SOIL SEED BANK OF A THEMEDA GRASSLAND 
55 
nants and may have been mistaken for Isolepis 
species. Specimens referred to Vulpia bromoides 
may have included some Vulpia myuros forma 
megalura, but most appeared to be of the former 
species. The Juncus bufonius/capitatus group 
included the native J. bufonius and the exotic 
J. capitalus, and equal numbers of both were 
assumed in native/exotic comparisons. 
RESULTS 
Excluding glasshouse contaminants, 41 her¬ 
baceous species including 17 natives and 24 exo¬ 
tics germinated from the soil seed bank (Table 
1). No ferns or woody plants were recorded. Exo¬ 
tics comprised 59% of species but 91% of in¬ 
dividuals (Table 2). A small number of species 
dominated the soil seed bank: Vulpia bromoides 
accounted for 61% of individuals, and the three 
exotics V. bromoides, Romulea rosea and Air a 
cupaniana together accounted for 81%. In con¬ 
trast, 24% of species were represented by only 
one or two individuals. Four species identified 
from the seed bank had not previously been re¬ 
corded from the reserve: one native, Crassula 
decumbens , and three exotics, Cerastium 
glomeratum, Sagina procumbens and a Vicia 
species. 
For all but three species, the number of viable 
seeds in species-poor and species-rich Themeda 
grassland was not significantly different (p > 
0.05). Viable seeds (and established plants) of V. 
bromoides and Aira cupaniana were signifi¬ 
cantly more abundant in species-rich than in 
species-poor grassland, and viable seeds (and 
flowering culms) of T. triandra were more abun¬ 
dant in species-poor than in species-rich grass¬ 
land (p < 0.05). Thirty species were recorded 
from the seed bank of species-poor grassland 
Group 
Individuals 
Number % 
Species 
Number % 
Annuals 
10113 
80 
24 
59 
Perennials 
2496 
20 
17 
41 
Exotics 
11506 
91 
24 
59 
Natives 
1103 
9 
17 
41 
Monocotyledons 
11608 
92 
15 
37 
Dicotyledons 
1001 
8 
26 
63 
Total 
12609 
— 
41 
— 
Table 2. General characteristics of soil seed bank of a 
long-grazed Themeda triandra grassland. 
and 33 from that of species-rich grassland (Table 
1). Both seed banks were similar in composition, 
with 25 species (66% of seed bank species) occur¬ 
ring in both. Of the 22 native species recorded 
from the vegetation of species-rich but not spe¬ 
cies-poor grassland, only four appeared in the 
seed bank of species-poor grassland: Calocepha- 
lus citreus, Helichrysum apiculatum, Isolepis 
species and Juncus bufonius. 
DISCUSSION 
The seed bank of Themeda triandra grassland in 
the Derrimut Grassland Reserve possessed 
characteristics typical of most seed banks, 
including high spatial variability, poor corre¬ 
spondence with the standing vegetation, domi¬ 
nation by one or two species of annual mono¬ 
cotyledons, and a relatively low representation 
by perennials (see e.g. Roberts 1970, Donelan & 
Thompson 1980, Rabinowitz 1981, Reichman 
1984, Schenkeveld & Verkaar 1984, Thompson 
1986). Its composition was very similar to that 
of seed banks of annual grasslands in inland Cal¬ 
ifornia (Heady 1956, 1977, Major & Pyott 1966, 
Bartolome 1979), due to the abundance of exotic 
annual grasses, including^//*#, Briza and Vulpia. 
Before these species were introduced, Victorian 
and inland Californian grasslands were domi¬ 
nated by perennials, T. triandra and Stipa 
pulchra respectively (Willis 1964, Heady 1977, 
Bartolome et al. 1986). Seed densities could not 
be compared accurately, due to different sam¬ 
pling and germination procedures, but those 
obtained here (from 4,601 to 28,337 viableseeds 
per m 2 ) were of the same order of magnitude as 
those recorded from California (Major & Pyott 
1966). Up to 670,000 seeds of Vulpia bromoides 
per m 2 occurred in dense stands of Vulpia in 
California (Bartolome 1979), and comparable 
densities might be expected in dense stands of V. 
bromoides at Derrimut. 
It has been suggested that the demise of T. 
triandra under grazing may have been caused by 
an inherently low rate of seed production 
(Groves et al. 1973, Groves & Williams 1981) 
but this appears unlikely since, in this study, T. 
triandra was the fifth most abundant species in 
the seed bank, producing up to 1,375 viable 
seeds per m 2 . 
Species that did not germinate in the glass¬ 
house cannot be presumed absent from the seed 
bank, since sampling and germination pro¬ 
cedures may not have been appropriate for all 
species. Although sampling intensity was suffi¬ 
cient to indicate general characteristics of the 
