Records of the Western Australian Museum 17: 231-233 (1995). 
Short communication 
An experimental evaluation of habitat choice in three species of goby 
(Pisces: Gobiidae) 
H.S. Gill 1 and P. Humphries 2 
1 School of Biological and Environmental Sciences, Murdoch University, South Street, Murdoch, 
Western Australia 6150 
2 Inland Fisheries Commission, 127 Davey Street, Hobart, Tasmania 7000 
The ways in which fish species at similar trophic 
levels are able to coexist have been examined 
extensively (see review by Ross 1986 and 
references therein). However, it is only recently 
that attempts have been made to use experimental 
methods to identify the influence that one species 
may have on the habitat choice of another (e.g., 
Werner and Hall 1976, 1977; Wiederholm 1987; 
Magnhagen 1988a, b). While studies of the effects 
of one species of fish on another in the field have 
the advantage of providing data on those species 
in their natural environment, the ability to regulate 
and manipulate conditions in the laboratory means 
that laboratory studies can be used to elucidate the 
role played by individual factors in influencing 
particular types of behaviour (Crowder 1986). For 
example, the use of artificial seagrass in laboratory 
trials provides cover without the confounding 
effects brought about by the animals and plants 
usually associated with living seagrass (Bell el al. 
1985). 
Two dominant habitat-types occur in the 
shallows of Wilson Inlet, a seasonally closed 
estuary in southwestern Australia, namely 
unvegetated sand and beds of the aquatic 
macrophyte Ruppia megacarpa (Lukatelich et al. 
1987), the latter of which may be dense or patchy. 
The three most abundant demersal species of 
teleost which occupy these habitats are gobies, each 
of which shows a greater degree of association 
with either beds of Ruppia or bare sand 
(Humphries et al. 1992; Humphries and Potter 
1993). Thus, whereas Favonigobius lateralis shows a 
strong association with bare sand, Afurcagobius 
suppositus and Pseudogobius olorum are more often 
associated with dense and patchy Ruppia habitats 
respectively. Since the above species are all found 
in the same area in Wilson Inlet, but are associated 
with different habitat types, a series of laboratory 
experiments were designed to determine whether 
the choice of habitat by each of the three goby 
species is influenced by the presence of either or 
both of the other species. 
The three species were collected from Wilson 
Inlet and were maintained in 20 1 glass aquaria at 
ambient photoperiod (13.5 h light/10.5 h dark), 
temperature (20°C) and salinity (ca 30 °/ 00 ) and fed 
brine shrimp daily. The experiments were 
conducted in a square tank comprising 1 m long 
and 0.5 m high sides. Washed sand was placed in 
the tank to a depth of 5 cm and covered with water 
to a depth of 35 cm. Artificial seagrass was placed 
in one half of the tank and the other half was left 
bare. To ensure even illumination, two fluorescent 
lights were positioned above the tank, these were 
perpendicular to the boundary of the artificial 
seagrass and bare sand. The artificial seagrass, 
which was designed to resemble Ruppia megacarpa, 
was constructed from 40 cm lengths of olive-green 
curling ribbon. Each blade was split into four 
longitudinal strips and stapled on to a wire mesh 
at a density of 670 'blades' m 2 , a density which 
simulates the dense Ruppia habitat. Prior to 
experimentation fish were acclimated to the 
experimental tank for approximately 18 hours, 
during which time they were not fed. To determine 
the intervals at which readings should be made, a 
total of 42 individuals of each species (in groups of 
6 individuals) were observed in isolation and the 
length of time they took to make a 'conspicuous 
movement' within the experiment tank was noted. 
A 'conspicuous movement' was one where a fish 
moved the equivalent of half the length of the tank, 
i.e., it could have moved from one type of habitat 
to another. Results showed that, on average, 78% 
of individuals of each species made a 'conspicuous 
movement' within 5 minutes. 
The number of fish in each habitat was recorded 
every 5 minutes for one hour in both the morning 
and afternoon and once every hour for the six 
hours between these readings. After each hourly 
reading, the fish's choice of habitat was assessed in 
response to a disturbance designed to simulate a 
predator. 
All trials were replicated twice, with the 
combinations randomly assigned to particular days 
to remove time effects. In the single species 
experiments, the trials were conducted with 
densities of 10, 20 and 30 individuals and in those 
with two or more species, the trials involved 10 
