Griffiths et al.: Fish recolonization in temperate Australian rockpools 



. 645 



pool fish assemblages. Therefore, the composition of spe- 

 cies in newly recolonized rockpools is probably depen- 

 dent upon the relative abundances of species in nearby 

 rockpools. Species having the highest local abundances, 

 such as B. cocosensis and E. rufopileus, are therefore 

 more likely to be the primary recolonizers because va- 

 cant habitats have a higher probability of being located 

 by these species during high-tide excursions throughout 

 the intertidal zone (also see Polivka and Chotkowski, 

 1998). These species are also versatile and can exploit 

 a range of microhabitats and, as a result, can occupy 

 almost any rockpool within the intertidal zone (Griffiths 

 et al., 2003). This is particularly true for B. cocosensis. 

 In contrast, less abundant species such as H. whiteleggi 

 often occupy more specific, and perhaps less abundant, 

 microhabitats such as algal cover (see Marsh et al.. 

 1978; Bennett and Griffiths, 1984) that may require 

 longer periods to locate than more abundant habitats, 

 such as cobble-covered substratum. 



Processes regulating fish assemblages 



The structure of multispecies assemblages can be 

 regarded as being regulated by either deterministic or 

 stochastic processes (see Grossman, 1982). Assemblages 

 regulated by deterministic processes generally occur in 

 environments where conditions are constant or fluctuate 

 consistently over time. The structure of these assem- 

 blages is generally predictable. This can be maintained 

 through a number of factors including partitioning of 

 resources in finite supply (Schoener, 1974; Behrents, 

 1987) and interspecific competition, which prevents any 

 single species being competitively dominant (Buss and 

 Jackson, 1979). 



In contrast, assemblages regulated by stochastic pro- 

 cesses generally exist in unpredictable environments. 

 Here, the resources are available on a random or pe- 

 riodic basis, which prevents superior competitors from 

 dominating the assemblage (Sale, 1977, 1978). The suc- 

 cess of particular species can be compared to winning 

 a "lottery" for living space (Sale, 1977, 1978, 1982). 

 Consequently, stochastically regulated assemblages are 

 generally species rich (Sale, 1977). 



Rockpool fish assemblages are often persistent for 

 lengthy periods, even after catastrophic natural dis- 

 turbances, such as hurricanes (Moring, 1996), and con- 

 tinual experimental eliminations (Grossman, 1982; 

 Collette, 1986). For example, Collette (1986) found two 

 species — Pholis gunnellus and Tautogolabrus adsper- 

 sus — to be dominant over 19 years of study in two New 

 England rockpools, whereas the rank of dominant spe- 

 cies in the rockpools of Barbados showed no evidence 

 of change over six years (Mahon and Mahon. 1994). 

 Similar stability and persistence were evident in the 

 present study, where B. cocosensis, E. rufopileus and 

 G. elevata were consistently the highest ranked species 

 in each collection for all three studies, regardless of 

 the period between sampling. This finding may indi- 

 cate that deterministic processes probably regulate the 

 Bass Point fish assemblage. If this is the case, it may 



seem ironic because the intertidal zone is subjected to 

 a high frequency of stochastic events. It would be easy 

 to assume that such events could eliminate fishes from 

 rockpools and thus leave microhabitats for other species 

 to exploit. This kind of process has been documented for 

 some sessile intertidal invertebrate assemblages that 

 rely on the availability of vacant substrata for success- 

 ful recruitment of larvae (see examples by Raffaelli 

 and Hawkins. 1996). However, the locomotory capabili- 

 ties and morphological and physiological adaptations of 

 resident intertidal fishes allow them to cope with such 

 disturbances by being able to cope temporarily with ad- 

 verse conditions (Martin. 1995). As a result, the abun- 

 dance of resident species may be little affected under 

 normal disturbance regimes. 



Conclusions 



The results of this study have significantly increased an 

 understanding of the patterns of recolonization of rock- 

 pools by fishes and some of the processes that underpin 

 these patterns. Such an understanding of recoloniza- 

 tion processes may improve our ability to predict the 

 consequences of significant natural and anthropogenic 

 disturbances on not only the fish assemblages but also 

 on other intertidal community assemblages that may be 

 maintained by the presence offish (see Coull and Wells, 

 1983; Connell and Anderson, 1999). 



On a more technical note, the recolonization rates 

 observed in the present study may provide insight for 

 other researchers aiming to stud}' natural temporal 

 variation of rockpool fish assemblages by minimizing 

 the possibility of confounding effects of sampling. This 

 may be particularly important for long-term monitoring 

 programs, such as for marine protected areas (MPAs). 

 that may require detection of changes in community 

 structure over time. Finding sufficient numbers of simi- 

 lar-size pools at a single location for monitoring can be 

 difficult: therefore repeated visits to the same rockpools 

 may often be required. For southeastern Australian 

 rockpools. we feel that a period of one to three months 

 is required before resampling the same rockpools with 

 the methods employed in this study. Although fish were 

 not returned to rockpools immediately after sampling 

 in the present study, we feel that this practice may 

 significantly increase recolonization rates. However, 

 the results of the present study should not provide a 

 foundation for studies using other defaunation methods, 

 such as anesthetics or ichthvocides. because other fac- 

 tors, such as chemical residues remaining in rockpools, 

 may complicate fish recolonization patterns. Further 

 investigation into these other factors will be necessary 

 in the future. 



Acknowledgments 



We sincerely thank Jade Butler and Alan Griffiths for 

 help with fieldwork. This paper is partly based upon 



