Dormancy and germination characteristics of herbaceous species in the seasonally dry tropics of northern Australia.

This study investigated changes in dormancy and germination over 8 months for 23 common species (annual and perennial grasses, legumes and other dicotyledons) from herbaceous communities in northern Australia. Seeds were exposed to three storage treatments: relatively constant laboratory conditions, an oven with fluctuating temperatures similar to those found on the soil surface (25/60°C), or exposed on the soil surface at Townsville. There were wide ranges of initial levels of dormancy (9-100%), rates of change of dormancy and response to the different storage conditions showing that species with several types of dormancy characteristics are able to coexist in these communities. The general trend in dormancy levels was a decline with time with the rate of decline greatest for seeds exposed on the soil surface and least for those stored in the laboratory. The species were divided into groups based on dormancy levels in seeds on the soil surface during the late dry and mid wet seasons. The dormancy characteristics of the groups were related to the ecology of the species in the groups. There was an approximately linear increase in germination rate (i.e. a decrease in the number of days to 50% of final germination) over time for all storage treatments; rates for seeds on the soil surface increased more rapidly than those of seeds in laboratory and oven samples.

INTRODUCTION

The optimum time for seed germination in strongly seasonal climates is as early in the growing season as possible to gain resources for growth and reproduction but not so early that survival is unlikely ( Rathcke & Lacey 1985). Thus a successful species requires some mechanism or mechanisms to prevent germination before, but not at, the optimum time so that the probability of encountering poor growing conditions in the period after germination is reduced ( Angevine & Chabot 1979).

In the seasonally dry tropics most germination occurs early in the wet season ( Torssell & McKeon 1976; Mott 1978, 1980; Marks 1983a, b; McIvor & Gardener 1991). In these environments, species need to avoid germination on small falls of rain during the dry season as mortality can be complete during subsequent dry periods ( McKeon 1985; McKeon ). Seeds then need to germinate early in the wet season as later‐establishing seedlings are less likely to survive competition from established perennial plants and older seedlings ( Ross & Harper 1972; Cook & Ratcliff 1984, 1985). Surface soil temperatures frequently exceed 50°C during the dry season in tropical areas ( Mott 1978; Mott ; McKeon ) and exposure to such high temperatures is an important dormancy‐breaking mechanism in some tropical grasses and Stylosanthes species ( Mott 1978; Mott ; Hacker 1984; Hacker ; Hacker & Ratcliff 1989). Herbaceous communities can contain many more species than those assessed already, so this study was undertaken to compare the changes in dormancy and germination of seed of a wide range of native and exotic species (based on Henderson 1997) collected at two sites typical of the seasonally dry tropics. The seed was stored under relatively constant laboratory conditions (control), exposed on the soil surface to simulate field conditions, or kept in an oven with fluctuating temperatures similar to those found on the soil surface (25/60°C). The oven treatment approximates the temperature environment experienced by seed in the field ( Mott 1978; McKeon ) but excludes other effects such as wetting/drying and fungal infections. Seed size has been related to seed germination and persistence and has been used to predict seed longevity (e.g. Grime 1989; Thompson ; Bekker ). Germination rate may control germination in tropical areas ( McKeon ) so relationships between seed weight and germination rate were also examined to assess the utility of seed size as a predictor.

METHODS

Species and collection sites

Fully mature dispersal units (hereafter referred to as seeds) of 21 species were collected in April–May 1983 from pastures in either the Townsville–Woodstock (indicated by W after the species name) area of north Queensland or from Katherine (K) in the Northern Territory ( Table 1). Seeds of three species (Digitaria ciliaris, Heteropogon contortus and Hyptis suaveolens) were collected from both areas and commercial samples of Stylosanthes hamata and Stylosanthes scabra were also included giving a total of 23 species and 26 seed lots.

