Kulbicki and Wantiez: Estimates of fish stocks by shrimp trawl and visual survey off New Caledonia 



669 



Trawling 



The RV Vauban (24 m long) towed a shrimp trawlnet 

 of the semiballoon, floridian type with a 14-m headrope 

 and a 2-cm mesh codend. The vertical opening was 

 1.2 m and the distance between the otterboards was 

 7 m at a speed of 2 knots. Towing speed varied from 

 2 to 2.5 knots. 



The hauls were each 800 m in length ( + 80 m). The 

 net was submerged immediately after completion of the 

 visual census at a distance of 400 m from the first buoy 

 of the transect. The track of the haul was along the 

 transect line and the net was retrieved 200 m after the 

 second buoy of the transect line. All fish caught were 

 sorted to species (Rivaton et al. 1990). All fish were 

 counted and weighed. 



For the calculation of density D (fish/ ha), no catch- 

 ability coefficient was used. Therefore, 



D = X (n,/S) 

 1=1 



where p = number of species caught, 



nj = number of fish of species i, and 

 S = surface area of the haul (7 m x 800 m = 

 5600 m- = 0.56 ha). 



Biomass (kg/ha) B was estimated in a similar fashion, 



B = X w,/S) 



i = l 



where w, = weight of species i (kg). 



Results 



Species 



A total of 82 species were either caught by trawls or 

 recorded by divers (Table 1). All the common species 

 known to occur in trawls in St. Vincent Bay (Kulbicki 

 and Wantiez 1990) were found in the present study 

 except for the Leiognathidae (ponyfishes). This family, 

 represented by eight species in the bay, is character- 

 ized by large fluctuations in trawl catches depending 

 on season and locality. In the present survey, Lei- 

 ognathus rivulatus which is normally an uncommon 

 species in the catch, was the major Leiognathidae. 



Table 1 indicates that the trawls caught more species 

 (64) than were seen on the transects (51). This could 

 be due to the larger area covered by the trawls (~5600 

 m-/trawl versus 700-1000 m- per transect). Most 



cryptic species, such as Scorpaeniforms, Platycepha- 

 lidae (flatheads), flatfishes, and small Balistidae (trig- 

 gerfishes) (including the filfishes Paramonacanthus 

 japonic2is and Pseudaiutarius nasicornis) were poor- 

 ly represented in the visual transects but caught by the 

 trawls. Conversely, small species such as Apogonidae 

 (cardinalfishes) and Pomacentridae (damselfishes) could 

 be detected on the transects but were too small to be 

 retained by the net. Large, fast-swimming species such 

 as Carangidae, Serranidae, or Scombridae were seen 

 on the transects but evaded the trawls. Gobiidae, which 

 are burrowing species, were also seen during the dives 

 but absent from the trawl catch. 



A total of 32 species were detected by both methods 

 (Table 1). Among these, only eight species were seen 

 or caught in more than 50% of the samples for both 

 methods [Saurida undosquamis, Synodus hoshinonis, 

 Leiognathus rivulatus, Lethrinus nematacanthus, Upe- 

 neus tragula, Upeneus sp. aff. asymetricus, Pristotis 

 jerdoni, and Cantkigaster compressa). These eight 

 species will subsequently be referred to as "main spe- 

 cies." Except for Synodus hoshinonis and Leiognathus 

 rivulatus, all these fish are among the 15 most frequent 

 species in the trawl catch of St. Vincent Bay (Kulbicki 

 and Wantiez 1990). 



The number of species per station was statistically 

 larger (t test at a = 0.05) for the trawls than for the 

 transects (Table 2). However, a chi-square test for 

 independent samples (Siegel and Castellan 1988) 

 indicates that the ranking of the stations is not sig- 

 nificantly different (at a = 0.05) between methods. 



Fisfi size 



Only fish lengths were used to compare fish size be- 

 tween the two methods. Table 3 indicates that visual 

 censuses always yielded larger lengths than trawls, 

 except in the case of Pristotis jerdo7i i. However, the 

 lengths are significantly different (t test at a = 0.05) 

 for only four species (Table 3), and the two methods 

 give mean lengths which are less than 2.8 cm apart. 

 Due to the fact that fish length was estimated in 2-cm 

 size classes for transects and 0.5-cm size classes for 

 trawls, one should be cautious about the significance 

 of the differences observed. 



In a review on the problems of visual transects, 

 Harmelin-Vivien et al. (1985) indicate that fish size is 

 usually underestimated by this method. In the present 

 study, if this were the case, it would mean that the 

 larger fish (within a given species) are able to evade 

 the trawl. In our opinion this is not the only reason, 

 and the discrepancy of fish length between transects 

 and trawls is probably also due to an overestimate of 

 length by the divers. 



