COMPARISON OF METHODOLOGIES FOR DETERMINATION
OF CONTENT AND DIGESTIBILITY OF HEMICELLULOSE
AND CELLULOSE IN TROPICAL GRASS HAYS
SANDRA L. RUSSO
A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL
OF THE UNIVERSITY OF FLORIDA IN
PARTIAL FULFILLMENT OF THE REQUIREMENTS
FOR THE DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
now its your turn...
The author would like to express her appreciation to Dr. G. 0.
Nott for serving as chair and academic advisor for the program of
study. The author is sincerely grateful to Dr. John E. Moore,
director of the research, for guidance and support, for always having
time and usually patience and for providing research facilities and
equipment. Appreciation is extended to Drs. Donald Graetz, Joseph
Conrad and 0. C. Ruelke, members of the supervisory committee, for
their time and advice. The statistical assistance of Dr. Ramon
Littell is gratefully acknowledged.
Appreciation is extended to Dr. Murray Gaskins for putting his
time and laboratory equipment at the author's disposal. Mr. John
Funk and Dr. John Moore provided invaluable assistance at the
Nutrition Lab in designing and setting up the experiments.
Thanks are due to Drs. Donald Graetz, Ray Varnell, G. O. Mott,
Ken Quesenberry, C. Y. Ward, David Buchanan, Art Hansen and Elon
Gilbert for providing financial assistance via a series of jobs,
assistantships and contracts.
The author appreciates the intellectual and non-intellectual
support of fellow graduate students and staff at the Nutrition Lab.
In particular, Carlos Chaves, Hari Hartadi and Steve Abrams laid the
groundwork for the dissertation.
The weekly Social, Agricultural and Food Scientists meetings
kept the mission of feeding the world's hungry through cooperation
as well as research at the front of the author's thoughts and is most
Finally, for the continued support and sometimes, the sacrifices,
of her husband, Allen Cook, the author is especially indebted.
TABLE OF CONTENTS
ACKNOWL E DGMETS . . . .
LIST OF TABLES . . . . .
LIST OF FIGURES . . . .
ABSTRACT . . . . . .
CHAPTER I. INTRODUCTION . .
CHAPTER II. LITERATURE REVIEW .
. . . . . . . iii
. . . . . . . v ii
. . . . . . . ix
. . . . . . . x
. . . . . . . 1
S . 3
Forage Quality . . . . . . . . .
Digestibility . . . . . . . . . .
Intake . . . . . . . . . . . .
Relationship Between Digestion and Intake . . . .
The Structure of Plant Cell Walls . . . . . .
Hemicellulose . . . . . . . . . .
Hemicellulose: Methodology . . . . . . .
Alkali extraction . . . . . . . .
Acid hydrolysis . . . . . . . . .
Hemicellulose: Structure . . . . . . .
Hemicellulose in Plants . . . . . . . .
Hemicellulose Digestion in the Ruminant . . . .
In vitro studies . . . . . . . . .
In vivo studies . . . . . . . .
CHAPTER III. CHEMICAL COMPOSITION OF TROPICAL GRASS HAYS.
Introduction . . . .
Experimental Procedures . .
Results and Discussion. . .
Neutral detergent fiber. .
Holocellulose. . . . .
Hemicellulose. . . . .
Cellulose. . . . . .
Lignin . . . . . .
Summary . . . . . .
CHAPTER IV. IN VITRO AND IN VIVO DIGESTIBILITY OF
TROPICAL GRASS HAYS . . . . . . . .
Introduction . . . . . .
Experimental Procedures . . .
Results and Discussion . . . .
In vitro digestibility . . .
In vivo digestibility . . .
Summary . . . . . .
. . . . . .52
. . . . . 54
CHAPTER V. PREDICTION OF FORAGE QUALITY FROM CHEMICAL AND
IN VITRO ANALYSES . . . . . . . . .
Introduction . . . . . .
Experimental Procedure . . . .
Results and Discussion . . . .
Intake and digestibility . .
Prediction of forage quality from
Conclusion . . . . . .
Summary . . . . . .
. . . . . .65
. . . . . 66
. . . . . 67
laboratory analyses 68
. . . . . 76
CHAPTER VI. GENERAL DISCUSSION . . . . . ... .77
APPENDIX. . . . . . . . . ... . . . .83
LITERATURE CITED. . . . . . . . . ... . 99
BIOGRAPHICAL SKETCH . . . .. .
. . . . . .108
LIST OF TABLES
1 Abbreviations used in the dissertation. . . . ... 32
2 Analysis of variance of chemical composition of
forty hays. . . . . . . . ... ..... 34
3 Analysis of variance of method differences. . . ... 35
4 Correlation coefficients (r) between variables
obtained by two methods . . . . . .... 36
5 Analysis of variance of in vitro data . . . ... .56
6 Comparison of in vitro digestion in rumen fluid and
in rumen fluid plus pepsin. . . . . . . ... 57
7 Digestion coefficients of chemical fractions in vitro
and in vivo . . . . . . . .. .. .. .. 59
8 Chemical composition of feces . . . . .... 61
9 Relationship (r2) between in vivo and in vitro
digestion coefficients for six hays . . . .... 63
10 Relationship (r2) between in vivo values and laboratory
analyses, obtained with thirty-nine different hays. . 69
11 Relationship (r2) between in vivo values and in vitro
analyses, obtained with thirty-nine different hays. 70
12 Coefficient of determination (R2), probability (P)
and standard error of the estimate (s vx) for the
repression of in vivo intake values oh several
independent variables . . . . . . . ... 72
13 Coefficient of determination (R2), probability (P)
and standard error of the estimate (s ) for the
regression of in vivo digestibility values on
several independent variables . . . . . .. 73
APPENDIX TABLE PAGE
14 Chemical composition of individual hays . . .. 83
15 In vitro digestion of individual hays . . . .. 85
16 Chemical composition and in vitro digestion of
individual hays . . . . . . . . . 87
17 In vivo intake and digestibility data . . . .. 89
18 Means of neutral detergent fiber by age and hay . 91
19 Means of holocellulose by age and hay . . . .. 92
20 Means of hemicellulose by age and hay . . . .. 93
21 Means of cellulose by age and hay . . . . .. 94
22 Means of lignin by age and hay. . . . . . 95
23 Analyses of variance of feces samples and in vivo
digestibility . . . . . . . .... . 96
24 Least squares means of in vivo intake per unit body
weight (g/kg body weight) . . . . . . .. 97
25 Least squares means of in vivo digestibility data . 98
LIST OF FIGURES
1 Change in NDF content by weeks and hay. . . ... 38
2 Change in NDF(H) content by weeks and hay .... .39
3 Change in holocellulose content by weeks and hay. 41
4 Change in alkali-soluble hemicellulose by weeks
and hay . . . . . . . . .. . 43
5 Change in Van Soest hemicellulose by weeks and hay. 44
6 Change in alkali-insoluble cellulose by weeks and
hay . . . . . . . . . . . . 47
7 Change in Van Soest cellulose by weeks and hay. .. 48
8 Change in chlorite lignin by weeks and hay. ... 49
9 Change in permanganate lignin by weeks and hay. .. 50
Abstract of Dissertation Presented to the Graduate Council
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy
COMPARISON OF METHODOLOGIES FOR DETERMINATION
OF CONTENT AND DIGESTIBILITY OF HEMICELLULOSE
AND CELLULOSE IN TROPICAL GRASS HAYS
SANDRA L. RUSSO
Chairman: Dr. G. O. Mott
Major Department: Agronomy
Experiments were conducted to: 1) fractionate cell walls of
tropical grasses into: holocellulose, cellulose and hemicellulose;
2) determine in vitro and in vivo digestibility of cell wall
components; and 3) study the relationship between these components
and quality of forages, as measured by feeding hays to sheep.
One cultivar of Cynodon dactylon (Coastcross-l), two experimental
lines of Digitaria species (X46-2 and X124-4) and two cultivars
of Paspalum notatum (Argentine and Paraguay) cut at four ages (2,
4, 6 and 8 weeks of regrowth) were used in the experiments.
Samples of each hay were analyzed for organic matter (OM),
crude protein (CP) and neutral detergent fiber (NDF). NDF frac-
tions (cellulose, hemicellulose and lignin) were determined by
two methods: 1) conventional Van Soest detergent analyses
(CELLV, HEMV, PLIG respectively): and 2) classical fractionation
using NaCO.2 and alkali (CELL, HC, LIG respectively). There were
significant differences between fractions obtained by the two
methods. HC remained the same or decreased slightly with maturity
but HEI.V declined. CELL and CELLV showed similar patterns,
increasing up to 6 weeks, then decreasing.
Samples were fermented in rumen fluid for 48 hr and digested
in acid-pepsin for 48 hr or fermented in rumen fluid only for 48
hr and residual NDF was fractionated as above. In vitro digesti-
bility of NDF and HC was increased by further digestion in acid-
pepsin while CELL digestibility showed minimal change. CELL diges-
tibility was generally higher than HC digestibility. Digestion
coefficients for each fraction were similar in vivo and in vitro
(r = .79 to .92).
NDF, CP and PLIG were the best single chemical analyses for
prediction of intake and digestibility, but any in vitro analysis
gave better estimates of in vivo digestibility. Two and three
variable models improved prediction of organic matter intake (OMI)
and digestible OMI.
Hemicellulose obtained by different methods was not the same.
Digestion of cell walls in acid-pepsin was due to breakdown of
hemicellulose and possibly, dissolution of lignin-hemicellulose
bonds. Further studies on chemical bonding, physical degradabi-
lity and chemical structure of the cell wall appear necessary
before being able to predict animal performance from laboratory
The world food shortage is most critical in that portion of the
world known as the tropics, which is faced with rising populations
and diminishing food supplies. The limited availability of land
suitable for arable cropping has focused attention upon the grass-
lands which occupy fifty percent of the earth's land surface. These
lands are able to provide food for people by the use of ruminant
animals to convert pasture into meat and milk.
The structural carbohydrates in pastures and forages are an
important source of energy to ruminants and, in turn, to people.
The amount and proportions of structural carbohydrates in the plant
cell wall determine the degree to which the ruminant can utilize
this energy. Changes in amounts and types of cell wall constituents
are responsible for differences in nutritive value of forages.
The cell wall polysaccharides are comprised of fibers of organ-
ized, crystalline cellulose in a matrix of hemicelluloses which are
associated physically and chemically with lignin (Jones, 1976).
Lignin influences digestibility of forages and has been considered
the primary structural inhibitor of quality in tropical grasses
(Moore and Mott, 1973). Cellulose and hemicellulose fractions are
also responsible for variations in digestibility, since they contain
the largest proportion of the indigestible substances in the cell
wall (Van Soest, 1975).
The role, structure and composition of hemicellulose(s) vis
a vis forage quality and nutritive value is not well understood.
Hemicellulose is often defined as the resultant of a certain set of
operations and it has been convenient to estimate hemicellulose
as the difference between acid detergent fiber (ADF) and neutral
detergent fiber (NDF). Acid extraction may not, however, remove
material that is hemicellulosic in nature (Bailey and Ulyatt, 1970;
Morrison, 1980). The entire hemicellulosic fraction has been
treated as a unit since no nutritionally meaningful fractionation
has been found and since hemicellulose digestibility is so closely
related to that of cellulose (Van Soest, 1975). Questions arise,
however, as to whether this grouping together of structural carbo-
hydrates is biologically rational, given the complexity of living
plants and of ruminant digestion.
The objectives of this research were: 1) to fractionate the
cell walls of tropical grasses into three components: holocellulose,
cellulose and hemicellulose; 2) to determine the in vitro and in
vivo digestibility of the three components; and 3) to study the
relationship between these components and the quality of the
forages in vivo.
The best measure of the nutritive value of a forage is animal
performance. Moore and Mott (1973) have defined forage quality as
output per animal, being a function of voluntary intake and digesti-
bility when the forage is fed ad libitum as the sole source of feed.
Intake and digestibility may be limited by the rate and extent of
forage digestion in the rumen, which, in turn, are related to the
amount of cell wall constituents (CWC).
Under the best of conditions, all forages would eventually be
tested in long-term production trials on pasture (Mott and Moore,
1970) having been preceded by a few years of agronomic and laboratory
evaluations. It is often not possible for researchers to conduct
pasture studies. Instead, trials with penned or caged animals are
used to measure voluntary intake and digestibility. Even these
animal trials may be beyond the means of some researchers who have to
rely exclusively on laboratory evaluation of forages. Fortunately,
enough data has been accumulated to show high correlations between
some laboratory methods and in vivo performance, e.g., the two stage
in vitro rumen digestion procedure (Tilley and Terry, 1963), and
digestibility. Unfortunately, a laboratory procedure that predicts
voluntary intake with acceptable accuracy has not yet been developed.
Thus, there is a continuum ranging from what would be the ideal way
to measure forage quality (animal response) to a vast array of ways
to predict forage quality in the laboratory which may or may not
be acceptable under all conditions.
A number of laboratory methods to estimate in vivo digestibility
have been devised. However, if only one method is to be used to
predict digestibility over a wide range of forages, then the method
to choose is the two-stage in vitro digestion procedure (Oh et al.,
1966; Marten and Barnes, 1980).
Digestibility is very important in determining the quality of
a forage and is more a function of plant factors than of animal
factors. It is now accepted that as a plant matures, its digestibi-
lity decreases due to increasing lignification and changes in plant
morphology. The cell contents (CC) of forages are considered to be
completely digestible while the CWC show variable digestibilities.
Riewe and Lippke (1970) found that tropical species accumulate dry
matter more rapidly which can be directly associated with an increase
in CWC. While temperate and tropical grasses have similar amounts of
CWC at early stages of growth, the amount of CWC and its composition
changes more rapidly with maturation in tropical species. This
increase in CWC may or may not decrease digestibility in tropical
grasses more than such an increase would in temperate grasses.