Table 1

Collection site and dispersal unit weight of 26 seed collections tested

Species	Collection site	Dispersal unit weight (mg)
Native annual (or short‐lived perennial) grass
Brachiaria subquadripara (Trin.) Hitchc.	Woodstock	1.5
Dactyloctenium radulans (R.Br.) P. Beauv.	Townsville	0.2
Exotic annual (or short‐lived perennial) grass
Chloris inflata Link	Woodstock	0.2
Digitaria ciliaris (Retz.) Koeler	Katherine	0.9
Digitaria ciliaris	Woodstock	0.9
Echinochloa colona (L.) Link	Woodstock	1.0
Eleusine indica (L.) Gaertn.	Townsville	0.4
Pennisetum pedicellatum Brunken	Katherine	3.3
Themeda quadrivalvis (L.) Kuntze	Woodstock	1.8
Native perennial grass
Aristida latifolia Domin	Townsville	1.0
Bothriochloa decipiens (Domin) C.E. Hubb.	Woodstock	1.0
Chrysopogon fallax S.T. Blake	Townsville	4.3
Heteropogon contortus (L.) P. Beauv. ex Roem. & Schult.	Katherine	1.7
Heteropogon contortus	Woodstock	1.9
Sporobolus indicus (Buse) Baaijens	Woodstock	0.1
Themeda triandra Forssk.	Townsville	2.4
Exotic perennial grass
Bothriochloa pertusa (L.) A. Camus	Woodstock	0.7
Cenchrus ciliaris L.	Woodstock	3.2
Chloris gayana Kunth	Woodstock	0.4
Urochloa mosambicensis (Hack.) Dandy	Woodstock	1.8
Exotic annual (or short‐lived perennial) legume
Stylosanthes hamata (L.) Taub.	Commercial	2.9
Stylosanthes scabra Vogel	Commercial	2.0
Other exotic annual (or short‐lived perennial) dicotyledons
Gomphrena celosioides Mart.	Townsville	3.0
Hyptis suaveolens (L.) Poit.	Katherine	5.5
Hyptis suaveolens	Woodstock	4.2
Sida acuta Burm. f.	Katherine	3.1

The Townsville–Woodstock and Katherine areas are typical of the seasonally dry tropics which are characterized by a hot ‘wet’ season and a warm ‘dry’ season. Average annual rainfall is 1160 mm at Townsville and 860 mm at Woodstock with 84% falling during the wet season (December–April). Maximum temperatures range from 31 to 33°C in December to 25°C in July. Average annual rainfall is 970 mm at Katherine with 92% falling during the wet season (November–March). Maximum temperatures range from 38°C in December to 30°C in July.

Seed storage and preparation

After collection, the seeds were stored at 22–23°C and 55–60% relative humidity in paper envelopes. The seeds were cleaned, and the awns removed from Aristida latifolia, Heteropogon contortus, Themeda quadrivalvis and Themeda triandra. Abnormal and empty seeds were discarded and samples of 50 seeds, all apparently sound, were counted. Seed weights were measured for four of these samples (i.e. 200 seeds).

Storage treatments

There were three storage treatments.

1. Laboratory. The seeds were stored in paper envelopes at 22–23°C and 55–60% relative humidity.

2. Oven. The seeds were stored in paper envelopes in an oven with a diurnal temperature fluctuation of 25/60°C.

3. Soil. The seeds were placed in fine, nylon mesh envelopes which were pegged to the soil surface in a level area, cleared of all vegetation and exposed to full sunlight, at the Davies Laboratory, Townsville.

For each storage treatment, there were four, fully randomised replicates of each seed lot. The seeds were placed in the three locations on 4 July 1983, approximately 2 months after collection. The storage period covered the mid–late dry season (July–November) and early–mid wet season (December–February).

Seed tests

At the time of placement and 4, 8, 12, 16, 20, 24, 28 and 32 weeks later, a sample of each seed lot from each replicate (i.e. 26 seed lots × three storage treatments × four replicates each of 50 seeds) was retrieved and tested for germination and dormancy. For four species (Bothriochloa pertusa, Heteropogon contortus, Themeda triandra and Urochloa mosambicensis), additional samples were retrieved after 36, 44, 64 and 96 weeks.

