grazing management and tussock distribution in elephant grass

Grazing management and tussock distribution inelephant grass

L. E. T. Pereira, A. J. Paiva, E. V. Geremia and S. C. da Silva

E.S.A. Luiz de Queiroz (ESALQ), University of S~ao Paulo, Piracicaba, SP, Brazil

Abstract

Soil occupation capacity via lateral expansion of tus-

socks in elephant grass (Pennisetum purpureum Schum.)

may be associated with basal tillering. As grazing man-

agement alters the proportion of basal and aerial tillers

in a tiller population, the hypothesis of this work was

that grazing management affects tussock size and dis-

tribution with implications for plant population stabil-

ity. The objective of this study was to evaluate the

tiller population stability index, the proportion of basal

and aerial tillers, tussock size, and the frequency of

tussocks and bare ground in rotationally managed ele-

phant grass cv. Napier. Treatments resulted from the

combination of two post-grazing heights (35 and

45 cm) and two pre-grazing conditions (95% and

maximum canopy light interception during regrowth –LI0�95 and LIMax) and were allocated to experimental

units (850 m2 paddocks) according to a 2 9 2 factorial

arrangement in a randomized complete block design,

with four replications. Measurements were taken from

January 2011 to April 2012. The post-grazing height

treatments affected the tiller population stability

index, but did not influence the pattern of tussock dis-

tribution. On the other hand, the different grazing fre-

quencies (targets of LI pre-grazing) altered the pattern

of tussock distribution and the proportion of bare

ground. In general, the tiller population stability index

and frequency of tussocks were higher and the fre-

quency of bare ground lower on swards managed with

the LI0�95 target relative to those managed with the

LIMax target, regardless of the post-grazing height

used, indicating a larger soil occupation capacity of

plants under the more frequent defoliation regime.

Such responses were associated with larger population

of basal tillers and highlight the importance of tiller

category and perennation pathway in defining pat-

terns of plant growth and tussock distribution.

Keywords: canopy light interception, frequency of

defoliation, tiller population stability index, sward

surface height, tillering

Introduction

Tall-tufted tussock-forming species represent the main

growth form among the tropical grasses with higher

potential for herbage production utilized in South

America. However, knowledge on how environmental

factors and management affect the horizontal struc-

ture and lateral expansion of tussocks or how grazing

affects the soil occupation capacity of those plants is

scarce. This growth form is characterized by the com-

pact arrangement of genetically identical tillers (Briske

and Butler, 1989). These tillers, grouped into anatomi-

cally connected subunits independent in terms of

carbon and nutrients assimilation and allocation

(including hormones), form clusters known as inte-

grated physiological units (IPUs) (Watson, 1986).

A large number of IPUs, when associated with

increased basal area of tussocks, have been demon-

strated to enhance the competitive ability of Schizachy-

rium scoparium (Michx. Nash) (Derner et al., 2012).

Ryel et al. (1994) showed that up to 50% of the sward

carbon acquisition potential might be lost when plants

are organized in tussocks relative to when they are

evenly distributed in the area, mainly due to within-

tussock shading, a condition that is not fully com-

pensated by side lighting (Caldwell et al., 1983). In

this sense, the physiological integration influences the

capacity of tall-tufted tussock-forming species to

exploit resources within their environment and inter-

act with neighbours (Watson, 1986; Derner et al.,

2012), affecting their pattern of soil occupation and

distribution in the area.

According to Castillo et al. (2003), maintenance of

small tussocks associated with high tiller population

density and short distances between tussocks favours

Correspondence to: S. C. da Silva, ESALQ/USP, Av. P�aduaDias, 11, C.P. 09 – 13418-900 - Piracicaba, S~ao Paulo,

Brazil.

E-mail: [email protected]

Received 30 January 2014; revised 11 June 2014

doi: 10.1111/gfs.12137 © 2014 John Wiley & Sons Ltd. Grass and Forage Science 1

Grass and Forage Science The Journal of the British Grassland Society The Official Journal of the European Grassland Federation

the occupation of the area by plants and prevents the

occurrence of weeds in Spartina densiflora Brongn

swards. In this way, tillering is an important compo-

nent of tussock growth and expansion, determining

the efficiency of soil surface occupation through varia-

tion in the frequencies of colonized and bare ground

areas in the pasture. Grazing, through leaf area

removal, alters the quantity and the quality of the

light reaching the sward base (Deregibus et al., 1985),

interfering with the dynamics of tiller replacement

(Pereira et al., 2013b) and affecting the balance

between the processes of tiller appearance and death,

which are represented by the tiller population stability

index (Bahmani et al., 2003) and have consequences

for sward structure and herbage accumulation (Pereira

et al., 2013a).

In elephant grass cv. Napier (Pennisetum purpureum

Schum. cv. Napier), defoliation frequency did not

change total tiller population density, but altered the

proportion of basal and aerial tillers (Pereira et al.,

2013a,b). Swards subjected to more frequent defolia-

tion usually show a larger contribution of basal tillers,

while those subjected to less frequent defoliation show

a larger number of aerial tillers per support unit (basal

tillers + decapitated tillers), an analogous growth strat-

egy to that associated with the increase in tiller num-

ber per IPU (Derner et al., 2012). As elephant grass is

a tall-tufted species, aerial tillers do not develop roots

and therefore rely on the parent tiller for support and

water and nutrient uptake (Briske, 1986). So, the

increase in the proportion of aerial tillers in the tiller

population may contribute to larger occupation of lat-

eral spaces along the vertical profile of the sward

(Stuefer, 1998) – aerial space between tussocks – rela-

tive to the horizontal occupation of the soil surface.

On the other hand, basal tillers would be responsible

for increasing the basal area of tussocks and reducing

the sharing of assimilates and nutrients among tillers

within an IPU, promoting more uniform distribution

of tillers within tussocks and of tussocks in the area, a

condition that would favour carbon acquisition (Ryel

et al., 1994) and could augment the growth potential

of plants. So, management strategies that favour aerial

tillering in tall-tufted tussock-forming species could

adversely affect the dynamics of tussock growth and

expansion, the capacity of plants to exploit soil surface

and increase the risk of erosion and invasion by

weeds.