Minson and McLeod (1970) showed temperate species to be 12.8 percent-
age units higher in IVDIMD than tropical species; however, Russo
(1976) did not find differences of such magnitude despite higher
CWC. Tessema (1972) noted that the overall rate of decline in
digestibility was more rapid with tropical grasses than with temperate
grasses, that hemicellulose fluctuated considerably in the tropical
grasses and that tropical grasses had higher concentrations of
lignin. Moore and Mott (1973) have suggested that the apparent
lower quality of tropical grasses may be due to lower protein values,
which will depress intake, and higher fiber values, which will depress
With respect to ruminant nutrition, the detergent analysis
system of Van Soest (1967) which separates forages into CC and CWC
is useful for forage quality analysis. The CC are soluble and
almost totally available to the ruminant while the CWC are completely
insoluble, with their digestibility depending on microbial action.
The CWC appear to have the most influence on the digestibility and
intake of forages. The higher CWC in tropical grasses places in-
creased emphasis on digestibility, especially with respect to its
influence on voluntary intake.
Intake is controlled by two mechanisms: distention and meta-
bolic. In forage-fed ruminants, the distention mechanism exerts
the major effect on intake. Many other factors, of course, affect
voluntary intake and have been reviewed by Capote (1975) and Golding
Crampton et al. (1960), Conrad (1966) and others have proposed
the following concepts relating to the distention theory in ruminants:
(1) probably a specific degree of rumen load reduction is the major
determinant of hunger: (2) the rates of forage cellulose and hemicell-
ulose degradation are directly related to the rate at which the rumen
load is reduced; and (3) the time it takes for the rumen load to be
reduced (when hunger recurs) is characteristic of the specific
forage involved. Most researchers now agree thac the rumen is
where the distention control over forage is exerted. Distention
mechanism is determined by (1) rumen fill and (2) retention time
(Weston, 1966: Thornton and Minson, 1972). Van Soest's "hotel
theory" (1975) suggests that the plant cell walls, not the dry
matter of the forage, are responsible for gut fill and only when the
cell walls are degraded or move out of the rumen, is gut fill re-
duced. Retention time is affected by the rates of digestion and
passage (Jones and Bailey, 1974; Waldo et al., 1972).
Relationship Between Digestion and Intake
The detergent systems of Van Soest separate the fibrous portion
of a forage into cellulose, hemicellulose and lignin. Other clas-
sification systems exist: Minson (1976) separated the CWC into a
potentially digestible fraction and an indigestible fraction, while
work by Abrams (1980) suggests there may be at least two types of
digestible cell walls. How these cell walls are utilized by an
animal i3 determined primarily by their rate of passage through
the digestive tract.
Chemical composition and structure of the cell walls are prob-
ably the most important factors influencing rates of digestion and
passage (Akin et al., 1974; Van Soest, 1965). The chemical bonds
holding the cell walls together as well as the structure of each
cell wall component (lignin, cellulose and hemicellulose) needs to
be elucidated. Lignin, for example, may provide a physical barrier
to the breakdown of the cell walls (Dekker, 1976). The rate of
breakdown is regulated by the organizational structure of the cell
walls (Akin et al., 1974: de la Torre, 1974).
Intake and digestibility can be seen then, to be linked to-
gether with respect to forage quality. Some alternative definitions
of forage quality use the combination of intake and digestibility,
such as voluntary intake of digestible energy (Heaney, 1970). The
goal is to predict digestible nutrient intake, i.e., animal perfor-
mance, without conducting long-term pasture production trials. Few
attempts have been successful, especially with tropical grasses. As
mentioned earlier, laboratory methods currently available can predict
forage digestibility but not voluntary intake. Characterization of
the chemical structure of the plant cell wall, especially in tropical
grasses, would help our understanding of the factors affecting volun-
The Structure of Plant Cell Walls
The plant cell wall is of two types, a thin primary wall and a
thicker secondary wall. The primary wall is laid down by undifferen-
tiated cells that are still growing. The primary wall is transformed
into a secondary wall after the cell has stopped growing. The
primary cell walls of many higher plants appear to have very similar
structures (Talmadge et al., 1973) but this is not true of secondary
Every living cell, whether plant or animal, is surrounded by a
complex membrane made up of lipids and proteins called the plasma-
lemma. In almost all plant cells, the plasmalemma forms a cell wall
which is rigid or semi-rigid and is chemically distinct in that the
structurally important components are polysaccharides (Preston, 1974).
Cellulose, hemicellulose, pectic polysaccharide, structural protein
and lignin have been identified as the major components of the plant
cell wall. Some of these components, e.g. lignin, are not poly-
saccharides, nevertheless, the essential and sudden change at the
plasmalemma surface is from a basically lipo-protein to a basically
polysaccharide formation (Preston, 1974).
Usually only one of these polysaccharides is crystalline and
cellulose predominates. In primary cell walls, the crystalline
cellulose fibrils form a core surrounded by a non-crystalline region
consisting of other cellulose molecules, non-cellulosic polysaccharides
and glycoprotein (Talmadge et al., 1973). In secondary walls, the
cellulose fibrils are aggregated into ropelike structures called
The non-cellulosic polysaccharides making up the bulk of the
primary cell wall have been defined by their presence in chemical
fractions. The two major non-cellulosic fractions are the pectic
polysaccharides obtained by extracting cell walls with boiling
water, EDTA, or dilute acid (Aspinall, 1970) and the hemicelluloses
solubilized by the subsequent extraction of the same walls with
alkali (Whistler and Richards, 1970). In the secondary wall, also,
it appears that the extractable non-cellulosic wall compounds
surrounding the microfibrils are hemicellulosic (Preston, 1974).
But even further, some of the hemicelluloses cannot be removed
without degradation of the microfibrils and are therefore integrally
mixed with them (Preston, 1974). It has been suggested (Bauer et al.,
1973) that the structural function of the hemicellulosic polysaccha-
rides is to interconnect the cellulose fibrils and the pectic poly-
saccharides of the plant cell wall, and that this function is based
on the ability of the hemicellulosic polysaccharides to bind non-
covalently to cellulose and to bind covalently, through glycosidic
bonds at their reducing ends, to the pectic polysaccharides.
Hemicelluloses are often implicitly or explicitly defined as the
cell wall and intercellular polysaccharides that can be extracted by
alkali from higher land plant tissues that are, or were, lignified
(Wilkie, 1979). Hemicellulose is often extended to include certain
carbohydrates found in cereal endosperm. Jermyn (1955) wrote "it
cannot be too strongly emphasized that 'cellulose' and 'hemicellulose'
are normally determined as the resultants of certain sets of opera-
tions, rather than as chemically defined species." (p.197) Hemicellulose
has, thus, variable definitions depending on the methods or proce-
dures used to isolate it. A more liberal definition of hemicellulose
refers to all of the types of polysaccharides (found in plants)
other than celluloses, starches and fructans (Wilkie, 1979).
It is necessary to point out that the majority of literature
on hemicellulose and its relationship to forage quality and ruminant
nutrition deals with the Van Soest hemicellulose obtained by subtract-
ing ADF from NDF. Work as early as 1970 (Bailey and Ulvatt) points
out the problems inherent in this definition of hemicellulose. Dis-
cussion of methodology differences follows. To avoid confusion, Van
Soest hemicellulose will be distinguished from alkali-soluble
non-endospermic hemicellulose (as isolated by classical methodology).
Hemicellulose was first isolated by Schultze in 1891 (Wilkie,
1979) using dilute alkali to extract polysaccharides from plant
tissues. Work on hemicellulose, thereafter, although becoming more
sophisticated, continued to use alkali to separate the compound
from plant tissue. The continued search for a simpler, more rapid
procedure led, however, to a second extraction scheme using acid
hydrolysis, either on delignified plant tissue (Routley and Sullivan,
1958) or on whole plant tissue (Van Soest, 1965). Alkali extraction
and acid hydrolysis are two considerably different methods of
arriving at a presumed similar result. Closer examination of these
two general methodologies appears warranted.
In grasses, some of the hemicellulose is covalently bonded to
lignin (Morrison, 1974). Earlier, the relationship of lignin and
hemicellulose was not well understood and hemicelluloses were
extracted from lignified tissues of grasses by direct treatment
with alkali (Wilkie, 1979). However, it has now become common
practice to delignify the plant tissue with sodium chlorite and
acetic acid according to the procedure first outlined by Whistler
et al. in 1948. The delignified plant tissue, termed holocellulose,
is primarily composed of hemicellulose and cellulose. The holocellu-
lose is then treated with dilute alkali under nitrogen to extract
It may be possible to extract some of the hemicellulose from
holocellulose with hot or cold water. The alkali solutions, either
sodium or potassium hydroxide, may range in concentration from 4 to
24% w/v. Dilute alkali extracts only part of the hemicellulose.
If exhaustive extractions are required, the concentrations of alkali
can be increased from a low concentration to a higher concentration
to extract more of the hemicellulose (Wilkie, 1979). Nevertheless,
even with this exhaustive extraction, not all of the hemicellulose
will be accounted for (Gaillard, 1958).
With filtration or centrifugation, a residue and a supernatant
are obtained; the supernatant containing the hemicellulosic materials.
With some diligence, as many as three fractions can be derived from
the supernatant. Acidification of the supernatant with acetic acid
at 0-2 C will precipitate the A fraction (Gordon and Gaillard, 1976)
which is composed primarily of long chain xylans with small amounts
of arabinose and uronic acid (Bailey, 1973). The polysaccharides
remaining in solution can be precipitated by ethanol (Gordon and
Gaillard, 1976) or recovered by dialysis and freeze-drying (Bailey,
1973) to obtain the B fraction. The B fraction contains smaller
molecular weight xylans with many side chains and more complex mole-
cules made up of galactose, glucose and rhamnose (Bailey, 1973).
These are sometimes further broken down into linear B and branched
B fractions based on their separation by iodine precipitation
(Gaillard, 1965). The ethanol in this supernatant can be evaporated
and a C fraction recovered by dialysis for 5-7 days, followed by
freeze-drying. The C fraction has not been well studied but is
composed primarily of lignin, arabinose, galactose, glucose and
xylose (Gordon and Gaillard, 1976). In addition to iodine precipi-
tation, individual mono- or polysaccharides may be isolated from
these fractions by copper precipitation, gel filtration or quater-
nary ammonium salt precipitation (Bailey, 1973). The sugars thus
isolated can then be identified by any number of methods ranging
from the colorimetric (Dubois et al., 1956) to chromatographic
(paper, thin-layer, gas or high performance liquid).
It becomes apparent, then, that examination of hemicellulose
and its components can very rapidly become a major undertaking,
particularly when it is desired to elucidate exact amount or
structures of components. Work along these lines thus falls on
a few hardy souls upon whom the rest of the researchers become
dependent for answers. It is also apparent that such procedures
have little value in a situation where a large number of samples
need to be analyzed.
A fair approximation of the hemicellulosic material in a plant
can be obtained by treating depectinated plant tissue with boiling
N acid for 2 3 hours (Bailey, 1973), hot water (Routley and Sullivan,
1958) or cold 72% sulfuric acid overnight followed by 2 hours
boiling (Sullivan, 1966). The sugars in the hydrolysate may then
With the advent of Van Soest's detergent analysis scheme, it
became an easy matter to isolate and work with the cell wall and its
constituents. The use of neutral detergent on plant tissue leaves a
residue which is essentially nitrogen-free and is composed of lignin
and the structural polysaccharides, cellulose and hemicellulose. The
use of acid detergent on a forage was originally intended as a pre-
paratory procedure for lignin (Van Soest, 1975) but has commonly
been used to obtain cellulose and lignin values by using the acid
detergent lignin (ADL) procedure (Van Soest and Wine, 1968) and to
obtain a hemicellulose value by subtracting ADF from NDF.
The Van Soest procedures are widely used because they are rapid,
simple and not extremely expensive. The isolation of the cell walls
by neutral detergent is a chemically, as well as structurally, rational
procedure and is a convenient starting point for further work on cell
wall constituents. Daughtry (1976) for example, combined the use of
neutral detergent for cell wall preparation with boiling in 1 N
H2SO4 to obtain what was termed a hemicellulose hydrolysate.
It is with the other Van Soest procedures, ADF and ADL, that
problems arise. ADF minus lignin and silica, for example, is a close
estimate of cellulose (Van Soest and Wine, 1968) but has been
found to be contaminated with hemicellulosic residues (Bailey and
Ulyatt, 1970; Morrison, 1980). ADL, on the other hand, gives lignin
values quite different from lignin values obtained on the same plant
sample by other methods. Lignin, like hemicellulose, is difficult
to define and becomes defined by the procedure used to isolate it.
Many, and sometimes imaginative, adaptations have been made on the
Van Soest procedures in an attempt to get fractions that behave in a
uniform manner. However, since the cell wall fractions are not uni-
form, it is unlikely that a simple evaluation procedure will isolate
a recognizably uniform entity in all forages at all ages under all
Recalling that hemicellulose becomes defined by the procedure
used to isolate it, these two different approaches-- solubility in
alkali and hydrolysis in acid --can be seen to, in effect, act as a
check on each other. The former, by elucidating the chemistry and
structure of hemicellulose piece by piece, acts as a baseline
against which the results of the more rapid, latter methodology can
be checked for accuracy. It is with a very critical eye that infor-
mation on "hemicellulose" should be examined and compared because,
as has been seen, these hemicelluloses may or may not be representa-
tive of the hemicelluloses present in plants.
