696 



Fishery Bulletin 102(4) 



Foregut contents were analyzed as in Stehlik (1993). 

 The foregut of each crab was removed and preserved in 

 7095 ethanol. After opening the foregut, we estimated 

 fullness of the gut (from c /c to 100%) visually, and prey 

 items were identified to the lowest possible taxon. The 

 proportion of the total volume of the foregut contents con- 

 tributed by each prey taxon was estimated visually — a 

 less labor-intensive modification of the methods of Wil- 

 liams (1981), Hyslop (1980), and Steimle et al. (1994). 

 The volume of each prey taxon was multiplied by the 

 percentage of gut fullness. Combining all foreguts, the 

 volumes of prey taxa were listed in descending order. 

 The top 12 prey categories on the list (with the excep- 

 tion of "unidentified" and nonexclusive categories such as 

 Mollusca) were selected for use in most of the subsequent 

 analyses. Foreguts that did not contain prey in any of the 

 12 categories were dropped from numerical analyses. 



The dietary data were grouped in turn by predator 

 species, sex, size class, and collection stratum, and the 

 mean percentage volumes of each of the 12 mutually 

 exclusive prey categories were calculated. For graphic 

 representation of ontogenetic differences in diet, blue 

 and rock crabs were grouped for convenience into 20- 

 mm CW classes, and lady crabs were grouped in 10-mm 

 CW classes because of their smaller size range. For 

 numerical analyses, two maturity classes were used. 

 We used Mann-Whitney tests to compare diets between 

 sexes within predator species and between maturity 

 stages within predator species. The test statistic was a 

 chi square approximation. 



Group average cluster analysis was used to graph the 

 separation of diets by species, sexes, maturity stages, 

 and strata by using the 12 prey categories as dependent 

 variables. A Bray-Curtis similarity matrix was gener- 

 ated for each of the groupings, cluster analysis was 

 performed by using Systat® (version 10. SPSS Inc., Chi- 

 cago, IL), and dendrograms were generated by using the 

 Bray-Curtis values as distance measures (Romesburg. 

 1984; Marshall and Elliott, 1997). A percent similarity 

 level was chosen a posteriori that generated a reason- 

 able number of classes. 



Analysis of similarity (ANOSIM) was used to test 

 for statistical significance of dietary differences among 

 predator species and for sexes within species. Analysis 

 of dissimilarity (SIMPER) was used to determine which 

 prey taxa contributed most to the differences between 

 species pairs (Clarke and Warwick, 1994). 



Spatial, temporal, and trophic niche breadth and over- 

 lap indices were calculated from the number per tow 

 (1992-94) and diets (June 1991-June 1992) of each 

 crab species and sex. Temporal niche and overlap were 

 calculated by month for combined years. Female rock 

 crabs were dropped from consideration of trophic niche 

 overlap due to low sample size. 



Niche breadth (Colwell and Futuyama, 1971; Mar- 

 shall and Elliot, 1997) is a measure of exploitation 

 within a particular resource (for example, substrates 

 or prey taxa within an estuary by a species). Niche 

 breadth values are relative and can be compared only 

 within one study. The highest value corresponds to the 



broadest niche, or to habitat or a diet generalist rather 

 than to a specialist. Niche breadth (S) was calculated 

 by the formula of Colwell and Futuyama (1971), and 

 modified for measuring trophic niche breadth according 

 to Hines et al. (1990): 



B = 1 / £( p k r from./ = 1 to n . 



where p hl = N k j I Y k ip ki is the proportion of crabs of 

 species k associated with resource state 



j)\ 



j = resource states (months, strata, diet cat- 

 egories); 

 n = number of resource states; 

 N kj = catch per tow of species k at resource state 



j; and 

 Y k = catch per tow of species k over all resource 

 states. 



When trophic niche breadth was calculated, 



N h = total volume of diet category j consumed by preda- 

 tor k ; 



Y k = total volume of all diet categories consumed by 

 predator k. 



Niche overlap is a measure of the joint use of a resource 

 by two species (Colwell and Futuyama, 1971). Niche 

 overlap (C Al ) between species h and i was calculated 

 by the following formula (Colwell and Futuyama, 1971; 

 Hines et al., 1990): 



C/,,=l-0.5(Xlp,,,-;V)fi-on V = lto». 



where p h/ and p„ are calculated in the same manner as 

 Pk, above. 



This index ranges from (no overlap) to 1 (complete over- 

 lap) and is independent of sample size and differential 

 resource availability (Eggleston et al.. 1998). 



Results 



Temperature and salinity 



Bottom water temperature in the study area followed 

 a temperate seasonal cycle. The range during 1992-94 

 was from to 26.6 C. Using the monthly mean tempera- 

 ture below or above 10°C, and migration cycles of the 

 crabs, we grouped the months into two seasons: winter 

 (November through April) and summer (May through 

 October). The mean temperature in the winter months 

 1992-94 was 5.5°C, and that for summer was 18.9°C. 

 Temperature nearest the estuary mouth was usually a 

 few degrees lower in summer months and higher in the 

 winter months each year, compared with the average 

 throughout the estuary. 



Bottom salinity in the study area ranged from 15.0 to 

 33.5 ppt. The majority of stations had salinities between 