Elephant grass is a bunchgrass with erect stems

forming well-defined tussocks (Alcantara and Bufarah,

1983). It is characterized by a high potential for herb-

age production and comprises a significant source of

forage for domestic herbivores in tropical areas of

South America, particularly in intensive dairying sys-

tems. The abundant tillering and significant participa-

tion of aerial tillers in tiller population (Corsi et al.,

1996) suggest that grazing management strongly

affects the growth dynamics and plant’s capacity to

colonize the area, as it affects the proportion of basal

and aerial tillers (Pereira et al., 2013a). If the soil occu-

pation capacity via lateral expansion of tussocks is

related to basal tillering, increases in the proportion of

aerial tillers in a tiller population in response to defoli-

ation may cause negative impacts on the expansion

and growth of tussocks and, consequently, on the pro-

portion of the soil surface occupied by plants. Against

that background, the objective of this experiment was

to study the influence of frequency and severity of

defoliation on both tiller population stability index

and patterns of soil surface occupation of the peren-

nial tall-tufted grass Pennisetum purpureum Schum. cv.

Napier.

Material and methods

Experimental site and period

The experiment was carried out at E.S.A. ‘Luiz de

Queiroz’ (ESALQ), University of S~ao Paulo, Piracicaba,

SP, Brazil (22°430S, 47°250W and 554 m asl) on an ele-

phant grass pasture (Pennisetum purpureum Schum. cv.

Napier) established in 1970 on a Eutroferric Red Nito-

sol. Since its establishment, the area has been inter-

mittently grazed by dairy cows. The average soil

chemical characteristics of the 0- to 20-cm layer were

pH CaCl2 = 5�9; organic matter = 54�9 g dm�1; P (ion-

exchange resin extraction method) = 29�4 mg dm�3;

Ca = 117�5 mmolc dm�3; Mg = 35�6 mmolc dm�3;

K = 3�4 mmolc dm�3; H + A = 53�9 mmolc dm�3;

sum of bases = 148�9 mmolc dm�3; cation exchange

capacity = 190�6 mmolc dm�3; and base satura-

tion = 82%. The climate is subtropical with dry win-

ters. Average annual rainfall is 1328 mm (CEPAGRI,

2012). The average mean air temperatures during the

experimental period (January 2011 to April 2012) fol-

lowed the historical pattern of variation (1917–2012),with lower mean temperatures recorded in June 2011

(16�3°C) and higher mean temperatures recorded in

February 2012 (25�7°C).The experiment began in January 2011, after an

initial grazing of the experimental paddocks followed

by mowing to 35 cm, and was conducted until April

2012. Measurements were grouped into six seasons

of the year defined as follows: summer I (January to

March, 2011), autumn (April to June, 2011), winter

(July to September, 2011), early spring (October to

mid-November, 2011), late spring (mid-November to

December, 2011) and summer II (January to April,

2012). Summer I was characterized by greater rain-

fall than summer II. The rainfall began earlier (early

© 2014 John Wiley & Sons Ltd. Grass and Forage Science

2 L. E. T. Pereira et al.

spring) relative to the historical average (late spring),

with lower precipitation during February and

March 2012 (summer II). There were three periods

of soil water deficit: in February 2011 (summer I),

from July to the beginning of October 2011 (early

spring) and from March to mid April 2012 (late

summer II).

Experimental treatments and monitoring ofexperimental conditions

The experimental treatments resulted from the combi-

nation of two post-grazing heights (35 and 45 cm)

and two pre-grazing conditions (95% and maximum

canopy light interception during regrowth – LI0�95 and

LIMax) and were allocated to experimental units

(850 m2 paddocks) according to a 2 9 2 factorial

arrangement and a randomized complete block design,

with four replications.

During the first pasture-growing season (January

to April 2011), a total of 250 kg N ha�1 was applied,

with 80 kg ha�1 after mowing and 170 kg ha�1 in

split applications during the remainder of the season.

During the second pasture-growing season (November

2011 to March 2012), a total of 300 kg N ha�1 was

applied in split applications throughout the season.

Fertilizer was applied following grazing using a com-

mercial N:P:K formula (20:0:8). Because the grazing

interval was not constant (a consequence of the way

the experimental treatments were defined), the total

amount of N to be applied was divided by the number

of days of each growing season and a daily rate calcu-

lated. The amount of nitrogen in each application was

adjusted according to the length of the rest period of

each experimental unit.

Light interception was monitored using a LAI 2000

canopy analyser (LI-COR, Lincoln, NE, USA). The

readings were taken once per week during regrowth

until the value of 90% was reached, after which mea-

surements were collected every two days to ensure that

the LI0�95 and LIMax targets were precisely achieved.

Maximum canopy light interception (LIMax) was con-

sidered reached when the recorded value did not

change during two consecutive measurements, which

corresponded to an average of 98%. Actual values of

canopy light interception for the treatments LI0�95/35,LI0�95/45, LIMax/35 and LIMax/45 were 95�4, 95�6, 98�4and 98�2% respectively. Readings were taken from six

sampling areas per paddock which were considered to

be representative of sward condition at the time of

sampling based on visual assessment of height and

herbage mass. In each sampling area, one reading was

taken above the canopy and five at ground level, total-

ling six readings above the canopy and thirty at ground

level per experimental unit.

Sward height was monitored by taking 80 system-

atic readings along four transect lines (twenty readings

per line) in each paddock using a stick graduated in

centimetres. Sward height was measured from ground

level to the top of the leaf horizon, even at times of

the year when plants were reproductive and produced

taller flowering stems (Carnevalli et al., 2006; Da Silva

et al., 2009). Post-grazing heights were as planned on

the paddocks managed with the LI0�95 target but

remained above the target for the LIMax condition,

consequence of excessive stem elongation under those

conditions (detailed information in Pereira et al.,

2013a,b). Grazing was executed by lactating or non-

lactating mature dairy cows, or dairy heifers, using the

mob grazing method (Gildersleeve et al., 1987). The

number of animals grazing each paddock was adjusted

to ensure achievement of post-grazing height targets

within 10–12 h (day grazing only). The average graz-

ing interval during the experiment was 29, 28, 40 and

38 d for treatments LI0.95/35, LI0.95/45, LIMax/35 and

LIMax/45 respectively.

Measurements

Measurements of frequency of tussocks, bare ground

and weeds as well as tussock perimeter were carried

out at the beginning of a regrowth cycle once every

season of the year using a 20-m-long nylon string

with marks every two metres. The string was placed

along four transect lines per paddock in a zigzag for-

mat. At each point, readings were taken to identify

what was seen under each mark into the following

categories: tussock, bare ground or weeds. A total of

forty points per paddock were evaluated. Tussocks in

each point had their perimeter measured at ground

level using a metric tape. Frequency of tussocks, bare

ground and weeds was calculated as proportion of the

total number of reading points. To describe patterns of

variation in tussock size and distribution, the average

perimeter and the corresponding coefficient of varia-

tion were calculated.