Hemicellulose was originally thought to be a precursor of cellu-
lose, hence the name hemi cellulose. While the reasoning has been
proven wrong, the name persists. Cellulose is a linear polymer of
8-1-4 linked D-glucose units; repeating unit 4-0-$-D-glucopyranosyl-D-
glucopyranose, cellobiose (Bailey, 1973). Hemicellulose is composed
of two major groups of polysaccharides: pentosans and non-cellulosic
Bailey (1973) reviewed the literature on chemical composition
of hemicellulose from which the following information is taken. The
pentosans are primarily based on xylose. They appear to be linear
chains of 8-1-4 linked D-xylopyranose units; repeating unit 4-0-8-D-
xylanopyranosyl-D xylopyranose, xylobiose. To these chains are
attached L-arabinose, D-glucuronic acid or its 4-0-methyl ether,
D-galactose and possibly D-glucose. These sugars are usually single
side chains but may be connecting points for further branching, par-
ticularly via L-arabinose. Here also, is where the linkages with
lignin polymers occur via ferulic acid, other phenolic acids or
their esters (Wilkie, 1979; Morrison, 1980).
The hexosans in forage grasses are primarily mannans, either
glucomannans or galactoglucomannans. These are linear chains of
8-1-4 linked D-glucose and D-mannose. The glucomannans are usually
free of other sugars, the galactoglucomannans have a-linked galactose
side chains. There has also been some evidence (Fraser and Wilkie,
1971; Wilkie and Woo, 1976) of the presence of linear chains of
s-linked D-glucose units which are not cellulosic. The glucans
have both 8-1-3 and 8-1-4 linkages. The ratio of the 1-3 to 1-4
linkages in the glucans in any one tissue falls as the plant matures
(Buchala et al., 1972). A more thorough discussion of the structural
features in hemicelluloses can be found in a review chapter by
Albersheim and coworkers (Bauer et al., 1973; Keegstra et al.,
1973; Talmadge et al., 1973) working with dicotyledonous plants
have developed a cell wall model. The cellulose fibers are linked
together by other polysaccharides, in particular, hemicellulosic
xyloglucans which completely cover the cellulose fibrils (Albersheim,
1978). Their preliminary work on grasses suggests that a model of
the cell wall for monocotyledonous plants will be similar to that
for dicots. The dominant hemicellulosic polysaccharide in primary
cell walls of monocots is an arabinoxylan rather than a xyloglucan
(Albersheim, 1978). A modified cell wall model (Bailey et al., 1976)
proposes that much of the hemicellulose and all of the pectin is not
linked to other CWC and only some of the hemicellulose is bonded to
glycoprotein and cellulose. They also point out that cellulose is
not the only crystalline structure and that hemicellulosic xylan is
capable of forming a crystalline structure.
Hemicellulose in Plants
Hemicelluloses comprise from 10 30% of the total dry matter
in temperate and tropical grasses (Bailey, 1973) and from 30 60%
of the cell wall in tropical grasses (Hartadi, 1980; Bailey and
Connor, 1973; Russo et al., 1981). While it is difficult to compare
hemicellulose values reported by various researchers due to differ-
ences in definitions and procedures, it appears that grasses contain
more hemicellulose than legumes (Bailey, 1973). Further, the
hemicellulose in grasses is different structurally from legumes.
Gaillard (1965) reported that the hemicellulose A fraction of
grasses was primarily an arabinoxylan while in legumes it was a
xylan with some uronic acids. Legumes seem to have more uronic
acids than grasses and some additional sugars such as rhamnose which
are not present in grasses (Gaillard, 1965; Collings and Yokoyama,
1979). Lignin has long been noted to be higher in legumes while
seemingly having less effect on the digestibility of legumes than it
does on grasses. Gordon and Gaillard (1976) suggest that this
difference could be due to the different types of hemicellulose
present in grasses and legumes and the subsequent differences in
The hemicellulose content of a plant changes with growth and
maturity. The most marked change in hemicellulose composition in a
grass is in the growth period before changing from a vegetative to a
floral morphology (Bailey and Connor, 1973). It is expected that
hemicellulose content will increase with maturation. Bailey (1973)
reported on several studies with temperate grasses where the increase
in hemicellulose with age was associated with in increase in stem
tissue. For tropical grasses, however, seasonal changes did not
appear so great, the hemicellulose content remaining about the same
in spite of increasing cell wall content and increasing proportion
The greatest change in hemicellulose content occurs during
growth. The leaves of speargrass (Heteropogon contortus) at vegeta-
tive, early seed and dormant stages had 23, 34 and 36% hemicellulose
(of the total dry matter); the stem in early-seed and dormant
stages had 37 and 42% hemicellulose (Blake and Richards, 1971).
Reid and Wilkie (1969a) found that in older, mature plants, there was
little apparent change in the hemicellulose of either leaves or stems.
Mature stems of rye, barley, orchardgrass, oats, ryegrass, timothy
and wheat had values of 23 28% for hemicellulose (Wilkie, 1979)
while mature leaves of oats, barley, wheat, kikuyugrass and pampas-
grass had values of 21 37% for hemicellulose.
Monosaccharide analysis (Buchala and Wilkie, 1973) of the hemi-
cellulose fraction found that with increasing maturity, there was an
increase in amount of D-xylose in all tissues (stems and leaves),
L-arabinose and D-glucose decreased and D-galactose remained constant
or varied slightly.
The hemicellulose changes with growth may be linked to formation
of the secondary wall. When a plant cell wall stops elongating, it
produces a thicker secondary wall and at this stage, lignin begins
to be synthesized into the wall along with continuing synthesis of
structural carbohydrates found in the primary wall (Morrison, 1979).
It is at this stage that lignin and hemicellulose become linked to
each other. With respect to cellulose, its structure does not change
during growth. Hemicellulose, on the other hand, does change its
structure during growth and with increasing lignification (Morrison,
1979). These changes have sometimes unexpected effects on digesti-
bility, which will be discussed in a later section.
Bailey (1973) reported that in temperate grasses the cellulose
content is greater than or equal to the hemicellulose content but,
in tropical grasses, the hemicellulose content is greater than or
equal to the cellulose content. A further difference between tem-
perate and tropical grasses is in monosaccharides present in the
hemicellulose fraction. Ojima and Isawa (1968) showed that two
tropical grasses (Paspalum notatum and Cynodon dactylon) had xylose:
arabinose:glucose:galactose ratios of 50:15:30:2 while two temperate
grasses (Phleum pratense and Lolium perenne) had ratios of the same
sugars of 74:12:11:2.
It is difficult to separate out environmental effects. The
possibility exists that the differences in CWC of temperate and
tropical grasses may be due to a temperature effect (Cooper and
Tainton, 1968). Anatomical and photosynthetic differences do exist.
Wilson and Ford (1971) studied three grasses (Panicum maximum,
Setaria sphacelata and Lolium perenne) under a controlled environ-
ment and found that the effect of increasing temperature on CWC is
not linear but passes through a maximum; this maximum may be due to
hemicellulose rather than cellulose differences (Bailey, 1973).
Reid and Wilkie (1969b) found little difference in monosaccharide
composition of hemicelluloses of oat plants grown indoors or out,
in light or in dark. Hemicellulose content and composition are not
constant and are apparently influenced by environmental conditions.
Almost every study on hemicellulose contains values for cultivars,
species and genera and many attempts have been made to relate hemi-
cellulose values to intraspecific or intrageneric variations. The
results can be very confusing and probably of little value. As
Wilkie pointed out (1979), "it is not possible to determine invariant,
quantitative values for total hemicelluloses . and there are
no self-evident parameters upon which to base such quantitative
comparisons, particularly at the taxa level." (p. 251) Fraser and
Wilkie (1971) re-emphasized the danger of comparative studies, che
danger arising when the procedure used in isolating materials for
study is either undesirable selective or inadequately reproducible.
With so many hemicellulose procedures in use, it appears that compa-
rative studies should be limited to within locations or laboratories.
Hemicellulose Digestion in the Ruminant
Ruminants obtain a great portion of their dietary energy from
the digestion of cell wall polysaccharides when they are fed forage
diets. The digestibility of the CWC is very crucial in determining
the quality of the forage. Intake, coo, is dependent on the break-
down of the CWC in the rumen. This digestion involves initially
enzymatic hydrolysis by various rumen microorganisms to release
sugars. These sugars can then be fermented to volatile fatty acids.
It is at the initial stages of hydrolysis of the various cell wall
polysaccharides that differences occur and when such properties as
degree of lignification (Bailey et al., 1976) or chemical structure
of the polysaccharides affect the digestion of the CWC.
In vitro Studies
Hemicellulose and its component monosaccharides can be separated
from plant tissue by chemical methods. Earlier in vitro work on hemi-
cellulose degradation in the rumen used these isolated fractions to
determine digestibility of hemicellulose. Gaillard (1965) found
that legume hemicellulose was more resistant to digestion than grass
hemicellulose; of the linear A and B fractions, A was more resistant;
and of the B fractions, the branched B fraction was more resistant
to digestion than the linear B. Sullivan (1966) also showed that
grass and legume hemicelluloses differed in their digestibilities
and suggested that it may be due to a difference in the chemical
heterogeneity of the xylans. Waite and others (1964) found that
during growth in timothy, ryegrass and orchardgrass, the xylan frac-
tion containing the uronic acids became less digestible. Dekker and
Richards (1973) found the glycans in hemicellulose to be more diges-
tible than xylan or uronic acid.
Gaillard (1965), Waite et al., (1964) and Bailey (1967) working
with temperate grasses, found the pectins, arabinans and galactans to
be extensively digested in young plants but incompletely digested in
older plants or in hay. They also concluded that xylans are incom-
pletely digested in mature plants and hay.
Other work has shown that rather than the chemical heterogeneity
of the hemicellulose being responsible for lower digestion, that the
incomplete digestion was due to the presence of lignin. Beveridge
and Richards (1975) studying speargrass, showed no significant
difference between the hemicellulose composition of fresh samples
and of samples digested for 72 hrs in the rumen. In their earlier
work (Dekker, Richards and Playne, 1972) the hypothesis had been
that branching would cause resistance to digestion but since there
were no differences between the original and "resistant" hemicellu-
loses in the later study, they attributed the poorer digestibility
to lignification. Bailey and MacRae (1970) isolated "digestion-
resistant" hemicelluloses which were capable of further digestion
after they had been delignified with NaCIO. Dehoritv et al. (1962)
delignified timothy, alfalfa and orchardgrass by ball-milling and
found that hemicellulose fermentation increased. The increases
became greater with maturity and, again, they offered the theory
that lignin provided a physical barrier preventing degradation of
the hemicellulose molecules.
In studies using isolated strains or species of rumen hemicellu-
lolytic bacteria, Dehoritv (1973) found that fermentation of hemicell-
ulose occurred in two stages. The first stage involved one species
of bacteria which degraded or depolymerized the hemicellulose but
did not utilize it. In the second stage, another type of bacteria
which had limited degrading abilities was able to utilize hemicell-
ulose that had been degraded by the first group. Dehority (1973)
also found that cellulolytic bacteria could degrade hemicellulose.
These synergistic systems may be oversimplifications but the apparent
specificity of the rumen bacteria provides another explanation for
variabilities in hemicellulose digestion.
Another avenue of in vitro studies is the use of depectinated
cell walls and mixtures of mold cellulase and hemicellulase to
study the patterns of hemicellulose digestion. With the enzyme
systems, hemicellulose hydrolysis ,as shown to differ between rye-
grass varieties and the extent of hydrolysis decreased with
maturity (Bailey and Jones, 1971; 1973). Jones and Bailey (1972)
found that prolonged drying did not affect the rate or extent of
Studies on isolated hemicelluloses, its components or enzymes
from isolated rumen microorganisms give only a limited idea of what
is going on in the rumen. More recent research has been on total
plant hemicellulose and in vivo studies of hemicellulose digestion
in the rumen.
In vivo Studies
Beever et al. (1971) studied the sites of carbohydrate digestion
by fitting sheep with rumen cannulas and re-entrant cannulas in the
proximal duodenum and terminal ileum. Their results showed that 90%
of the cellulose and about 70% of the hemicellulose (in ryegrass)
was digested before entering the small intestine. Very little hemi-
cellulose digestion occurred in the small intestine itself but
approximately 30% of the hemicellulose was digested in the cecum
and colon. These values are averages for fresh, dried and ensiled
ryegrass. Other experiments with sheep also showed a significant
post-ruminal digestion of hemicellulose especially when the forage
was ground and pelleted (Thomson et al., 1972).
Van't Klooster and Gaillard (1976) showed that distal to the
duodenum in dairy cows hardly any digestion of hemicellulose
occurred on diets of hay, grass and/or concentrates. Digestion of
cellulose was almost completed before the small intestine.
In a review of several digestion studies, Ulyatt et al. (1975)
found that the proportion of hemicellulose digested in the large
intestine was higher than that of cellulose. An analysis of these
and other results (Bailey et al., 1976) showed that in forages
as the hemicellulose increases, the percent digested in the rumen
increases. The differences noted in all of these studies in the
contribution of the lower gastrointestinal tract to digestion of
hemicellulose may be due to species differences between cattle and
sheep, to the type of feed used or to differences in techniques.
Overall, the in vivo studies show that hemicellulose digestion
decreases with maturity and increasing lignification and that hemi-
cellulose is less digestible than cellulose. Models for cell wall
carbohydrates and their fate in the ruminant animal (Smith et al.,
1971; McLeod and Minson, 1974; Abrams, 1980) suggest the existence
of at least one digestible cell wall fraction and one indigestible
fraction, generally accounted for by the presence or absence of
lignin. Morrison (1979) discusses the physical theory vs. the chem-
ical theory. The physical theory proposes that the lignin and hemi-
cellulose form a complex which has a "cage" effect; in young tissue,
the complex is not fully developed and digestion is easily accom-
plished by the rumen microbes but in older tissue, the "bars" of
the "cage" are closer together and enzyme access is restricted.
