LEBER AND GREENING: COMMUNITY STUDIES IN SEAGRASS MEADOWS 



trawl during each tow exceeded that sampled with 

 the crab scrape by a factor of 4.6, mean numbers of 

 individuals collected in scrape samples were signifi- 

 cantly greater than those in trawl samples in 44% 

 of the 16 scrape-trawl comparisons (Table 2). The 

 trawl was a significantly more effective collecting 

 device for number of individuals of fishes (Table 2; 

 April, July, and October fishes), but interaction terms 

 were significant for April and October analyses (see 

 Interactions, below). Mean numbers of individuals 

 were greater in trawl, than in scrape, samples in two 

 other cases (Fig. 1, January and July decapods in 

 night samples); however, scrape-trawl differences on 

 those dates were nonsignificant (Table 2). The crab 

 scrape was clearly the better gear for sampling 

 epibenthic individuals. 



Species numbers were never significantly greater 

 in trawl, than in scrape, samples (Fig. 1). In contrast, 

 the crab scrape collected significantly more species 

 than the trawl in 75% of the scrape-trawl com- 

 parisons (Table 2). Because the scrape often sampled 

 greater numbers of individuals than the trawl, the 

 presence of more species in scrape, than in trawl, 

 samples may be simply a sampling phenomenon. By 

 chance alone, one would expect to encounter more 

 rare species in larger samples. Using rarefaction 

 analysis (Simberloff 1978), we have factored out the 

 influence of sample size on species number for a 

 better comparison of scrape vs. trawl sampling ef- 

 fectiveness (Fig. 2). Eight of the 12 cases in which 

 the scrape sampled significantly more species than 

 the trawl (Table 2) can be attributed to a sampling 

 phenomenon; there were generally more species in 

 scrape samples because so many more individuals 

 were collected in each scrape tow. However, it is clear 

 in Figure 2 that the greater numbers of decapod 



species in January and July scrape samples, and fish 

 species in April and October scrapes, represent real 

 differences in the catch effectiveness of these gears 

 for species within these two taxa. 



Factor 2: Day vs. Night 



Day-night differences were clear. None of the com- 

 bined (scrape-trawl) daytime collections contained 

 significantly more species or individuals than night 

 collections. But nocturnal samples contained signifi- 

 cantly more individuals than daytime samples in 

 69%, and more species in 62%, of the 16 day-night 

 comparisons (Table 2). 



Interactions 



Significance of an interaction term indicates 

 dependence of one factor upon the other; in this case, 

 when sampling differences between scrape and trawl 

 exist but are dependent upon time of day. Scrape- 

 trawl vs. day-night interactions were significant in 

 8 of the 32 ANOVAs in Table 2. For these eight cases, 

 either the trawl sampled better only at night for a 

 certain taxon/month combination (one of the eight 

 interactions), or the scrape sampled better only dur- 

 ing the day (five of the eight cases), or both of these 

 events occurred (two of the eight cases, scrape was 

 better during the day but the trawl was better at 

 night). 



Although fish were taken in greater abundances 

 by the trawl on three of the four sampling dates, 

 interactions were significant on two of those dates 

 (April and October, Table 2). With the exception of 

 July collections, fish were equally as abundant in day- 

 time scrape samples as in trawls (see Figure 1). 



Table 2.— Two-way ANOVA, F-values. Underlined values indicate trawl samples significantly larger, all other significant values 

 are scrape samples. All significant day-night values indicate night significantly larger than day samples. 



= P < 0.05. 

 = P< 0.01. 

 = P< 0.001. 



445 