The rates of tiller appearance, death and survival,

necessary to calculate the sward stability index used in

this study, were generated using data from a detailed

set of tillering dynamics measurements of a concomi-

tant experiment in the same area (Pereira et al., 2013a,

b). These were performed using plastic coloured rings

to tag tillers in each of three tussocks per paddock. Tus-

socks were chosen in areas that represented the aver-

age sward condition at the time of tagging (visual

assessment of sward height, herbage mass, tussock size

and distribution) and remained the same throughout

the experiment. Between February and March 2011,

time of the first tagging procedure, all tillers on each

chosen tussock were counted and received a white


Tussocks distribution in elephant grass 3

plastic ring. After every grazing, surviving tillers were

counted, including aerial tillers (non-rooted tillers orig-

inated from axillary buds), new tillers counted and

tagged with a different colour, and dead ones counted

and their tags removed. Vegetative tillers were consid-

ered dead when they were brown, withered and had

no live leaves inside the sheath. Reproductive tillers

were classified as dead when the defoliated stem was

brown, sapless and did not support new daughter til-

lers. Missing tillers were also classified as dead.

For each tiller generation (coloured ring), calcula-

tions were performed to estimate the relative rates of

tiller appearance (TAR = number of new tillers on tag-

ging date t/total number of tagged tillers on date t-1)

and death (TDR = number of dead tillers on tagging

date t/total number of tagged tillers on date t-1). The

combined effect of tiller appearance and death on tiller

population was evaluated using a tiller population

change index (stability index – P1/P0), used by Bah-

mani et al. (2003) and calculated as P1/P0 = TSR 9 (1+TAR), where P0 is the tiller population at the begin-

ning of a given observation period and P1 the total

population at the end of that period. TSR and TAR are

the relative rates of tiller survival and appearance of

the total tiller population respectively. Because the

duration and dates of the grazing cycles for individual

paddocks were variable, a consequence of the way tar-

gets pre-grazing were defined (95 and maximum can-

opy light interception during regrowth), observation

periods varied within and between treatments. Values

of TAR and TDR were then recalculated for a standard

30-day observation period (tiller/tiller, 30-day inter-

val) assuming a constant daily survival rate over the

interval, as described by Bahmani et al. (2003) for

dealing with a similar situation, and TSR calculated as

1-TDR. The above equation requires the assumption

that tillers will not produce daughter tillers in the first

month of their life. Population density is constant

when P1/P0 = 1 and would be decreasing or increasing

when P1/P0 is smaller or larger than one respectively.

Statistical analysis

Analysis of variance was performed using the Mixed

Procedure of SAS� (Statistical Analysis System), version

8�2 for Windows�. The choice of the covariance

matrix was made using the Akaike Information Crite-

rion (AIC) (Wolfinger, 1993), and the analysis per-

formed considering LI pre-grazing, post-grazing height,

season of the year and their interactions as fixed

effects and blocks as a random effect (Littell et al.,

1996). Season of the year was treated as a repeated

measure, and when appropriate, treatment means

were calculated using the ‘LSMEANS’ statement and

comparisons made using Student’s t-test at 5% proba-

bility. Only the significant effects are reported in

results. Because tiller population can vary throughout

the year with seasonal fluctuations in incident radia-

tion and herbage mass (Matthew et al., 1995; Bahmani

et al., 2003) independently of treatment effects and it

can affect tussock size and distribution, a principal

components analysis (PCA) was performed on the cor-

relation matrix of a data set comprising sward height

pre-grazing; tiller population stability index; total,

basal and aerial tiller population; tussock perimeter;

frequency of tussocks and of bare ground to gain fur-

ther insight and identify possible functional associa-

tions between them. Coefficients for the PCs were

then submitted to ANOVA to help to describe and to

interpret treatment effects (Jolliffe, 1986).

Results

Proportion of basal and aerial tillers and tillerpopulation stability index

The proportion between basal and aerial tillers in the

tiller population was affected by the LI pre-grazing 9

season of the year interaction (P < 0�0001). Differencesbetween means of LI pre-grazing treatments were

recorded from autumn onwards, with swards managed

with the LI0�95 target showing larger proportions of

basal tillers than those managed with the LIMax target.

With this treatment, aerial tillers accounted for more

than 73% of total population of tillers (Figure 1). The

tiller population stability index (SI) varied with LI pre-

grazing (P < 0�0001), post-grazing height (P = 0�005),season of the year (P < 0�0001) and the LI pre-grazing

9 season of the year (P < 0�0001) as well as the post-

grazing height 9 season of the year (P = 0�0201) inter-actions. Swards managed with the LI0�95 target showed

SI values equal to and/or larger than 1 during summer

I, autumn, late spring and summer II. On the other

hand, swards managed with the LIMax target showed SI

values smaller than 1 throughout the experimental per-

iod, with lowest values recorded for both LI treatments

during winter (Figure 2a). For the post-grazing height

treatments, SI values remained smaller than 1 through-

out the experimental period, with lowest values

recorded during winter. Swards managed at 45 cm

showed larger SI than those managed at 35 cm during

summer I and summer II (Figure 2b).

Pre-grazing height and average tussockperimeter

The sward height pre-grazing varied with LI pre-graz-

ing (P < 0�0001) and season of the year (P < 0�0001).Values were larger on swards managed with the LIMax

target (126 cm) relative to those managed with the



LI0�95 target (84 cm). For both LI targets, lowest val-

ues were recorded during autumn and winter

(Table 1). The average tussock perimeter varied with

season of the year (P < 0�0001) and with the LI pre-

grazing 9 season of the year interaction (P = 0�0317).Values of tussock perimeter decreased as the experi-

ment progressed for both targets of LI pre-grazing,

with differences between them being recorded only

during winter, when swards managed with the LI0�95target showed larger tussock perimeter than those

managed with the LIMax target (Table 1). From the

beginning (summer I) to the end (summer II) of the

experiment, there was a reduction in tussock perime-

ter (Table 1) and variability (range of variation) for

both LI targets, a pattern that was reinforced by the

variations in skewness and kurtosis (Table 2). Skew-

ness was positive throughout the experimental period

that, in association with high kurtosis values, indicates

higher frequency of larger tussocks (mean value larger

than the median) at the beginning of the experimen-

tal period. As the experiment progressed, skewness

decreased, with lower values recorded in summer II

for both pre-grazing LI targets, indicating increased

frequency of smaller tussocks (smaller difference

between mean and median) and larger uniformity of

tussock distribution (Figure 3) among perimeter cate-

gories (kurtosis close to zero and smaller standard

deviation; Table 2).