The chemical theory suggests that in young tissue, lignification is
at a minimum and hemicelluloses are not very complex; with maturity,
complexity of the structure increases. Morrison concluded that
both theories may be involved; i.e., in youn: grass, the proportion
of side chains is the major factor controlling digestion while with
older tissue the lignin content is the dominant factor.
There are still many gaps in our knowledge of hemicellulose in
the plant and its fate in the ruminant, especially with respect to
the tropical grasses. The literature on hemicellulose digestibility
is often contradictory but there is the possibility that significant
hemicellulose digestion occurs after the rumen and that animal
performance is related to the digestibility of hemicellulose.
Because of the high hemicellulose content of tropical grasses, the
acid-soluble hemicellulose from the Van Soest method may not be an
acceptable estimate of the hemicellulose in these plants. An alter-
nate method of measuring hemicellulose is by alkali extraction from
plant cell walls. The hemicellulose obtained in this manner may be
more representative of the hemicellulose in the plant than the hemi-
cellulose obtained by the Van Soest analysis. Alkali soluble hemi-
cellulose could be a major factor influencing forage quality in
tropical grasses. More information on the hemicellulose fraction
of tropical grasses may lead to a better understanding of the
factors influencing quality. This dissertation investigates cell
wall content and digestibility in tropical grasses, with special
emphasis on hemicellulose.
CHEMICAL COMPOSITION OF TROPICAL GRASS HAYS
TWhile the present official method of forage analysis is the
proximate analysis system first developed in the late 1800's, the
Van Soest (1965) system which partitions forages into cell contents
(CC) and cell wall constituents (CWC) is now extensively used for
evaluation of forages. Use of neutral detergent reagent on a forage
leaves a residue which is essentially nitrogen-free and is composed
of the structural components of the plant, the CWC (Van Soest, 1965;
Van Soest and Wine, 1968). The CC, which were removed by the
neutral detergent reagent, contain the sugars, starches, lipids
and proteins. The CC are essentially digestible by the ruminant
while the CWC have incomplete and differential digestibilities
(Van Soest, 1975; Abrams, 1980). The Van Soest system of forage
analysis is biologically rational and it has been recommended for
grading hays by the American Forage and Grassland Council (AFGC)
(Rohweder et al., 1976; Barnes and Marten, 1979).
Nevertheless, in spite of the widespread acceptance and use of
the Van Soest system, problems still remain, particularly with res-
pect to tropical and subtropical grasses (Moore and rMott, 1973). Due
to physiological and photosynthetic differences and environmental
influences, tropical grasses are higher in CWC than temperate
grasses (Cooper and Tainton, 1968; Wilson and Ford, 1971; Moore and
Mott, 1973). Prediction equations such as proposed by the AFGC rank
the tropical grasses very low due to their high CWC, yet animal per-
formance does not necessarily reflect the predicted low quality of
Due to these anomalies, questions arose as to whether the Van
Soest method, which had been developed on temperate forages, was
suitable for analysis of tropical grasses. While the separation of
CC and CWC appears rational (Van Soest, 1975), the further fraction-
ation of the CWC using acid detergent reagent to obtain cellulose,
hemicellulose and lignin, warrants further examination. The break-
down of the CWC by the acid detergent reagents appears incongruent
with earlier, classical work on hemicellulose, for example, which
used alkaline conditions to fractionate the CWC. Resolution of the
differences between the Van Soest hemicellulose and hemicellulose
obtained via classical methodology may lead to a clearer under-
standing of the factors affecting forage quality in tropical grasses.
The objective of this study was to compare the cell wall frac-
tions of tropical grass hays obtained by the Van Soest analysis with
cell wall fractions obtained by classical methodology, involving
an alkali extraction procedure.
One cultivar of Cynodon dactylon ('Coastcross-l' bermudagrass),
two experimental lines of Digitaria species (X124-4 and X46-2) and
two cultivars of Paspalum notatum ('Paraguay' and 'Argentine'
bahiagrass) grown in an upland sandy soil in north-central Florida
during the spring-summer were used in this experiment. At the
beginning of the growing season a complete fertilizer was applied to
provide 72 kg N, 72 kg P205 and 72 kg K20/ha. Another nitrogen fer-
tilization at a rate of 66 kg N/ha was applied during July. All
plots were staged during July and August by mowing and removing the
growth; thereafter, two field replications of the 2, 4, 6 and 8 week
regrowths were harvested and artificially dried in a portable wagon
at approximately 50 C. The dry hays were stored in woven bags and
chopped prior to feeding.
Voluntary intake and digestibility were determined simultaneously
by feeding the grasses ad libitum to mature wethers. Each hay was
fed to at least cwo (and often three) wethers in separate trials,
with no wether receiving the same hay twice. Animals were treated
for internal parasites, fed a standard hay and weighed before each
trial. The seven day collection period was preceded by a fourteen
day preliminary period. Each animal was housed in a wooden pen with
slotted floors and had access to water, trace mineralized salt and
deflourinated phosphate at all times. Ad libitum feeding was
achieved by allowing a feed refusal (orts) of 200-300 g of hay
daily. Daily samples were taken of the hay offered during the
trial. The orts were removed and weighed daily. Feces were collected
in canvas bags, weighed daily and 20% kept for further analyses.
The waste (hay dropped through the floor of the cages) was weighed.
All samples were dried at 50 C. Each sample was pooled by animal,
ground to pass a 4 mm screen in a Wiley mill, mixed and a
representative portion reground through a 1 mm screen. A composite
hay sample was made by mixing proportional amounts from each animal
fed that specific hay. Animals were weighed at the end of the
trial. The composite hay samples were used for the laboratory
analyses in this experiment.
The hay samples were analyzed for dry matter (DM) and organic
matter (OM) (A.O.A.C., 1975). Neutral detergent fiber (NDF) was
determined as outlined by Goering and Van Soest (1970) using
sintered glass crucibles for filtration. NDF residues and all
subsequent residues were dried at 55 C in a forced air oven over-
night and at 55 C in a vacuum oven for 1-3 hrs.
Hartadi (1980) determined NDF (using glass wool for filtration)
termed NDF(H), acid detergent fiber (ADF), permanganate lignin (PLIG),
hemicellulose (HEMV), cellulose (CELLV) and crude protein (CP)
(A.O.A.C., 1975; Goering and Van Soest, 1970).
The NDF residues were delignified with sodium chlorite (NaCIO2)
according to a procedure adapted from Whistler et al. (1948). An
Eberbach heated shaking water bath set at 70 C was fitted with
clamps to hold six 500 ml Erlenmeyer flasks and was placed in an
explosion proof hood. Rubber tubing, attached to a manifold, was
directed above each flask to deliver CO2 continuously in order to
displace C102 formed during the reaction. NDF residues were placed
in the flasks and NaCIO2 equivalent to the weight of the sample and
100 ml of 1% acetic acid were added to each flask. The weight of
NDF residue transferred was determined by weighing the sintered
glass crucible before and after the transfer. Flasks were placed in
the water bath and shaken for 30 minutes. Flasks were removed from
the bath in order to add an additional .5 gm of NaCIO2. The flasks
were returned to the bath and shaken for an additional 15 minutes.
The flasks were then removed from the water bath to a rack above the
bath, continually flushed with CO2 and allowed to cool. Enough
ascorbic acid was added to stop the reaction (color changed from
bright yellow to pale yellow or white). Each sample was filtered
on a sintered glass crucible, and the residue obtained was termed
holocellulose (HOLO); the soluble portion was termed lignin (LIG).
The holocellulose residues were further fractionated to cellu-
lose and hemicellulose by the use of KOH as suggested by Whistler
and Gaillard (1961), Bailey et al. (1976) and Wilkie (1979). Holo-
cellulose residues were transferred to 100 ml glass bottles with
plastic screw-on caps. Weights were determined by difference, as
for NDF. Fifty milliliters of 5% KOH was added and the bottles
flushed with N2. Up to 20 samples at a time were shaken inter-
mittently in an Eberbach shaker for 24 hrs. After 24 hrs, an addi-
tional 9.5 gm KOH was added to each bottle (to bring the final
solution up to 24% KOH), the bottles were again flushed with N2 and
shaken an additional 24 hrs. The samples were filtered on sintered
glass crucibles fitted to 250 ml side-arm Erlenmever flasks. The
residue was washed with 10% acetic acid and water and was termed
cellulose (CELL). The filtrate was immediately neutralized to pH 7
by the addition of acetic acid and refrigerated. The filtrate
contained the hemicellulosic portion of the cell walls (HC). CW
components were expressed as the percentage of NDF.
A list of abbreviations is included in Table 1. Descriptions
of hays and composition data are listed in Appendix Tables 14 17.
Statistical analysis (S.A.S., 1979) included the general linear
model procedure and simple correlations.
Results and Discussion
Analyses of variance for the chemical composition of the hays
are presented in Table 2. Cultivar and age were the main effects.
Age generally did not have a consistent linear effect on chemical
parameters. For all parameters, hay (cultivar) and age were
significant. Significant interactions of hay x age occurred only
for NDF, NDF(H), HC, HEMV, CELL and PLIG. Except for HOLO and LIG
(which sum to 100 for each sample), the R2 values were .89 .94,
showing that the effects of hay and age explained most of the
variation in chemical composition.
One of the objectives of this experiment was to determine if
the HC, CELL and LIG values were different from HEMV, CELLV and
PLIG or, in other words, wasthe NaC10O-KOH method different from the
Van Soest analyses. The two procedures (NaC102-KOH vs. Van Soest)
were compared by analysis of variance of hemicellulose, cellulose
and lignin (Table 3). For all 3 variables, methods were different.
The overall means for NaClO2-KOH vs. Van Soest's analysis, respec-
tively, were hemicellulose 51.8 vs. 45.2; cellulose 40.8 vs. 43.5;
and lignin 7.4 vs. 8.6 (% of NDF).
The correlation between values obtained for the various chemical
fractions by the two methods are in Table 4. High correlations
Table 1. Abbreviations used in the dissertation.
ADF Acid detergent fiber, % of DM
CC Cell contents, 7 of DM
CELL Alkali-insoluble cellulose, % of NDF
CELLD Cellulose digestibility, in vivo, %
CELLV Van Soest cellulose (ADF PLIG), % of NDF
CP Crude protein, % of DM
CWC Cell wall constituents, % of DM
DM Dry matter, %
DNDF Digestible neutral detergent fiber, % of DM
DOM Digestible organic matter, % of DM
DOMI Digestible organic matter intake, g/kg 75/day
HC Alkali soluble hemicellulose, % of NDF
HCD Hemicellulose digestibility, in vivo, %
HEMV Van Soest hemicellulose (NDF ADF), % of NDF
HOLO Holocellulose, % of NDF
HOLOD Holocellulose digestibility, in vivo, %
IVCELLD In vitro cellulose digestion, rumen fluid only, %
IVCELLDP In vitro cellulose digestion, rumen fluid plus pepsin, %
IV`HCD In vitro hemicellulose digestion, rumen fluid only, %
IVHCDP In vitro hemicellulose digestion, rumen fluid plus pepsin, %
IVHOLOD In vitro holocellulose digestion, rumen fluid only, %
IVHOLODP In vitro holocellulose digestion, rumen fluid plus pepsin, %
IVNDFD In vitro neutral detergent fiber digestion, rumen fluid
IVNDFDP In vitro neutral detergent fiber digestion, rumen fluid plus
Table 1 continued.
LIG Lignin (100 HOLO), % of NDF
NDF Neutral detergent fiber, % of DM
NDFD Neutral detergent fiber digestibility, in vivo, %
NDF(H) Neutral detergent fiber (Hartadi, 1980), % of DM
NDFI Neutral detergent fiber intake, g/kg W'75/day
OM Organic matter, % of DM
OMD Organic matter digestibility, in vivo, %
OMI Organic matter intake, g/kg W'75/day
PLIG Permanganate lignin, % of DM
o o o
0 0 cO
0 0J 0
0 0 (
0 0 0
- oi oo
o o oo
o o o
o o Lfl
o 0 r
0 0 0
o 0 0
0 0 C
o o o
o 0 0
o 0 0
,T Cn <1
3= < 3Q
r r- 0
cy 0 c-
0yi C 0 i*-
Table 3. Analysis of variance of method differences.a
df Hemicelluloseb Cellulose Lignin
Hay (H) 4 .0001 .0001 .1364
Age (A) 3 .0001 .0001 .0001
H x A 12 .0001 .0034 .0193
Method (M) 1 .0001 .0001 .0001
H x M 4 .0001 .0029 .0001
A x M 3 .0111 .0026 .0001
A x H x M 12 .2122 .3154 .0836
C.V. 2.6 2.7 8.4
S.D. 1.3 1.2 .7
Mean 48.5 42.1 8.0
aThe two methods are NaC0l2-KOH
Values expressed as percentage
and the Van Soest
Table 4. Correlation coefficients (r) between variables obtained
by two methods.a,b
NDFc HOLO HC CELL LIC
NDF(H)d .96*** .15 .10 -.06 -.15
HEMV .03 .12 .71*** -.73*** -.12
CELLV -.04 -.24 -.86*** .86*** .24
PLIG .55*** .03 -.01 .02 -.03
aThe two methods are NaClO1-KOH and the Van Soest analysis.
Abbreviations are found in Table 1.
CAnalyzed by Van Soest method and recovered on sintered glass.
Analyzed by Van Soest method and recovered using glass wool.
existed between the two hemicellulose fractions (r=.71) and the two
cellulose fractions (r=.86) but there was no relationship between
the lignin values (r=-.03).
Differences due to method of analysis are more readily apparent
when presented in graphic form. The actual numerical values for the
means of each chemical analyses, grouped by age and cultivar, may
be found in Appendix Tables 18 22.
