no 



Fishen/ Bulletin 100(1) 



Figure 2 



Dendrogj-am resulting from a cluster analysis on the mean 

 volumetric contribution of major food categories by size 

 class of lobsters from the Magdalen Islands. The bottom 

 graph shows the joining distance at each step. The vertical 

 dashed line indicates the cut-off value for clusters, selected 

 because of the sudden increase in joining distance. 



were found in more than TCS of stomachs (Fig. 4). Bi- 

 valves were still found in a large proportion of stomachs 

 {8T7c) but accounted for a low proportion (0.18) of volume. 

 Gastropods, polychaetes, and macroalgae also occurred 

 frequently but accounted for only a small fraction of stom- 

 ach volume. Pagurids and lobsters were found in few stom- 

 achs but contributed >0.'2 of stomach volume. 



The grouping of the largest lobsters, 65-77 mivi CL, had 

 rock crab as the most important food item (specific abun- 

 dance=0.55; occurrence=86' 7 ). Lobsters, pagurids and fish 

 contributed a large proportion of stomach volume when 

 they were eaten, but these prey were ingested by <20'^( 

 of lobsters. Gastropods, flesh, bivalves, polychaetes, and 

 macroalgae were found in a large proportion of stomachs 

 but occupied a small proportion of the volume of these 

 stomachs. 



Overall, bivalves, rock crab, and flesh were the only food 

 items that each accounted for >0.1 of stomach volume for 

 the whole sample (Table 2). For these food items, a signifi- 

 cant linear relationship existed between volumetric con- 

 tribution and lobster CL, the latter explaining 68*7? to 929c 

 of the variability in volume (Fig. 5). Regi-ession of volumet- 

 ric contribution on lobster CL produced a negative slope 

 for bivalves and flesh, and a positive slope for rock crab. 

 Similarly, strong linear or nonlinear relationships existed 

 between percent occurrence of these three food items and 

 lobster CL (Fig. 5). Furthermore, large lobsters tended to 

 eat larger rock crabs than small lobsters, as evidenced 

 by the significant positive linear relationship between the 

 CW of rock crabs found in lobster stomachs and lobster CL 

 (Fig. 6). 



Figure 3 



Results of the factor analysis on the correlation matrix ol 

 volumetric contribution of major food categories by size 

 class of lobsters from the Magdalen Islands. See text for 

 factor loadings. Three clusters identified in Figure 2 are 

 shown inside ellipses; the other size classes constitute 

 the fourth cluster. 



Discussion 



Data 



Stomach content analysis is a useful method for the inves- 

 tigation of the natural diet of animals, even though the 

 lack of distinctive hard parts in some prey and differential 

 digestibility of soft and hard body parts limits the spec- 

 trum of food items that can be recognized and can lead 

 to biased perception of the relative importance of the food 

 items. We took care to process lobsters as quickly as pos- 

 sible after collection, thus attenuating the effects of differ- 

 ential digestibility, and we examined only intermolt and 

 nonovigerous lobsters, thus reducing sources of diet vari- 

 ability associated with molt cycle and female reproduc- 

 tive status (e.g. Weiss, 1970: Ennis, 1973). In addition, our 

 study was conducted over a small area where the various 

 lobster size classes were evenly distributed; therefore all 

 lobsters potentially could access the same food. We rec- 

 ognize that our volumetric contribution index underesti- 

 mates the importance of predominant food items, owing to 

 correction for stomachs with multiple food items and total 

 scores >10. However, this was a minor problem because 

 analyses using uncorrected values revealed that the vol- 

 umetric contribution of the three main food items was 

 underestimated by no more than 2-5'^< and that relation- 

 ships to lobster size class were unchanged. Therefore, we 

 are confident that the dietary differences among the lob- 

 ster size classes that we detected are real and that they 