Patterns of tussock distribution

The frequency of tussocks varied with LI pre-grazing

(P < 0�0001), season of the year (P = 0�0001) and

with the LI pre-grazing 9 season of the year

(P < 0�0001) and the post-grazing height x season of

the year (P = 0�0002) interactions. The frequency of

tussocks increased from summer I to summer II on

swards managed with the LI0�95 target, with highest

values recorded in early spring. On swards managed

with the LIMax target, the frequency of tussocks

decreased from the beginning of the experiment in

summer I, with lowest values recorded in late spring

followed by summer II. Differences between the tar-

gets of LI pre-grazing were recorded from early spring

onwards, when swards managed with the LI0�95 target

showing higher frequency of tussocks than those

managed with the LIMax target. Differences between

post-grazing height treatments were recorded only in

summer I, when swards managed at 35 cm showed

lower frequency of tussocks than those managed at

61·0

39·0

41·5

16·6 17·127·0 25·4

14·2

21·5 25·237·0

38·0 28·1

78·5 74·863·0 62·0

71·9

0·0

10·0

20·0

30·0

40·0

50·0

60·0

70·0

80·0

90·0

100·0

Porp

ortio

n of

tille

r cla

ss (%

)

P < 0·0001

58·5

83·4 82·973·0 74·6

85·8

0·010·020·030·040·050·060·070·080·090·0

100·0

Summer I Autumn Winter Early spring Late spring Summer II

Prop

ortio

n of

tille

r cla

ss (%

)

Season of the year

P < 0·0001

(a)

(b)Figure 1 Proportion (%) of basal

and aerial tillers in the tiller population

of elephant grass cv. Napier subjected

to strategies of rotational stocking

management characterized by the

targets of 95% (a) and maximum

canopy light interception (b) during

regrowth from January 2011 to April

2012. White and black bars represent

the proportions of aerial and basal

tillers respectively. P-values of LI

pre-grazing x season of the year

interaction are included. For both

variables n = 8, df = 69.



45 cm. The higher values of frequency of tussocks

were recorded during autumn, winter and early spring

for both the 35- and 45-cm targets (Table 3).

The frequency of bare ground varied with LI pre-graz-

ing (P < 0�0001) and with the LI pre-grazing 9 season of

the year (P < 0�0001) and the post-grazing height 9 sea-

son of the year (P = 0�0342) interactions. Swards man-

aged with the LI0�95 target showed a higher frequency of

bare ground during summer I relative to the remaining

seasons of the year, the difference disappearing from

autumn until the end of the experiment in summer II.

On the other hand, the frequency of bare ground

increased throughout the experiment on swards man-

aged with the LIMax target, with highest values recorded

in summer II. There was no difference between the

targets of LI pre-grazing during summer I and winter.

0·00

0·20

0·40

0·60

0·80

1·00

1·20

1·40

1·60

Popu

latio

n st

abili

ty in

dex

LI 0·95 LI MaxP < 0·0001

0·00

0·20

0·40

0·60

0·80

1·00

1·20

1·40

1·60

Summer I Autumn Winter Early spring Late spring Summer II

Popu

latio

n st

abili

ty in

dex

Season of year

35 cm 45 cmP = 0·02

(a)

(b)

Figure 2 Tiller population stability

index (Pf/Pi) in elephant grass cv.

Napier subjected to strategies of

rotational stocking management

characterized by combinations of two

post-grazing heights (35 and 45 cm)

and two pre-grazing conditions (95%

and maximum canopy light inter-

ception during regrowth) from January

2011 to April 2012. P-values of LI pre-

grazing x season of the year (a) and

post-grazing height x season of the

year (b) interactions are included.

Dotted line represents Pf/Pi = 1. Bars

represent mean � s.e.m., n = 8, df =69.

Table 1 Pre-grazing sward height (cm) and tussock perimeter (m) of elephant grass cv. Napier subjected to strategies of rota-

tional stocking management characterized by the pre-grazing conditions of 95% and maximum canopy light interception during

regrowth (LI0�95 and LIMax) from January 2011 to April 2012. F- and P-values of LI pre-grazing x season of the year interaction

are included. For both variables, mean � s.e.m, n = 8 and df = 68.

Season of year

Pre-grazing height (cm) Tussock perimeter (m)

LI0�95 LIMax LI0�95 LIMax

Summer I 94�7 � 4�81 136�9 � 4�81 3�21 � 0�069 2�89 � 0�069Autumn 81�9 � 4�81 121�7 � 5�16 2�85 � 0�151 2�59 � 0�151Winter 71�4 � 4�81 102�4 � 4�81 2�78 � 0�121a 2�39 � 0�121 b

Early spring 86�7 � 4�81 123�4 � 4�81 2�74 � 0�145 2�80 � 0�145Late spring 86�8 � 4�81 135�6 � 4�81 2�39 � 0�125 2�33 � 0�125Summer II 82�4 � 4�81 135�9 � 4�81 2�09 � 0�146 2�34 � 0�146F-value 1�66 2�62P-value 0�15 0�0317Means followed by different lower case letters in rows are different.



During the remaining seasons of the year, the frequency

of bare ground was lower on swards managed with the

LI0�95 target. In relation to the post-grazing height treat-

ments, there was no difference between seasons of the

year for swards managed at 45 cm. For swards managed

at 35 cm, the frequency of bare ground was highest in

summer I and lowest in winter, with no difference

between the remaining seasons of the year. Differences

between post-grazing height treatments were detected

only during summer I, when higher values were

recorded on swards managed at 35 cm (Table 3).

The frequency of weeds varied with season of the

year (P = 0�0139) and with the LI pre-grazing 9 sea-

son of the year interaction (P = 0�0139). Differences

between LI pre-grazing treatments were recorded only

during autumn, when swards managed with the LI0�95target showed higher values than those managed with

the LIMax target. The frequency of weeds was smaller

Table 2 Summary statistics for tussock perimeter (cm) in elephant grass cv. Napier subjected to strategies of rotational stocking

management characterized by the pre-grazing conditions of 95% and maximum canopy light interception during regrowth (LI0�95and LIMax) from January 2011 to April 2012.