Neutral Detergent Fiber
NDF values were higher than NDF(H) values (Figures 1 and 2),
presumably due to differences in procedure. NDF(H) was filtered on
Gooch crucibles packed with glass wool, which is more rapid than
filtering on sintered glass crucibles, as the NDF samples were. NDF
and NDF(H) analyses were conducted by two different analysts, at
different times. Disregarding the numerical values of the two
determinations, it can be seen that the lines follow the same
pattern, all cultivars increasing in NDF from 2 to 4 weeks, then
leveling off or decreasing. Overall, in both cases, the two digit-
grasses are different from the bahiagrasses and the bermudagrass.
The two analyses were highly correlated (Table 4). The NDF residues
were used for subsequent fractionation in the present study.
It has often been stated that maturity is the greatest factor
affecting forage quality. In many instances, the effect of age is
linear; i.e., with increasing maturity, some factor increases and/or
decreases. Well-known effects of maturity have been the increase in
cell wall components and the decrease in crude protein and digesti-
bility. The data presented here, however, generally do not exhibit
/ / .. o. ..-
2 4 6 8
W E E S
Figure 1. Change in NDF content by weeks and hay.
O- -Argentine bahiagrass, 0- -0Paraguay bahiagrass,
*---*X46-2 digitgrass, ----' X124-4 digitgrass,
A. A Coastcross-1 bermudagrass.
Figure 2. Change in NDF(H) content by weeks and hay.
0- -0 Argentine bahiagrass, 0- -0Paraguay bahiagrass,
i----* X46-2 digitgrass, -----*X124-4 digitgrass,
A. Coastcross-1 bermudagrass.
I ,- I
a linear trend with maturity. One known event, extensive insect
damage (striped grass looper, Mocis latipes) at 4 weeks to the
digitgrasses and bahiagrass, may provide partial explanation for
lack of linearity.
Holocellulose, the delignified plant cell wall composed prima-
rily of hemicellulose and cellulose, made up the major portion of
the NDF fraction, ranging from 90.45 94.36% of NDF (Appendix
Table 19). Holocellulose values, and all subsequent chemical compo-
sition values, were expressed as percentage of NDF rather than DM.
Holocellulose content of X46-2, X124-4 and Argentine cultivars
were not different; Coastcross-1 was different from the former three
cultivars and Paraguay fell between Coastcross-l and the rest (Figure
3). Four of the cultivars showed an eventual decline in holocellu-
lose content from 2 to 8 weeks but the interim pattern was erratic.
Earlier plant analysts felt that lignin interfered with the
analysis of structural carbohydrates (Whistler et al., 1948).
Adaption of wood chemistry methods led to the use of NaC 02-KOH
to de-lignify plant material. Packett et al. (1965) indicated that
polysaccharide structures are left intact after delignification
provided the procedure is not continued to reduce lignin values to
or less. Wilkie (1979), however, stated that delignification
may oxidize some reducing-end residues and cause partial depolymeri-
zation. Whether or not hemicellulose is bonded to lignin is still
uncertain although work by Albersheim et al.(Bauer et al., 1973;
Keegstra et al., 1973; Talmadge et al., 1973) on dicotyledonous
2 4 6 8
Figure 3. Change in holocellulose content by weeks and hay.
) - -Argentine bahiagrass, 0- -*Paraguay bahiagrass,
*----*- X46-2 digitgrass, *---T X124-4 digitgrass,
G OCoastcross-1 bermudagrass.
plants suggest that there is bonding between hemicellulose and lignin.
Hemicellulose has been extracted from plants without prior delignifi-
cation (Bailey et al., 1976) but the completeness of the extraction
is unclear (Gaillard, 1958). It was felt on these tropical grass
hays, which are high in CWC and lignin, the best procedure would
be delignification prior to further analyses.
The holocellulose values, as an end in themselves, did not give
a clearer or better picture of the plant than did NDF values, which
are easier to obtain. The erratic pattern with maturity suggests
some caution should be used in interpretation of the LIG data
Two methods of analyzing for hemicellulose were used NaC10 -KOH
and the Van Soest detergent system. Figures 4 and 5 present the
changes observed in hemicellulose content by cultivar and age (see
Appendix Table 20 for means). HC values, except in the case of
Coastcross-l, decreased only slightly with maturation which agrees
with other work on hemicellulose in tropical grasses (Bailey, 1973).
HEV values, on the other hand and again excepting Coastcross-l,
showed an overall greater decline with maturity.
HC values were higher as a percentage of NDF than were HEMV
values. With both methods, X124-4 was different from the other
grasses. The two bahiagrasses and X46-2 were similar in HC content
compared to the other grasses while Coastcross-1 was different from
the rest. The bahiagrasses and Coastcross-1 formed a similar group
with respect to HERV values, with the digitgrasses differing from
each other and from the rest of the grasses.
2 4 6 8
Figure 4. Change in alkali-soluble hemicellulose by weeks and hay.
- --Argentine bahiagrass, 0- -Paraguay bahiagrass,
*-- X46-2 digitgrass, ----*X124-4 digitgrass,
s r\ ^-**' """
2 4 6 8
Figure 5. Change in Van Soest hemicellulose by weeks and hay.
0- -Argentine bahiagrass, - Paraguay bahiagrass,
*--- X446-2 digitgrass, ----'-X124-4 digitgrass,
.A Coastcross-1 bermudagrass.
Hemicellulose has been defined as the cell wall and inter-
cellular polysaccharides that can be extracted from higher land
plant tissues that are, or were, lignified (Wilkie, 1979). More
simply, hemicellulose refers to all of the types of polysaccharides
found in plants other than cellulose, starch and fructans.
The preponderance of literature on hemicellulose deals with
the Van Soest hemicellulose obtained by subtracting ADF from NDF.
Work as early as 1970 (Bailey and Ulyatt) points out the problems
inherent in this definition, viz., that there are hemicellulosic
sugars present in the ADF residue (see also Morrison, 1980).
The hemicellulose content of a plant changes with growth and
maturity and is expected to increase with maturity. Bailey (1973)
reported on several studies with temperate grasses where the increase
in hemicellulose content with age was associated with an increase in
stem tissue. For tropical grasses, however, seasonal changes are
not as marked, the hemicellulose content remaining about the same in
spite of increasing CWC and increasing proportion of stems.
Hemicellulosic materials extracted by alkali may precipitate on
neutralization and mild acidification at 0 2 C to form hemicellu-
lose-A. The polysaccharides remaining in solution can be precipitated
with ethanol to form hemicellulose-B. Repeated attempts to obtain
a hemicellulose-A fraction failed. Scant information in the litera-
ture as reviewed by Bailey (1973) showed that hemicellulose-A
comprises 1.0 19% of DM in Heteropogon contortus leaves and stems
respectively, 1.4 3.2% of DM in Digitaria decumbens and .1 .4%
of DM in Setaria sphacelata. It would seem then, that the hemicellu-
lose-A fraction is generally low in tropical grasses, thus providing
partial explanation for our failure to obtain the fraction. HC
values presented here are representative of the hemicellulose-B
fraction which consists primarily of xylans.
Cellulose obtained by either method (Figures 6 and 7) showed
remarkably similar patterns (see Appendix Table 21 for means). In
both instances, Coastcross-1 was very different (having the lowest
amount of cellulose) from the other grasses; X46-2, Paraguay and
Argentine cultivars were not different, and X124-4 had the highest
amount of cellulose. CELL, collected on the sintered glass crucibles,
was a pure white fibrous substance very similar in appearance to
CELLV obtained by the ADF procedure.
Lignin values (Figure 8) were calculated as 100 HOLO. These
lignin values are very different from the PLIG values (Figure 9;
Appendix Table 22). Lignin, like hemicellulose, is defined by the
method used to isolate it. The completeness of delignification of
NDF was tested by staining holocellulose with phloroglucinol-HCI
(a test for lignin). No stain was present in the holocellulose but
it is likely that the material oxidized by the NaCIO2 included more
than lignin. Obtaining a lignin value in this manner is not
recommended due to the possibility of additional products forming
during the oxidation of lignin from the cell walls by NaCIO2. Only
Coastcross-1 had a different amount of LIG chan the other grasses
while PLIC amounts were different for each cultivar.
Figure 6. Change in alkali-insoluble cellulose by weeks and hay.
()- -0 Argentine bahiagrass, O- -(Paraguay bahiagrass,
--- -X46-2 digitgrass, *----- X124-4 digitgrass,
A- Coastcross-l bermudagrass.
2 4 6 8
Figure 7. Change in Van Soest cellulose by weeks and hay.
0 -0 Argentine bahiagrass, O- -OParaguay bahiagrass,
*-- 46-2 digitgrass, *'----- X124-4 digitgrass,
A. .ACoastcross-1 bermudagrass.
2 4 6 8
Figure 8. Change in chlorite lignin by weeks and hay.
0- -Argentine bahiagrass, - -OParaguay bahiagrass,
--- *X46-2 digitgrass, ---- X124-4 digitgrass,
. OCoastcross-1 berumudagrass.
9 / '. ........... O
2 4 6 8
WE E S
Figure 9. Change in permanganate lignin by weeks and hay.
O- -OArgentine bahiagrass, 0- - Paraguay bahiagrass,
S-- X46-2 digitgrass, *----- X124-4 digitgrass,
. Coastcross-1 bermudagrass.
Q. .Qroastcross-l bermudagrass.
One cultivar of Cynodon dactylon (Coastcross-l), two experimen-
tal lines of Digitaria species (X46-2 and X124-4) and two cultivars
of Paspalum notatum (Argentine and Paraguay) cut at four ages each
(2, 4, 6 and 8 weeks regrowths) were evaluated by chemical analyses.
Composite samples for each hay were analyzed for organic matter and
neutral detergent fiber. NDF fractions (cellulose, hemicellulose,
lignin) were determined by two methods: one, the conventional Van
Soest detergent analysis (CELLV, HEMV, PLIG); the other a classical
fractionation using hypochlorite and alkali (HOLO, CELL, HC, LIG).
Values obtained by the two methods used in the analyses were
significantly different. As percentages of NDF, HC values remained
the same or decreased slightly with age for each cultivar while
HEMV values showed a greater decline with maturity. CELL and CELLV
values had similar patterns, increasing up to 6 weeks, then decreasing
slightly. Correlation between HC and HEMV was .71 and between CELL
and CELLV was .86. Further studies of these fractions to determine
their digestibility and value in prediction equations are in subse-
IN VITRO AND IN VIVO DIGESTIBILITY OF TROPICAL GRASS HAYS
Ruminants derive a great portion of their dietary energy from
the digestion of cell wall polysaccharides when they are fed forage
diets. The digestibility and intake of CWC is crucial in determining
forage quality. Differences in breakdown of the CWC in the rumen are
due, in part, to differences in chemical structure of the polysaccha-
Early in vitro work on hemicellulose degradation in the rumen
used isolated fractions to determine digestibility of hemicellulose.
Two general theories arose from these studies. One theory attributes
low hemicellulose digestibility to chemical heterogeneity of the
hemicellulose structure (Caillard, 1965; Sullivan, 1966; Dekker and
Richards, 1973). Other work has shown that rather than the chemical
heterogeneity of the hemicellulose being responsible for lower
digestion, that the incomplete digestion was due to the presence of
lignin (Bailey and MacRae, 1970; Beveridge and Richards, 1973).
Morrison (1979) suggests that both theories may be involved, i.e.,
in young plants the chemical structure of the hemicellulose inter-
feres with digestion processes while in older plants, increasing
lignification causes lower digestibility.
Studies on isolated hemicelluloses and its components give
only a limited idea of what is occurring in the rumen. Beever et
al. (1971) in in vivo studies, showed that 70% of the hemicellulose
digestion occurred before entering the small intestine, very little
hemicellulose digestion occurred in the small intestine and approxi-
mately 30% of the hemicellulose digestion occurred in the cecum and
colon. A review of several digestion studies (Ulyatt et al.,1975)
found that the proportion of hemicellulose digested in the large
intestine was higher than that of cellulose. Bailey et al. (1976)
found that as hemicellulose content increased, the percent digested
in the rumen decreased. Ulyatt and Egan (1979) found 77 to 94% of
digestible hemicellulose apparently digested in the rumen, and 6 to
23% in the intestines. Some explanation of the confusing in vivo
results may be provided by Gaillard and Richards (1975) who suggest
the presence of a lignin-carbohydrate complex which passes from the
rumen in solution, is precipitated in the lower pH of the abomasum,
undergoes some redissolution at later stages in the lower gastroin-
testinal tract, and emerges in the feces.
An intermediate step, then, between studying the digestibility
of isolated hemicellulose and other CWC fractions and the complex
in vivo studies appears to be in vitro studies on whole plant tissue,
followed by analyses of the various fractions.
The objectives of this study were to determine the in vitro
and in vivo digestibilities of various fractions of tropical grass
One cultivar of Cynodon dactylon (Coastcross-l bermudagrass),
two experimental lines of Digitaria (X124-4 and X46-2) and two
cultivars of Paspalum notatum (Paraguay and Argentine bahiagrass)
harvested after 2, 4, 6 and 8 weeks of regrowth with two field
replications were used in this experiment (Abrams, 1980). The
grasses were artificially dried and chopped.
The grasses were fed ad libitum as chopped hay to mature wethers
to determine simultaneously voluntary intake and digestibility of
organic matter (OM) and neutral detergent fiber (NDF) (Abrams, 1980).
Voluntary intakes were reported on the basis of metabolic body
weight (g/kg W 75/day). Each sample was pooled by animal, ground
to pass a 4 mm screen in a Wiley mill, mixed and a representative
portion reground through a 1 mm screen. A composite hay sample was
made by mixing proportional amounts from each animal fed that specific
hay. Feces were collected in canvas bags, weighed daily and 20%
kept for further analyses. The composite hay samples and some indi-
vidual feces samples were used for the analyses in this study.