Any seeds which showed evidence of prior germination were counted and removed and the remainder germinated in Petri dishes in a germination cabinet with a diurnal temperature fluctuation of 30/25°C and a 12‐h day. Optimal conditions for germination are not known for most species but 30/25°C is in the optimum range for many tropical species ( Ellis ). Germinated seeds were counted and removed daily for 14 days. Any seeds which did not germinate during that period were dissected and classified as dormant if the embryo and endosperm were firm and intact, or as dead if the tissues were pulpy and had begun to decay ( Mott 1978). There was some field germination after 16 weeks and it is possible some seeds in these samples classified as dead had previously germinated. This gave four classes of seed: field‐germinated, germinable, dormant and dead. (For the two legumes, hard seeds were also counted separately but have been included in the dormant class.)

Meteorological data

The seeds exposed on the soil surface were located 10 m from the Davies Laboratory meteorological station where rainfall, and maximum and minimum temperatures were recorded daily.

Data analyses

The data for the proportions of germinable and dormant seeds were analysed using factorial (seed lot × storage treatment × retrieval time) anova after arcsin transformation to normalize the variances. Plots of cumulative germination over the 14‐day test period were all sigmoid shaped so germination rate was estimated by fitting a sigmoid curve to the data, and calculating the number of days required to reach 50% of the final value. These values were analyzed by analysis of variance after logarithmic transformation. When the F statistic in the anova was significant, the treatment means were compared by least significant difference (LSD, P = 0.05). All analyses were performed in GENSTAT ( Payne ).

RESULTS

There were only 8 mm of rainfall during the first 12 weeks of storage but 17 mm fell in the 2 weeks prior to the sampling at 16 weeks ( Fig. 1). There were significant falls between all subsequent samplings with 225 mm during weeks 25–28, and 200 mm during weeks 29–32. Maximum temperatures increased from 23 to 25°C at the time of placement to 30–33°C during weeks 17–32. Minimum temperatures showed a similar increase from 10 to 13°C to 20–24°C. For the seeds exposed for a longer period, the second wet season commenced during week 75.

Figure 1

Weekly mean maximum and minimum temperatures and rainfall during the experiment. The arrows indicate when seed samples were retrieved and tested.

Germination and dormancy

Figure 2 shows examples of the changes in the proportions of different seed categories under different storage conditions. To reduce the volume of data presented, six species which cover the range of responses were chosen.

Figure 2

Changes in the percentage germinable (), dormant () (including hard), dead () and apparently field‐germinated () seed of six species stored in a laboratory at 22–23°C, in an oven at 25/60°C, and exposed on the soil surface at Townsville. Weeks 0–16 correspond with the dry season, and weeks 17–32 with the wet season.

For seeds stored in the laboratory or in the oven, there was generally little change in the proportion of dead seed with storage time. Under these conditions there were three patterns of change in the proportion of germinable seeds: little or no change (e.g. Stylosanthes hamata in the laboratory and Eleusine indica in both regimes), a gradual increase (e.g. Chloris gayana in both laboratory and oven) or a rapid initial increase followed by a period with little or no change (e.g. Brachiaria subquadripara, Aristida latifolia and Heteropogon contortus in both regimes). As there was little change in the proportion of dead seed, the change in the proportion of dormant seeds was the reverse of that in germinable seeds.

The changes for seeds on the soil surface were markedly different to those of seeds in the laboratory or oven. The proportion of dead seeds generally increased slowly during the dry season and then more rapidly after the wet season commenced. For the proportions of dormant and germinable seed there were three general patterns. For one group of species, there was little change in the numbers of germinable or dormant seed until rain fell between weeks 14 and 16. After this, both components declined but some seeds remained dormant at the final sampling (e.g. Stylosanthes hamata). In a second group, the proportion of germinable seeds increased (and dormant decreased) during the dry season and the numbers of germinable seeds declined rapidly during the wet season. Within this group, the species could be divided into those with some dormant seed at the final sampling (e.g. Brachiaria subquadripara) and those with no seeds remaining dormant (e.g. Aristida latifolia, Heteropogon contortus, Chloris gayana). In the third group, high initial dormancy rates declined sharply with the onset of the wet season but germinable seed percentages remained low (e.g. Eleusine indica).