Season of the year

Pre-grazing

target

Number of

observations Skewness Kurtosis CV (%)

STD

deviation*

Summer I LI0�95 169 0�93 1�23 49�5 1�77LIMax 180 1�34 2�68 47�9 1�54

Autumn LI0�95 186 0�55 0�58 38�1 1�11LIMax 178 0�59 2�50 37�0 0�97

Winter LI0�95 189 1�14 2�27 46�9 1�32LIMax 178 0�29 0�54 39�7 0�95

Early spring LI0�95 208 0�37 �0�01 33�7 0�92LIMax 149 0�91 1�96 36�6 1�00

Late spring LI0�95 189 0�79 2�19 37�2 0�90LIMax 151 0�31 �0�19 31�4 0�74

Summer II LI0�95 188 0�23 �0�65 37�2 0�77LIMax 144 0�29 �0�24 30�6 0�70

*Standard deviation.

0

10

20

30

40

50

60

Freq

uenc

y (%

)

0

10

20

30

40

50

60

0

10

20

30

40

50

60

0

10

20

30

40

50

60

0 to 1 1 to 2 2 to 3 3 to 4 4 to 5 5 to 6 6 to 7 > 7

Freq

uenc

y (%

)

Caterogies of tussock perimeter (m)Caterogies of tussock perimeter (m) Caterogies of tussock perimeter (m)

0

10

20

30

40

50

60

0 to 1 1 to 2 2 to 3 3 to 4 4 to 5 5 to 6 6 to 7 > 70

10

20

30

40

50

60

0 to 1 1 to 2 2 to 3 3 to 4 4 to 5 5 to 6 6 to 7 > 7

(a) (b) (c)

(d) (e) (f)

Figure 3 Frequency distribution of tussock perimeter categories (cm) in elephant grass cv. Napier subjected to strategies of

rotational stocking management characterized by the pre-grazing conditions of 95% and maximum canopy light interception dur-

ing regrowth (LI0�95 and LIMax) throughout the year: (a) summer I, (b) autumn, (c) winter, (d) early spring, (e) late spring and (f)

summer II. White and grey bars represent LI0�95 and LIMax respectively.



than 10% regardless of targets of LI pre-grazing and

season of the year, and for that reason, it will not be

discussed in this study.

Principal components analysis

Although eight PCs were available from the analysis,

the first three explained 77�4% of the total variation

in the data set comprising sward height pre-grazing;

tiller population stability index; total, basal and aerial

tiller population; tussock perimeter; frequency of tus-

socks and frequency of bare ground, and were associ-

ated with treatment and season of the year effects

(Figure 4 and Table 4).

The first PC (PC1) explained 39�6% of the total

variation and was related to a plant size 9 density

relationship characterized by a contrast of sward

height pre-grazing and frequency of bare ground on

one side and tiller population density (total and basal

tillers) and frequency of tussocks in the other. The

magnitude of changes associated with that mechanism,

however, is affected by interactions between season of

the year and LI pre-grazing (P < 0�0001) and post-graz-

ing height (P = 0�0124). In general, swards managed

with the LI0�95 target showed larger coefficients than

those managed with the LIMax target throughout the

experiment. Differences between coefficients for post-

grazing heights were detected only during summer I,

when swards managed at 45 cm showed larger PC coef-

ficients than those managed at 35 cm.

The second PC (PC2) explained another 25�5% of

the total variation and was related to tiller categories

and tiller population. There was a contrast of total tiller

and aerial tiller population density on one side and tiller

population density of basal tillers and population stabil-

ity index in the other. Coefficients were influenced

by LI pre-grazing (P < 0�0001), post-grazing height

(P < 0�0001) and season of the year (P < 0�0001), with

larger values observed for swards managed with the

LIMax target, at 35 cm and during winter respectively.

During the experimental period, lower coefficients were

observed during summer I and late spring, with interme-

diate values during the remaining seasons of the year.

The third PC (PC3) explained an additional 12�2%of the total variation and indicated a relationship

between tiller population composition and tussock

expansion characterized by a contrast of population

density of basal tillers and population stability index on

one side and sward height pre-grazing and tussock

perimeter on the other. This was associated with a LI

pre-grazing 9 season of the year interaction (P =

0�0030). During summer I and summer II, coefficients

were larger for swards managed with the LIMax relative

to those managed with the LI0�95 target.

Table 3 Frequency of tussocks and of bare ground (mean � s.e.m) in elephant grass cv. Napier subjected to strategies of

rotational stocking management characterized by combinations of two post-grazing heights (35 and 45 cm) and two pre-grazing

conditions (95% and maximum canopy light interception during regrowth – LI0�95 and LIMax) from January 2011 to April 2012.

F- and P-values of LI pre-grazing 9 season of the year interaction are included. For both variables, n = 8 and df = 68.

Season of the year

Pre-grazing targets Post-grazing targets

LI0�95 LIMax 35 cm 45 cm

Frequency of tussocks (%)

Summer I 51�0 � 1�65 54�7 � 1�65 48�0 � 1�65b 57�7 � 1�65aAutumn 58�1 � 3�28 55�3 � 3�28 56�2 � 3�28 57�2 � 3�28Winter 58�4 � 2�21 55�6 � 2�21 56�6 � 2�21 57�5 � 2�21Early spring 65�0 � 1�72a 53�5 � 1�81b 58�4 � 1�72 60�0 � 1�81Late spring 59�1 � 2�34a 47�2 � 2�34b 53�7 � 2�34 52�5 � 2�34Summer II 60�2 � 1�83a 48�8 � 1�83b 55�6 � 1�83 53�3 � 1�83F-value 10�47 5�57P-value <0�0001 0�0002

Frequency of bare ground (%)

Summer I 43�0 � 2�00 38�1 � 2�00 44�9 � 2�00a 36�3 � 2�00bAutumn 31�9 � 2�39 b 40�0 � 2�39 a 34�7 � 2�39 37�2 � 2�39Winter 34�1 � 2�43 38�1 � 2�43 33�4 � 2�43 38�7 � 2�43Early spring 31�2 � 1�88 b 43�3 � 1�88 a 36�6 � 1�88 38�0 � 1�88Late spring 35�3 � 2�15 b 45�0 � 2�15 a 39�4 � 2�15 40�9 � 2�15Summer II 31�7 � 1�88 b 46�6 � 1�88 a 38�5 � 1�88 39�9 � 1�88F-value 5�58 2�57P-value 0�0002 0�034Means followed by different lower case letters in rows are different.