Half gram samples of each hay were fermented in rumen fluid and
buffer for 48 hr, followed by 48 hr in acid-pepsin (Moore et al.,
1972; Moore and Hott, 1974; 1976), refluxed in neutral detergent
solution (Goering and Van Soest, 1970) and filtered on sintered glass
crucibles. The NDF residues were saved for further analyses. This
treatment was termed the rumen fluid plus pepsin (RF+P) treatment.
Another set of hay samples (.5 gm each) were fermented in rumen fluid
for 48 hr, refluxed in neutral detergent solution (Goering and
Van Soest, 1970), filtered on sintered glass crucibles and the NDF
residue saved for further analyses. This treatment was termed the
rumen fluid (RF) treatment. Each of the RF and RF+P residues were
analyzed for holocellulose (HOLO), hemicellulose (HC), cellulose
(CELL) and lignin (LIG) as outlined in Chapter III.
Fecal samples from the in vivo studies of 2 wk and 8 wk
Argentine bahiagrass, Coastcross-1 bermudagrass and X46-2 digitgrass
were analyzed for NDF, HOLO, HC, CELL and LIC (Chapter III), and acid
detergent fiber (ADF) (Goering and Van Soest, 1970). In vivo
digestion coefficients were calculated for these fractions. Abbre-
viations are listed in Table 1. Descriptions of hays and in vivo
data are listed in Appendix Tables 14 17.
Statistical analyses were performed using the statistical
analysis system (SAS, 1979) computer package for general linear
models, t-tests and correlation procedures.
Results and Discussion
In Vitro Digestibility
Analysis of variance of the in vitro data is presented in
Table 5. Values for in vitro digestion of individual hays are in
Appendix Table 15. In vitro digestibility of all fractions decreased
with increasing maturity in all hays. RF and RF+P digestibilities
were compared by t-tests (SAS, 1979) (HO:RF+P RF = 0) for each
fraction (NDF, HOLO, HC and CELL) by hay and weeks of regrowth
- r- I
0 -- 1
0 0 1
0 0 1
17 cni (
>, Ol, V,
4 C \O
mo r~- <
0 00 00
0 r- Or>
rn CM4 1
r- 4 a
CM Cm CM4 C I
0 0 0 0
00 r- 0 -1
o -4 to CM
o No i
to ej I
r- -T --T o i
o 0 0
-N CN C4 CN
o o oC 1 1
N- II I
SN 0 3 I
3 .0 CM
0 r- 00
NDF digestibility was greater (P values ranged from .14 to
.0007, Table 6) in RF+P for all hays and regrowths. The standard
method of determining NDF digestibility (Goering and Van Soest, 1970)
uses only rumen fluid for 48 hr. While the overall increase in NDF
digestibility ranged from 2.2 to 5.7 percentage units (Table 7), for
individual hays there were differences in NDF digestibility of up to
7 percentage units. In most instances, CELL was more digestible
than HC, whether digested in RF or RF+P. HOLO and CELL digestibility
increased in RF+P only for Paraguay and Coastcross-1 and either
decreased or showed minimal change in the other grasses. HC digesti-
bility was not different for X46-2 but showed an increase of 2.1 to
3.0 percentage units for the other grasses. For individual hays,
there were increases in HC digestibility of up to 12 units by use
By weeks of regrowth, NDF digestibility increased with RF+P
except at 8 weeks. HOLO and CELL showed no significant difference
in digestibility. HC exhibited the greatest difference in digesti-
bility at 2 weeks; the difference decreased with increasing maturity.
NDF and HC digestibility decreased with age (Table 7 and Appendix
Table 16) which is consistent with work on temperate grasses (Gaillard,
1965; Waite et al., 1964; Bailey, 1967).
It seems, then, that further digestion in vitro of the CWC
occurs under acid-pepsin conditions. This digestion, for the most
part, is not attributable to further cellulose digestion but rather,
to hemicellulose digestion and dissolution of lignin. Whether this
increase in hemicellulose digestibility is due to solubilization by
00 cn 01%
00 0 "
o r CM
%D Lrn V
% 0 0
en %D t-
lZ r- 0Z
M. 1 y
CV) c" 0
eo n i
* e *
0r0 I 0
r- % L,
4 N H
o Ln %D r- CV %D
iA in Lrn %D in V
-.7 D '.
C-') '.D -.
'.0 -j 't-
c0 %o L*
N& 4-1 u -1
u (M c 1
0) I 0) C 0)
3 %D &o co :
IT $4 0
(" x < 0 -4
En r-( Cl(
-l 4J U .;
0) 1 C a0) C-
0) I ) m 0) I
oX < u o
the acidic conditions and whether the hemicellulose would re-precipi-
tate under neutral conditions (as suggested by Gaillard and Richards,
1975) is not known. The majority of the in vivo studies have used
HEMV as the measure of hemicellulose. Our earlier experiments suggest
that HEIXV and HC may not be the same entity. Thus, studies showing
that the greatest proportion of "hemicellulose" digestion occurs in
the rumen may be misleading, as indicated by these in vitro results.
In Vivo Digestibility
Fecal samples collected during the feeding trials of these
tropical grass hays were used in further analyses. Argentine bahia-
grass, Coastcross-1 bermudagrass and X46-2 digitgrass were chosen
from each of the three genera at 2 wk and 8 wk of age. NDF, HOLO, HC,
CELL, LIC and ADF values are found in Table 8. Analysis of variance
of the chemical composition of the fecal samples is in Appendix
Compared to the chemical composition of the hays (Appendix
Table 14), fecal samples were lower in NDF, HOLO and CELL, higher
in LIG and were about the same or slightly higher in HC content.
In vivo intake information was used in conjunction with the feces,
orts and waste information to calculate digestion coefficients for
NDF, HOLO, HC and CELL (Table 7; Abrams, 1980). Analysis of variance
for the digestion coefficients is in Appendix Table 23. The values
are compared with in vitro digestibility data on the same hays in RF
and RF+P (Table 7). At 2 weeks, in vivo digestion coefficients were
similar to in vitro for X46-2 digitgrass and Argentine bahiagrass.
For these two grasses, also, digestion coefficients were higher for
Table 8. Chemical composition of feces.
NDFb HOLO ADF HC CELL LIG
X46-2 61.3c 89.9a 37.5b 57.2a 34.4a 10.1b
Argentine 67.8b 86.2b 38.9b 56.2a 30.la 13.8a
Coastcross-1 73.4a 86.7ab 43.4a 53.9a 32.8a 13.3ab
X46-2 73.9c 86.6b 44.2a 50.5a 46.la 13.4a
Argentine 76.5b 84.3b 44.3a 49.4a 31.5b 15.7a
Coastcross-1 82.la 90.3a 45.5a 54.2a 36.0ab 9.7b
aNumbers within a column at each age followed by different letters
are significantly different from each other (Duncan's multiple range).
Abbreviations may be found in Table 1.
cellulose than for hemicellulose at 2 and 8 weeks. The Coastcross-1
bermudagrass had higher in vivo digestion coefficients. Digestibi-
lity of the various fractions was high at an early stage of growth.
Not unusually, the situation changed with maturity; at 8 weeks, the
in vivo digestion coefficients were higher than the in vitro,
ranging from at least 2 to 14 percentage points higher. The degree
of relationship (or correlation) between the in vivo and in vitro
values was high (Table 9) despite lower in vitro values at 8 weeks.
Implications are that in vitro digestibility information is more
representative of in vivo performance at early stages of plant
growth and that with maturity, the in vitro digestibility values
still accurately reflect in vivo but the absolute values will be
lower than in vivo values. The correlation of HCD with IVHCDP was
higher than with IVHCD, implying that acid-pepsin may have some
effect in vivo.
One cultivar of Cvnodon dactvlon (Coastcross-l), two cultivars
of Digitaria species (X46-2 and X124-4) and two cultivars of Paspalum
notatum (Argentine and Paraguay) cut at four ages each 2, 4, 6 and
8 weeks of regrowth, were evaluated by in vitro digestion of neutral
detergent fiber (NDF) in rumen fluid (RF) and RF plus acid-pepsin
(RF+P) and by chemical analyses of NDF fractions. In addition,
sheep feeding trials uere run. Composite samples of each hay were
analyzed for NDF, holocellulose (HOLO), hemicellulose (HC), cellulose
(CELL) and lignin (LIG). Fecal samples of 2 and 8 week Coastcross-l,
Table 9. Relationship (r2) between in vivo
coefficients for six hays.
and in vitro digestion
Variables in Model r P
NDFDa vs. IVNDFD .91 .0033
NDFD vs. IVNDFDP .92 .0025
HOLOD vs. IVHOLOD .88 .0058
HOLOD vs. IVHOLODP .84 .0094
HCD vs. IVHCD .87 .0065
HCD vs. IVHCDP .91 .0034
CELD vs. IVCELD .84 .0100
CELD vs. IVCELDP .79 .0180
aAbbreviations may be found in Table 1.
Argentine and X46-2 grasses were analyzed for NDF, HOLO, HC, CELL,
LIG and acid detergent fiber (ADF).
In vitro digestibility in RF and RF+P of NDF, HOLO, HC and
CELL decreased with increasing maturity in all hays. In vitro
digestion of cellulose was generally greater than that of hemicellu-
lose. In vitro digestibility of INDF and hemicellulose was increased
by further digestion in RF+P while holocellulose and cellulose di-
gestibility showed minimal change.
Fecal samples were lower in NDF, HOLO and CELL, higher in LIG
and about the same or slightly higher in HC content. In vivo
digestion coefficients were calculated for each of the fractions
(NDF, HOLO, HC and CELL) and showed trends similar to in vitro
PREDICTION OF FORAGE QUALITY FROM CHEMICAL AND IN VITRO ANALYSES
Forage quality must be related to animal performance. Moore
and Mott (1973) defined forage quality as output per animal and as
being a function of voluntary intake and digestibility of nutrients
when the forage is fed alone, ad libitum, to a specified animal.
When long-term production trials are not feasible as is often the
case, researchers turn to laboratory analyses of forage quality.
Many and varied are the laboratory procedures devised to predict
forage quality. These may be grouped into chemical and in vitro
methods. Thus far, chemical methods have not been overly successful
in predicting forage quality but the in vitro methodology, especially
with respect to predicting in vivo digestibility, has been more
successful. In a review of the many forage evaluation methods in
popular use, Barnes and Marten (1979) reported that the in vitro
system was superior to chemical analyses for predicting in vivo
digestibility of forages. Intake is much more difficult to predict
from laboratory analyses but Rohweder et al. (1976) have suggested
the use of neutral detergent fiber (NDF) to predict intake.
This study was conducted to examine the relationship between
the Van Soest chemical analyses, another chemical analysis procedure
(NaC102-KOH, see Chapter III) and/or in vitro digestion with in vivo
intake and digestibility of organic matter (OMI) and NDF in 39
tropical grass hays of three genera harvested after 2, 4, 6 and 8
weeks of regrowth.
One cultivar of Cvnodon dactvlon (Coastcross-1 bermudagrass),
two experimental cultivars of Digitaria (X46-2 and X124-4 digitgrass)
and two cultivars of Paspalum notatum (Argentine and Paraguay bahia-
grass) harvested after 2, 4, 6 and 8 weeks of regrowth were used in
this experiment. Hays were chopped and fed ad libitum to mature
wethers to determine intake and digestibility as outlined in
Chapter III. A composite hay sample was made by mixing proportional
amounts of hay from each animal fed that particular hay, was ground
through 4 mm and 1 mm screens of a Wiley mill and was used for the
The composite hay samples were analyzed for DM, OM, NDF, holo-
cellulose (HOLO), hemicellulose (HC, HEMV), cellulose (CELL, CELLV)
and lignin (LIG, PLIG) as outlined in Chapter III. In vitro fermen-
tation in rumen fluid (RF) for 48 hrs and in vitro fermentation in
RF for 48 hrs followed by 48 hrs in acid pepsin (RF+P) were both
followed by refluxing in neutral detergent solution to determine
in vitro NDF, HOLO, HC and CELL digestion as outlined in Chapter IV.
A list of abbreviations is presented in Table 1.
Statistical analyses were performed using the statistical ana-
lysis system (SAS, 1979) computer package for general linear models
and correlation procedures.
Results and Discussion
Intake and Digestibility
Individual values of in vivo intake and digestibility are in
Appendix Tables 17, 24 and 25. Chaves (1979), Abrams (1980) and
Hartadi (1980) presented discussions of the relationship between
intake and digestibility in vivo of these hays. They suggested that
digestibility was not the primary factor controlling intake of these
hays. Furthermore, there was a poor relationship between intake and
digestibility (OMI vs. OMD, r = .47; Chaves, 1979). Therefore, it
is unlikely that a single laboratory procedure will prove to be a
good predictor of in vivo performance.
Another reason for skepticism concerning the ability of a single
laboratory procedure to predict forage quality lies in the effect of
maturity on the in vivo values compared to the effect of maturity on
the laboratory values. In Chapter III, it was reported that maturity
did not have a consistent effect on the chemical parameters, i.e.,
the effect of age was not linear. Expectations are that, with
maturity, some factors decrease (e.g., crude protein) and some
increase (e.g., NDF) but the hays used in this study did not follow
the expected pattern. However, the in vitro and in vivo values did
follow the expected pattern of decline with maturity in most of
the hays (see Appendix Tables 15 and 16). Argentine bahiagrass and
Coastcross-1 bermudagrass did not change in OMI and changed very
little in OMD but the other three grasses decreased from 2 8 weeks
in OMI by 4 8 units (g/kg W.75/day) and decreased in OMD by 10 12
percentage units (Abrams, 1980).
