Griffiths et al.: Fish recolonization in temperate Australian rockpools 



639 



Table 2 



Results of repeated-measures ANOVAs for significant 

 differences in numbers of species and number of individ- 

 uals (/m :! I caught at Bass Point during two short-term 

 recolonization studies among sampling intervals (time) 

 (within-subjects factor), seasons (spring and autumn) 

 and sites (among-subjects factors). Both numbers of 

 species and individuals data were log 1(1 (.v+l) trans- 

 formed before analysis, which removed heteroscedas- 

 ticity in the data. Mauchly's criterion for sphericity of 

 variances was violated for number of species (P=0.025); 

 therefore the analysis was performed with Greenhouse- 

 Geisser-adjusted degrees of freedom. Mean squares 

 (MS) and significance levels are shown and significant 

 results are given in boldface. * = P<0.05; ** = P<0.01; 

 *** = P<0.001. 



Source 



Number of 

 species 



df MS 



Number of 

 fish 



df MS 



Among subjects 

 Season tSe) 

 Site(S) 

 SxSe 

 Residual 



Within subjects 

 TimeiT) 

 TxS 

 TxSe 

 TxSxSe 

 Residual 



1 



3 



3 



24 



2.16 

 6.47 

 2.16 

 6.47 

 51.75 



Mauchlv's criterion W 



14.94 

 46.80* 



3.66 

 9.38 



18.03** 



2.93 

 2.54 

 1.89 

 1.41 



0.569* 



1 1292.24 



3 2782.52** 

 3 63.53 



24 370.27 



3 

 9 

 3 

 9 



72 



825.70** 

 247.99 



94.86 

 137.84 



79.71 



0.625 



and A. costatus, varied considerably among sampling 

 intervals for each study. This result probably reflects 

 their generally low abundances, because differences in 



Table 3 



Results of repeated-measures ANOVAs for significant 

 differences in numbers of species and number of indi- 

 viduals (/m 3 l caught at Bass Point during the long-term 

 recolonization study among sampling intervals (within- 

 subjects factor) and sites (among-subjects factors). Both 

 numbers of species and individuals data were log 10 (.v+l) 

 transformed before analysis, which removed heterosce- 

 dasticity in the data. Mean squares ( MS ) and significance 

 levels are shown and significant results are given in bold. 

 ** = P<0.01. 



Number of 

 species 



Source 



df 



MS 



Number of 

 individuals 



MS 



Among subjects 



Site(S) 3 86.93 176.18 



Residual 12 27.49 169.67 



Within subjects 



TimeiT) 2 3.26 451.38** 



TxS 6 4.44 111.82 



Residual 24 2.84 55.45 



Mauchly's criterion W 0.622 0.705 



ranks can be a result of a few incidences of low indi- 

 vidual counts. 



The mean number of the six most abundant recolo- 

 nizing species showed considerable variability in space 

 and time. For the short-term study, densities of these 

 species differed significantly among sites and among 

 time intervals or at least for higher order interactions 

 containing these effects (Table 5). No definitive conclu- 

 sions could be made regarding the effects of defaunation 

 on these species because short-term recolonization pat- 

 terns for each species were clearly variable within and 

 among seasons (Fig. 4). However, the mean number of 

 fish was generally highest in initial samples and low- 



