310 



Fishery Bulletin 101(2) 



be negligible over the fishing period. However, the initial 

 population biomass, stock density, and potential yield could 

 have been overestimated if there was immigration into the 

 study area from adjacent habitats. Although the study site 

 is characteristic of the habitat in which P. filamentosus ag- 

 gregates (Ralston et al., 1986), the data relate to the peak 

 spawning period of this species on the nearby Seychelles 

 Bank. If there was an inward flux offish to a spawning site 

 during the time that fishing was conducted, the biomass 

 would have been overestimated. Conversely, it would have 

 been underestimated if there was emigration out of the 

 area. Because of the short time frame over which fishing 

 took place, it is assumed in the present study that the popu- 

 lation was closed and any such flux was insubstantial. 



Inconstant catchability is perhaps the greatest potential 

 source of error in applying methods of estimation based on 

 secular change in catch per unit of effort (Ricker, 1975). 

 The Leslie model assumes that catchability (q) is constant 

 over the fishing period. However, it is often found that the 

 first few units of effort cause a rapid depletion of more 

 vulnerable, faster, and more aggressive fish, and an accom- 

 panying rapid change in catch per unit of effort or other 

 abundance indices. After this initial removal, the remain- 

 ing fish have effectively lower q values, so that q declines 

 progressively as depletion proceeds. There may even be a 

 large pool of fish with (? = for some reason, and this pool 

 will not be sampled by the depletion process. Thus, the 

 general effect of varying catchability among individuals is 

 that the estimate oiq is biased upwards (and consequently 

 the initial biomass estimate is biased downwards), so that 

 underestimates of biomass in the order of 30% to 50% are 

 not unlikely (Hilborn and Walters, 1992). The use of an 

 adjusted cumulative catch is intended to compensate for 

 the decline in catchability during each time interval (Chap- 

 man, 1961). Nonetheless, as suggested by Cowx (1983), es- 

 timates should be treated with care. 



In commercial fishing operations, fishermen may main- 

 tain their catch rates as abundance drops by targeting the 

 remaining high concentrations of fish. As a result, catch- 

 ability can be underestimated and the initial stock size 

 overestimated. The data used in the present study showed a 

 relatively constant catch rate for the first six days of fishing, 

 followed by a dramatic reduction for the remaining seven 

 days. This pattern may be result of an overall .shift in loca- 

 tion of the fishing units. However, because the individual 

 dories were not equipped with position-fixing equipment, 

 there were no data collected on the precise location of indi- 

 vidual catches. Furthermore, the phenomena could equally 

 have been caused by a change in the behavioral charac- 

 teristics of the fish. A sudden drop in water temperature, 

 caused by localized upwelling or an increase in planktonic 

 prey abundance (or by both), could have contributed to the 

 decline in catch per unit of effort. The patterns observed 

 in the data suggest that they do not conform well to the 

 assumptions of a depletion model and therefore limit the 

 integrity of the biomass estimates. Position-fixing capability 

 on each of the dories and an independent means of deter- 

 mining behavioral characteristics, such as an underwater 

 video system, would have helped to elucidate the cause of 

 the irregular decline in catch per unit of effort. 



A problem with the use of depletion estimators in the 

 case of multispecies applications is that the catchability of 

 each of the component species may not remain constant in 

 relation to each other over the period that fishing is con- 

 ducted. Interactions between species, such as the competi- 

 tion for baits, can alter catchability, e.g. Rothschild ( 1967). 

 An increase in the catchability of a subordinate species 

 may occur as a result of the removal of a more competitive 

 species, as shown to be the case with the deepwater snap- 

 pers Pristipomoides auricitla and Pristipomoides zonatus 

 (Polovina, 1986). Pristipomoides filamentosus comprised 

 80.1% of the total catch from the fishing grounds on Saya 

 de Malha Bank. Of the remaining species in the catch, 

 there was none that was considered abundant enough to 

 bias the catchability estimate through species interactions, 

 such as competition for bait. 



Polovina ( 1986) showed that after over half of the initial 

 biomass of the deepwater snapper Pristipomoides zonatus 

 had been removed, there was very little change in size com- 

 position with cumulative catch. Despite this finding, size- 

 specific behavior has been considered to affect catchability 

 and could be a potential source of error in the estimate of 

 "q" determined in the present study. Given the need for 

 studies of the effects of fishing on commercially important 

 tropical species (Russ, 1991) and recent developments in 

 aging Pristipomoides filamentosus (Hardman-Mountford 

 et al., 1996), future studies should examine the implica- 

 tions of intensive line fishing on the age structure of snap- 

 per populations. 



The major reproductive peak for P. filamentosus on the 

 edge of the Seychelles Bank, 400 km to the northwest of 

 Saya de Malha Bank, was found to occur between Febru- 

 ary and April (Mees, 1993). If the reproductive peak for the 

 stock at Saya de Malha Bank occurred at the same time, 

 then the catcher boats may have been targeting spawning 

 aggregations. If this was the case, the population biomass 

 and subsequently stock density at the study site would be 

 lower during periods of the year when there is a reduction 

 in spawning activity Furthermore, P. filamentosus tends to 

 aggregate in shoals in upcurrent localities and near under- 

 water headlands and promontories (Ralston et al., 1986). 

 In addition to being the possible behavior of spawning 

 aggregations, this behavioral characteristic explains why 

 P. filamentosus comprised 80A7r of the total catch of what 

 is basically a multispecies fishery. 



Errors may have occurred in the calculation of the 

 width of the 55-130 m depth band. If the gradient of 

 the seabed in this depth range was steeper than that 

 estimated, the biomass density would be greater and vice 

 versa. A bathymetric survey would have improved the 

 precision of the surface area calculations and biomass 

 density estimates. 



Nevertheless, the biomass density obtained in the pres- 

 ent study (2364 kg/km-') is of the right order compared 

 to that of Mees (1993), where the mean initial biomass 

 density off! filamentosus on three banks in the Seychelles 

 was 2987 kg/km- for populations that had not previously 

 been exploited. Likewise, the estimated potential yield of 

 567 kg/km^ per year derived in the present study compared 

 well to the maximum sustainable yield of 717 kg/km- per 