To enable a comparison of all species, the percentages of germinable and dormant seeds in the initial and final samples, plus those on the soil surface at the end of the dry season (week 16) are shown in Table 2.

Table 2

Changes in the proportions of germinable and dormant (including hard) seeds in samples stored under different regimes for varying periods

	Germinable seeds			Dormant seeds
	Init. week	Lab	Oven	Soil	Soil	Init.	Lab	Oven	Soil	Soil
	0	32	32	16	32	0	32	32	16	32
Species	(%)	(%)	(%)	(%)	(%)	(%)	(%)	(%)	(%)	(%)
Native annual grass
Brachiaria subquadripara	26b	73ab	84a	72ab	8c	74a	25b	12b	18b	21b
Dactyloctenium radulans	0a	1a	4a	0a	0a	100a	98ab	96ab	100a	84b
Exotic annual grass
Chloris inflata	2b	47a	44a	2b	0b	98a	2c	1c	64b	0c
Digitaria ciliaris (K)	8c	43b	78a	45b	0c	92a	56b	19c	49b	0d
Digitaria ciliaris (W)	4b	76a	64a	79a	0b	95a	23b	30b	19b	2c
Echinochloa colona	1c	8bc	49a	17ab	0c	99a	92ab	51c	74b	6d
Eleusine indica	2a	2a	0a	5a	0a	98a	98a	99a	93a	26b
Pennisetum pedicellatum	0b	22a	33a	4ab	0b	96a	42b	36b	85a	1c
Themeda quadrivalvis	32b	71a	69a	89a	0 c	68a	26bc	29b	7c	0d
Native perennial grass
Aristida latifolia	2b	38a	71a	74a	0b	98a	58b	27c	16c	0d
Bothriochloa decipiens	2c	74a	70ab	54b	0c	94a	14bc	15b	13bc	2c
Chrysopogon fallax	41a	62a	69a	52a	0b	47a	23b	23b	22b	1c
Heteropogon contortus (K)	0c	83a	64b	65b	0c	88a	13cd	24b	17bc	1d
Heteropogon contortus (W)	0b	50a	47a	27a	1b	76a	32bc	39bc	42b	17c
Sporobolus indicus	2b	47a	73a	73a	0b	98a	53b	28b	9c	0c
Themeda triandra	9ab	22a	16a	22a	0b	76a	67ab	73a	50b	9c
Exotic perennial grass
Bothriochloa pertusa	0b	7ab	5ab	22a	0b	76a	53b	27b	53b	4c
Cenchrus ciliaris	5c	16bc	32b	55a	6c	89a	56b	41b	13c	14c
Chloris gayana	0b	35a	57a	51a	0b	53a	29ab	3c	12bc	1c
Urochloa mosambicensis	0b	2ab	8ab	15a	0b	100a	97ab	92ab	85b	34c
Exotic legume
Stylosanthes hamata	11b	11b	51a	14b	0b	79a	81a	35b	74a	66a
Stylosanthes scabra	18b	14b	44a	14b	0c	47a	43a	3b	33a	9b
Other exotic dicotyledons
Gomphrena celusoides	47a	66a	72a	55a	0b	52a	33a	36a	44a	4b
Hyptis suaveolens (K)	86a	95a	77a	98a	1b	9ab	5ab	22a	1b	0b
Hyptis suaveolens (W)	7bc	26ab	17ab	41a	0c	54a	32ab	23b	0c	1c
Sida acuta	0b	1b	36a	0b	0b	99a	96a	61b	100a	47b

Values for germinable or dormant seeds of a species followed by the same letter are not significantly different (P > 0.05). K, collection from Katherine; W, collection from Woodstock.

The dormancy level in the initial samples varied from low in the Katherine sample of Hyptis suaveolens to complete in Dactyloctenium radulans and Urochloa mosambicensis. No seeds germinated in the initial samples of eight species.