Discussion

For tall-tufted tussock-forming grasses such as ele-

phant grass, the proportion of areas with tussocks and

those of bare ground may represent an important indi-

cator of the effect of grazing management on pasture

persistence and productivity. The results from this

experiment demonstrate that defoliation frequency

interfered with the proportion of tiller categories

(basal and aerial) in tiller population (Figure 1) and

the capacity of plants to use paddock area (Table 4

and Figure 4). For similar tussock perimeters

(Table 1), swards managed with the LI0�95 target

showed higher tiller population stability index from

early spring onwards (Figure 2), higher frequency of

tussocks and lower frequency of bare ground (Table 3)

relative to those managed with the LIMax target. These

differences in patterns of plant distribution between

the targets of LI pre-grazing demonstrate that duration

of the regrowth period (e.g. grazing interval) strongly

affects tiller replacement by modifying the perenna-

tion pathway with consequences for tussock’s

expansion and distribution and their capacity to use

paddock area, interfering with the stability of plant

population (PC1, PC2 and PC3; Table 4).

Swards managed with long regrowth intervals are

subjected to large variations in quantity and quality of

light within the canopy (Caldwell et al., 1983). This

results in stem elongation and excessively tall swards

(Table 1), interfering with tillering (Deregibus et al.,

1985) through reduction in activation of basal buds

and reduction in expansion and/or new tussock for-

mation (PC1 and PC3; Table 4). Taller plants with lar-

ger tussock size (PC3, Table 4), a condition normally

associated with higher herbage mass of swards man-

aged with the LIMax target (Pereira et al., 2013a),

result in a larger proportion of the paddock area with

bare ground, indicating that tussocks are further apart.

Increased distance between individuals in a plant com-

munity has been considered as evidence of competi-

tion for some kind of resource (Nobel, 1981). These

findings were corroborated by the results from the

PC1 that showed a contrast between sward pre-graz-

ing height/frequency of bare ground on one side and

population density of basal tillers/frequency of tus-

socks on the other, which is possibly a consequence of

interclonal interference (Briske and Butler, 1989)

triggered by intense competition for light in swards

managed with the LIMax target (Figure 4). They also

indicate that there may be a compensation mechanism

involving tussock size and tussock frequency

analogous to the tiller size x density compensation

commonly described in temperate (Matthew et al.,

1995; Sackville Hamilton et al., 1995) and tropical

(Sbrissia et al., 2001, 2003; Sbrissia and Da Silva,

2008) grazed swards.

In a concomitant experiment in the same experi-

mental area, Pereira et al. (2013a,b) demonstrated that

–5

–4

–3

–2

–1

0

1

2

3

4

5

–5 –4 –3 –2 –1 0 1 2 3 4 5

PC 2

(25·

5%)

PC 1 (39·6%)

Figure 4 Plot of coefficients for principal components 1 and 2 (PC1 and PC2) based on the correlation matrix for sward sur-

face height pre-grazing, tiller population of basal, aerial and total, tiller population stability index, tussock perimeter and frequency

of tussocks and of bare ground for elephant grass cv. Napier subjected to strategies of rotational stocking characterized by the

pre-grazing conditions of 95% and maximum canopy light interception during regrowth (LI0�95 and LIMax) management from Janu-

ary 2011 to April 2012. Open and closed symbols represent pre-grazing targets (LI0�95 and LIMax, respectively), while triangles

and circles represent post-grazing heights (35 and 45 cm respectively).



the regrowth period affected the growth pattern of

plants by changing the balance between tiller catego-

ries in tiller population. In that experiment, shorter re-

growth periods, represented by the LI0�95 target,

favoured larger density of basal tillers. Even though

the number of aerial tillers did not differ between tar-

gets of LI pre-grazing, the number of aerial tillers per

support unit (basal tillers + decapitated tillers) was lar-

ger on swards managed with longer regrowth inter-

vals, represented by the LIMax target. In that context,

swards managed with the LIMax target used a growth

strategy analogous to the one based on increasing the

number of tillers per IPU and not on increasing the

number of IPUs. Such a strategy is a consequence of

stem elongation and elevation of growing points into

the grazing horizon of the swards managed with the

LIMax target, as demonstrated by the taller pre-grazing

heights recorded for that management strategy

(Table 1; PC1 and PC3 of Table 4). Although the

increase in number of individuals per IPU may be an

effective growth strategy in the short term for maxi-

mizing the exploitation of light patches along the ver-

tical profile of the sward (Stuefer, 1998), it is not an

effective way for horizontal colonization of the soil

surface (Table 3) and maintenance of population sta-

bility (Figure 2). The reduction in frequency of tus-

socks and the increase in the frequency of bare

ground in swards managed with the LIMax target

(Table 3) are the result of the reduced tiller population

stability, which could possibly favour the beginning of

a degradation process under long regrowth interval

management conditions over long periods.

Growth strategies based on an increased number of

IPUs and/or increased number of individuals per IPU,

in addition to interfering with tussock size and soil

surface occupation, lead to modification of the canopy

light interception efficiency with potential impacts on

sward regrowth and herbage accumulation. The effi-

ciency of light interception and therefore its potential

for herbage production depend on the sward leaf area

index (LAI), but for a given LAI, it depends on the

spatial heterogeneity of the leaf area density. In this

sense, the frequency and size of tussocks and tiller cat-

egory (basal or aerial) may determine the formation of

leaf clusters, leading to high level of mutual shading

and intertussock areas with low leaf area density.

Oker-Blom and Kellom€aki (1983) demonstrated that

grouping of shoots decreased canopy light interception

substantially and that within-plant shading may be

considerably greater than between-plant shading.

Under those conditions, the light environment affect-

ing individual plants or tussocks is determined by the

structure of the tussock itself.

Throughout the experimental period, there was a

change in the horizontal structure of the swards char-

acterized by the reduction in average tussock perime-

ters (Table 1) and variability regardless of the grazing

management strategies used (Table 2; Figure 3). Dur-

ing summer I, 90% of the tussocks had perimeters

between 1�23 and 6�08 m, and standard deviations for

LI0�95 and LIMax (Table 2) were, respectively, 1�77 and

1�54. In summer II, the same proportion of the

tussocks had perimeters between 1�00 and 3�40 m,

and standard deviations decreased to 0�77 and 0�70 for

LI0�95 and LIMax, respectively, highlighting the reduc-

tion in the number of large tussocks and indicating

greater uniformity in tussock size as the experiment

progressed. According to Derner et al. (2012), the

number of tillers per IPU and the number of IPUs

comprising a tussock determine the size and architec-

ture of tussocks. With the increase in tussock size and

age, the IPUs become physically separated because of

the death and decomposition of older tiller genera-

tions, death of tillers in the centre (central die-back)

Table 4 Coefficients of principal components based on the

correlation matrix for sward surface height pre-grazing, tiller

population of basal (TPDb), aerial (TPDa) and total (TPDt),

tiller population stability index, tussock perimeter and fre-

quency of tussocks and of bare ground for elephant grass cv.