Prediction of Forage Quality from Laboratory Analyses
Simple correlations between in vivo data and laboratory
analyses were conducted (SAS, 1979). Only the best correlations
for each in vivo parameter are presented in Table 10. The highest
correlation obtained between OMI and all laboratory analyses was
with CP (r2=.20). DOMI had the highest correlation with NDF
(r2=.46) and [NDFI had the highest correlation with PLIG and CP
(r =.06). Intake has always been difficult to predict by laboratory
analyses alone. Differences in voluntary intake are often attributed
to rate of digestion and in vitro digestibility systems generally
are more accurate at predicting intake than are chemical analyses.
Table 11 presents correlations between in vivo and in vitro values.
Two in vivo intake variables, OMI and DOMI, had significant corre-
lations with all of the in vitro analyses; r values ranging from
.13 to .56 with significance probabilities of P5.10 to 5.001, respec-
tively. NDFI, which did not vary (Appendix Table 24), was not
predicted well, or at all, by any of the in vitro analyses.
OMD and NDFD had the highest correlations with NDF (r-=.59 and
.56, respectively) while DNDF had the highest correlation with PLIG
(r-=.23) (Table 10). Aside from NDF and PLIG, there were no chemical
analyses which accurately predicted in vivo digestibility. However,
the in vitro analyses generally gave highly significant correlations
with OMD and NDFD; r- values ranging from .61 to .81 (Table 11).
Correlations between in vitro analyses and DNDF, although significant,
Table 10. Relationship (r ) between in vivo values and
analyses, obtained with thirty-nine different hays.
Variables in Model
Variables in Model r P
NDFI vs. PLIG
NDFI vs. CP
aAbbreviations may be found in Table 1.
Table 11. Relationships (r ) between in vivo values and in vitro
analyses, obtained with thirty-nine different hays.a
In vivo Variables
DOM I JDFI
.18** .77*** .54*** .02
.20*** .79*** .56*** .02
IVHOLOD .15* .72*** .47*** .01
IVHOLODP .19** .71*** .52*** .01
.16** .73*** .49*** .01
.20*** .67*** .52*** .02
IVCELLD .17** .71*** .50*** .02
IVCELLDP .13* .61*** .42*** .004
aAbbreviations may be found in Table 1.
Interestingly, in almost every instance, r values for each
chemical fraction's in vitro digestibility in RF and RF+P (IVNDFD
and IVNDFDP, for example) were not different. In Chapter III,
it was shown that there was a significant increase in digestion
in the RF+P treatment for the NDF and hemicellulose fractions,
while holocellulose and cellulose digestibility showed minimal
change. The similar in vivo correlations with the RF and RF+P
in vitro data make laboratory decision making easier. Since there
is little difference in the two methods (RF vs. RF+P) insofar as
predicting in vivo data is concerned, the method of choice would
be RF, which is less time, equipment and reagent consuming.
Furthermore, there was no advantage to using a more complicated
procedure for residue analysis than the NDF procedure.
Tables 12 and 13 present the models obtained by general linear
models procedure (SAS, 1979) that best fitted the in vivo intake
and digestibility data. A rational, rather than an empirical or
multiple regression, approach was used to choose independent
variables for each model. Independent variables were chosen on
the basis of significant correlation with the in vivo measurement
of interest (Tables 10 and 11). In most cases, in vitro digestion
values in RF only were used unless RF+P values were much higher.
In addition, some variables were included on the basis of their
presumed biological role in the structure of the cell wall. Cell
wall structure and components play a crucial role in determining
intake and digestibility of tropical grasses.
Table 12. Coefficient of determination (R2), probability (P) and
standard error of the estimate (s ) for the regression of in
vivo intake values on several independent variables.
Dependent Independent 2 P s
Variable Variable(s) y-x
OMIa,b NDF PLIG .24 .0078 6.42
OMI NDF HC PLIG .25 .0166 6.44
OMI NDF CELL PLIG .26 .0150 6.43
OMI NDF HC CELL PLIG .25 .0347 6.51
OMI NDF HC HC2 .23 .0269 6.54
OMI HC CP PLIG .25 .0185 6.47
OMI IVNDFDP IVHOLODP .20 .0183 6.57
OMI TVNDFDP IVHOLODP TVHCDP .21 .0366 6.61
OMI IVNDFDP IVHOLODP IVHCDP IVCELD .21 .0779 6.70
DOMT NDF PLIG .53 .0001 4.89
DOMI NDF HC PLIG .53 .0001 4.94
DOMI NDF CELL PLIG .53 .0001 4.94
DOMI NDF HC HC2 .47 .0001 5.30
DOMI HC CP PLIG .42 .0002 5.47
DOMI IVNDFDP PLIC HC .58 .0001 4.71
DOMI IVNDFD IVHOLOD .55 .0001 4.81
DOMI IVNDFD IVHOLOD IVHCD .55 .0001 4.87
DOMI IVNDFD IVHOLOD TVCELD .55 .0001 4.93
NDFI HC PLIG .07 .2824 4.85
NDFI HC CP PLTG .08 .4106 4.90
NDFT IVNDFDP HC PLIG .07 .4677 4.92
aAbbreviations may be found in Table 1.
bIntake expressed as g/kg W 75/day.
Table 13. Coefficient of determination (R2), probability (P) and
standard error of the estimate (s ) for the regression of in
vivo digestibility values on several independent variables.
Dependent Independent 2
Variable Variable(s) y.x
OMDa,b NDF CP .61 .0001 3.62
OMD NDF CP PLIG .65 .0001 3.47
OMD NDF PLIG .65 .0001 3.42
01D NDF HC .59 .0001 3.69
OMD NDF CELL .60 .0001 3.68
OMD NDF HC PLIG .66 .0001 3.44
OMD NDF HC CELL .43 .0001 3.74
OMD NDF HC CELL PLIG .66 .0001 3.49
NDFD CP PLIG .43 .0001 5.14
NDFD NDF CP .59 .0001 4.38
NDFD NDF PLIG .64 .0001 4.08
NDFD NDF CP PLIG .64 .0001 4.14
NDFD CP PLIG HC .54 .0001 4.67
NDFD NDF HC PLIG .66 .0001 4.02
NDFD NDF HC CELL PLIG .67 .0001 4.05
NDFD NDF HC HC2 .58 .0001 4.46
DNDFC HC CELL .19 .0225 3.50
DNDF HC PLIG .40 .0001 3.02
DNDF HC CELL PLIG .42 .0002 2.99
DNDF PLIG CELL .42 .0001 2.96
DNDF CP PLIG .32 .0011 3.22
DNDF CP PLIG HC .40 .0004 3.05
DNDF NDF CP PLIG .32 .0035 3.26
DNDF NDF HC CELL PLIG .43 .0006 3.02
aAbbreviations may be found in Table 1.
Digestibility expressed as %.
CDigestible NDF expressed as % of DM.
Going from a model containing one independent variable to a
model containing more than one independent variable did not produce
better or more acceptable predictions of intake. OMI ranged from
46.7 to 79.7 g/kg W'75/day (Appendix Table 24). Four of the "best"
equations for predicting OMI included NDF and PLIG. R2 values and
standard error of the estimate, respectively, were .24 and 6.42
(NDF and PLIG), .25 and 6.44 (NDF, HC and PLIG), .26 and 6.43
(MDF, CELL and PLIG) and .25 and 6.51 (NDF, HC, CELL and PLIG). DOMI
(ranging from 25.6 to 54.0 g/kg W 75/day) was best predicted by two
and three variable models that included an in vitro value in the
model, as to be expected for the prediction of digestible OM intake.
The best equation for predicting DOMI included IVNDFDP, PLIG and HC
(R = .58 and s =4.71), but this three variable equation was not
much better than the equation using only IVNDFDP (R =.56).
NDFI was still predicted poorly by any equation, be it single
or multiple variable, probably because there is little variation in
NDFI (see Appendix Table 24). As R2 values increased slightly,
standard error of the estimate increased so that the best predictor
of NDFI remained crude protein alone (Table 10).
OtD and IDFD were predicted quite well by any and all of the
single variable in vitro equations (Table 11). Adding more variables
to the model decreased the R2 and increased s for OMD and NDFD.
Single variable models of in vitro data used for predicting DNDF had
R2 values ranging from .14 to .38. The best two variable model for
predicting DNDF included PLIG and CELL (R =.42) and was not improved
by adding another variable to the model (HC, CELL and PLI, R=.42).
by adding another variable to the model (HC, CELL and PLIG, R =.42).
The best four variable model included NDF, HC, CELL and PLIG (R =.43)
but the standard error of the estimate increased with the addition
of more variables to the model statements.
The results of the studies indicated that NDF, CP and PLIG
were the chemical analyses most closely associated with the digesti-
bility and intake of tropical grasses. New methodology developed to
analyze cell walls did not contribute to the prediction of in vivo
performance. Most of the in vitro procedures gave fair to good
predictions of OMI, OMD, DOMI, NDFD and DNDF but not of NDFI. Two,
three and four variable models increased the R2 values for OMI, OMD
and DOMI and decreased the R2 values slightly for DNDF and more so
for NDFD. NDFI could not be predicted by any single or multiple
variable model. It is concluded that we still do not have an
accurate picture of the structure and function of the cell wall in
tropical grasses. The difficulty of predicting in vivo animal
performance remains a great challenge in evaluation of forage
quality. Twenty of these hays were analyzed at the Pennsylvania
State University by near infrared reflectance spectroscopy (NIRS)
(Russo, 1981, unpublished data). Using large multiple regression
models and two wavelengths, NIRS gave better estimates of forage
quality than any of the chemical and in vitro analyses conducted
in this study.
One cultivar of Cynodon dactylon (Coastcross-1 bermudagrass),
two cultivars of Digitaria species (X46-2 and X124-4) and two culti-
vars of Paspalum notatum (Argentine and Paraguay bahiagrass) cut at
four ages each (2, 4, 6 and 8 week regrowths) were evaluated by
chemical analyses and in vitro NDF digestibility in rumen fluid (RF)
and RF plus acid pepsin (RF+P). In addition, sheep feeding trials
were conducted. Composite samples for each hay were analyzed for
organic matter, NDF and holocellulose (HOLO). Hemicellulose (HC,
HEMV), cellulose (CELL, CELLV) and lignin (LIG, PLIG) were determined
by two different methods. Information obtained was used to generate
prediction equations for sheep in vivo intake and digestibility.
NDF, CP and PLIG were the best single chemical analyses for
prediction of intake and digestibility. OMI and NDFI were not pre-
dicted as well as DOMI. Any of the in vitro analyses were better
for prediction of in vivo digestibility (OiD, NDFD and DNDF) than
were chemical analyses. Two and three variable models which included
NDF and PLIG improved prediction of OMI and DOMI.
Forage quality reflects the potential of animal production
from forage by use of the available nutrients in the forage. Forage
quality may be defined as animal performance (Moore and Mott, 1973).
Analysis of forage quality in the laboratory has almost a hundred
year history and gaps still remain in our knowledge of how the
forage plant is put together, how it functions and how it affects
Essentially all of the forage evaluation procedures have been
developed in temperate climates using temperate forages. Not
unlike the problems associated with transferring Western agricul-
tural expertise to developing countries is the problem of trans-
ferring forage evaluation schemes generated on temperate forages
to tropical forages. Sometimes it works, sometimes it does not.
Tropical grasses are rather different from temperate grasses.
Tropical grasses have a much greater photosynthetic capability, are
more efficient in water use, have different carbon pathways and
have different anatomies. The first product of photosynthesis is
carbohydrates and tropical grasses have specialized plastids in
the sheath surrounding the vascular bundles for starch conversion
and storage. Tropical grasses are higher in cell wall and cell wall
components than temperate grasses.
What is the cell wall? It is a highly ordered macromolecular
multilayered network made up of structural carbohydrates (cellulose
and hemicellulose), lignin and protein. Dividing plants into cell
contents and cell walls (as does the Van Soest forage evaluation
system) provides a rational basis for discussion of forage quality.
Cell contents are essentially completely digestible by the ruminant,
cell walls have incomplete digestibilities. A number of cell wall
models have been proposed. Albersheim and others (Bauer et al.,
1973; Keegstra et al., 1973; Talmadge et al., 1973) propose that the
cell wall is made up of long crystalline microfibrils (cellulose)
surrounded by hemicellulose, which is hydrogen bonded to the cellu-
lose and covalently bonded to the other cell wall components (pri-
marily lignin). Covalent bonds are more difficult to break than
are hydrogen bonds. Bailey and others (1976) propose that all of
the crystalline polymer need not be cellulose, but could be hemicell-
The role of hemicellulose in the plant has been established
as structural but hemicellulose may exist not only in the cell wall
and have been found in the endosperm of cereal plants, for example
(Wilkie, 1979). Hemicellulose is added to the wall continuously and
is increased by increases in temperature and light. While cellulose
composition and structure does not change with maturity or under
environmental influences, hemicellulose does. Not enough information
exists, but it is likely that each genera, each cultivar, perhaps
even each plant, and certainly, different parts within a plant have
different kinds and amounts of hemicellulose. Hemicellulose content
and concentrations are variable, not constant.
The results of this research indicate that "hemicellulose" is
an elusive term. Van Soest hemicellulose is certainly not the same
hemicellulose obtained in the NaCIO2-KOH procedure despite presumed
similarities in the cellulose fraction. Lack of expertise prevented
further examination of the sugars making up the hemicellulose but
some initial observations were made. One of the experimental digit-
grasses, X46-2, which is "high" in forage quality by most standards,
appeared very low in xylans. Xylans are usually highly ordered,
even crystalline and difficult to digest (Daughtry, 1976). Tropical
grasses are generally higher in xylans than are temperate grasses
(Ojima and Isawa, 1968; Bailey, 1973) and xylans may be responsible
for lower quality of tropical grasses. This experimental line of
digitgrass should be examined for xylan content and compared with
other grasses used in this study.