For seeds stored in the laboratory, over 90% of the seeds of Dactyloctenium radulans, Echinochloa colona, Eleusine indica, Urochloa mosambicensis and Sida acuta remained dormant after 32 weeks. In all other species except the two Stylosanthes, dormancy levels decreased and germination increased although the effects were not always significant.

Oven storage did not lead to significant changes in the percentage of dormant seeds in samples of Dactyloctenium radulans, Eleusine indica, Themeda triandra and Urochloa mosambicensis but dormancy levels decreased and germination increased (relative to initial levels) in other species. The percentage germination at 32 weeks was significantly higher in oven stored than laboratory stored seed for five species (Digitaria ciliaris (K), Echinochloa colona, Stylosanthes hamata, Stylosanthes scabra, Sida acuta (K)) and values were similar in the two storage regimes for the remainder of the species except Heteropogon contortus (K) which showed the reverse pattern.

In the field, after 16 weeks there was almost complete breakdown of dormancy in Themeda quadrivalvis, Sporobolus indicus and Hyptis suaveolens but dormancy remained at the initial levels in Dactyloctenium radulans, Eleusine indica, Stylosanthes hamata and Sida acuta. After 32 weeks in the field, 22 species had no germinable seeds remaining and germination exceeded 1% in Brachiaria subquadripara and Cenchrus ciliaris only. Some species had considerable dormant seed in the field at this time with more than 40% of the seeds of Dactyloctenium radulans, Stylosanthes hamata and Sida acuta in this category.

For the species studied over the longer period, seed of Heteropogon contortus and Urochloa mosambicensis exposed on the soil surface had a second, smaller peak of germination around week 64 ( Table 3). After 96 weeks exposure, there was negligible germination and little dormant seed of any species stored on the soil.

Table 3

Changes in the proportions of germinable and dormant seeds in samples of four perennial grasses stored under different regimes for varying periods

		Germinable		Dormant
		seeds			seeds
Week	Lab.	Oven	Soil	Lab.	Oven	Soil
Bothriochloa pertusa
36	56a	46a	0a	7c	10c	5b
44	7b	12b	0a	50ab	50b	20a
64	2b	0b	1a	57a	57ab	8b
96	6b	0b	0a	41b	72a	1b
Heteropogon contortus
36	68a	47a	0b	10c	14c	3a
44	67a	31b	5a	8c	56b	1b
64	33b	0c	6a	24b	56b	1b
96	23b	0c	1b	38a	72a	1b
Themeda triandra
36	16a	11a	0a	70a	81a	4a
44	9ab	8ab	0a	70a	73a	1b
64	17a	0b	0a	71a	86a	1b
96	5b	0b	0a	81a	76a	0b
Urochloa mosambicensis
36	1a	2a	0b	98a	97a	35a
44	2a	2a	1b	97a	98a	28b
64	1a	0b	15a	99a	100a	2c
96	0a	0b	0b	99a	100a	2c

Values for germinable or dormant seeds of a species within a treatment followed by the same letter are not significantly different (P > 0.05).

Germination rate

There was an approximately linear increase in germination rate (i.e. a decrease in the number of days to 50% of final germination) with time for all storage treatments ( Fig. 3). The laboratory and oven samples changed slowly (0.044 units/week) but the seeds on the soil surface changed more rapidly (0.104 units/week).

Figure 3

Mean (±SE) germination rates (days to reach 50% of final germination) of seeds stored for different periods of time in a laboratory at 22–23°C (▪), in an oven at 25/60°C (○), and exposed on the soil surface at Townsville (●).

To illustrate the differences between species, Table 4 shows the values for seeds on the soil surface at the commencement of the wet season after 20 weeks’ exposure. There were large differences between species with half the seeds of Stylosanthes hamata, Themeda quadrivalvis and Aristida latifolia germinating within 16 h compared with over 4 days for Sida acuta and Hyptis suaveolens (K). Apart from Digitaria ciliaris (W) and Hyptis suaveolens (K), seeds exposed on the soil surface germinated faster than seeds stored in the laboratory.

