Zeller et al.: Small-scale fishery catches for US island areas in the Western Pacific 



275 



1950 to 2.9 kg by 2002 (Table 5). Given that over 99% 

 of the human population of CNMI lives around the 

 three main islands, the catch per reef habitat area was 

 assessed for both the entire CNMI reef area (476 km^i 

 and also for the reef areas of the three main islands 

 (331.2 km2). Thus, between 1950 and 2002, estimated 

 annual catch per km^ reef area appears to have de- 

 clined from 1.0 t to 0.4 t, and from 1.4 t to 0.6 t for the 

 entire CNMI reef area (476 km-) and main islands reef 

 areas (331.2 km-), respectively (Table 5). 



The historic fisheries catches for American Samoa, as 

 re-estimated by Zeller et al. (2006a), indicated a poten- 

 tial decline of about 79% in small-scale fisheries catches 

 of nonpelagic species between 1950 and 2002 (Fig. 2D; 

 modified from Zeller et al., 2006a). There was also a 

 7-fold difference between the re-estimated catches and 

 the reported data for the 1980-2002 time period. 



Discussion 



Local and regional fisheries experts often acknowledge 

 that they are aware of the limited nature of much of 

 the official data, but rarely are willing or able to quan- 

 tify the missing catches. Our re-estimation makes the 

 potential scale of under-reporting of total extractions of 

 marine resources evident. Specifically, our study illus- 

 trates not only the potential discrepancy by a factor of 

 4.55 between what was reported and what may have 

 been caught (for the period of data reporting), but also 

 indicates the potential scale of declines (77% overall 

 for all areas combined) in total catches over the last 50 

 years. Although the historic catch estimates proposed 

 here obviously do not represent a formal stock assess- 

 ment, they are useful as baselines of potential historic 

 patterns and trends in fisheries catches. 



Regarding our comparison of catch data to those from 

 official, reported fisheries, we acknowledge that most 

 fisheries statistics were originally designed as an eco- 

 nomic development and monitoring tool, where there 

 was a common focus on commercial catches (with the 

 exception of Guam). Nevertheless, reported data are 

 being increasingly used to present national and global 

 fisheries conditions and status and trends of resources. 

 Thus, the under-representation of likely total catches 

 as indicated here may lead directly to an erroneous 

 interpretation of the status of fisheries within the U.S. 

 flag-associated islands. Significantly, the situation of 

 under-reporting contributes to the continued margin- 

 alization of small-scale fisheries (Pauly, 1997), and the 

 ongoing under-valuation of the direct and indirect eco- 

 nomic and social contribution of noncommercial (e.g., 

 subsistence, and increasingly recreational) fisheries to 

 the economic well-being of these islands (Zeller et al, 

 2006b). Such underestimations of catch histories may 

 also have repercussions for the move towards ecosys- 

 tem-based fisheries management. 



The general approach used here, which relies on an- 

 chor points of data obtained from a variety of peer- 

 reviewed and non-refereed data sources, moderated by 



conservative assumptions, and interpolated for missing- 

 data years, results in catch estimates that accounted 

 for all fisheries sectors. We acknowledge that our esti- 

 mates clearly are not statistically rigorous in the sense 

 of approximating "true" time-series values, which are 

 obviously not known. However, given our conservative 

 approach to estimation, the present estimates are less 

 wrong than the current default of reporting zero catch 

 for fisheries sectors not considered in official figures. 

 Ignoring the catches of noncommercial sectors of fish- 

 eries in the U.S. flag-associated island areas of the 

 western Pacific has likely resulted in a skewed picture 

 of the historic catch trends, as well as the magnitude 

 of catches for nonpelagic, near-shore resources in these 

 islands. 



Catch estimation procedures such as ours are associ- 

 ated with high data uncertainty; this is the nature of 

 alternative, non-standardized data sources. The pau- 

 city of data for the earlier periods was an acknowl- 

 edged shortcoming to our approach; nevertheless, our 

 approach is based on the best data and information 

 available. We endeavored to remain conservative in 

 our estimation throughout the period of examination; 

 thereby incorporating a precautionary aspect into the 

 data. Our conservative approach can be placed into 

 context by the following consideration. 



The re-estimation of catches for Guam, as undertaken 

 here, indicates a decline in catches of 86%, and a 2.5-fold 

 discrepancy between the re-estimated catches and the 

 reported statistics over the time period for which DAWR 

 reported data exist (1965-2002). The validity of the dif- 

 ferences between reported and re-estimated catches is 

 supported by the observation that, at least for the earlier 

 periods, the catch data as reported by our sources (and 

 forming the reported data) were "probably several times" 

 less than the actual yields (Zeller et al.'^). 



Concerns about our approach to the unreported catch- 

 es can be placed into perspective through an alterna- 

 tive, albeit less rigorous estimation (Zeller et al.^). In 

 1977, 38.6% of households in Guam were considered to 

 have at least one family member who fished, and mean 

 monthly catch per surveyed household was 32.7 kg, or 

 392 kg/year. With an average of 5 people per household 

 and a population of 110,000 in 1977 for Guam, these fig- 

 ures imply 22,000 households (110,000 people/5 people 

 per household), of which 38.6% (i.e., 8492 households) 

 had active fishermen. These actively fishing households 

 alone could thus have caught 3,328,864 kg in 1977 

 (8492 households with catch rate of 392 kg). Accounting 

 for pelagic fish in their catch (45.8% of reported catches 

 in 1977 were caught with pelagic gear), this calculation 

 would imply a nonpelagic catch of 1,804,244 kg for 1977 

 (3,328,864 kg x [1-0.458]). This admittedly very indirect 

 estimate is 2.76 times our total reconstructed nonpe- 

 lagic catch estimate of 654,345 kg for 1977, and 12.6 

 times the DAWR reported catch of 143,220 kg. Thus, 

 this indirect approximation supports our contention 

 that our re-estimation approach was conservative, and 

 total catches in the earlier periods were considerably 

 higher than those of the reported data. 