In vitro digestibility results were illuminating. NDF and
hemicellulose digestibility increased in acid-pepsin conditions,
cellulose digestibility did not increase. Further breakdown of the
cell wall, then, in the lower gastrointestinal tract, could be due
to breakdown of the hemicellulose. Hemicellulose has generally
been presumed to have a low digestibility, with lignification
postulated as the limiting factor but a distinction should be made
between hemicellulose unavailability due to structural organization
of the cell wall carbohydrates or interference by cellulose, ligni-
fication, acetylation, etc. (Morrison, 1979). McLeod and Minson
(1974) propose that the cell wall carbohydrates have two fractions -
lignin-free and lignified the lignin-free being totally digestible,
the lignified fraction being undigestible. Other digestion models
(Waldo et al., 1972; Abrans, 1980) are similar. Accurate procedures
for determining hemicellulose and lignin must be developed in order
to answer some of these questions and to prove some of these theories.
It would seem that if the proposed cell wall models and digestion
models were fairly accurate, using both models would allow one to
state that hemicellulose can be "digested" from cellulose and
within the hemicellulose polymer itself but that pieces of the
hemicellulose polymer remain covalently bonded to lignin and are
more resistant to digestion. Bailey and Ulyatt (1970) and Morrison
(1980) intimate these conclusions also. To extrapolate further,
Chaves (1979) suggests that bridges are formed between the cell
wall carbohydrates and the lignin by phenolic acids which can
deleteriously affect microbial digestion.
In tropical grasses, hemicellulose, lignin and phenolic acids
are higher than in temperate grasses. Are these, then, the components
responsible for low quality in tropical grasses? Can the plant
breeder do anything about it? Breeding for low lignin brought about
the development of brown midrib corn, academically interesting but
of little use to the farmer. Is low hemicellulose or low lignin
content really desirable in a forage plant?
Forage quality analysis and prediction of intake and digestibi-
lity still has researchable areas. This research should be carried
further to determine monosaccharide composition, relate the composi-
tion to animal performance and then devise a rapid method of cell
wall monosaccharide determination. The physical and chemical forces
holding the various cell wall components together and what it takes
in terms of energy, microbes, enzymes, acids or whatever to break
those bonds could be examined. Anatomy and histology of forages as
related to intake and digestibility should be examined and the poly-
saccharides and monosaccharides in the various structures be
determined. Infrared reflectance spectroscopy may alter procedures
used by many forage testing laboratories as it becomes more accurate
and accessible for routine use. That may be for the better, allowing
more basic research to be conducted on forage plants.
So eeeOee O O OO ONWe Oeeee e e Oe
%D 0 )O' 4TL* T 00r r C C4 * C ** * **C L) ** * * **- r4 .0 C
omoocI -4L"r-4-Im o
CM r4 O 00 '~ v( ) 4 O C ] C ML O CM
mn d T m 's m' l l)I l -T(Tm s-T 'T 'T
-l T CM W % C O
CMq -T -.:T C14 W0 M O'
C4 CM i-v v1:) 0 0
N~> rI) u Wr L in n
rNrN~r ecln O
oC0o om m0*m m
nO %c C r-
- r-l O l-4O
CM 0 T \000 i-
* * * a-in i
%oD r C% LnM r- t~
0N 0\ 0^ 0^0 CN m CNm
0 0 O~~~
, , .J ~
%0 %\0 0 % C
* * * _
c ooWmr-r- OOcM
r- r~- CM M M Mo r-c
00 0oo oo t~ 0' oo 0o r- P-
Cl -i 4r- r-I
S0 0 0 0* 0* 0*
r~lJh 001O o
r-4 0 T 0 0 0 0w
-.: in 'T -1 CMq in r~
o * * *
M 44 -4
3 4 -C4 X
&0 C I
CB o '~
,-1 % C0 %D
S41J U u
0i C4 4-J AJ CM
C Q) I 0 0 I
04 -:c x 0 0 x
-H c *H -H ',4 r: 5H -H 5A r t cO H r= 5 -r- I -H E -' H '
( (- (m m r-4 r4 CZi -1 -1 (3 (m -i (i 0 0 (C r- (-1 (Mi -1 (C 0 0 m (
41 a ma jl 4- a ca 4J cl a 4 a 4 -1 a ca rl 4 1 u a 1a Mu MJ 41 -C -0 4o 4 1 41
*rP- P- *d 4 -r-4 P -HL0 H -,A P. O H *,-I P. P. -H 0 0 T-H Ow 4 -I -H 3. Cu T-H 0 0 H
QrAQ mm QA QQ -1UG AA r-4 C4 C4 C4 C4 (y) (y) Trl rl rdLr Lr.%
17-.:T OO %D %Dn O "
m C% o 0 o 'TO O % -C
o. CN .% ..C.C.(N *
r- o0- CM
r- r- r-
=J M4 -T
60 1.. CM
MO () T- I
0) CI I
(1) (a -T
-C -i cCM
,i C ,.--
r i-4 V) CM4
T3 Ln in) '-,
CM CM4 r-1 l 00 CM
%0oY a o-4- CM4 V
~~.D~~ 04 -T -. 4 ~C -.7
1 C C C C
N Du % D 00
T* ** 7
-T T ^T--
CO '0 .
00 J ",-0
rN .-4 -n
m o O
--1 .-- o0
oO -o r
(Ni 10 .40
r- cl r) r-
CO z CT, M C1M M c1 O
W r-4C1 0 m ND U
Sr.o o o ^ n U' n 'L.
1- c -i 0 0oM I 0Dn 0D
m T CNO N ON
04 CNI 0 (N 01 1 4
c7%C7"CNCN 7" 7"(T"C7
,- Cr00 N- O C0
CO CO C- IO CM C-
r 0oo 0 r- 0 ooo
00 0 -1 Fn -7 0 -.0
0' a' 0' a' 0' 0- 0'
a' LA C -7~
C-1 C- 00 -4 40
=1 4-1 -3- IT
o00 C I I
ex C4 -7 C.
E E -- -1-1
Q. Ca *r- *r-I
- 40 *c- *I--
c= 1 -4.-
O-N LI rC I
0% 01 ON CN
CO Q I
EE E -'
r-4 ,-4 TO 0 0 0 0
mo ca 1.J *C -L -a -v
0- Q. *-4 O O O 0
CO 0 *-4 >. >. >C >C
C Cr- uu u
o 0 --in '%. r -T Ln 'D r-
, ? COOCO
a,% -IT C
- n rON ND O DO
ON W (m3 M- o) C-4 r~-
c< _; cT4 r4 4 r0 0
* * * * *
0 ~v ~ m
4T- '-I in
%D rl -\D %
Lor)o %oi CT\
c4T N 1
ins % sD
r-- ) CM1
" *." '. '.
r (O m C 0o IT
in) Lni i) u
* * * **~6
%D C4 -00 0
SOc CM %D CO
s 4r O so
C-4 -t n Mo0 %D in
%Di c n -I t 'T %D D I
* * * **hlr
oDoOr OMm W
- 1 0 C i IT w r
1- %D in Ln sO0 -IT I-
* *0 * *
- -m . mmmmm T T n Ln
\O n n \O D Ln
IT ) m.O
- ,- ..
.. u' <'
U 1Lf r-:
- SO M 003 0 CM
r'- in ITa '-4 %.0 so ITa
r~, %.0< c in "M %D m r-I r-. in
O r-| m O ON % -1 i-1 in
r4i 11; -~ r~- 1o r11> cT r4i c4 1o co
r" r" r~- %~.0 o10 r~- -1 ini m %Do <
r, Wc o r,-4 M mm o M in
*n n%0\ W) i %D W) LnD
IT w> %o 0 rr o -1 r~-
mbO N0 HN
inr) l C14sDCM
C4 _l; l r~
sIso in nin
" m r .Do M .0O -
in %.o a,,M -- -- oo
% nICT inO in m r- in
,.D r-. 'D a, LCn Cr 'D0 r-. 00 ON
-7'- N 00 3*1 00 r-- a' 0 Co 4 C14
LI Lf4 1.0 Lt') U') LI) L4 iLr) Lf) ur) Lr
If' r)-) -7 1.0 00oI r-7 Lf)14 C7% C7%a
00 1.0 '.D IT 00 Co '-4 r- ir) -7 r-
.1 '. Co '. .1 '. .0 .4 .'
Lf4 Lf4 1.0 1.0 Lf4 Lo) Lf4 Lo) d) Lf4 -17
' %D0 -7 Ir) C4 r-- a7% Co r-- Lr
r- o o -4 r- r-Coor- o a)
-IT -.7 1.0 1.0 17 Lf4 Lf4 Lo) ir) Lf4 -Zr
C)oCo Lo) a,. 0. 1.0 C( N a' -7 N
e-- ON '0 %D0 a'% 1-4 r- -7 C. IT
Co Co o. Co Lf4 r- '.0 U) -7 Co r-
17 17 %D0 Lf4 17 -.7 LI) V) Lf4 17 -.7
'.0 C11 a C' ci 17 .. i I, 3' F ) '.0
.~C . .' Co .- a' a'
O 0) .-4 (3N -4 -7 r- 0o Co C. a
L') 124 %.0 U) ir) U') Lf4 Lf4 .fl Lf4 -7
%D0 a'% -4 -7 Co C C'n '0 C) 7 -7
Lf4 -7 '-4 Lr) Co --7 ~-4 3*1 C '7
'-4 cN eD a', o. -4 'o0 '. c7% a')C
124 t) '.0 124 Lf4 L(4 L,) 12) V24 -T -.
M~ a) C. 124 C C. 3*1 C'n .0 CN -7
a'4 3') r.F-3-4 00 m V. F-1 '4 a' Co
-.7 -7 %0 124)-.7 124 124 12) 124 -7 -.7
12n C. -4 -T '0 (N (NC '.0 Cn4
00 cn 0.C 010r ) -Co co
a' :7: Co N- C. 3' L4 Lf r Co F-
-7 -.7 124 124 Ln '1 7 Lt4 124 .24 IT7 17
Co C. -4 Cn L) 12 .0 F-- -7 424 '.0 F--
.or r-r-r- r- r- o~o a)o~o
0o i-A CM4 r- t~-
00 "M CM4 Ulf CM
NhC14 00 0
nCM ooo CM
110 r~- P- %
m00-1 ooC 00
00 C 19 CM ril
CT^ 00 r- r--l r--l
CM4"CM M~ M
1-4 ( 3-T cM oo
CM r~ i O rs
ON C0 c') 0o r^
rLi r- r- 110 en
r-i 00 n oo r rn
moo en ene
in n Ni
ON r NM r r- I' -i r- sTN
O~ .NOO meeee
%00CC0 00%0- 0
%0 r. 00 00 r~. r. r~. r- r~. r~- r~- fl- 0
110 Lo i
0I 00 r~
CT^ D ^D
0w % "m r- r- I D
* . .O. .
mC oCm rI. CI
rn r cI '.0 r1-
mCM wC 0I T- ui0 I-
wM oo 1 -1 C mI o
* * * **r~r
11 1 N
ITr -4 r-
en -4 ry-
m U\) T\o
CM Co CM
C00 CY 00
r00 C) CIT 00
Ir I 0 I
r-M 1<10 i- C)
<3 \) 110 0 '-q
r- *c4 *a
T I'D%0 I
%0 co-- T\
17 IT -:T
110 110 r-A OD i-A
en i-I %.o i) i-4
* * *
'a- \ f- in
r~. o ^ r~ r~
o Lr co -".
oNM 0 r
-T 'O0 Lt -7, .T
st e try
0- Ltr~ r-3 rn-
Lr i Lin L t
' 1o 0 -r '.- .0 i 0
r-'. C4 r~-'. r -: O0
NrQ o ~ Oo
* * * **)Q
mn om o mO
in .n M- 0 1-1 '"
r 0rQr ci NrQ
00 C cn ir cM 00
Cen n Cn .-i in
rN 00 r- 'T caj rc
4 c rm -C4 cJ N-4
-.7 7T -t -3 T -T
an o in -T7 0-.
N N- NO C 0
0 cn \' i r -
IT -T -T .T -T
-n e Ci-
SCO 0- 0 Nr- CO 0-
o0 0 -T C 0 Lr).
o m 0so
Co o rn n oZ rN- T iU) 'o r-
W nC - -q wC C-4C m r- mOQCD '4\
r-r Dr -%D %0P-%0Lr lu0P 0V V l Z10 DV -r Lr) .0O0
m~~oom ~ e~~cm oor-~~m-
% i iv v 4 i: c L a m o a % 4 i r r T -
44 444 m mU cN CN C'41 0
m I z z V T z z I I T : L U I I .
V)~~~ ~ ~ ~ ~ C% C)"L) 0o % v)- 0r- 1rr 0 D 0C r-C -T%
% % % 0 V % )L o ) r)tr)
1 C C r 19- C q q uJ-,
CN -Tr- rq c a ri % r m T L C) ) 4raNCv r C 0
r-%D%0r-P-%0% 0 0 r)Lr %o oL U) l Ul % % U)% % V L) D r
m C 0 r mLn m .41r-0 D 0 n - D C L) -4C1
. . . . . . . I *
,T1 M CLr r L) r4 14C1 L) r L) 0 1001. r-ic4 ;
r-1. % r % U) n D D D D n n n n n ) ) D 0 r)U' -T r)Lr I
C1 L) o N :)-icn m r)r-ooc)c- m -, L) o %o o r-lM T r
r-Ir-l 1 C C4 4 N M M M M -. -T -T ) f)U) f)Lr