Table 4

Germination rates (number of days required to achieve 50% of final germination value) of seeds of herbaceous tropical species at the commencement of the growing season following exposure on the soil surface for 20 weeks

	Germination rate
Species	Soil	Lab.
Stylosanthes hamata	0.60 (0.21)	1.3
Themeda quadrivalvis	0.65 (0.22)	1.9
Aristida latifolia	0.68 (0.22)	3.5
Heteropogon contortus (K)	0.83 (0.26)	2.4
Echinochloa colona	0.88 (0.27)	4.6
Urochloa mosambicensis	0.93 (0.28)	5.5
Bothriochloa decipiens	1.18 (0.34)	3.0
Sporobolus indicus	1.20 (0.34)	2.5
Brachiaria subquadripara	1.20 (0.34)	3.0
Themeda triandra	1.40 (0.38)	2.3
Heteropogon contortus (W)	1.45 (0.38)	3.0
Digitaria ciliaris (K)	1.45 (0.39)	3.0
Chloris gayana	1.50 (0.39)	3.4
Chloris inflata	1.50 (0.40)	3.5
Chrysopogon fallax	1.53 (0.40)	3.2
Cenchrus ciliaris	1.63 (0.42)	4.9
Digitaria ciliaris (W)	1.68 (0.43)	1.7
Eleusine indica	1.90 (0.46)	7.0
Stylosanthes scabra	1.93 (0.45)	5.4
Bothriochloa pertusa	2.50 (0.51)	4.5
Gomphrena celosioides	2.80 (0.57)	4.4
Hyptis suaveolens (W)	3.53 (0.65)	6.8
Pennisetum pedicellatum	4.00 (0.69)	5.9
Sida acuta	4.50 (0.74)	7.8
Hyptis suaveolens (K)	4.53 (0.74)	4.6
Dactyloctenium radulans*

The transformed (log [x + 1]) values are shown in parentheses; the least significant difference for the transformed values = 0.11 (P = 0.05). For comparison, germination rates of seeds stored in the laboratory for the same period are also shown. *No seeds of Dactyloctenium radulans germinated.

There was a weak negative correlation (r 2 = 0.35, P = 0.002) between germination rate and seed size.

DISCUSSION

There were wide ranges of both initial levels of dormancy and rates of change of dormancy in these species under different storage conditions. Thus species with several types of dormancy characteristics coexist in these communities as previously described for temperate and desert communities ( Angevine & Chabot 1979; Baskin & Baskin 1988).

The general trend in dormancy levels was a decline with time with the rate of decline greatest for seeds exposed on the soil surface and least for those stored in the laboratory. The 26 seed lots have been divided into four groups based on dormancy levels (expressed as a percentage of the initial values) after 16 (late dry season) and 32 weeks (mid wet season) exposure on the soil surface. On each occasion the species were divided into groups of high (above median value) and low (below median value) dormancy. The median values were 50% for week 16 (dry season) and 4% for week 32 (wet season).

Group 1

Group 1: species with high dormancy during both dry (mean = 81%) and wet (mean = 32%) seasons: Dactyloctenium radulans, Echinochloa colona, Eleusine indica, Bothriochloa pertusa, Heteropogon contortus (W), Themeda triandra, Urochloa mosambicensis, Stylosanthes hamata, Stylosanthes scabra, Gomphrena celosioides and Sida acuta.