Napier subjected to strategies of rotational stocking manage-

ment characterized by combinations of two post-grazing

heights (35 and 45 cm) and two pre-grazing conditions (95%

and maximum canopy light interception during regrowth)

from January 2011 to April 2012.

Variables

Principal component

number*

PC1 PC2 PC3

Sward surface height �0�44 0�16 0�41TDPb 0�41 �0�74 �0�59TDPa 0�29 0�56 �0�04Perimeter 0�13 �0�22 0�71TDPt 0�38 0�50 �0�05Stability index 0�13 �0�53 �0�44Frequency of tussocks 0�43 �0�18 0�31Frequency of bare ground �0�43 0�19 �0�17Eigenvalue 3�16 2�04 0�98% of variation explained 39�6 25�5 12�2Level of significance (P-value)

LI pre-grazing (LI) <0�0001 <0�0001 <0�0001Post-grazing height (PGH) NS <0�0001 NS

Season of the year (S) <0�0001 <0�0001 <0�0001LI 9 PGH NS NS NS

LI 9 S <0�0001 NS 0�003PGH 9 S 0�012 NS NS

LI 9 PGH 9 S NS NS NS

NS, not significant. *Light and dark grey highlight contrasts

among variables in individual PCs.



(Wan and Sosebee, 2000) and increased recruitment

of tillers from the periphery (PC3, Table 4) of large

tussocks (Olson and Richards, 1988; Changgui and

Sosebee, 2002). This favours the physical separation of

the IPUs (Derner et al., 2012) and the formation of

new small tussocks (Figure 3). This also indicates a

period of adaptation of almost 8 months before swards

reached a stable structural condition in early spring

associated with the defoliation regimes used, an

important aspect to be taken into consideration when

planning and carrying out grazing experiments in field

conditions.

Targets of post-grazing height could not be met in

LIMax from the beginning of the experiment, with the

difficulty increasing as the experiment progressed. This

was also reported in an experiment using the same

experimental protocol with Mombac�a guinea grass

(Panicum maximum Jacq. cv. Mombac�a) managed with

the LIMax target (Carnevalli et al., 2006). After 95%

canopy light interception, competition for light within

the canopy increases and stem elongation is enhanced

relative to leaf elongation, resulting in increased sward

height pre-grazing (Table 1). The fact is usually associ-

ated with reduction in rates of leaf accumulation and

increase in rates of senescence (Da Silva et al., 2009).

The post-grazing height treatments influenced the til-

ler population stability index (Figire 2b), but did not

affect the pattern of tussock distribution. As a result,

definition of post-grazing targets for elephant grass cv.

Napier should also take into account characteristics of

ingestive behaviour of the grazing animals, as the

post-grazing condition could interfere with herbage

intake and animal performance in spite of the lack of

differences in the horizontal structure of swards.

Recent experimental evidence demonstrates that the

rate of herbage intake remains relatively stable from

the beginning of grazing until the removal of approxi-

mately 40–50% of sward pre-grazing height, but

declines sharply when stubble height is reduced fur-

ther (Carvalho et al., 2009; Fonseca et al., 2012).

The results from this experiment demonstrate that

grazing management alters, in the short term, the per-

ennation strategy used by elephant grass cv. Napier

(basal or aerial tillers) and interferes with the stability

of tiller population, the size and distribution of tus-

socks as well as the proportion of bare ground of the

paddocks. More frequent defoliation, represented by

the LI0�95 target, resulted in higher stability of tiller

population, higher frequency of smaller tussocks and

lower frequency of bare ground, regardless of post-

grazing height used (Figure 4), all of these responses

being associated with a high population of basal tillers.

Such a condition is suggestive of higher accumulation

of leafy herbage, with potential positive impacts to

animal performance and productivity.

Acknowledgments

To S~ao Paulo Research Foundation and National

Council for Technological and Scientific Development

for the sponsorship provided to develop the project,

and to Dr. Gilles Lemaire for the discussions and sug-

gestions to improve the manuscript.

References

ALC AN TARA P.B. and BUFARAH G. (1983) Plantas

forrageiras: gram�ıneas e leguminosas. [Forage plants: grasses

and legumes] 2nd edn. S~ao Paulo: Nobel.

BAHMAN I I., THOM E.R., MAT THEW C., HOOPER R.J. and

LEMA I R E G. (2003) Tiller dynamics of perennial

ryegrass cultivars derived from different New Zealand

ecotypes: effects of cultivar, season, nitrogen fertilizer,

and irrigation. Australian Journal of Agricultural Research,

54, 803–817.BR I SK E D.D. (1986) Plant response to defoliation:

morphological considerations and allocation priorities.

In: Joss P.J., Lynch P.W. and Williams O.B. (eds)

Rangelands: a resource under siege, pp. 425–427.Cambridge: Cambrige Univ. Press.

BR I S K E D.D. and BUT L ER J.T. (1989) Density-dependent

regulation of ramet populations within the bunchgrass

Schizachyrium scoparium: interclonal vs. intraclonal

interference. Journal of Ecology, 77, 963–974.CALDWEL L M.M., DEAN T.J., NOWAK R.S., DZUREC R.S.

and R I CHARDS J.H. (1983) Bunchgrass architecture,

light interception, and water use efficiency: assessment

by fiber optic point quadrats and gas exchange.

Oecologia, 59, 178–184.CARNEVA L L I R.A., DA S I L VA S.C., BUENO A.A.O.,

UEBE L E M.C., BUENO F.O., HODG SON J., S I L VA G.N.

and MORA I S J.P.G. (2006) Herbage production and

grazing losses in Panicum maximum cv. Mombac�a under

four grazing management. Tropical Grasslands, 40, 165–176.

CARVALHO P.C.F., TR I NDADE J.K., DA S I LVA S.C.,

BREMM C., MEZ ZA L I R A J.C., NAB I NG ER C., AMARAL

M.F., CARAS SA I I.J., MART I N S R.S., GENRO T.C.M.,

GONC�ALVE S E.N., AMARAL G.A., GONDA H.L., PO L I

C.H.E.C. and SANTO S D.T. (2009) Consumo de

forragem por animais em pastejo: analogias e

simulac�~oes em pastoreio rotativo. [Herbage intake by

grazing animals: analogies and simulations under

rotational stocking management]. In: Da Silva S.C.,

Pedreira C.G.S., Moura J.C. and Faria V.P. (eds)

Simp�osio sobre manejo da pastagem - Intensificac�~ao de

sistemas de produc�~ao animal em pasto, pp. 61–93.Piracicaba: FEALQ.

