76 Christopher King Beachy 



to development ( x early stage = 5.43, x stage 45 = 5.96, n = 12, 

 Ts = 21, ns). This result was assumed to hold true for D. ochrophaeus 

 and E. wilderae. This assumption was not required for D. aeneus 

 and A. maculatum because these eggs were collected at early 

 developmental stages. 



Interclutch variation was quantified by taking a mean egg 

 diameter for all clutches at, or prior to, Harrison stage 30 (see 

 above discussion on changes in size). These means were pooled 

 according to species, and a CV was calculated for each species. 

 These species CVs were squared and subjected to pairwise 

 F-tests (Lewontin 1966). Kaplan (1987) showed that Bombina 

 orientalis (Boulenger, 1890) females can produce clutches of 

 different mean egg sizes. By assuming that this is the case for the 

 species used in my study (i.e., the variation represented in a sample 

 of the population might mirror the variation introduced by a single 

 female in her lifetime), one can set predictions that are similar to 

 those for intraclutch variation. 



To determine the relative contributions of intraclutch and 

 interclutch variation to the overall variance, egg size data for each 

 species were subjected to a one-way ANOVA, with individual 

 clutches as the factor. Relative contributions of intraclutch and 

 interclutch variation to overall variance for each species were 

 calculated using the factor and error sum-of-squares of the 

 ANOVA table (Sokal and Rohlf 1981). 



I analyzed data with StatView512+TM following the 

 methods of Sokal and Rohlf (1981). In all analyses a = 0.05. 



RESULTS 



My hypothesis was that the variation in egg size would be 

 greatest in A. maculatum, the temporary pool breeder, and the lowest 

 in D. aeneus, which is not constrained by habitat variability and 

 thus should exhibit the greatest degree of canalization. Variation in 

 egg size should be intermediate in the other three species. Descriptive 

 statistics of egg size for the five species under study are presented 

 in Table 1. 



Intraclutch Variation 



Coefficients of variation of all clutches were analyzed with a 

 model I one-way ANOVA (Sokal and Rohlf 1981), with species as 

 treatment. Significant differences were found in intraclutch CV among 

 species. A Fisher's PLSD a posteriori test was employed to 

 determine the nature of the differences. The prediction that species 

 using ephemeral larval habitats will display larger variation in egg 

 size was not supported. Of all species, D. aeneus exhibited the 