Apart from Themeda triandra, this mixed group of annual and perennial grasses, legumes and dicots are all good colonizers of bare areas and a number are major weeds in Australia and other tropical areas –Echinochloa colona, Eleusine indica and Sida acuta ( Holm ; Kleinschmidt & Johnson 1977; Mott 1980; Akobundu & Agyakwa 1987; Martin 1996). Dactyloctenium radulans is an annual grass which increases under heavy stocking ( Anderson 1993; Roberts & Silcock 1993) and can be a weed of cultivated areas and roadsides ( Tothill & Hacker 1983). Urochloa mosambicensis is a perennial grass sown in pastures but which is naturalized in northern Australia and can be a problem in cultivation ( Kleinschmidt & Johnson 1977). Stylosanthes hamata and S. scabra are important sown legumes in semiarid tropical areas and the high hard seed level is an important adaptation to the variable climate experienced in these areas. Gomphrena celosioides is a common weed of disturbed areas and often invades overgrazed pastures ( Kleinschmidt & Johnson 1977; Anderson 1993). Although all species in this group had high dormancy levels during the wet season and some may survive in the soil for a number of years (e.g. Eleusine indica; Hawton & Drennan 1980), very few seeds of Urochloa mosambicensis and Themeda triandra exposed on the soil surface for a further 64 weeks were dormant at the end of the period.

Group 2

Group 2: species with high dormancy in dry season (mean = 77%) and low dormancy in wet season (mean = 1%): Chloris inflata, Pennisetum pedicellatum.

These two annual grasses are common colonizers of waste land and can be weeds in cropping areas ( van Rijn 1968; Kleinschmidt & Johnson 1977; Mott 1980; Martin 1996). This dormancy pattern results in the retention of the seed bank until the start of the wet season when most of the seed becomes germinable thus providing for mass germination events due to the high seed production of these species ( McIvor ).

Group 3

Group 3: species with relatively low dormancy in dry season (mean = 20%) but relatively high dormancy in wet season (mean = 22%): Brachiaria subquadripara, Cenchrus ciliaris.

Brachiaria subquadripara is an annual or short‐lived perennial which invades legume‐based pastures ( Gillard & Fisher 1978; Gillard ; Tothill ). Cenchrus ciliaris is a strongly perennial introduced grass widely sown in Australia and now naturalized in central inland Queensland ( Kleinschmidt & Johnson 1977) and invading along river banks into mesic habitats in the arid zone ( Humphries ). The retention of some dormancy during the wet season suggests some seeds of these species would persist in the soil, although Silcock & Smith (1990) showed Cenchrus ciliaris seed on or in the soil lost most of its viability within two years.

Group 4

Group 4: species with low dormancy during both dry (mean = 20%) and wet seasons (mean = 1%): Digitaria ciliaris (K) and (W), Themeda quadrivalvis, Aristida latifolia, Bothriochloa decipiens, Chrysopogon fallax, Heteropogon contortus (K), Sporobolus indicus, Chloris gayana, Hyptis suaveolens (K) and (W).

This group contains predominantly perennial grasses, five native and one sown (Chloris gayana, which is now naturalised in central and southern Queensland), with two annual grasses and the dicot, Hyptis suaveolens. Mott (1978) showed seeds of a number of perennial native grasses were dormant at maturity but that dormancy was broken by the beginning of the following wet season. The low dormancy levels in the three weedy species in this group (Digitaria ciliaris, Themeda quadrivalvis and Hyptis suaveolens) is surprising although Marks & Nwachuku (1986) found there were few viable seeds of Digitaria ciliaris after storage in the soil for a wet season. These results suggest persistent seed banks will not be important for species in this group which rely on long‐lived plants (perennial grasses) or frequent seed input (annual species) for long‐term survival.

Only one seed lot was used for most species but for the three species with collections from both Katherine and Woodstock there was some variability between species in the grouping of the collections from the two sites. Seeds for both collections of Hyptis suaveolens and Digitaria ciliaris were in the same group but the dormancy patterns for the two Heteropogon contortus collections were different. The sample of Heteropogon contortus collected from Katherine had lower dormancy than the Woodstock collection in both the late dry (17 vs 42%) and mid wet seasons (1 vs 17%). Such variation between collections could be expected as Heteropogon contortus is a variable species ( Tothill 1966; Tothill & Hacker 1976) and a higher level of dormancy would be expected in a collection from an area with more variable climate (e.g. Woodstock) than one from an area with more reliable climate such as Katherine ( McCown 1981).

