44 Z. Woldu, M.A. Mohammed Saleem Agriculture, Ecosystems and Environment 79 2000 43–52
about 20 of the agricultural gross domestic product of the country Mengistu, 1997. The livestock are
herded together and grazed on communal pastures, private land and stables depending on the time and
season of the year. Cattle are the most important live- stock species in the Ethiopian agriculture and they
are kept mainly for traction and milk production. Grazing is being expanded to very steep slopes and
marginal lands, as more land is cultivated to compen- sate for the diminishing soil fertility and to meet the
ever-increasing food demands.
The species composition and productivity of the pasture of the common grazing lands are highly influ-
enced by the species of the livestock, the intensity of grazing, climatic and edaphic factors. The seasonal
stocking rate on the central plateau of Ethiopia is very high for the ecological carrying capacity of the
grassland Woldu, 1986. The grassland communities are predominantly of Pennisetum sphacelatum–
Commelina africana type and could develop to Andro- pogon abyssinicus–Hyparrhenia arrhenobasis type if
grazing intensity is relaxed Woldu, 1986.
Grasslands in general are stressful environments. Defoliation, uprooting, trampling and desiccation are
the important stresses in grasslands under high graz- ing pressure. The role of livestock on the grassland
vegetation can be seen from two main perspectives. The most obvious one is the influence on the species
composition and reduction in the above and below ground biomass. The less obvious but equally im-
portant is the maintenance and distribution of the biodiversity. Free grazing livestock move to any ac-
cessible site and their feeding behaviour or foraging choice may generate a pattern of association between
the plant species and the livestock. This relationship helps in seed dispersal which may be manifested in
spatial and temporal differences in ground cover and patchiness. Seed morphology may influence whether
seeds survive the mastication and the digestive en- zymes of the ingesting animal. Those that survive
through the animal gut germinate faster as dormancy is broken by the abrasion, scarification and wetting of
the digestive system of the animal. On the other hand, seeds retained in the animal gut for a long period may
be induced to germinate and be killed while in the gut Janzen, 1982a, b. Deposition of seeds with manure
gives a better chance of survival since a source of ni- trogen is readily available for early seedling growth. It
appears, therefore, that the animal gut filters out those species by enhancing their germination while in the
gut and by providing favourable conditions when they are released. It is with this understanding that an ex-
periment was designed to study the species composi- tion of manure seed bank at different times of the year
and compare these with the species composition of natural grasslands under different grazing pressures.
The specific objectives of this study were 1 to de- termine the species composition and richness of germi-
nating seeds in manure seed bank, 2 to compare the species composition of manure seed bank and those of
the natural grassland under different grazing pressure, 3 to assess the role of livestock in the maintenance
of the biodiversity of the grassland vegetation through the spatial transfer of seeds in their manure, and 4
to assess the use of manure in improving soil fertility.
2. Materials and methods
2.1. Experimental site Two complementing experiments were conducted at
Ghinchi Research Station, a local research base, where the International Livestock Research Institute ILRI
is working with a consortium of institutions on natural resource management issues. Ghinchi is located in the
Ethiopian highlands at 2290 m asl 80 km west of Addis Ababa 9
◦
02
′
N; 38
◦
07
′
E. The experiments include manure seed bank pot trials and grazing experiments
in six different sites. 2.2. Manure seed bank pot trials
For the manure trials, the cattle herd of two farm- ers were monitored as they grazed different parts of
the landscape at different times of the year. Manure was collected each day from the grazing areas, and a
composite sample was taken from the weekly collec- tions and air dried. Five replicates were made from
each composite sample, and 123.3g of the sample was spread over a pot 18.0 cm — top diameter, 11.5 cm —
bottom diameter and 15.5 cm — height half filled with 1.90 kg of sterilised sand. The plant species growing in
the pots were recorded until all seedlings had emerged. The seedlings were identified to the species level as
much as possible. The manure experiment in the pots
Z. Woldu, M.A. Mohammed Saleem Agriculture, Ecosystems and Environment 79 2000 43–52 45
was run for two complete years, August 1996–July 1997 and August 1997–July 1998. The plant species
and other data for each replicate of the week were entered in access database. Relevant information was
extracted using the structured query language.
2.3. Grazing experiments To study the influence of grazing on the grass-
land vegetation, six sites with varying slopes ranging from 0–8 were selected to represent the area where
the livestock of the community grazed. Each site had 10 m × 10 m plots where a grazing was totally pre-
cluded NOG, b moderately grazed MDG — 1.8 Tropical Livestock Unit Month — TLUMha and c
very heavily grazed VHVG — 4.2 TLUMha see also Mwendera et al., 1997. The NOG and MDG
plots were enclosed by barbed wire fences and there was no fencing around VHVG plots. The door of the
fence around MDG was opened for 3 days in a week for the free grazing livestock of the farming commu-
nity. Grazing took place during all days of the week in VHVG. Data on species composition, per cent cover
of three most dominant species and bare ground were collected at 30 days interval from the second week of
September to the second week of December. This pe- riod is shortly after the summer rain season when plant
growth rate is maximal and most plants bear inflo- rescence. During the long dry period January–May
and the summer rain period June–September most species in VHVG and some in MDG and NOG are not
identifiable. Data were collected for three consecutive years from 1995–1997.
Only the list of species in NOG and VHVG were considered for the present article. The result which
includes MDG will be communicated in another pub- lication.
2.4. Data analysis The weekly manure-seedling data from the five
replicates were pooled for each month and a data matrix was constructed for each year with the species
in the rows and months on the columns. The species composition of the manure M was compared with
data on the grazing experiment plots using chi square χ
2
. The relative occurrences of the four important grassland families were calculated for each year of
Fig. 1. Monthly average rainfall in Ghinchi Research Station, Ethiopia in 1996 and 1997.
the grazing experiment and manure seed bank data. The relative composition of the annual and peren-
nial species in each experiment was also calculated. Rainfall data of 1996 and 1997 was obtained from
Ethiopian Agricultural Research Meteorological Ser- vices and the monthly average rainfall of the 2 years
were plotted Fig. 1.
Principal Components Analysis PCA was per- formed using Syntax to find the influence of time on
the species composition of the manure seed bank. The PCA is an ordination technique which involves
extracting the same number of axes of variations as in the rows or columns of the input data matrix
in decreasing order of importance. Scatter diagram scattergram of the two axes showing the highest
variations provides a two dimensional display of the underlying pattern in the original data matrix see
Orloci, 1978.
3. Results and discussion