CAS T I L LO J.M., RUB I O -CASA L A.E., LUQUE T.,

F I GU EROA M.E. and J IMNEN E Z -N I E VA F.J. (2003)

Intratussock tiller distribution and biomass of Spartina

densiflora Brongn in a invaded salt marsh. Lagascalia, 23,

61–73.CEPAGRI (2012) Centro de pesquisas meteorol�ogicas e

clim�aticas aplicadas �a agricultura. [Center of Applied



Climatic and Meteorological Research in Agriculture]

Campinas: UNICAMP. Available at: http://www.cpa.

unicamp.br/outras-informacoes/clima_muni_436.html

(accessed 27 December 2012).

CHANGGU I W. and SOSEBE E R.E. (2002) Tiller

recruitment and mortality in the dryland bunch grass

Eragrostis curvula as affected by defoliation intensity.

Journal of Arid Environments, 51, 577–585.CORS I M., DA S I L VA S.C. and FAR I A V.P. (1996)

Princ�ıpios de manejo do Capim-Elefante sob pastejo.

[Grazing management principles of elephant grass]. In:

Peixoto A.M., Moura J.C. and Faria V.P. (eds) Pastagens

de Capim-Elefante: utilizac�~ao intensiva, pp. 51–70.Piracicaba: FEALQ.

DA S I LVA S.C., BUENO A.A.O., CARNEVA L L I R.A.,

UEBE L E M.C., BUENO F.O., HODG SON J., MAT THEW

C., ARNOLD J.C. and MORA I S J.P.G. (2009) Sward

structural characteristics and herbage accumulation of

Panicum maximum cv. Mombac�a subject to rotational

stocking managements. Scientia Agricola, 66, 8–19.DEREG I BU S V.A., S �ANCHE Z R.A., CASA L J.J. and TRL I CA

M.J. (1985) Tillering responses to enrichment of red-

light beneath the canopy in humid natural grassland.

Journal of Applied Ecology, 22, 199–206.DERNER J.D., BR I S K E D.D. and POL LE Y H.W. (2012)

Tiller organization within the tussock grass

Schizachyrium scoparium: a field assessment of

competition–cooperation tradeoffs. Botany, 90, 669–677.FON SECA L., MEZ ZA L I RA J.C., BREMM C., F I L HO R.S.A.,

GONDA H.L. and CARVALHO P.C.F. (2012)

Management targets for maximizing the short-term

herbage intake rate of cattle grazing in Sorghum bicolor.

Livestock Science, 145, 205–211.G I LD ER S L E EV E R.R., OCUMPAUGH W.R., QUE S ENBERR Y

K.H. and MOORE J.E. (1987) Mob-grazing of

morphologically different Aeschynomene species. Tropical

Grasslands, 21, 123–132.JO L L I F F E I.T. (1986) Principal component analysis 1st edn.

New York: Springer-Verlag.

L I T T E L L R.C., MI L L I K EN G.A., S TROUP W.W. and

WOLF I NG ER R.D. (1996) SAS system for mixed models.

Cary, NC: SAS Institute.

MAT TH EW C., LEMA I R E G., SACKV I L L E HAM I L TON N.R.

and HERNANDE Z GARAY A. (1995) A modified self-

thinning equation to describe size/density relationships

for defoliated swards. Annals of Botany, 76, 579–587.NOBE L P.S. (1981) Spacing and transpiration of various

sized clumps of a desert grass Hilaria rigida. Journal of

Ecology, 69, 735–742.OKER -BLOM P. and KEL LOM €AK I S. (1983) Effect of

grouping of foliage on the within-stand and within-

crown light regime: comparison of random and

grouping canopy models. Agricultural Meteorology, 28,

143–155.OL SON B.E. and R ICHARDS J.H. (1988) Spatial

arrangement of tiller replacement in Agropyron

desertorum following grazing. Oecologia, 76, 7–10.PERE I R A L.E.T., PA I VA A.J., DA S I L VA S.C. and

GEREM I A E.V. (2013a) Components of herbage

accumulation in elephant grass cvar Napier subjected to

strategies of intermittent stocking management. Journal

of Agricultural Science, First view article, 1–13.PERE I R A L.E.T., PA I VA A.J., GEREM IA E.V. and DA

S I L VA S.C. (2013b) Regrowth patterns of elephant grass

(Pennisetum purpureum Schum.) subjected to strategies

of intermittent stocking management. Grass and Forage

Science, Early view article, 1–10.RYE L R.J., BEY SCH LAG W. and CALDWEL L M.M. (1994)

Light field heterogeneity among tussock grasses:

theoretical considerations of light harvesting and

seedling establishment in tussocks and uniform tiller

distributions. Oecologia, 98, 241–246.SACKV I L L E HAM I L TON N.R., MAT THEW C. and

LEMA I R E G. (1995) In defence of the -3/2 boundary

rule: a re-evaluation of self-thinning concepts and

status. Annals of Botany, 76, 569–577.SBR I S S I A A.F. and DA S I LVA S.C. (2008) Tiller size/

density compensation in marandu palisadegrass swards.

Revista Brasileira de Zootecnia, 37, 35–47.SBR I S S I A A.F., DA S I LVA S.C., CARVA LHO C.A.B.,

CARNEVA L L I R.A., P I N TO L.F.M., FAGUNDE S J.L. and

PEDRE I RA C.G.S. (2001) Tiller size/population density

compensation in Coastcross grazed swards. Scientia

Agricola, 58, 655–665.SBR I S S I A A.F., DA S I LVA S.C., MAT THEW C., CARVALHO

C.A.B., CARNEVA L L I R.A., P I N TO L.F.M., FAGUNDE S

J.L. and PEDRE I RA C.G.S. (2003) Tiller size/density

compensation in grazed Tifton 85 Bermudagrass swards.

Pesquisa Agropecu�aria Brasileira, 38, 1459–1468.S TUE F ER J.F. (1998) Two types of division of labour in

clonal plants: benefits, costs and constraints. Perspective

in Plant Ecology, Evolution and Systematics, 1, 47–60.WAN C. and SOSEB E E R.E. (2000) Central dieback of the

dryland bunchgrass Eragrostis curvula (weeping

lovegrass) re-examined: the experimental clearance of

tussock centres. Journal of Arid Environments, 46, 69–78.WATSON M.A. (1986) Integrated physiological units in

plants. Tree, 1, 119–123.WOLF I NG ER R.D. (1993) Covariance structure selection in

general mixed models. Communications in Statistics

Simulation and Computation, 22, 1079–1106.



grazing management and tussock distribution in elephant grass

Documents