In tropical areas where high evaporation rates mean that the soil surface may remain moist for only a short period following rain, germination rate may be the major attribute controlling subsequent establishment ( McKeon ). Almost all species germinated faster after exposure on the soil surface than after storage for the same time in the laboratory. McKeon (1984) found a similar result with Stylosanthes hamata, Stylosanthes humilis and Digitaria ciliaris which all germinated faster after exposure on the soil surface to rainfall. The wide range in germination rates between species means there could be large differences in establishment after small falls of rain; some species had almost complete germination within 1 day whereas others required more than 4 days to reach 50% germination. Rapid germination may particularly increase the likelihood of establishment of smaller seeds in seasonally dry environments ( Murali 1997) and in this experiment there was a weak trend for germination rate to decline as seed size increased. Similar relationships have been found with other suites of species or genotypes from the seasonally dry tropics ( Krishnasamy 1986; Agboola 1996; Murali 1997) but not necessarily in temperate (e.g. Counts & Lee 1991) or cultivated species (e.g. Imrie 1972; Newell & Bludau 1993).

All but seven of the 26 seed lots had either low initial dormancy or significantly reduced dormancy after 16 weeks exposure on the soil surface. This is just prior to the median start of the wet season at Townsville ( McCown 1981). For three species (Dactyloctenium radulans, Eleusine indica and Gomphrena celosioides), dormancy level did not alter during storage in the oven also, suggesting dormancy‐breaking is complex and not simply the result of high temperatures. Silcock found virtually no germination of Dactyloctenium radulans even after 8 years’ storage in a laboratory unless the seed was scarified; treatment with concentrated sulphuric acid can also increase germination of this species ( Silcock & Williams 1975). Dormancy of Eleusine indica is decreased by gibberellic acid and potassium nitrate and can be broken by mechanical or acid scarification ( Hawton & Drennan 1980; Kanzler & van Staden 1984). Dormancy in the other four species (two Stylosanthes species, Pennisetum pedicillatum and Sida acuta) responds to high temperatures ( Mott 1980; Mott ) and it is possible the temperatures in the field were not sufficiently high to affect them; Mott (1980) found Pennisetum pedicillatum seed in the hotter environment at Katherine was nondormant by the end of the dry season and in this study there was a significant decline after 32 weeks in the oven.

For the four perennial grasses stored for 96 weeks, there was negligible germinable or dormant seed for any species after two wet seasons exposure on the soil surface. This result is consistent with previous studies which have shown these species and a number of other common perennial grasses, usually have small soil seed banks ( McIvor 1987; Howden 1988; McIvor & Gardener 1991, 1994). Removal of a species from the community for more than 1 or 2 years is likely to leave no seed available for re‐establishment and could result in permanent changes to the community unless the species is able to recolonize from elsewhere or is deliberately reintroduced.

CONCLUSIONS

We draw three major conclusions from this study. First, the dormancy and germination characteristics of the species depend on storage conditions with the changes for seeds stored on the soil surface markedly different to the changes for seeds stored in the laboratory or oven. Other authors working both with temperate ( Grime ; Baskin & Baskin 1988; Washitani & Masuda 1990) and with tropical species ( Elberse & Breman 1989) have concluded that field behaviour can be related to laboratory germination and response to storage conditions. From this study, the most useful treatments will be those where the storage conditions are as similar as possible to the conditions actually experienced by seeds in the field, for example the soil storage treatment in this experiment. Second, there are a range of dormancy responses in species from these communities. This is true even within a plant type, for example all four dormancy groups were represented in the annual grasses. The major native perennial grasses had very low to low levels of dormancy after exposure on the soil surface during the dry and wet seasons but a number of weedy annual grasses and dicots (except Hyptis suaveolens) had medium to high dormancy levels at the end of the dry season which were maintained to different degrees during the following wet season. Third, provenance or seed collection site (they were confounded in this experiment) may affect seed dormancy patterns.

Acknowledgements

We thank the late Dr C. J. Gardener for his contributions to experimental design and data collection.