriculture Handbook. 379. U.S. Gov. Print Office, Washington, DC. Han, L.X., and M.D. Casler. 1999. Theoretical inbreeding at selectively neutral loci in uniparental mass selection and recurrent selection with polycrossing of selected plants. Crop Sci. 39:1009–1015. Infrasoft International. 1995. NIRS3 version 3.10. Routine operation, calibration development, and network system management manual. Infrasoft International, Port Matilda, PA. Karn, J.F., and J.M. Krupinsky. 1983. Chemical composition of intermediate wheatgrass affected by foliar diseases and stem smut. Phytopathology 73:1152–1155. Karn, J.F., J.M. Krupinsky, and J.D. Berdahl. 1989. Nutritive quality of foliar disease resistant and susceptible strains of intermediate wheatgrass. Crop Sci. 29:436–439. Lenssen, A.W., E.L. Sorensen, G.L. Posler, and D.L. Stuteville. 1991. Resistance to anthracnose protects forage quality of alfalfa. Crop Sci. 31:147–150. Mertens, D.R. 1987. Predicting intake and digestibility using mathematical models of ruminal function. J. Animal Sci. 64:1548–1558. Morrison, I.M. 1980. Hemicellulosic contamination of acid detergent residues and their replacement by cellulose residues in cell wall analysis. J. Sci. Food Agric. 31:639–645. Oberheim, R.L., C.C. Berg, R.T. Sherwood, and K.E. Zeiders. 1987. Yield and quality of forage from orchardgrass selected for resistance to purple leaf spot. Crop Sci. 27:673–676. Osbourn, D.F., R.A. Terry, G.E. Outen, and S.B. Cammell. 1974. The significance of a determination of cell walls as the rational basis for the nutritive evaluation of forages. p. 374–380. In V.G. Iglovikov and A.P. Movsissyants (ed.) Proc. XII Int. Grassl. Congr. Vol. III, part 1, Moscow. 11-20 June 1974. State Committee of Science and Technology, Moscow. Reid, R.L., G.A. Jung, and W.V. Thayne. 1988. Relationships between nutritive quality and fiber components of cool season and warm season forages: a retrospective study. J. Animal Sci. 66:1275–1291. Ride, J.P. 1983. Cell walls and other structural barriers in defense. p. 215–236. In J.A. Callow (ed.) Biochemical plant pathology. Wiley & Sons, Chichester, UK. Robertson, J.B., and P.J. Van Soest. 1981. The detergent system of analysis and its application to human foods. p. 123–158. In W.P.T. James and O. Theander (ed.) The analysis of dietary fiber in food. Marcel Dekker, Inc., New York. Rowe, D.E., and R.R. Hill, Jr. 1984. Theoretical improvement of autotetraploid crops: interpopulation and intrapopulation selection. USDA, Technical Bulletin No. 1689. Ruiz, T.M., E. Bernal, C.R. Staples, L.E. Sollenberger, R.N. Gallaher. 1995. Effect of dietary neutral detergent fiber concentration and forage source on performance of lactating cows. J. Dairy Sci. 78: 305–319. SAS Institute Inc. 1997. SAS/STAT Software: changes and enhancements through release 6.12. SAS Institute Inc., Cary, NC. Sherwood, R.T., and C.C. Berg. 1991. Anatomy and lignin content in relation to resistance of Dactylis glomerata to Stagonospora leaf spot. Phytopathology 81:1401–1407. Summers, C.G., and W.D. McClellan. 1975. Effect of common leafspot on yield and quality of alfalfa in the San Joaquin Valley of California. Plant Dis. Reporter 59:504–506. Surprenant, J., D.K. Barnes, R.H. Busch, and G.C. Marten. 1988. Bidirectional selection for neutral detergent fiber and yield in reed canarygrass. Can. J. Plant Sci. 68:705–712. Webb, D.H., F.W. Nutter, Jr., and D.R. Buxton. 1996. Effect of acid detergent lignin concentration in alfalfa leaves on three components of resistance to alfalfa rust. Plant Dis. 80:1184–1188. Wilson, J.P., R.N. Gates, and W.W. Hanna. 1991. Effect of rust on yield and digestibility of pearl millet forage. Phytopathology 81:233–236.



更多相关文章:
更多相关标签:

All rights reserved Powered by 甜梦文库 9512.net

copyright ©right 2010-2021。
甜梦文库内容来自网络,如有侵犯请联系客服。zhit325@126.com|网站地图